CFD-ACE V2010.0 Modules Manual Part2
CFD-ACE V2010.0 Modules Manual Part2
CFD-ACE V2010.0 Modules Manual Part2
0
Modules Manual, Part 2
iii
CFD-ACE_V2010.0_Modules_Manual_Part2
iv
Modules
Introduction
About ESI
ESI Group is a world-leading supplier and pioneer of digital simulation software for prototyping and
manufacturing processes that take into account the physics of materials. ESI Group has developed an
extensive suite of coherent, industry-oriented applications to realistically simulate a product’s behavior
during testing, to fine-tune manufacturing processes in accordance with desired product performance,
and to evaluate the environment’s impact on product performance. ESI Group’s products represent a
unique collaborative and open environment for Simulation-Based Design, enabling virtual prototypes to
be improved in a continuous and collaborative manner while eliminating the need for physical prototypes
during product development. The company employs over 700 high-level specialists worldwide covering
more than 30 countries. ESI Group is listed in compartment C of NYSE Euronext Paris. For further
information, visit www.esi-group.com.
1
CFD-ACE_V2010.0_Modules_Manual_Part2
EUROPE
Czech Republic
MECAS ESI s.r.o.
Brojova 2113/16
326 00 Pilsen
CZECH REPUBLIC
Phone: +420 377 432 931
Fax: +420 377 432 930
France
ESI Group ESI Group
5, Parc du Golf 20, Rue Du Fonds Pernant
13 856 Aix-en-Provence Cedex 3 ZAC De Mercieres
FRANCE 60471 Compiègne
Phone: +33 (0)4 42 97 65 30 FRANCE
Fax: +33 (0)4 42 97 65 39 Phone: +33 (0)3 44 30 43 60
Fax: +33 (0)3 44 86 87 77
ESI France ESI France
Le Récamier Sales & Technical Headquarters
70, rue Robert Parc d'Affaires SILIC
69458 Lyon Cedex 06 99 rue des Solets
FRANCE BP 80112
Phone: +33 (0)4 78 14 12 00 94513 Rungis cedex
Fax: +33 (0)4 78 14 12 02 FRANCE
Phone: +33 (0)1 49 78 28 00
Fax: +33 (0)1 46 87 72 02
Germany
ESI GmbH ESI GmbH
Sales & Technical Headquarters Werner-Eckert-Str. 6
Mergenthalerallee 15-21 81829 München
65760 Eschborn GERMANY
GERMANY Phone : +49 (0)89 451 0888 0
Phone: +49 (0) 6196 9583 0 Fax : +49 (0)89 451 0888 18
Fax: +49 (0) 6196 9583 111
ESI GmbH
Kruppstr. 82-100 / ETEC V5-105
45145 Essen
GERMANY
Phone: +49 (0)201 125 072 0
Fax: +49 (0)201 125 072 24
Italy The Netherlands
ESI Italia srl ESI Group Netherlands
Headquarters Radex Innovation Centre
Via San Donato 191 room 4.57
40127 Bologna Rotterdamseweg 183 C
ITALY 2629 HD Delft
Phone: +39 0516335577 THE NETHERLANDS
+39 0516335578 Phone: +31 (0)15 268 2501
Fax: +39 0516335601 Fax: +31 (0)15 268 2514
2
Modules
Spain
ESI Group Hispania, S.L. ESI Group Hispania, S.L.
Headquarters Regional Office
Parque Empresarial Arroyo de la Vega c/ Valencia, 63 (Oficinas AGORA)
C/ Francisca Delgado, 11. 08015 Barcelona
Planta 2ª SPAIN
28108 Alcobendas Phone: +34 93 508 51 72
Madrid Fax: +34 93 508 51 71
SPAIN
Phone: +34 91 484 02 56
Fax: +34 91 484 02 55
Switzerland
Calcom ESI
Parc Scientifique EPFL / PSE-A
CH-1015 Lausanne
SWITZERLAND
Phone: +41 (21) 693 2918
Fax: +41 (21) 693 4740
United Kingdom
ESI UK ESI UK
1 Robert Robinson Avenue CEME Innovation Centre
The Magdalen Centre Marsh Way
Oxford Science Park Rainham
Oxford OX4 4GA Essex
UNITED KINGDOM RM13 8EU
Phone: +44 (0) 1865 784 830 UNITED KINGDOM
Fax: +44 (0) 1865 784 826 Phone: +44 (0) 2085 965 095
Fax: +44 (0) 2085 967 058
ASIA
China
ESI ATE
Room 16A, Base F
Fu Hua Mansion
No. 8 Chaoyangmen North Avenue
Beijing 100027
CHINA
Phone: +86 (10) 6554 4907
Fax: +86 (10) 6554 4911
India
ESI Software Pvt. Ltd ESI India
No. 24-25, 27th Cross Sales & Technical Branch
Banashankari II Stage Indrakrupa #17
Bangalore 560 070 100 Feet Ring Road
INDIA 3rd Phase, 6th Block,
Phone: +91 80 4181-8400 BSK 3rd stage
Fax: +91 80 4181-8405 Bangalore 560 085
INDIA
Phone: +91 80 4017 4747
Fax: +91 80 4017 4705
3
CFD-ACE_V2010.0_Modules_Manual_Part2
Japan
ESI Japan Ltd. ESI Japan Ltd.
Headquarters & Sales Division Technical Office
16F Shinjuku Green Tower Bldg. 15F Shinjuku Green Tower Bldg.
6-14-1, Nishi-Shinjuku 6-14-1, Nishi-Shinjuku
Shinjuku-ku Shinjuku-ku
Tokyo 160-0023 Tokyo 160-0023
JAPAN JAPAN
Phone: +81 3 6381 8490 Phone: +81 3 6381 8494
Fax : +81 3 6381 8488 Fax : +81 3 6381 8489
ESI Japan Ltd. ESI Japan Ltd.
Kansai Branch Office Chubu Branch Office
Advance Esaka 5F 9F Daisan Horiuchi Bldg. 4-6-23,
8-10 Toyotsu-cho Meieki, Nakamura-ku, Nagoya-shi
Suita-shi Aichi 450-0002
Osaka 564-0051 JAPAN
JAPAN Phone: +81 52 589 7100
Phone: +81 6 6330 2720 Fax: +81 52 589 7001
Fax: +81 6 6330 2740
Malaysia South Korea
ESI Group South-East Asia Hankook ESI
12, Jalan Dato Haji Harun, 157-033, 5F MISUNG Bldg., 660-6,
Taman Taynton, Cheras Deungchon-3Dong, Gangseo-ku,
56000 Kuala Lumpur Seoul
MALAYSIA SOUTH KOREA
Phone: +60 (12) 6181014 Phone: +82 2 3660 4500
Fax: +82 2 3662 0084
AMERICAS
United States of America
ESI North America ESI North America
32605 West 12 Mile Road 12555 High Bluff Drive
Suite 350 Suite 250
Farmington Hills, MI 48334 San Diego, CA 92130
USA USA
Phone: +1 (248) 381 8040 Phone: +1 (858) 350 0057
Fax: +1 (248) 381 8998 Fax: +1 (858) 350 8328
ESI North America ESI North America
6851 Oak Hall Ln. 6767 Old Madison Pike
Suite 119 Suite 600
Columbia, MD 21045 Huntsville, AL 35806
USA USA
Phone: +1 (410) 381 4245 Phone: +1 (256) 713 4700
Fax: +1 (410) 309 5943 Fax: +1 (256) 713 4799
ESI North America
5201 Great America Pkwy
Suite 320
Santa Clara, CA 95054
USA
Phone: +1 (408) 492 0507
4
Modules
Brazil
ESI South America
Av. Pedroso de Morais, 1619 cj.312
São Paulo - SP CEP 05419-001
BRAZIL
Phone/Fax: +55 11 3031 6221
Mobile: +55 11 8134 9098
AFRICA
Tunisia
ESI Services Tunisia
Technopole Borj-Cedria
Route Touristique Soliman
2050 Borj-Cedria
TUNISIA
Copyright Information
™©1997-2010 by ESI-Group
All rights reserved. Published 2010.
This information is the confidential and proprietary product of ESI-Group. Any unauthorized use,
reproduction, or transfer of this manual is strictly prohibited. Subject to limited distribution and restricted
disclosure only.
CFD-ACE™, CFD-ACE+™, CFD-CADalyzer™, CFD-VIEW™, CFD-GEOM™, SimManager™, CFD-
VisCART™, CFD-TOPO and CFD-FASTRAN™ are registered trademarks of ESI-Group.
TM
5
CFD-ACE_V2010.0_Modules_Manual_Part2
6
Modules
• CFD-TOPO - predicts the transport, chemistry, etch and deposition of semiconductor materials on
the microscopic scales
• SimManager - uses the CFD-ACE+ package to perform parametric and optimization studies
using various parameters (e.g. geometrical parameters, boundary values, etc.)
New Features and Improvements
CFD-ACE-SOLVER 2010.0
CFD-ACE+ Solver V2010.0 includes all new features and improvements developed in the last 20 months,
i.e. since the V2009.0 release in December 2008. The following sections summarize these developments.
In earlier versions, the Discrete Ordinate Method (DOM) Radiation model that was available to CFD-
ACE+ users was, more precisely, the Sn Discrete Ordinate Method (or SnDOM). While the Sn quadrature
schemes that were available with this model were S4, S6, S8 and S12 (referring respectively to 24, 48,
80, 168 ordinate directions used in radiation computations for a 3D simulation), S4 was the default and
could be changed only via special DTF updates.
In V2010.0, the option to choose the Sn quadrature scheme (S4, S6, S8 and S12) is made available in
CFD-ACE+ GUI.
In V2010.0, a new DOM methodology, called Control Angle Finite Volume Method (CAFVM) has also
been implemented. In this method, since the computation directions can be user specified, it requires the
specification of number of cells and (in the polar and azimuthal directions, respectively) for the
discretization of a unit sphere.
• 3D:
• 2D:
• 2D Axisymmetric:
7
CFD-ACE_V2010.0_Modules_Manual_Part2
For multi-dimensional reactor scale models to inform real time control for plasma processing, such as to
determine settings like electrode voltage, requires accurate coupling between circuit parameters and
plasma characteristics. In V2010.0, an electrical circuit coupling to Berkeley SPICE 2g6 has been
developed to address these types of needs as well as enable a more general investigation of the
relationship between the external circuit and plasma properties in radio frequency discharges. A ‘loose
coupling’ methodology allows the CFD-ACE+ Plasma solver to account for external circuit interactions to
a greater degree of generality.
The coupled SPICE-ACE+ option is available in the MO->Adv tab for CCP Plasma only. On activating the
circuit model, the user needs to select a netlist input file for the circuit. Additionally, the following settings
are needed from the user:
• Electric BC: While existing options will continue to be supported, the “External Circuit” option will
now be available for the Fixed Potential type BCs. Here the user needs to indicate the node
number that would correspond to this reactor BC as denoted in the netlist input file. In this
manner, the circuit legs at different reactor surfaces can be “hooked”.
The SPICE-ACE+ option is mutually exclusive to the capacitor-only circuit model available prior to
V2010.0. However, backward compatibility is maintained: in order to revert to the previous capacitor-only
circuit model, special DTF updates with the following 2 lines are necessary:
V_float int 1 1
Cb_per_unit_area double 1 C
where C is the single capacitance in F/m2 that would otherwise be specified in the GUI and applied to all
sinusoidal BCs.
If different capacitances are specified at N different sinusoidal BCs, the following update is required:
V_float int 1 1
RF_Vsin_Cb_Num int 1 N
RF_Vsin_Src_Freq double N f1 f2… fN
RF_Vsin_Cb_per_Area double N C1 C2… CN
Where f 1 , f 2 , ... f N are the RF frequencies of the sinusoidal BCs and C 1 , C 2 , ..C N are the capacitances in
2
F/m .
Capacitively Coupled Plasmas (or CCPs) are often used in the semiconductor industry to sustain etch
and deposition processes involved in the manufacture of chips. The physics of “stochastic heating” plays
an important role in such discharges, especially when lower pressures and/or higher frequencies are
used. The stochastic heating model that has been available in CFD-ACE+ for the past few years is
empirical and requires user input that is not readily available. In V2010.0, a new model has been added
for stochastic heating in Capacitively Coupled Plasmas. This model is more general and has been made
the default. In this model, an analytical formula published by Lieberman et al, is used and a novel ‘self-
aware’ approach is implemented which allows CFD-ACE+ to automatically detect regions in the plasma
where stochastic sheath heating occurs.
8
Modules
This model is based on the work of E Kawamura, I Lee and co-workers at the University of California at
Berkeley. The power due to stochastic heating is calculated using the following formulation:
Where V 1 is the integrated electric field (or the electric potential difference) from the local plasma cell to
the powered electrode, m is the electron mass, e is the electron charge, is the highest angular driving
frequency at the powered electrode and T e is the electron temperature.
This model can be activated under MO/Advanced when CCP plasma is selected. This new
implementation allows CFD-ACE+ to detect the sheath regions adjoining the sinusoidal powered
surfaces. The stochastic heating power is then computed using the formula above.
In earlier versions, when using the STS radiation model, there was no option to save intermediate data
(such as geometric factors, auxiliary data, etc.). Hence when a case was restarted, all this calculation had
to be done again. The calculation of this intermediate data consumes a considerable amount of time,
especially for large cases. The time consumption is further increased as the level of accuracy is raised
(low < moderate < high < extremely high).
To avoid this re-calculation of intermediate data in the case of restart, a new feature has been introduced
in V2010.0. When the solver stops after the first run, a new file with the name <filename>_sts.rst will be
created in the working directory which stores the necessary intermediate data.
At the time of restart, this file, <filename>_sts.rst, will be read and used as input – thus avoiding the need
to re-calculate the intermediate data.
In the absence of <filename>_sts.rst file in the working directory, an error message stating “STS Restart
Read Error: Error Reading STS Restart File. Will Generate STS Geometric Data” will be displayed in the
out file and the Solver will calculate the intermediate data again and continue with the simulation.
CFD-ACE+ provides two different approaches for modeling the time-averaged effects of fans, rotors,
propellers, and other similar devices: (i) the Fan Blade Model approach, and (ii) the Fan Curve Model
approach. The Fan Blade Model approach is based on first-principles geometrical and physical modeling
of the effects of rotating blades, while the Fan Curve Model approach is a purely phenomenological one
in which a correlation is imposed between the pressure head and the shear stress across the fan plane
with the local volume flow rate or the local flow velocity across the fan plane.
In previous versions of CFD-ACE+, the user was able to supply the correlations for the Fan Curve Model
in two different forms: (i) via an input file; and (ii) using a piecewise-linear profile input into the CFD-ACE+
GUI. In V2010.0, two new forms of specifying these correlations have been added: (i) a fifth order
polynomial function (relating the pressure head or the shear stress across the fan plane with the normal
velocity across the fan plane), and (ii) a constant value (of the pressure head or shear stress across the
fan plane). In addition, in V2010.0, the layout and organization of the panels through which the user
supplies the needed input for the Fan Curve Model have been re-organized in the CFD-ACE+ GUI. The
two new input specification forms for the Fan Curve Model correlations accord more closely with the
forms in which performance data are commonly supplied by manufacturers of fans and other such
devices, and should therefore improve the usability and the convenience of the Fan Model in CFD-ACE+.
9
CFD-ACE_V2010.0_Modules_Manual_Part2
By default, CFD-ACE+ applies a no-slip boundary-condition on all wall boundaries. This default treatment
is the most appropriate one for simulations in which the continuum assumption of fluid dynamics holds
accurately. If, however, a simulation contains regions in which the continuum assumption does not hold
accurately (such as when the flow is rarefied or when the local geometric dimensions are comparable
with the mean free path), it is more accurate to use a non-continuum boundary condition on the walls.
CFD-ACE+ allows users to apply a slip-regime (non-continuum) boundary condition on walls using the
Slip Walls Model. This model can be activated under the MO Tab (MO -> Adv -> Slip Walls), and it affects
the solutions for both the flow and the heat transfer modules of CFD-ACE+.
In earlier versions of CFD-ACE+, the “Slip Walls” Model, when activated, would be applied to all wall
boundaries present in a simulation. In V2010.0, the “Slip Walls” Model has been generalized so that it can
now be applied to individual wall boundaries that are chosen by the user under the BC Tab. This
generalization enables users to model computational domains with different flow regimes in the same
simulation.
The ability to apply the Slip Walls Model on a wall-by-wall basis is implemented as follows: if the Slip
Walls Model is activated (under the MO Tab, MO -> Adv -> Slip Walls), then the Slip Walls Model
boundary condition subtype is applied by default to all wall boundaries and to all solid-fluid interfaces in
the model. This behavior is the same as in previous versions of CFD-ACE+. In V2010.0 and onwards,
however, users can then choose to change the default Slip Model wall boundary condition subtype on any
individual wall in the usual way through the BC Tab. The Slip Wall Model boundary condition subtype can
also be applied to moving and rotating walls.
The Wall Boundary Condition in V2010.0 has been extended by the creation of three sub-types which can
be selected by users from the GUI. These three sub-types are as follows: (i) the no-slip sub-type; (ii) the
inviscid sub-type; and (iii) the slip (or partial-slip) sub-type. Additional sub-types of the Wall Boundary
Condition type are also planned for future versions of CFD-ACE+.
The no-slip sub-type applies the regular treatment for Wall boundary conditions in CFD-ACE+ and is
identical to the default treatment for Wall boundary conditions in previous versions of CFD-ACE+. The
inviscid sub-type is a new boundary condition treatment in CFD-ACE+. This boundary condition allows
users, for example, to simulate the confining effects of boundaries without also imposing their viscous
effects, and has applications and benefits in various settings. This new inviscid wall boundary condition
sub-type is treated by allowing the wall boundary to impose a constraint only on the component of the
local flow velocity vector that is normal to the wall boundary, and to impose no constraint on the
component of the local flow velocity vector that is tangential to the wall boundary. This new boundary
condition is therefore equivalent to a symmetry boundary condition and also to a zero-shear stress
boundary condition for the flow module. The new inviscid sub-type boundary condition is available for all
flow simulations unless the Slip Walls Model is activated: in that case, users can only choose between the
no-slip sub-type and the non-continuum (partial slip) sub-type for the wall boundary condition type.
In earlier versions of CFD-ACE+, a symmetry boundary condition could have been used to achieve the
effects of an Inviscid wall sub-type. However, in a multi-physics simulation, a symmetry boundary
condition is applied to all modules, whereas the user may only wish to apply this boundary condition to
the flow module. The new inviscid sub-type allows the user to do just that; namely, to circumvent the
limitation associated with symmetry boundary conditions, and to give users the flexibility to apply an
inviscid wall boundary condition subtype for the flow module and set different boundary conditions for
other modules.
The new sub-types for wall boundary conditions in V2010.0 of CFD-ACE+ are applied as follows: all wall
boundaries are by default assigned in the CFD-ACE+ GUI the standard no-slip boundary condition (which
imposes the condition that the local fluid velocity and temperature are identical, respectively, to the local
wall velocity and temperature). Users can then visit each wall boundary condition under the BC Tab and
change the sub-type of that wall boundary condition to the inviscid sub-type. In this way, users can create
a model with an arbitrary combination of no-slip and inviscid wall boundary condition sub-types. The
10
Modules
arrangement described above also ensures the most convenient and trouble-free handling with respect to
back-compatibility (for models that were created in earlier versions of CFD-ACE+).
It should be noted that the inviscid wall boundary condition sub-type applies the “full slip” condition only
for the flow module, and not to any other module. Thus, for example, phenomena such as a temperature
jump associated with slip walls cannot be modeled through this feature. The inviscid wall boundary
condition sub-type imposes inviscid behavior only on wall boundaries. Therefore, users should not expect
to solve the Euler equations when using this sub-type; in CFD-ACE+, the fluid is still modeled as a
viscous fluid, and the Navier-Stokes Equations (not the Euler Equations) are the equation being solved to
model fluid flow.
The CFD-ACE+ Solver has long had the capability to output summary data containing the integrals of the
mass flux over surfaces. These surfaces include inlet and outlet boundaries, fluid-fluid interfaces, wall and
thin-wall boundaries, and solid-fluid interfaces. At wall and thin-wall boundaries and at fluid-solid
interfaces, mass fluxes only arise when fluid mass is being generated or destroyed at the surface,
possibly as a result of chemical reaction or some other mass transfer mechanism. The summary data
also contains the rates at which mass is generated or destroyed within volume sub-regions of the
computational domain, possibly due to the presence of sources or sinks of mass in those sub-regions.
The integral of the mass flux over a surface (which evaluates to the total mass flow rate through that
surface) can either be used directly as a primary solution result (for example, if a user were performing a
simulation primarily to determine the mass flow rate across some portion of a device), or as an additional,
secondary solution result. The mass flow balance summary data also contains an overall summary of the
net balance (or total sum) of the rates of mass generation and destruction in the computational domain
and the rates of mass flow across its boundaries, and this net balance, relative to the mass inflow or
outflow rate or some other appropriate scaling reference, can be used as one measure of the extent of
convergence of a solution.
In V2010.0, the mass flow balance summary output has been improved and generalized to produce more
detailed and convenient summary data. Relative to preceding versions of the CFD-ACE+ Solver, the most
important changes in the mass flow balance summary output are as follows:
• The boundary-by-boundary output is now given in terms of Surface ID values, which have
replaced the Key values that were used in preceding versions of the CFD-ACE+ Solver. The
boundary-by-boundary output is still also given in terms of the names of boundaries, just as in
preceding versions of the CFD-ACE+ Solver. Surface ID values provide a new means of
identifying and tracking surfaces and volumes in the ACE+ Suite, and they will eventually fully
supplant the Key values, which have heretofore served this identification and tracking role. Thus,
the user is still able to name boundaries in the CFD-ACE+ GUI and to have those names
reflected in the summary output, but the user will now see Surface ID values instead of Key
values in the output summary. For users who are accustomed to or prefer using Key values, a
mapping is also optionally output with the summary data to show the correspondence between all
the Key and Surface ID values for all the boundaries in a model.
• The mass flow rates across fluid-fluid interfaces are now included in the total mass flow balance
summary output.
• The mass flow rate across any grouping of individual boundaries can now be output as a single
summary flow rate for that grouping. Thus, users can now, during the simulation set-up process in
the CFD-ACE+ GUI, compose groups of boundaries, name these groups, and then view
summary mass flow rates across all the elements in these groups as if all these elements have
been agglomerated into a single boundary. For example, a user can collect all the individual inlets
of a flow apparatus together into a group and then have the total mass flow rate across all these
inlets reported in the mass flow balance summary as a single mass flow rate value for this group.
• The mass flow balance summary output is now largely independent of any domain decomposition
carried out for parallel processing or other purposes. In particular, unlike the situation with earlier
11
CFD-ACE_V2010.0_Modules_Manual_Part2
versions of the CFD-ACE+ Solver, if a boundary is split by, for example, a domain decomposition
operation, the mass flow rate summary will report a single line for this boundary, with the mass
flow rate summed across all the fragments of this boundary, regardless of, say, the residence of
these fragments on different processors in a parallel computation. The independence cannot be
complete because a domain decomposition process can create, eliminate, or modify existing
fluid-fluid interfaces in a computational domain, so the summary output for any such interfaces
may be accordingly affected.
• The mass flow balance summary data is now output to a separate file, named
“modelname.MASSUM”, instead of being output to the “modelname.out” file. Summary data for
volume sub-regions and for overall mass balance is still output to the “modelname.out” file as in
preceding versions of the CFD-ACE+ Solver.
• The output is now organized and formatted in a way that is more convenient for human reading
and script processing.
The output of the mass flow rate balance summary (and the output of the Key to Surface ID
correspondence mapping if this is desired too) can still be activated and controlled through the CFD-
ACE+ GUI in a similar manner to that with preceding versions of CFD-ACE+, but the user now has the
option of outputting the mass flow balance summary with a frequency independent of that for the other
outputs from the solver.
The CFD-ACE+ Solver has long had the capability to output summary data containing the integrals of the
energy flux over surfaces, where this flux is comprised of contributions from conduction, convection, and
radiation. The surfaces across which the energy flux is computed, integrated, and output include inlet,
outlet, wall and thin-wall boundaries, and fluid-solid interfaces. The summary data also contains the rates
at which energy is generated or destroyed within volume sub-regions of the computational domain,
possibly due to the presence of sources or sinks of energy in those sub-regions. Such sources and sinks
may be user-defined, and may also originate from phenomena and models in the Flow, Heat, Chemistry,
Spray, Electro-Magnetics, Plasma, or other modules of CFD-ACE+. The integral of the energy flux over a
surface (which evaluates to the total energy flow rate through that surface) can either be used directly as
a primary solution result (for example, if a user were performing a simulation primarily to determine the
heat flow rate across some portion of a device), or as an additional, secondary solution result. The energy
flow balance summary data also contains an overall summary of the net balance (or total sum) of the
rates of energy generation and destruction in the computational domain and the rates of energy flow
across its boundaries, and this net balance, relative to the energy inflow or outflow rate or some other
appropriate scaling reference, can be used as one measure of the extent of convergence of a solution.
In V2010.0, the energy flow balance summary output has been improved and generalized to produce
more detailed and convenient summary data. Relative to preceding versions of the CFD-ACE+ Solver,
the most important changes in the energy flow balance summary output are as follows:
• The boundary-by-boundary output is now given in terms of Surface ID values, which have
replaced the Key values that were used in preceding versions of the CFD-ACE+ Solver. The
boundary-by-boundary output is still also given in terms of the names of boundaries, just as in
preceding versions of the CFD-ACE+ Solver. Surface ID values provide a new means of
identifying and tracking surfaces and volumes in the ACE+ Suite, and they will eventually fully
supplant the Key values, which have heretofore served this identification and tracking role. Thus,
the user is still able to name boundaries in the CFD-ACE+ GUI and to have those names
reflected in the summary output, but the user will now see Surface ID values instead of Key
values in the output summary. For users who are accustomed to or prefer using Key values, a
mapping is also optionally output with the summary data to show the correspondence between all
the Key and Surface ID values for all the boundaries in a model.
• The energy flow rate across any grouping of individual boundaries can now be output as a single
summary flow rate for that grouping. Thus, users can now, during the simulation set-up process in
12
Modules
the CFD-ACE+ GUI, compose groups of boundaries, name these groups, and then view
summary energy flow rates across all the elements in these groups as if all these elements have
been agglomerated into a single boundary. For example, a user can collect all the individual fluid-
solid interfaces separating a solid object from the surrounding flow field into a group and then
have the total energy transfer rate across all these interfaces reported in the energy flow balance
summary as a single energy flow rate value for this group.
• The energy flow balance summary output is now largely independent of any domain
decomposition carried out for parallel processing or other purposes. In particular, unlike the
situation with earlier versions of the CFD-ACE+ Solver, if a boundary is split by, for example, a
domain decomposition operation, the energy flow rate summary will report a single line for this
boundary, with the energy flow rate summed across all the fragments of this boundary, regardless
of, say, the residence of these fragments on different processors in a parallel computation. The
independence cannot be complete because a domain decomposition process can create,
eliminate, or modify existing fluid-fluid interfaces in a computational domain, so the summary
output for any such interfaces may be accordingly affected.
• The energy flow balance summary replaces what was previously called the heat flow summary
(to reflect a more appropriate term and the added generalization made in this release), and this
summary is now output to a separate file, named “modelname.ENGSUM”, instead of being output
to the “modelname.out” file. Energy flow balance data related to volume sub-regions of the
computational domain and to the overall energy balance is still output to the “modelname.out” as
in preceding versions.
• The output is now organized and formatted in a way that is more convenient for human reading
and script processing.
The output of the energy flow rate balance summary (and the output of the Key to Surface ID
correspondence mapping if this is desired too) can still be activated and controlled through the CFD-
ACE+ GUI in a similar manner to that with preceding versions of CFD-ACE+, but the user now has the
option of outputting the energy flow balance summary with a frequency independent of that for the other
outputs from the solver.
In V2010.0, a new set of Output Control options have been introduced that allow independent control over
the frequency with which information is written to the DTF file - Restart data (cell-center solution) and
Graphic data (nodal solution) components - and the frequency with which information is written to the
output and summary files.
The user gets to specify a general output frequency (as in previous versions) and then can override this
setting, if desired, individually and independently for Restart, Graphic and Summary data. These override
options/controls thus allow the user to manage, more efficiently, what gets written out during a simulation.
These options/controls are available for both steady state and transient simulations.
If the surface tension coefficient, σ, varies along an interface separating two immiscible fluids, this
variation will generally generate a resultant force that is tangential to the interface, and this tangential
force typically causes flow along the interface and the layers of fluid attached to it. This surface-tension-
induced tangential force and motion are called Marangoni Effects. The component of the surface-tension
force that is tangential to the interface is in addition to the component that is normal to the interface. This
normal component generally exists whenever there is surface tension, regardless of whether it varies,
and typically causes a pressure jump across the interface and motion perpendicular to the interface.
A Marangoni Effects Model has been enabled in V2010.0. This model can be activated under the surface
tension model options within the MO panels for the VOF Module. Variations in the surface tension
coefficient that lead to Marangoni Effects in CFD-ACE+ can arise from dependence of the surface tension
13
CFD-ACE_V2010.0_Modules_Manual_Part2
coefficient on a wide variety of solution variables or material properties, including the temperature,
species concentrations, or any other variables accessible through user subroutines.
In V2010.0, improvements have been made in the VOF Module to the face-pressure interpolation and to
the accuracy with which data is exchanged across inter-processor boundaries. These two sets of
improvements result in greater accuracy in the transport of the VOF function across parallel interfaces
under gravitational fields and surface-tension forces, leading to improved accuracy and performance for
parallel computations with the VOF Module.
In V2010.0 and earlier versions, domain decomposition can be performed by executing the
dtf_decompose script which will decompose the DTF file into the specified number of zones using the
specified method. This decomposition process runs only on one processor, i.e. it is a serial process. In
V2010.0, a new CFD-Decompose script has been introduced which will allow users to perform domain
decomposition in serial and parallel modes.
Usage: CFD-Decompose [options]
Minimum options:
CFD-Decompose --dtf=model.DTF --numZones=Nzones
Decompose sim 1 (assumed) into Nzones (using serial code).
Minimum options for a parallel run:
CFD-Decompose --dtf=model.DTF --numZones=Nzones --dmp --
numProcs=Nprocesses
Decompose sim 1 (assumed) into Nzones using Nprocesses in DMP mode locally
(current directory is working directory, running only on local machine).
For more information regarding this script and the options available with it, please refer to the CFD-ACE+
V2010.0 User Manual.
Over the past few releases, CFD-ACE+ has supported several MPIs (Message Passing Interfaces)
including MPICH, HP-MPI and other native MPIs. In these previous versions, the MPI libraries have been
tightly tied to CFD-ACE-SOLVER requiring a different executable for each MPI library (e.g., CFD-ACE-
SOLVER-MPICH-MPI, CFD-ACE-SOLVER-HP-MPI, etc…). Starting with V2010.0, MPIs are called
dynamically - a single solver executable calls the chosen MPI library at runtime.
Also, the older parallel script cfd-ace-mpi is no longer supported. Users should use CFD-SOLVER
introduced in V2009.4.
Support for Single Processor Run
In previous versions, the parallel script (cfd-ace-mpi or CFD-SOLVER) used a MPI executable (e.g., CFD-
ACE-SOLVER-HP-MPI) even if the number of parallel process was just one (serial). This resulted in
certain additional overhead related to MPI. In V2010.0, this additional overhead related to MPI is
removed. Most users would not see any difference.
Usage example:
CFD-SOLVER -dtf filename.DTF
Support for HP-MPI on all Platforms
14
Modules
HP-MPI has been supported on Linux platforms for a while and on Windows platforms since V2009.4. In
V2010.0, HP-MPI continues to be supported on all platforms.
Usage example:
CFD-SOLVER -dtf filename.DTF -num 2 -hpmpi
Support for MPICH2 on Linux Platforms
In V2010.0, MPICH 2 support has been added on all Linux platforms. MPICH 2 can be invoked through
CFD-SOLVER script using the –mpich2 option.
Usage example:
CFD-SOLVER -dtf filename.DTF -num 2 -mpich2
Users do not have to install any additional software. However, there are a few additional steps necessary
to use MPICH 2 properly.
Verify that a secret word is registered
MPICH 2 requires a secret word to distinguish the mpd processes of different users. This secret word
should be saved in a file .mpd.conf within your home directory with read/write permissions only for the
owner and not accessible by anyone else. This can be any word and doesn’t have to be your password.
This should be done on all nodes.
From a command prompt, issue the following commands:
$ echo ‘‘MPD_SECRETWORD=xxxxxx’’ > $HOME/.mpd.conf
$ chmod 600 $HOME/.mpd.conf
Here “xxxxxx” is the secret word you have chosen.
If you are running as root, create at /etc/.mpd.conf with the same content and permissions.
Note: SELinux File Context Settings
Users with Security Enhanced Linux (SELinux) may experience some issues where certain CFD-ACE+
libraries are not allowed to load. To avoid this, SELinux should be configured to allow CFD-ACE+ libraries
to load. One method of doing this is shown below.
From a command prompt, issue the following command:
$ chcon -t textrel_shlib_t $ESI_HOME/2010.0/ACE_SOLVER/lib/*.so
Your system administrator may also be able to achieve this in a different way or disable SELinux.
In V2010.0, a gas phase reaction mechanism for low temperature plasmas sustained in SF6/O2 mixtures
has been added to the Database.
CFD-ACE-SOLVER 2009.4
In V2009.4, a new capability to automatically generate lookup tables (LUT) for non-Maxwellian electron
energy distribution functions (EEDFs) was implemented. This is pertinent to cases where the Plasma
module is activated but not the Kinetics module. This new capability allows most users to skip the manual
procedure of running a one-cell problem first and updating the case file as documented in the manual for
LUT generation. The old capability remains supported for backward compatibility and for advanced users.
15
CFD-ACE_V2010.0_Modules_Manual_Part2
In V2009.4, an additional algorithm was implemented for computing the forces and moments applied by
the pressure on a momentum resistance element. The forces and moments are computed in this
algorithm by summing the elemental forces and moments applied by the pressure only (that is, excluding
the shear forces) on each external face of each grid cell included in the momentum resistance region.
While this new algorithm is just as valid and accurate as the previous algorithm (from both a theoretical
viewpoint and a discretized implementation), it has the advantage of running in a small fraction of the time
of the previous algorithm for any practical case with a large number of cells. The user can select which
calculation algorithm is to be used for momentum resistance force and moment summary for a particular
simulation. Unless the user has some specific need to have the pressure and moment calculations
performed on a surface mesh set up around a momentum resistance, the user should always select the
new algorithm in preference to the older one.
In V2009.4, the fixed mass-flow-rate inlet boundary condition was extended to work in parallel-
computation mode, enabling this boundary condition to be applied freely to any boundary patch, without
regard to whether the computation will be run in parallel or serial mode, and without regard to the manner
in which domain partitioning may divide the boundary patch.
The Frontal Area projection tool, introduced in CFD-ACE+ Solver (and CFD-ACE+ GUI) in V2009.4,
calculates the projected area of a set of surfaces on a plane. A typical application of this tool would be to
compute the frontal area of a vehicle in external aerodynamics simulations for drag coefficient calculation.
CFD-ACE-SOLVER 2009.2
In V2009.2, a new model for charge leakage in dielectrics was implemented in the CCP Plasma Module
of CFD-ACE+.
In any plasma environment, the solid materials in contact with the plasma fall into one of two broad
categories: metals, or dielectrics. Due to their high conductivity, grounded metals do not retain electric
charge, whereas dielectrics can retain both bulk and surface electric charge. Prior to V2009.2, charge on
the dielectric surfaces bordering the plasma was generated or destroyed, accounting only for the
incoming flux of charged species. In some simulations, it may be desirable to account for the leakage of
electric charge through the volume of such a dielectric, if that dielectric happens to be partially conducting
or “lossy”. This capability was implemented in V2009.2. Given the conductivity of such a lossy dielectric,
the charge within the dielectric is updated using the charge conservation equation, in addition to the
electric potential, as part of the solution process. This feature is restricted to the CCP Plasma Model.
In the volume of the partially conducting dielectric material, the charge conservation equation is solved in
addition to the Poisson’s equation for the electrostatic potential. The volumetric charge is solved explicitly
at each computational cell using the standard finite volume method. The current density term is assumed
to be due to current conduction only. At the surface of the dielectric bordering the plasma, the surface
charge density calculation allows for charge leakage into the volume of the dielectric, in addition to the
flux of all charged particles. At metallic boundaries, zero charge and the applied electric potential are
enforced. The relevant input property for the lossy dielectric feature is the user-supplied electrical
conductivity in the lossy dielectric, specified in units of 1/Ohm-m.
The current model explicitly updates the volumetric charge inside the lossy dielectric material and so the
limitation that arises is that the time steps are constrained by the dielectric relaxation rate. The
computation will diverge if the user input for the electrical conductivity and the time-step are such that the
dielectric relaxation criterion might be violated, i.e., Δt CCP > ε/σ where ε is the permittivity and σ is the
conductivity of the lossy material. For instance, with a Δt CCP of 1ns, the upper limit for σ would be 8.85e-3
16
Modules
(1/Ohm-m), and so on. Conversely, one can determine the upper limit on Δt CCP given the conductivity. An
implicit method that circumvents the dielectric relaxation criterion may be implemented in the future.
In V2009.2, a Realizable K-ε Turbulence Model (hereinafter called the RKE Model) was implemented in
CFD-ACE+.
The RKE Model is most readily characterized and described as a variant of the Standard K-ε Turbulence
Model. In particular, the RKE Model can be derived from the Standard K-ε Model by the following: (i)
replacing the transport equation for the rate of dissipation of turbulent kinetic energy in the Standard K-ε
Model with a similar transport equation that models the dissipation rate based on the dynamic behavior of
the mean square vorticity fluctuation in the high turbulent Reynolds Number limit; and (ii) replacing the
eddy viscosity equation of the Standard K-ε Model with an eddy viscosity equation that ensures
satisfaction of the realizability constraints (for the turbulent normal and shear stress components). The
aim of these two modifications is to overcome some of the recognized defects of the Standard K-ε Model;
namely, (i) the inaccurate reflection of the turbulent length scales in the transport equation for the
dissipation rate; and (ii) the over-prediction of the eddy viscosity for high shear rates.
The RKE Model will give better predictions than the Standard K-ε Model for flows with high mean shear
rates and for flows with large-scale separations. These improved predictions should be in all turbulence-
dependent quantities, such as wall shear stresses, boundary-layer thicknesses, the locations of
separation and re-attachment points, and so on. On the other hand, the Standard K-ε Model will retain the
accuracy advantage over the RKE Model in flows with dominant boundary layers and weak or no
separations, since by its formulation, the model dissipation equation of the RKE Model is valid in the limit
of high turbulent Reynolds Number. Other than for the above, the RKE Model retains most of the
characteristics, properties, strengths, and weaknesses of the Standard K-ε Model, especially with respect
to the flow regimes for which it is most applicable, and with respect to convergence behavior and
computational resource requirements. A typical application for which the RKE Model should give better
predictions than the Standard K-ε Model would be separated flow around a car or bluff body.
By its formulation, the RKE Model must always be used with appropriate wall functions, such that the y+
value is no less than 30, and preferably closer to 60 or 80, and with the same guidelines and limitations
that apply for any High Reynolds Number K-ε Model.
The RKE Model can be activated from the GUI as a separate new turbulence model, and the chosen wall
function can also be selected from the GUI just as it would be for all other turbulence models that require
a wall function. The wall functions that can be chosen for the RKE Model are currently the following: (i)
Standard Wall; (ii) Two-Layer Wall; and (iii) Non-Equilibrium Model. The RKE Model runs in parallel
mode, and with all the other modules that are currently coupled with the turbulence module.
Prior to V2009.2, users could specify the direction vectors of an anisotropic resistance (porous media)
only as constants that remain fixed throughout a computation. Starting with V2009.2, users may invoke a
newly added user subroutine (udirection) to set (or to change) the direction vectors of the anisotropic
resistance on a local, cell-by-cell basis. This capability gives users more flexibility in specifying anisotropic
properties, including the flexibility to vary direction vectors depending on the solution or other global or
local runtime variables.
The UDIRECTION user subroutine is called on a cell-by-cell basis for each volume condition where the
user specifies or selects the “User Sub (udirection)” option for anisotropic properties in the GUI.
Other enhancements/improvements
17
CFD-ACE_V2010.0_Modules_Manual_Part2
• Prior to V2009.2, an Arbitrary Interface touching a Thin Wall (along an edge or at a point) was not
permitted. This limitation was removed in V2009.2, giving users more flexibility with modeling
typical applications in which Thin Walls and Arbitrary Interfaces occur in combination.
• A number of non-repeatable and sporadically-occurring problems on the MS-Windows 64-bit
platform, including stack overflow, memory corruption and mysterious crashes, were resolved in
V2009.2 by correcting the datatype of the hidden string length parameter in the Fortran/C
interface functions.
The Database Manager, which CFD-ACE+ users have access to, includes some new features and
improvements in V2010.0. These include a new shared level, new folder structure, additional
import/export capabilities, drag-and-drop support, new scripting functions, among others. For more
information on this, please refer to the Appendix at the end of this document.
Some features of CFD-ACE+ might require the user to group boundaries together (e.g. for the calculation
of the projected area of a collection of surfaces). In order to accommodate this procedure, a capability
has been added in CFD-ACE+ GUI V2010.0 to allow the grouping of boundaries by categories, where
each category pertains to a particular feature. These boundary groups can be created through the Model
Explorer, and will appear under the “Group Categories” branch of the simulation Geometry Tree.
18
Modules
For details regarding the features that use/require these group categories, please refer to the relevant
feature information in this document or in the User Manual.
The Frontal Area projection tool, introduced in CFD-ACE+ GUI (and CFD-ACE+ Solver) in V2009.4,
calculates the projected area of a set of surfaces on a plane. A typical application of this tool would be to
compute the frontal area of a vehicle in external aerodynamics simulations for drag coefficient calculation.
Starting with V2010.0, a ‘Projected Area’ group category has been introduced in CFD-ACE+ GUI for more
efficient use of this tool.
Using Help
This help system is arranged in two volumes:
Volume I - User Manual describes the CFD-
Volume II - Modules contains a section for each
ACE+ operations and features of the CFD-ACE-
of the CFD-ACE+ modules that appear in the
Solver which are module independent:
Problem Type (PT) Panel:
19
CFD-ACE_V2010.0_Modules_Manual_Part2
We recommend that you first read the User Manual Overview to learn the basics of how the CFD-ACE+
application works. Then review the remaining information in the User Manual and Modules that apply to
your application of interest for details on using each module or feature.
It is also worthwhile to review the Introduction, Applications, and Features sections of each module to
determine if they can help you to model your systems.
Getting Started
To execute the graphical software (once the environment and path has been set according to the
installation instructions that can be found on the CFD Portal) from the command line, enter one of the
following commands in a DOS window on Windows Systems or in a shell on Linux/UNIX systems:
• CFD-GEOM
• CFD-CADA
• CFD-VisCART
• CFD-ACE-GUI
• CFD-FASTRAN-GUI
• CFD-TOPO-GUI
• CFD-VIEW
• SimManager
The appropriate solver can be executed from CFD-ACE-GUI, CFD-FASTRAN-GUI, CFD-TOPO-GUI, or
SimManger. They can also be submitted from the command line using:
• CFD-ACE-SOLVER -dtf model.DTF
• CFD-FASTRAN-SOLVER -dtf model.DTF
• CFD-TOPO-SOLVER -dtf model.DTF
If multiple versions of the software have been correctly installed, then the old version can be executed
using: CFD-GEOM -runver 2006 (which will run version 2006 of GEOM).
Note your license file will dictate which applications you can execute.
Windows users that installed via CD will have short cuts under Start -> Programs -> ESI-Software. If
your software was received via ftp or the CFD portal, then you can create your own short cuts. To do so:
1. Create an ESI_Software folder typically under C:\Documents and Settings\All Users\Start
Menu\Programs
2. Copy the desired icons from the latest 20xx.x\UTILS\icons directory in the ESI_Software folder
3. In Windows Explorer, right click on the icon and select: Create Shortcut
4. Right click on the just created shortcut and select: Properties
5. Change the target to the desired application in the 20xx.x\UTILS\bin directory (for instance: CFD-
VIEW.exe)
20
Modules
Matteo.Palmonella@esi-
Africa
group.com
Asia Pacific
China [email protected]
Japan [email protected]
Korea [email protected]
Venugopal.Rallabhandi@esi-
India
group.com
Rest of Asia [email protected]
Europe
Eastern Europe [email protected] [email protected]
France [email protected]
Germany [email protected]
Italy [email protected]
Emilio.Mencia.de.Miguel@esi-
Spain
group.com
United Kingdom [email protected]
Matteo.Palmonella@esi-
Rest of Europe
group.com
Matteo.Palmonella@esi-
Middle East
group.com
Americas
North [email protected]
South [email protected]
21
CFD-ACE_V2010.0_Modules_Manual_Part2
• E-mail: [email protected]
• Telephone: 256-713-4750 (United States country code is 01)
When reporting a problem it is important to have the following information:
• CFD-ACE-GUI version number
• CFD-ACE-Solver version number
• Modules you were using
• Type of problem you were trying to solve
• Any error messages you may have received in the modelname.out file or screen
Electric Module
22
Modules
• Electric-Theory
• Electric-Limitations
• Electric-Implementation
• Electric-Symbol Definitions
• Electric-References
Electric Module Applications
Common uses of the Electric Module are given below. This is not a complete list and many more uses are
possible.
• Capacitance of metal/dielectric structures such as parallel plates, concentric spheres, etc.
• Electric fields due to distributions of space charge and equipotential contacts.
• Electrostatic loading of mechanical structures. The Electric Module calculates the electrostatic
pressure force which can be used with the Stress Module to determine the structural response to
electrostatic loads. A prime example of this would be micro systems such as a doubly clamped
beams, accelerometers, high frequency resonators, electrostatic torsional micro mirrors, linear
lateral resonator comb drives, angular resonator comb drives, and fluid damped beams.
• Electrostatic force on a conductive bath. In conjunction with the Free Surfaces (VOF) Module, the
Electric Module can calculate the force on a conductive fluid, addressing problems such as the
extraction of a conductive fluid from a bath.
• Joule heating.
• Current density as a source for a magnetic field calculation.
• Electrical field distribution in an electro-kinetically driven flow field as it occur some biochips and
bioMEMS devices.
• Simulation of nonuniform electric field (AC field) in a Dielectrophoretic (DEP) system.
• Electric field distribution in the electro-fluidic system to calculate pH value.
Electric Module Features
You can use the Electric Module as stand-alone or with other modules for multi-disciplinary design, e.g.,
electromechanical, electrothermal, electrochemical, electrofluidic, etc. The Electric Module has several
unique features:
• A multi-pole accelerated BEM solver which has a solution speed on the order of O(N) fast where
N is the number of faces and does not require volume meshing;
• An FVM solver that can effectively solve complex multi-physics and nonlinear problems
• Support both structured and unstructured meshes (tetrahedral, hexahedral, prisms, and general
polyhedrals)
• Coupling with other physics such as the Heat Transfer, Stress, Free Surfaces (VOF), and
Magnetic Modules.
• Coupling with dynamic problems such as moving deforming materials and moving space charges
(e.g., charged particles).
Electric-Theory
Electric Module
Theory-Introduction
23
CFD-ACE_V2010.0_Modules_Manual_Part2
Electrostatic fields are generated by a combination of stationary charges and applied potentials.
Electrostatics is governed by Gauss’ Law, which states that the net electric flux passing through a closed
surface is equal to the net charge enclosed by that surface. In differential form and mks units, Gauss’s
Law is written as:
∇• D = ρ (10-1)
where the vector D is the electric displacement flux density (C/m ) and ρ is the volume charge density,
2
3
(C/m ). The electric flux density, D, and electric field, E, are related by the constitutive relation:
D = εE (10-2)
where:
ε = εrεo (10-3)
In electrostatics, the electric field vector, E, is irrotational (i.e., the curl of the electric field is zero, ∇ x
E = 0) so a scalar electric potential, φ, can be defined as:
E = − ∇φ (10-4)
Inserting equation 10-2, equation 10-3, and equation 10-4 into equation 10-1 we get Poisson’s equation
for the electric potential:
∇• (ε r ε o ∇φ) = −ρ (10-5)
Equation 10-5 is solved by the electrostatic option of the Electric Module if the Electrostatics option is
selected. After the electric potential φ is solved using equation 10-5, CFD-ACE+ computes other
quantities that depend on φ. These other quantities include the electric field, surface charge, capacitance,
electrostatic pressure force, energy contained in the electric field, and the electric virtual force (energy
gradients).
Electric fields are calculated in all dielectric materials and in free space surrounding them. In
electrostatics, it is assumed that conductors have zero electric field (φ is constant). To specify that a
material is a conductor in the electrostatic option of the Electric Module the electrical conductivity has to
be set to Perfect Conductor.
The boundary conditions for φ are:
• at all metal contacts specify the electric potential or voltage
• on external boundaries apply zero voltage or surface charge density
• at the interfaces between two dielectrics
.
Where E n is the normal electric field at side 1 or 2. Apply the Ignore or Dielectric option.
The Electric Theory section includes:
• Theory-Electric Conduction (Currents in Conductors and Lossy Dielectrics)
• Theory-Leakage Currents
24
Modules
Theory-Electric Conduction
Electric Module
Theory-Electric Conduction (Currents in Conductors and Lossy Dielectrics)
The Electric Conduction option calculates the current density J in conductors and lossy dielectrics. The
fundamental equation governing the current flow is the electrical current continuity equation:
(10-6)
where J is the electric current density (A/m ) and ρ (C/m ) is the charge density. Substituting ρ of Gauss’
2 3
Law for the Electric Field ( equation 10-1) into the continuity equation (equation 10-6) and assuming the
time derivative and the divergence operators commute results in:
(10-7)
Next, equation 10-7 is written in terms of the electric field. Assuming the total current density J is due only
to conduction current (e.g., J = J c = σE), and all the medium are isotropic (e.g., D = εE) equation 10-7
becomes:
(10-8)
where σ is the electric conductivity measured in (Ω-m) (note 1/σ is the resistivity).
-1
The electrical resistivity 1/σ can be specified as either a constant value or a function of temperature:
(10-9)
where:
Finally, equation 10-8 is written in terms of the electric potential. Using the definition of the electric field in
terms of the electric potential and the magnetic vector potential:
25
CFD-ACE_V2010.0_Modules_Manual_Part2
and assuming that there are no time varying magnetic fields (i.e., = 0) results in E =- ∇φ which is
substituted into equation 10-8.
(10-10)
The general equation for the electric potential in a conductor or lossy dielectric is given by equation 10-
10.
There are three different options (approximations) for solving the conduction problem (equation 10-10):
DC conduction, AC conduction (one frequency), and time domain conduction.
After the electric potential φ is solved using equation 10-10 the Electric Module computes other quantities
that depend on φ. These other quantities include resistance or impedance, source currents (for the
Magnetic Module), joule heating sources (for the Heat Transfer Module) and body forces (for the Free
Surfaces (VOF) Module).
See Also
Electric Conduction-DC Conduction
Electric Conduction-AC Conduction (One Frequency)
Electric Conduction-Transient Conduction (Time Domain)
Electric Conduction-Dielectric Charge Leakage
Electric Module
Theory-Electric Conduction-DC Conduction
The DC Conduction option assumes steady state conditions ( −>0). Assuming steady state or dc
currents, equation 10-10 becomes a Laplace equation for the electric potential:
(10-11)
Electric Module
Theory-Electric Conduction-AC Conduction (One Frequency)
The AC Conduction option solves the current continuity equation assuming a sinusoidal steady state.
Converting the current continuity equation ( equation 10-10) into the frequency domain where the electric
potential becomes a complex quantity yields:
26
Modules
(10-12)
The electric potential of this equation is complex (i.e., ) so two equation need to be solved:
(10-13)
(10-14)
For many cases the dc conduction problem will suffice. For ω = 0.0 or where σ >> ωε, equation 10-12
reduces to the dc conduction problem ( equation 10-11). For a typical conductor σ ≈ 10 Ω m and ε ≈ ε o
6 -1 -1
= 8.85x10 C /Nm . So the DC conduction equation is appropriate for frequencies (f = ω/2π) on the order
-12 2 2
10
of 10 Hz or less.
In the AC conduction option the real and imaginary part of the electric potential in equation 10-12 are
solved.
See Also
Electric Conduction (Currents in Conductors and Lossy Dielectrics)
Electric Conduction-DC Conduction
Electric Conduction-Transient Conduction (Time Domain)
Electric Conduction-Dielectric Charge Leakage
Electric Module
Theory-Electric Conduction-Transient Electrical Conduction (Time Domain)
(10-15)
In the transient conduction option of the Electric Module, equation 10-10 is solved for the electric potential
φ. To achieve economical solution speed, a variable time step may have to be used.
See Also
Electric Conduction (Currents in Conductors and Lossy Dielectrics)
Electric Conduction-DC Conduction
Electric Conduction-AC Conduction (One Frequency)
Electric Conduction-Dielectric Charge Leakage
Electric Module
Theory-Leakage Currents
Imagine a capacitor with two dielectric materials between two contact electrodes connected to a charging
circuit (e.g., battery) as shown:
27
CFD-ACE_V2010.0_Modules_Manual_Part2
Electric Module
Theory-FVM and BEM Electrostatics Solution Techniques
Two different techniques are available for solving the electrostatic equation:
• FVM (Finite Volume Method) - The FVM requires a volume mesh and calculates the electric
potential (and electric field) at every cell (volume) in the mesh. The BEM requires only a surface
mesh (although volume information is required in specifying material volume conditions) and
calculates the electric potential only at boundary and interface faces.
• BEM (Boundary Element Method) - The BEM method is recommended for open boundary
problems, for problems with small surface/volume bodies, external fields, etc. FVM method is
recommended for enclosed systems, for large surface/volume bodies, for electrical current flows
in conductors, for electro-thermal simulations, etc.
Electric Module
Theory-Surface Charge Density and Total Charge
(10-16)
28
Modules
(10-17)
Electric Module
Theory-Capacitance
If an electrical potential is applied between two separated conductor plates, a charge +Q appears on one
plate and -Q on the other, and an electric field, E, is established between the plates.
Electric Field Due to a Voltage Difference on the Plates of a Parallel Plate Capacitor
The ratio between the separated charge (Q) and an electrical potential difference is called capacitance, C
(C = Q/V). The unit of capacitance is the Farad which is Coulomb/Volt.
Capacitance (Q/V) is a purely geometrical quantity determined by sizes, shapes and separation of
conductors. V by definition is the potential of positive conductor less that of negative conductor and Q is
the charge of the positive conductor. Capacitance is a positive quantity (C > 0).
In the Electric Module a total charge is calculated for every set of boundaries with a unique fixed
electric potential φ o . To determine the Q at a fixed potential boundary, one starts with the electric field
boundary conditions.
29
CFD-ACE_V2010.0_Modules_Manual_Part2
(10-18)
(10-19)
Equation 10-18 states that the discontinuity at an interface in the normal component of the displacement
flux is equal to the surface charge at the interface. Assume one region (2) is metallic so that the electric
field goes to zero there. The total charge on a contact is then given by equation 10-20. The total charge is
the sum at each face with a fixed potential (φ o ) of the incident displacement flux normal to the fixed
electric potential boundary (ε i E ni ) times the area of the face (A i ).
(10-20)
A capacitance is calculated for each set of faces with a unique potential boundary condition. First, the
total charge on each collection of faces with a unique fixed potential boundary condition given by equation
10-20 is calculated. Then for those face sets with φ ≠ 0.0 the ratio is calculated.
Electric Module
Theory-Capacitance Matrix for Multiple Conductors
In several applications (e.g., electronic interconnects of integrated circuits) large numbers of conductors
are arranged in close proximity to each other. In electronic integrated circuits made of conductive
materials (e.g., Al, Cu,...) separated by a dielectric (e.g., SiO 2 ), a large number of capacitances exist.
There may be desired capacitance effects or parasitic effects that have unwanted influence on circuit
performance. Imagine three conductors and a ground plate (see figure below). If one of the conductors
has an applied voltage of unity (e.g., V = 1) and others are grounded (e.g., V = 0). A capacitance matrix
can be formed as:
(10-21)
30
Modules
Q 1 = C 10 + C 12 + C 13 (10-22)
The capacitance matrix relates charges Q and voltages V for any number of conductors.
(10-
23)
The resistance R 12 between any pair of contacts 1 and 2 is calculated using Ohm’s Law:
R 12 = V 12 /I 12 (10-24)
where I 12 is the total current flow in the conductor from contact 1 to 2 and V 12 is the electric potential
difference between contact 1 and 2. The total current is obtain by summing the dot product of current
density with the area elements at the contacts (J²•An A ). The potential difference is obtained by V 12 = φ 1 -
φ2.
Electric Module
Theory-Electrostatic Pressure Force (Electrical/Structural)
The coupling between the electrostatic option in the Electric Module and the Stress Module is through
pressure forces. The Electric Module solves Poisson’s equation and uses the electric fields to calculate
the pressure force. The Stress Module uses the pressure forces as boundary conditions.
Electrostatic Pressure (P) is calculated using the following relation (Bachtold, 1997):
31
CFD-ACE_V2010.0_Modules_Manual_Part2
(10-25)
The electrostatic pressure on the conductor surface is always triggering for interfaces between two
dielectric materials with different electric permittivity values e1 and e2. Electrostatic pressure is given by:
(10-26)
Electric Module
Theory-Virtual Force (Electric/Free Surfaces)
A virtual force calculation was added to enable the coupling with the Free Surfaces (VOF) Module. Since
boundary conditions at the free surface are no longer directly associated with a mesh face, a force
calculation that did not rely on an available list of boundaries was needed. The solution was to calculate
the virtual force at a volume due to the varying electric energy in the region. The electric potential is used
to calculate a force at the surface of the liquid due to an electric field by solving
(10-27)
where ε is the electric permittivity. The virtual force per unit volume F virt is obtained by taking the gradient
of the energy per unit volume stored in the electric field. To couple the Electric Module with the Free
Surfaces (VOF) Module, VOF is given access to the electric material properties and recalculates them as
a function of fluid fraction. The conduction problem is then solved and the electric virtual forces at every
cell center are calculated by taking the gradient of energy stored in the electric field. For coupling to the
VOF method, the virtual force is applied only to cells which have a fractional content of the liquid.
Elsewhere in the domain the force is assumed zero. These virtual forces are passed as a body force to
the fluid model.
Electric Module
Theory-Electric Current Density (Electric/Magnetic)
In the DC conduction option, current sources can be calculated for the Magnetic Module. This is a useful
method for describing source currents in that you do not have to specify a direction of the current, it
enables the use of source currents of arbitrary geometry, and it allows the current density direction to
reflect any changes due to a moving grid.
J = σE (10-28)
32
Modules
Electric Module
Theory-Joule Heating (Electric/Heat Transfer)
In the DC conduction option of the Electric Module Joule heating sources are calculated when it is
coupled with the Heat Transfer Module. The heat generated in a conductor due to current flow through
conductive materials (Joule heating) is:
(10-29)
Theory-Peltier/Seebeck Effect
Electric Module
Theory-Peltier/Seebeck Effect
The thermoelectric Seebeck effect generates electrical potential difference in any conducting material
with temperature gradient. The value of this potential is proportional to temperature dependent material
Seebeck coefficient and temperature difference. This effect is not connected with material junction or
Joule heating effect.
When the electrical current flows through a material with non-zero Seebeck coefficient two additional
effects are observed. The reversible change in heat content at an interface between dissimilar
conductors, called Peltier effect, and change of heat content within conductor, called Thomson effect.
Seebeck Effect
Seebeck Effect
The temperature gradient within isolated conductor generates electrical potential. The value of this
potential is proportional to Seebeck coefficient of material and the temperature difference. The Seebeck
coefficient is temperature dependent.
where:
Peltier Effect
33
CFD-ACE_V2010.0_Modules_Manual_Part2
Peltier Effect
The flow of current through materials junction absorbs power proportional to relative Peltier coefficient π 21
and current value I. The direction of current flow and a sign of coefficient determine whether heat is
liberated or absorbed. The relative coefficient is equal to difference between absolute material Peltier
coefficients π 21 =π 2 - π 1 . The Peltier effect is a result of the change in the entropy of the electrical charge
carriers as they cross the junction.
where:
Thomson Effect
Thomson Effect
The flow of current through a material with temperature gradient absorbs power proportional to Thomson
coefficient β, the temperature difference, and the current value I.
where:
Coefficient Values
The electrical and thermal energy conservation equation for thermoelectric relationship leads to following
equation:
34
Modules
The zero net change in the enthalpy approximation leads to another equation:
Those two equations were used to find dependence between Peltier, Thomson, and Seebeck coefficients.
This leads to following formulas:
Therefore:
Electric Module
Theory-Peltier/Seebeck-ACE+ Implementation
Both Peltier and Thomson effect create additional heat sources for the Heat module.
The Peltier effect absorbs (generates) heat on the boundary between two materials. Those materials
should have different Seebeck coefficients. This effect is implemented as surface heat source. For each
face the difference between Seebeck coefficients on both sides is computed. Than the result value is
multiplied by face temperature and electrical current flowing through this face. The heat source is added
to adjoining cells with weighting factor of distance and thermal conductivity.
Therefore:
The Thomson effect absorbs (generates) heat inside material. The material should have temperature
dependent Seebeck coefficient. This effect is implemented as a volume heat source. The equation for this
effect was changed in the following way.
where:
35
CFD-ACE_V2010.0_Modules_Manual_Part2
The Electric module in ACE+ for DC conduction problem solves the Laplace equation:
where:
The introduction of thermoelectric Seebeck effect will add an additional source term to those equations:
where:
S - thermoelectric Seebeck coefficient
The first equation is rearranged to following form:
The right side of this equation is the additional source term for the Electric module. The second equation
(2) modifies the electric field equation in the same module.
Known Problem
Those two equations should not generate artificial currents after converged solution, since the terms
added to first and second equations are the same. Unfortunately, because S∇T is computed differently in
equations (1) and (2), due to different discretization schemes, those terms do not reduce themselves.
This problem generates artificially big value of E at local points and solution instability. The high local
current J=σE generates heat and prevents convergence in the Heat module. That effect influents mutually
∇T term in equations.
Electric Module
Theory-Peltier/Seebeck-Validation
36
Modules
The electric current flows from the right side of structure to the grounded left side. All other sides were set
to adiabatic conditions and were electrically isolated.
The structure has rectangular uniform cross-section.
2
The current was set to 1A per 1mm of cross-section. The structure in x direction is build from three
layers, counting from left, 5 mm metal conductor, 5 mm thermoelectric material, and 5 mm of metal
conductor.
The metal property was set to the following values:
•
8
Electrical conductivity 10 1/Ω-m
• Thermal conductivity 100 W/m-K
Thermal conductivity of the thermoelectric material was 1 W/m-K.
The Seebeck coefficient of the thermoelectric material was set to 200 µV/K.
7
For this case the electrical conductivity of the material was set to high value 10 1/Ω-m, to make the Joule
2
effect negligible (P J = R I = 0.5mW) comparing to Peltier effect
(P P =S I T = 0.2 mW/K * T = 60mW for 300K).
The thermal resistance of the thermoelectric is:
2
1/R T = 1 W/m-K * 1mm / 5mm = 0.2mW/K => R T = 5 K/mW
Therefore the temperature drop on the thermoelectric should be approximately:
∆T = P * R T = 0.2mW/K*(300K-∆T) * 5 K/mW = 300K-∆T => ∆T » 150K
The images below present the result of ACE+ simulation. The first one presents a temperature map, the
second the temperature graph.
37
CFD-ACE_V2010.0_Modules_Manual_Part2
The Seebeck coefficient of the material was 200 µV/K. For this case, the electrical conductivity of the
2
material was 80,000 1/Ω-m. The total Joule heating in this case (P J = R I = 62.5mW) is comparable to
the Peltier effect.
The analytical solution for 1D case gives following results:
• Temperature at x=5mm (left side of TE material) should be 304K
• Temperature at x=10mm (right side of TE material) should be 232K
• Maximum temperature in TE should reach 315K at x=6.3mm
The images below present the result of ACE+ simulation. The first one presents temperature map, the
second the temperature graph.
38
Modules
39
CFD-ACE_V2010.0_Modules_Manual_Part2
Electric Module
Theory-Peltier/Seebeck-Thermoelectric Effects Example
Verification Case 1
Two elongated blocks of thermoelectric materials with different Seebeck coefficients were connected
together at one end (right-hand side in Figure 1). The isothermal boundary condition of 400K was set for
the connected end (right-hand side in Figure 1) and the isothermal condition of 300K to opposite side of
the blocks (left-hand side in Figure 1).
40
Modules
41
CFD-ACE_V2010.0_Modules_Manual_Part2
Electric Module
Implementation-Grid Generation
The Electric Module can be applied to any geometric system (3D, 2D planar, or 2D axisymmetric). and
supports all grid cell types (quad, tri, hex, tet, prism, poly).
The general grid generation concerns apply, i.e., ensure that the grid density is sufficient to resolve
solution gradients, minimize skewness in the grid system, and locate computational boundaries in areas
where boundary values are well known.
Electric Module
Implementation-Model Setup and Solution-Introduction
42
Modules
CFD-ACE+ provides the inputs required for the Electric Module. Model setup and solution requires data
for the following panels:
• Problem Type
• Model Options
• Volume Conditions
• Boundary Conditions
• Coupling with Other Modules
• Initial Conditions
• Solver Control
• Output
Electric Module
Implementation-Model Setup and Solution-Problem Type
Click the Problem Type [PT] tab to see the Problem Type Panel. See Control Panel-Problem Type for
details.
Select Electric to activate the Electric Module.
Electric Module
Implementation-Model Setup and Solution-Model Options-Introduction
Click the Model Options [MO] tab to see the Model Options Panel. See Control Panel-Model Options for
details. The Model Options section includes:
• Model Options-Shared
• Model Options-Electric
• Model Options-Flow
• Model Options-Heat
Electric Module
Implementation-Model Setup and Solution-Model Options-Shared
There are no settings under the Shared tab that affect the Electric Module. See Control Panel-Model
Options for details.
Model Options-Electric
Electric Module
Implementation-Model Setup and Solution-Model Options-Electric-Introduction
The model options for the Electric Module are located under the Electric (Electr) tab.
43
CFD-ACE_V2010.0_Modules_Manual_Part2
44
Modules
Advanced tab in the Solver Control Panel in Electric Module Settings (for BEM Solver) Mode
• Unbounded - The Unbounded toggle specifies if the problem is bounded or unbounded. A
bounded problem assumes the boundary elements bound the solution space. An unbounded
problem assumes a solution is unbounded with the electric potential φ(∞) = 0. Unbounded is true
(toggle on) if the problem is an unbounded problem, false (toggle off) if the problem is bounded.
• Iterations - The iterations controls the maximum number of iterations to be made by the Electric
Module before exiting. The default value is 200.
• Interpolation Order - The Interpolation Order controls the interpolation level on a face where the
potential and/or gradient will be solved. Enter a value of 0 to specify uniform interpolation order, 1
to specify linear. An interpolation order greater than 0 provides accurate results but increases the
computational expense of the calculation drastically. Only uniform (0) and linear (1) interpolation
are currently available. In addition, linear interpolation is only available for 3 or 4 sided faces (i.e.,
triangles and quadrilaterals). If linear interpolation is specified for faces with more than four sides,
the code will use uniform order interpolation. Uniform interpolation introduces only one node/face
the at face center. Linear interpolation introduces n nodes for a n-sided face. Values less than
zero are changed to zero and values greater than 1 are changed to 1. The default value is 0.
• Expansion Order - Expansion Order controls the method of approximating the kernel function as
a series expansion. Currently only Taylor series expansion of the kernel function is available.
Multipole expansion is not implemented due to its lack of generality. In the future, other expansion
modes may become available. The default and only possible value is 2, any changes made here
will not affect the solution.
• Distribution Criteria - The Dist. Criteria is the multipole accuracy parameter and controls the
level of fast used in the Electric Module. A zero distribution criteria value is conventional BEM
(with no attempts made to speed up the algorithm.) The solution will have a high level of accuracy
but the solution speed will be very slow. As the parameter is increased from zero, the level of
optimization increases as attempts are made to speed the level of convergence. Mechanisms,
such as clumping together boundary elements into an equivalent boundary element, are used.
The optimization (larger Distribution Criteria values) provides faster solutions with a lower level of
accuracy of the solutions and an increase in the memory requirements to achieve a solution.
Values greater than one or less than zero are changed to 0.5. The default value of 0.5 is
recommended.
• Relative Permittivity (ε r ) - Relative Permittivity is available when the unbounded option is
selected. The default value is 1.0.
45
CFD-ACE_V2010.0_Modules_Manual_Part2
Electric Module
Implementation-Model Setup and Solution-Model Options-Electric-Capacitance Matrix
If you select the Capacitance Matrix option, CFD-ACE+ will calculate the capacitance matrix composed of
multi-solid conductors and output the text file DTF_model_name.cap. A pull-down menu appears enabling
you to choose the Boundary Trace Method or Volume Trace Method.
Single Capacitor
(10-30)
Where, V is the electric potential across the plates. The SI unit of capacitance is farad (F):
(10-31)
In a lot of cases, the farad is a very large unit. It is common to use smaller units such as microfarad (µF)
or picofarad (pF):
(10-32)
46
Modules
(10-33)
For a system composed of several conductors, the capacitance becomes a capacitance matrix.
Capacitance Matrix
It will often write out a symmetric matrix of a capacitor system:
(10-34)
(10-35)
(10-36)
47
CFD-ACE_V2010.0_Modules_Manual_Part2
i j
(2) calculate electric field by solving electrostatic potential
(3) calculate charge q j for each conductor
(4) calculate mutual C ij
This will take file1.DTF, process it and create a new file, file2.DTF, which has the new
volume conditions/cell groups and boundary and interface patches. It requires file1.DTF
be set up to run ACE+ in a particular mode (emag?).
New
interconnect_tracerT
48
Modules
Where:
Electric Module
Implementation-Model Setup and Solution-Model Options-Flow
If the flow module has been activated in conjunction with the electric module, Lorentz Forces can be
included. (The interaction of a conductor with the electric/magnetic field, J x B). To activate Lorentz
Forces, click on the Lorentz Force box located in the Flow tab.
When the Electric Module is used with the Flow Module, and the Magnetic Module is not used, you must
specify a constant magnetic field for the Lorentz force.
If the Magnetic Module is activated, the computed magnetic filed is used automatically by CFD-ACE+.
49
CFD-ACE_V2010.0_Modules_Manual_Part2
Electric Module
Implementation-Model Setup and Solution-Model Options-Heat
If you use the Heat Module with the Electric Module, the effects of Joule heating will be simulated for DC
conduction, AC conduction, or transient electric model options.
50
Modules
Electric Module
Implementation-Model Setup and Solution-Volume Conditions-Introduction
Click the Volume Conditions [VC] tab to see the Volume Condition Panel. See Control Panel-Volume
Conditions for details. Before volume conditions can be assigned, you must select an entity from the
Viewer Window or the VC Explorer. Before the Piezo tab will appear, you must select the Solid option
from the VC Setting Mode Properties menu.
When you select Properties from the Setting Mode pull down menu, the following properties appear:
Flow, Porous Media, Momentum Resistance, Heat, Species, General, and Stress. The panel options
change with each selection.
The Volume Conditions panel contains the following tabs:
• Volume Conditions-Phys
• Volume Conditions-Struct
• Volume Conditions-E/M (Electric/Magnetic)
• Volume Conditions-Piezo (Piezoelectric)
Electric Module
Model Setup and Solution-Volume Conditions-Phys
51
CFD-ACE_V2010.0_Modules_Manual_Part2
Electric Module
Implementation-Model Setup and Solution-Volume Conditions-Struct
The Volume Conditions Struct tab enables you to set the following options:
• Material Type: Linear
• Material: Isotropic or Anisotropic
• Local X-Axis Vectors: Constant with a11 and a12 values
• Young's Modulus: Constant with E1, E2, and E3 values
• Shear Modulus: Constant with G12, G13, and G23 values
• Poisson's Ratio: Constant with nu12, nu13, and nu23 values
• Thermal Expansion Coefficient: Constant with Alpha1, Alpha2, and Alpha3 values
52
Modules
Volume Conditions-Electric/Magnetic
Electric Module
Implementation-Model Setup and Solution-Volume Conditions-Electric/Magnetic-Introduction
53
CFD-ACE_V2010.0_Modules_Manual_Part2
The Volume Conditions E/M tab enables you to set the following options:
• Electrical Conductivity and Electrical Conductivity (Resistivity)
• Relative Permittivity
• Relative Permeability
• Space Charge
Electric Module
Implementation-Model Setup and Solution-Volume Conditions-Electric/Magnetic-Electric Conductivity
The E/M tab Electric Conductivity section enables you to select an evaluation method.
54
Modules
• Perfect Conductor - This method is only available when you choose Electrostatics in the MO
panel (see Model Options-Electric Tab-Electric Field Options for details) and Solid Properties in
the VC panel (see Volume Conditions-Introduction for details). This option tells the CFD-ACE-
Solver that this domain Electric Field does not exist and this domain will not be included in the
solution to Poisson’s equation. This is true for both FVM and BEM problems.
• Perfect Insulator - This method is available for Electrostatics only. This method tells the CFD-
ACE-Solver that this domain is a part of Poisson’s equation solution domain and electric field
does exist in this domain
• Constant - When this method is selected, the constant value of electrical conductivity must be
specified.
• Constant Resistivity - When this method is selected, the constant resistivity of material must be
specified. CFD-ACE-Solver calculates the electrical conductivity of material as inverse of
resistivity.
• Function of Concentration - This option is for fluids and only available when you choose
Chemistry in the PT and MO panel, Buffer Conductivity, and the Media as Liquid Phase. (Select
an entity from the Explorer List. In the MO panel, change the Chemistry Media to Liquid Phase. In
the VC Panel/VC Setting Mode, in the Properties pull-down menu, change the Properties to Fluid.
Then click the E/M tab and from the Electrical Conductivity (Resistivity) menu, select Function of
Concentration). This option should be used for Electro-Chemistry only.
• Resistivity Function of T - When this method is selected, the reference temperature,
Coefficient, and resistivity at reference temperature must be specified. The Electric Module,
calculates resistivity based on local temperature using the following relation.
(10-37)
(10-38)
The above option is available only if you have activated the Heat Transfer Module is activated.
• User Sub (uecond) - When you select this method, the solver expects a user subroutine uecond
that you have written. For details on how to write and compile user subroutines, see User
Subroutines.
Electric Module
Implementation-Model Setup and Solution-Volume Conditions-Electric/Magnetic-Relative Permittivity
The E/M tab Electric Conductivity section enables you to select the relative permittivity.
55
CFD-ACE_V2010.0_Modules_Manual_Part2
Only the relative permittivity (ε r ) of the total electric permittivity (ε r ε o ) is set in CFD-ACE+. The value used
for the permittivity of free space is ε o = 8.854187 x10 F/m or C /Nm . The relative permittivity defaults to
-12 2 2
• Constant- If you select this option, enter a permeability factor in the Relative Permeability field.
• User Sub(urpermeability) - If you select the User Sub (urpermeability) option, the solver expects
a user subroutine urpermeability that you have written. For details on how to write and compile
user subroutines, see User Subroutines.
Electric Module
Implementation-Model Setup and Solution-Volume Conditions-EM-Space Charge
The Electric/Magnetic Space Charge enables you to select the space charge evaluation method.
•
3
Density - The space charge (ρ) is specified as a density in units of C/m (i.e., per unit volume).
The space charge defaults to 0.0 for all cells in the grid.
• Total - The space charge (ρ) is specified as a total charge in units of C (i.e., for whole selected
volume). The space charge defaults to 0.0 for all cells in the grid.
Electric Module
Implementation-Model Setup and Solution-Volume Conditions-Piezoelectric
56
Modules
The Piezoelectric option is available only if you have chosen MO Panel->Electric Tab->Electric Field
Options->Piezoelectric. Choose this option only for those solid volumes where piezoelectric equations are
to be solved. Choosing this option displays one of the following panels (2D or 3D), depending on the
dimensions of the problem. Steps for using the panel are described below.
57
CFD-ACE_V2010.0_Modules_Manual_Part2
Constitutive Forms
From the Constitutive Forms pull-down menu, choose the stress-charge or a strain-charge option:
Stress-charge - with this option, structural stress and the electric flux density are expressed in terms of
the strain and the electric field vector as:
(10-39)
(10-40)
Strain-charge - with this option, strain and electric flux density are expressed in terms of the stress and
electric field:
58
Modules
(10-41)
(10-42)
where:
These equations represent the standard structural and electric field constitutive relations, with the addition
of the coupling terms in which an electric field {E} produces a stress {σ} (or a strain {γ}), and a strain field
{γ} (or a stress field {σ}) produces an electric flux {D}.
Conversion Between Stress Charge and Strain Charge
The stiffness matrix is obtained from information provided in the Struct section of the volume conditions
and thus need not be given here. In this section, for either formulation, you must enter the dielectric
matrix and the piezoelectric coupling matrix.
Inside the code, the calculations are done using the stress-charge formulation. The strain-charge
formulation is input, the stress-charge data is obtained from the following transformation.
(10-43)
(10-44)
Piezoelectric materials are typically anisotropic, in which material properties depend on direction. For
anisotropic materials, a local material coordinate direction for each volume is specified under the Struct
property tab. The dielectric matrix and piezoelectric coupling matrix are expressed in terms of that local
coordinate system. If the material was specified as isotropic, the local system is taken as the global
system.
Dielectric Matrix
The dielectric matrix is a diagonal matrix and only the diagonal terms need to be given. It is expressed in
terms of the local coordinate system. If the material was specified as isotropic, the local system is taken
as the global system.
Piezoelectric Coupling Matrix
59
CFD-ACE_V2010.0_Modules_Manual_Part2
The piezoelectric coupling matrix is a 6 x 3 matrix in 3D, and is 4 x 2 in 2D. The entries in this matrix must
correspond to the stress (or strain) vector and electric field vector as:
(10-45)
(10-46)
(10-47)
(10-48)
for 2D.
The input box for the piezoelectric coupling matrix contains fields for each entry of the matrix. The
subscripts 1 to 6 refer to the stress (or strain) components σ 11 through σ 12 as defined in Equation 10-45,
and the coordinate axes X, Y, and Z correspond to the electric field directions 1, 2, and 3, respectively
(Equation 10-46).
As an example, consider Equation 10-39, in which the stress vector is written in terms of the strain and
the electric field. Expanding the stress contribution from the electric field (the second term on the RHS)
gives:
(10-49)
(10-50)
(10-51)
(10-52)
(10-53)
(10-54)
The piezoelectric coupling matrix is typically sparse, reflecting the physical behavior of the material. As an
example, a typical strain-charge piezoelectric coupling matrix for quartz is:
60
Modules
From this matrix, it is seen that the electric field component E 1 produces strains γ 11 , γ 22 , and γ 23 , and
component E 2 produces strains γ 13 and γ 12 . The E 3 component of the electric field vector does not
contribute to the strain field.
The input for the piezoelectric coupling matrix for quartz is shown in the following figure:
61
CFD-ACE_V2010.0_Modules_Manual_Part2
For this material, the E 3 electric field component produces strains γ 11 , γ 22 , and γ 33 , component E 2
produces strain γ 23 , and E 1 produces γ 13 .
Note that these properties are defined in a local material coordinate system, which must be input relative
to the global system in the VC->Struct panel when the anisotropic material option is chosen. See the
Volume Conditions section of the Stress Module for a description of the anisotropic material axes.
Finally, note that some references define the piezoelectric coupling matrix as the transpose of the
definition used here. In that case, the matrix size from the reference would be 3x6 (for 3D) or 2x4 (for 2D).
When specifying the matrix values in the GUI, use the transpose of the referenced values.
Electric Module
Implementation-Model Setup and Solution-Boundary Conditions-Introduction
Click the Boundary Conditions [BC] tab to see the Boundary Conditions Panel. See Control Panel-
Boundary Conditions for details. To assign boundary conditions and activate additional panel options,
select an entity from the viewer window or the BC Explorer.
The Electric Module is fully supported by the Cyclic and Arbitrary Interface boundary conditions. (See
Cyclic Boundary Conditions or Arbitrary Interface Boundary Conditions for details on these types of
boundary conditions and instructions for how to implement them). The Electric Module does not support
the Thin Wall boundary condition feature.
All of the general boundary conditions for the Electric Module are located under the Electric (Electr) tab
and can be reached when the boundary condition setting mode is set to General. The boundary
conditions available in the Electric Module depend upon which Electric Field option is selected in the MO
page: Ignore, Surface Charge, Fixed Potential, Fixed Current, and Dielectric. These boundary conditions
are available on all computational boundaries, regardless of type (Inlet, Outlet, Wall, etc.).
The Electric Boundary Conditions include:
• Boundary Conditions-Ignore
• Boundary Conditions-Fixed Potential
• Boundary Conditions-Surface Charge
• Boundary Conditions-Fixed Current
• Boundary Conditions-Dielectric
• Boundary Conditions-Piezoelectric
Electric Module
Implementation-Model Setup and Solution-Boundary Conditions-Ignore
The Ignore boundary condition indicates that a set of faces has no electric boundary conditions and is
ignored by the Electric Module. This boundary condition is useful when multi-disciplinary problems are
being specified. Use the Ignore boundary condition to specify no electric boundary condition for a set of
faces that may have a boundary condition for another problem type. This boundary condition does not
require any constants to be specified. This option is available for Interfaces and for BEM on all boundary
types.
Electric Module
Implementation-Model Setup and Solution-Boundary Conditions-Fixed Potential
The Potential (Voltage) boundary condition specifies the electric potential at a cell faces. When you select
this option, enter the desired voltage (in volts) for the specified cell faces.
62
Modules
Voltage can be specified as constant, sinusoidal, UserSub(ubound), or parametric. See User Subroutines
and Control Panel-Boundary Conditions-Parametric for more information on using the UserSub and
Parametric options.
The Sinusoid boundary condition applies time varying voltage to the surface in the following form:
V 1 + V 0 sin(360f t + φ)
When you select this option, enter values for the voltages V 1 and V 0 in volts, the frequency f in Hz, and φ
the phase shift in degrees.
Electric Module
Implementation-Model Setup and Solution-Boundary Conditions-Surface Charge
The weighted (by the relative permittivity) electric field normal boundary condition specifies an electric
field normal to a cell faces weighted by the absolute permittivity. When you select this option, the Surface
Charge Density (ρs) field appears. Enter the charge density in V/m.
(10-55)
Electric Module
Implementation-Model Setup and Solution-Boundary Conditions-Fixed Current
You can specify current charge as UserSub(ubound) parametric. See User Subroutines and Control
Panel-Boundary Conditions-Parametric for more information on using the UserSub and Parametric
options.
This option is available only for DC Conduction Model option of the Electric Module. The current density
must be specified in units of Amp/m2. This is a normal to the boundary face.
(10-56)
Electric Module
Implementation-Model Setup and Solution-Boundary Conditions-Dielectric
When the Dielectric option is selected, faces where the voltage value is desired but the potential or its
gradient is not known are included in the calculation. CFD-ACE+ automatically computes a surface
charge for these boundaries. This option is valid for only CCP Plasma or the BEM solver (only at
interfaces).
A dielectric boundary condition is much like a conjugate interface in heat transfer. It is simply a flux
balance at the interface. The equation would be:
63
CFD-ACE_V2010.0_Modules_Manual_Part2
(1)
where
Electric Module
Implementation-Model Setup and Solution-Boundary Conditions-Piezoelectric
The only boundary conditions supported for piezoelectric analysis are zero gradient and specified (fixed)
potential. The fixed potential value may be specified as constant, sinusoidal, parametric, or through a
user subroutine.
Electric Module
Implementation-Model Setup and Solution-Coupling to Other Modules
The Electric Module is coupled to the Heat Transfer, Magnetic, Stress, and Free Surface (VOF) Modules.
For the Stress and VOF Modules the coupling is automatic. To couple the influence of the magnetic field
on the electric field, you must activate the src current (conduction) option under the MO/Magnet tab.
For the Stress Module, coupling is automatic when using Stress or an Implicit Pressure boundary
condition with the electrostatic option of the Electric Module. If an electrostatic force exists at the
boundary, it is automatically fed to the Stress Module.
Joule heating is automatically included when the electric module is coupled with the Heat Transfer
Module and the DC conduction, ac conduction, or transient option is specified. See Model Options.
Electric Module
Implementation-Model Setup and Solution-Initial Conditions
Click the Initial Conditions [IC] tab to see the Initial Conditions Panel. See Control Panel-Initial Conditions
for details.
The Initial Conditions can be specified as constant values or read from a previously run solution file. If
constant values are specified, you must provide initial values required by the Electric Module. The values
can be found under the Electric (Electr) tab and the only the initial electric potential needs to be specified.
Electric Module
Implementation-Model Setup and Solution-Solver Control-Introduction
Click the Solver Control [SC] tab to see the Solver Control Panel. See Control Panel-Solver Control for
details. The Solver Control panel provides access to the settings that control the numerical aspects of the
CFD-ACE-Solver.
64
Modules
Electric Module
Implementation-Model Setup and Solution-Solver Control-Iter
The Solver Control Iterations tab enables you to set Maximum Iterations, Convergence Criteria, and
Minimum Residual. For details on these fields, see User Manual-Control Panel-Solver Controls-Iterations.
Electric Module
Implementation-Model Setup and Solution-Solver Control-Spatial
The Solver Control Spatial tab enables you to view the type of spatial differencing being used by CFD-
ACE+.
65
CFD-ACE_V2010.0_Modules_Manual_Part2
Electric Module
Implementation-Model Setup and Solution-Solver Control-Solvers
The Solver Control Solvers tab enables you to select the linear equation solver to be used for each set of
equations.
Electric Module
Implementation-Model Setup and Solution-Solver Control-Relax
The Solver Control Relax tab enables you to select the amount of under-relaxation to be applied for each
of the dependent (solved) and auxiliary variables used for the electric equations.
66
Modules
Electric Module
Implementation-Model Setup and Solution-Solver Control-Limits
The Solver Control Limits tab enables you to set the minimum and maximum allowed variable values.
CFD-ACE+ will ensure that the value of any given variable will always remain within these limits by
clamping the value. Activating the Electric Module enables you to set limits for the electric potential
variable. See Control Panel-Variable Limits for details on how limits are applied.
Electric Module
Implementation-Model Setup and Solution-Solver Control-Adv
The Solver Control Advanced tab appears when you select the Boundary Element Method in the Model
Options Electric tab.
67
CFD-ACE_V2010.0_Modules_Manual_Part2
Advanced tab in the Solver Control Panel in Electric Module Settings (for BEM Solver) Mode
Electric Module
Implementation-Model Setup and Solution-Output-Introduction
There are no settings under the Output tab that affect the Electric Module. See Control Panel-Output
Options for details about the available output settings. The Output section includes:
• Summary Output
• Graphical Output
Electric Module
Implementation-Model Setup and Solution-Output-Graphical Output
Under the Graphics tab, you can select the variables to output to the graphics file (modelname.DTF).
These variables will then be available for viewng and analysis in CFD-VIEW. Activating the Electric
Module provides output of the variables listed below:
Variable Units
Ω m
-1 -1
Electrical Conductivity
Relative Permeability -
Relative Permittivity -
68
Modules
2
Surface Charge C/m
3
Electric Virtual Force Vector N/m
2
Electrostatic Pressure Force N/m
2
Conduction Current Density A/m
Electric Module
Implementation-Model Setup and Solution-Output-Summary Output
Under the Summaries tab, select the summary information to be written to the ASCII text output file
(modelname.out). Activating the Electric Module gives the option of Electric Flux Summary to be written
and for conduction problems, the current summary will be written to the output file. See Control Panel-
Summary Output for details on the general summary output options, including boundary integral output,
diagnostics and monitor point output.
Electric Module
Implementation-Post Processing
CFD-VIEW can post-process the solutions. When the Electric Module is invoked, the electric potential
field is usually of interest. The electric potential field can be visualized with surface contours and analyzed
through the use of point and line probes. complete list of post processing variables available as a result of
using the Electric Module is shown in the table.
69
CFD-ACE_V2010.0_Modules_Manual_Part2
Ω m
-1 -1
sig_i Imaginary part of conductivity
Ω m
-1 -1
sig_r Real part of conductivity
2 2
εo electric permittivity of free space (8.854187 x10-12) C /Nm
3
ρ space charge density C/m
resistivity Ωm
2
ρs surface charge density C/m
70
Modules
σ Ω m
-1 -1
electric conductivity
Q total charge C
f frequency 1/s
j -
imaginary number often viewed as
Magnetic Module
where:
2
A = the vector laplacian
µr = a multiplication factor
Equation 11-1 assumes an isotropic permeability µ and the Coulomb gauge ∇•A=0. The current sources
J tot can be user specified currents (J s ), displacement currents , currents calculated from the
solution of the electric conduction problem (σE), or eddy currents ( ). Equation 11-1 is solved on
a 2D (cartesian, cylindrical) or 3D unstructured mesh using the FVM. In 2D the source currents are
71
CFD-ACE_V2010.0_Modules_Manual_Part2
Detailed instructions on model preparation are presented in the Implementation section. The examples
section provides solutions for some sample problems. The Magnetic Module includes:
• Magnetic-Applications
• Magnetic-Features
• Magnetic-Theory
• Magnetic-Limitations
• Magnetic-Implementation
• Magnetic-Symbol Definitions
Magnetic Module Applications
Some of the common uses of the Magnetic Module are:
• Magnetic fields due to distributions of current carrying wires and time varying electric fields
• Electric fields due to time varying magnetic fields
• Eddy currents
• Lorentz force on current carrying wires
• Torque on magnetic moments (i.e., magnetizable materials and permanent magnets).
• Lorentz force for flows of electrically conducting fluids (MHD problems)
• Induction heating
• Plasma source such as inductively coupled plasmas
Magnetic Module Features
You can use the Magnetic Module as a stand-alone or with other modules (heat transfer, stress, flow,
electric, plasma) for multi-disciplinary design. You can take advantage of several unique features of the
Magnetic Module including:
• Common environment for magnetic and multi-physics design
• Direct model setup in CFD-MicroMESH or CFD-GEOM
• The Magnetic Module supports both structured and unstructured meshes (tetrahedral,
hexahedral, prisms, and general polyhedrals)
• The Magnetic Module is coupled with the heat transfer, stress, flow, and electric modules
• Lorentz Forces when coupled with flow and electric modules (MHD Problems)
• Stress Module - Coupling
• The Magnetic Module can be used for dynamic problems with moving deforming materials
Magnetic-Theory
Magnetic Module
Theory-Introduction
The Magnetic Module solves Maxwell’s equations using the magnetic vector potential A. The derivation
begins with Faraday’s Law and Ampère’s Law (with displacement current) respectively in differential form
and assuming mks units:
(11-2)
(11-3)
72
Modules
Using the definition of the magnetic vector potential, B = ∇xA and the constitutive relation, B = µH,
equation 11-2 and equation 11-3 become:
(11-4)
(11-5)
Using equation 11-4 and the vector identity ∇ x ∇ ϕ = 0 (where ϕ is any scalar) the electric field is
defined as:
(11-6)
(11-7)
Assuming the Coulomb gauge ( ∇•A = 0), writing the permeability as µ = µ r µ o , and using the constitutive
relation D = εE, equation 11-7 becomes:
(11-8)
Finally, substituting the expression for the electric field in equation 11-6 into equation 11-8 and writing the
permittivity as ε = ε r ε o , yields:
(11-9)
(11-10)
73
CFD-ACE_V2010.0_Modules_Manual_Part2
Various forms of equation 11-10 are solved in the Magnetic Module for the vector magnetic potential A.
The form of equation 11-10 that the Magnetic Module options solve are covered in the next section.
The current sources, J tot , in equation 11-1 represents the right hand side of equation 11-10 for the
frequency domain or the right hand side of equation 11-9 for the static or time domain and can be user
specified currents (J s ), displacement currents , currents calculated from the solution of the electric
conduction problem (σE), or eddy currents . Equation 11-1 is solved on a 2D (Cartesian,
cylindrical) or 3D unstructured mesh using the FVM. In 2D the source currents are scalars with assumed
Magnetic Module
Theory-AC Single Frequency
The AC Single Frequency option solves the complete form of the ac sinusoidal steady state form of
equation 11-10.
The possible source currents include user-specified currents (J s ), displacement currents
(−ε r ε o ω ¹
2
currents calculated from the solution of the electric conduction
problem or eddy currents .
Magnetic Module
Theory-DC and AC Single Frequency
The DC and AC Single Frequency option solves a DC steady state equation in addition to the AC single
frequency (see equation 11-10). The form of the DC equation is:
(11-11)
The possible DC source currents are conduction currents calculated from the conduction option in the
Electric Module (σ∇φ) or user-defined currents (J s ). All DC current sources are assumed steady state.
Magnetic Module
Theory-Static
The Static option of the Magnetic Module solves only the DC steady state equation 11-11.
This option is activated when the steady state condition is selected in the Shared tab.
Magnetic Module
Theory-Transient
The transient option solves the complete transient form of equation 11-9. This option is activated when
the transient condition option is selected from the Shared tab. The convective term, , is active
only if the Lorentz Force option is selected.
74
Modules
The possible source currents include user specified currents, J s , displacement currents, and
, currents calculated from the solution of the electric conduction problem (σ∇φ), or eddy currents
.
Magnetic Module
Theory-Calculating Auxiliary Quantities
The following sections describe how to calculate the auxiliary quantities that depend on the electric
potential:
• Magnetic Force On Current Carrying Wire
• Force on a High Permeability Material (Magnetic/Structure)
• Lorentz Force
• Induction Heating
Magnetic Module
Theory-Magnetic Force on Current Carrying Wire (Magnetic/Structure)
For regions where current sources exist, the incremental force due to a magnetic field on an incremental
volume element with a current density is calculated using:
dF = Jdv x B (11-12)
where:
dF = incremental force
B = magnetic field
dν = an incremental volume element
J = current density
Magnetic Module
Theory-Force on a High Permeability Material (Magnetic/Structure)
The coupling between the Magnetic Module and the structural mechanics model is through volume
forces. The Magnetic Module solves forms of equation 11-10 and uses the magnetic vector potential
fields to calculate the volume force. The stress model uses the volume forces as body force conditions.
For regions were magnetized material exist either the virtual force per unit volume or the torque on a
magnetic moment can be calculated.
Virtual Force
For regions were magnetized material exists, the virtual force per unit volume on the volume element is
calculated by taking the gradient of the magnetic energy density of the volume element, F virt = U mag . If
75
CFD-ACE_V2010.0_Modules_Manual_Part2
electric fields also exist in the region, the energy density of the electric field is included in the gradient
calculation. The instantaneous energy density of a magnetic field is:
Umag = ½B•H.
When the fields are in phasor form:
o
where the energy represents the average over one cycle, and the symbol " ” represents the complex
conjugate.
A magnetic moment force calculation has been added in addition to the virtual force calculation. A
material with a magnetic moment M placed in a magnetic field H will experience a force F given by:
(11-13)
The magnetic moment M is currently assumed proportional to the applied magnetic field (M = X m H). So
using the definition of a magnetic moment in terms of B and H, M = B/µ o − H and the definition of relative
permeability, µ r = (1 + X m ) equation 11-13 becomes
(11-14)
which gives the force on a magnetizable material in terms of the applied magnetic field
Magnetic Module
Theory-Lorentz Force (Coupling with Flow and Electric Modules)
The Lorentz Force option, is available when the Flow and Electric modules are used with the Magnetic
module. The Lorentz force, J x B is included as a source term in the momentum equation for the flow of
an electrically conductive fluid in a magnetic field. The current density J is given by:
J = σ(E+u x B) (11-15)
(11-16)
The last two terms damp the flow when the magnetic field B and velocity u are not aligned.
The current in the fluid is calculated from the conservation of charge. The divergence of the current
density is zero, giving:
(11-17)
76
Modules
This equation is solved for the electric potential, φ. The Magnetic Module was modified to include the term
·σ(u x B). The boundary conditions for the electrostatic module have been modified to allow for insulated
boundaries where the normal component of the current density vector is zero.
Magnetic Module
Theory-Induction Heating (Coupling with Heat Module)
A time varying magnetic field generates a time varying electric field which generates eddy current in
conductive materials. The flow of current through the conductor generates heat and is called inductive
heating. The inductive heating can be thought of as joule heating (J·E) with the conductive currents
(J = σE) generated by the time varying field . So assuming the electric field is due
only to the time varying magnetic field and the current is only a conduction current
the time averaged expression for joule heating becomes
(11-18)
where σ r is the real component of conductivity and ω is the radian frequency (2πf).
Magnetic Module Limitations
There are a few limitations in the current implementation of the Magnetic Module. These limitations may
be removed in later versions.
• No inductance calculation
• Cannot handle spatial variations in permeability in 3D (can handle 2D and axisymmetric cases).
• The thin walls boundary condition is not supported with the magnetic module.
Magnetic-Implementation
Magnetic Module
Implementation-Introduction
The Implementation section describes how to setup a model for simulation using the Magnetic Module of
the CFD-ACE-Solver. The Magnetic Implementation section includes:
• Grid Generation - Describes the types of grids that are allowed and general gridding guidelines
• Model Setup and Solution - Describes all of the Magnetic Module related inputs to the CFD-ACE-
Solver
• Post Processing - Provides tips on what to look for in the solution output
Magnetic Module
Implementation-Grid Generation
The Magnetic Module can be applied to any geometric system (3D, 2D planar, or 2D axisymmetric).
Furthermore all grid cell types are supported (quad, tri, hex, tet, prism, poly).
The general grid generation concerns apply, i.e., ensure that the grid density is sufficient to resolve
solution gradients, minimize skewness in the grid system, and locate computational boundaries in areas
where boundary values are well known.
77
CFD-ACE_V2010.0_Modules_Manual_Part2
Magnetic Module
Implementation-Model Setup and Solution-Introduction
CFD-ACE+ provides the inputs required for the Magnetic Module. Model setup and solution requires data
for the following panels:
• Problem Type
• Model Options
• Volume Conditions
• Boundary Conditions
• Initial Conditions
• Solver Control
• Output
Magnetic Module
Implementation-Model Setup and Solution-Problem Type
Click the Problem Type [PT] tab to see the Problem Type Panel. See Control Panel-Problem Type for
details.
Select the Magnetic to activate the Magnetic Module. The Magnetic Module label for the tabs is Magnet.
The Magnetic Module can be run alone or in conjunction with the other problem type options.
Magnetic Module
Implementation-Model Setup and Solution-Model Options
The Model Options [MO] panel provides options for specifying the parameters for each module activated
in the Problem Type panel, the model’s time dependency and other shared parameters. See Control
Panel-Model Options for details. Click on the MO tab, then select the Magnet tab to display the Magnetic
Module options. There are four options available for calculating Magnetic Potential.
• AC Single Frequency. Solves equation 11-10 for the vector magnetic potential A
• Transient. Solves equation 11-9 for the vector magnetic potential A. (To activate, you must
specify a transient problem.)
• DC and AC Single Frequency. Solves considering AC sinusoid and DC steady state conditions
• Static. Solves equation 11-11 for the vector magnetic potential A. (To activate, you must specify a
steady state problem in the Shared tab.)
The control parameters available for all options of the magnetic module are shown in the table below:
Parameter Use
78
Modules
Src Current
(Conduction) A toggle that is present only if DC conduction, AC conduction, or transient
option of the electric model is being solved in conjunction with the Magnetic
(Optional Source Module. Allows current densities calculated in the electric model to be used
Currents, activated as sources in the Magnetic Module (see the conduction terms in equation
when coupled with the 11-9 and equation 11-10).
electric module)
Magnetic Module
Implementation-Model Setup and Solution-Volume Conditions-Introduction
Click the Volume Conditions [VC] tab to see the Volume Condition Panel. See Control Panel-Volume
Conditions for details. Before any volume condition information can be assigned, one or more volume
condition entities must be made active by picking valid entities from either the Viewer Window or the VC
Explorer. The Magnetic Volume Conditions section includes:
• Volume Conditions-Properties
• Volume Conditions-Electric Conductivity
• Volume Conditions-Magnetic Field Sources
Magnetic Module
Implementation-Model Setup and Solution-Volume Conditions-Properties
Click on the VC tab, then select the Property option from the Settings Mode option menu. Select the
Volume Condition from the graphics window or the BC/VC Explorer (below the Graphics window). There
are three options available here: Solid, Fluid, or Block. Select either Fluid or Solid. This will determine a
particular volume whether solid or fluid. Specify the following properties for the Magnetic Module. The
following properties can be specified for both gas and solid volumes.
Property Usage
79
CFD-ACE_V2010.0_Modules_Manual_Part2
Magnetic Module
Implementation-Model Setup and Solution-Volume Conditions-Electric Conductivity
Method Comments
80
Modules
Magnetic Module
Implementation-Model Setup and Solution-Volume Conditions-Magnetic Field Sources
Current sources for the Magnetic Module are specified as volume conditions. Click the VC tab on the
control panel. Select Magnet from the Setting Mode menu. Click on the magnetic field source checkbox to
activate Sources. There are four Evaluation methods for the electromagnetic source: No Source, AC
Source, DC Source, and Permanent Magnetic Source. The direction of the current sources are always in
the z or φ direction for the 2d planar and 2D axisymmetric respectively. 3D current source cannot be
specified. Axisymmetric current sources with arbitrary axis of symmetry may be specified as 3D.
No Source
No source is specified for the volume set. This is the default.
DC Source
The option of introducing a DC source into a Coil volume is available when the magnetic field is being
solved as a Static field. It is also available when the AC+DC option has been selected in the MD/Magnet
panel. For a DC source you must specify the coil current magnitude and, for 3D, the axis of symmetry an
AC source. The procedure for setting these values for DC source is the same as for AC.
AC Source
Current - If you select the AC Single Frequency Magnetic Field Option in the Magnet tab, the Current
amplitude (in Amperes) is input for two dimensional geometries. For three dimensional geometries, the
Current Density must be specified in units of Amps per square meter. Phase Angle Imaginary Part of
Current or Current Density (3D) must be specified. For AC Single Frequency, the Current of each Electro
Magnetic Source will be automatically and proportionally adjusted to achieve the specified power
absorption in the plasma, if Total Power has been selected under the MD/Plasma Power Specification
option.
81
CFD-ACE_V2010.0_Modules_Manual_Part2
Frequency - If you have selected the AC Single Frequency Magnetic Field Option in the MD/Magnet
panel, the frequency of all Electro Magnetic Sources is fixed to the value specified under the MD/Magnet
Field Frequency window and cannot be modified under VC’s. If you have selected the Transient Magnetic
Field Option in the MD/Magnet panel, you can set a different frequency for each Electro Magnetic Source.
Current Direction - For 2D, axisymmetric geometries, the current may be axisymmetric or along the wire
direction (X-direction). To activate the current in the X-direction for axisymmetric and 2D plane
geometries, the following DTF update should be placed in the Special DTF Update window in CFD-ACE-
GUI.
DTF update for activating current in the X-direction for axisymmetric cases.
For 2D, plane geometries, it is assumed that the direction of the electric current is in the positive z
direction.
For three-dimensional geometries, six input fields (x 1 ,y 1 , z 1 , and x 2 ,y 2 , z 2 ) under Axial Direction option
will appear for specifying two points that define the axis of symmetry for the current. Inherent in this
method is the assumption that the coil and its associated current are axisymmetric. For non-
axisymmetrical coils, use the Electric Module to solve for the current in the coil as a function of voltage
drop between the coil terminals.
Permanent Magnetic Source
This option is available when one of the following options are selected from the Module panel: transient,
static, and DC+AC single frequency. Two evaluation methods are available with this option: No Source
and Constant. The Constant option requires that you specify the X-Y components of the magnetization
vector.
Magnetic Module
Implementation-Model Setup and Solution-Boundary Conditions
Click the BC tab, then select the Magnet tab folder to specify Magnetic Module related Boundary
Conditions. To assign boundary conditions and activate additional panel options, select an entity from the
viewer window or the BC Explorer. There are four boundary conditions available in the Magnetic Module:
Fixed Potential (cart), Extrapolation, Non Reflecting, and Perfect Conductor. The boundary conditions can
be specified for all components of A.
The Fixed Potential boundary condition specifies a fixed vector magnetic potential at a cell faces. When
you select this option, enter the desired fixed potential (in N/A) for the specified cell faces. This may
include all three directional components in 3D and real and imaginary components if in the frequency
domain. For two dimensional grids only one spatial component can be specified. You can use the
82
Modules
parametric option to enter the mathematical expressions instead of constant real values. For more
information, see Parametric Input.
For axisymmetric cases with the current along the wire (X-direction), the initial magetic vector potential
would be Ax(real) and Ax(imaginary).
Extrapolation
The Extrapolation boundary condition specifies that each component (spatial, real/imaginary) will have no
gradient in the direction of the face normal (i.e., ≠A [x,y,z][r,i] /≠n = 0).
Non-Reflecting
This is the outer-boundary condition for a n infinite domain. It is also called absorbing boundary condition
(ABC). When solving transient wave propagation problems, the wave will propagate out the domain with
out reflection.
Perfect Conduction
When the computational domain is in contact with a perfect conductor, the surface tangential electric field
is zero on the boundary. At this boundary, the magnetic vector potential has zero tangential component.
Magnetic Module
Implementation-Model Setup and Solution-Initial Conditions
Click the IC tab on tool bar. Select Initial Source to be Constant Values. Select Magnet tab folder to
specify Magnetic Module related Initial Conditions.
Constant
Magnetic Module
Implementation-Model Setup and Solution-Solver Control-Introduction
Click the Solver Control [SC] tab , then select following tab folders to specify Magnetic Module related
Solution Control parameters:
• Solver Control-Iterations
• Solver Control-Spatial Differencing
• Solver Control-Solver Selection
• Solver Control-Relax (Relaxation)
• Solver Control-Limits
• Solver Control-Output
83
CFD-ACE_V2010.0_Modules_Manual_Part2
Magnetic Module
Implementation-Model Setup and Solution-Solver Control-Iterations
The Iterations option enables you set the total number of solution iterations. The value should be set so
that enough solution iterations are done to reduce the residuals by five orders of magnitude.
Magnetic Module
Implementation-Model Setup and Solution-Solver Control-Spatial Differencing Scheme
Magnetic Module
Implementation-Model Setup and Solution-Solver Control-Solver Selection
Magnetic Module provides the Mag. Vec. Pot. variable to the Solvers panel. The Solvers option enables
you to select the linear equation solver used to solve the algebraic equations for each dependent
variable. The default selection is the CGS+Pre (conjugate-gradient-squared plus preconditioning) solver.
A standard CGS solver and an Algebraic Multigrid Solver (AMG) are also available. For all three solvers,
you can set the number of sweeps, or iterations, and the convergence criteria. In most cases, you can
use the default selections of the CGS+Pre solver with 500 sweeps for the Pressure Correction equation
and 50 for all other equations.
Magnetic Module
Implementation-Model Setup and Solution-Solver Control-Relax (Relaxation)
The Relaxation option sets the relaxation factors for each dependent variable and for the auxiliary
variables. When the Magnetic Module is selected the auxiliary variable is the Mag. Vec. Pot.
Relaxation is a constraint on the change of a dependent or auxiliary variable from one solution iteration to
the next. It is required to maintain the stability of the coupled, non-linear system of equations.
To increase the relaxation, or constraint, on a set of dependent variables, move the slider for that set of
variables to the right, and to decrease the relaxation, move the slider to the left. You can increase or
decrease the amount of constraint by a factor of 10 by clicking the downward facing arrow at the left of
the slider bar or the upward facing arrow at the right.
Magnetic Module
Implementation-Model Setup and Solution-Solver Control-Limits
The Limits option sets minimum and maximum values for certain dependent and auxiliary variables.
Magnetic Module provides the dependent variable Mag. Vec. Pot. to the Limits option.
Magnetic Module
Implementation-Model Setup and Solution-Output Introduction
The Output tab sets the output location for transient simulations and the output file frequency for steady
simulations.
For steady state simulations, the Output tab provides the Output Results check box and a pull down
menu with two options:
• End of Simulation - With this option, the output interval is set, by default, to the maximum
number of iterations specified in Iterations tab.
84
Modules
• Specified Interval - With this option, the Iteration Frequency field appears where you can enter
the number of iterations to use as the specified interval.
For transient simulations, the Output tab provides the Output Location check box and a pull down menu
with three options:
• Unique Filename - Writes a unique DTF file for every time step frequency specified.
• Unique Simulation - Stores the results in the same DTF file with unique simulations in it.
• Replace Simulation - Contains the latest simulation in the DTF.
You can specify time step frequency for writing output results. The default value is one.
Magnetic Module
Implementation-Model Setup and Solution-Output-Graphics Output
The following Magnetic Module variables are available for graphics output in DTF.
• Electric Conductivity (σ)
• Relative Permittivity (ε r )
• Relative Permeability (µ r )
• Magnetic Vector Potential (A)
• Magnetic Field (B)
• Power dissipation
• M torque force
Magnetic Module
Implementation-Post Processing
CFD-VIEW can post-process the solutions. A complete list of post processing variables available as a
result of using the Magnetic Module is shown in the table.
85
CFD-ACE_V2010.0_Modules_Manual_Part2
86
Modules
µr relative permeability -
2
µo permeability of free space (4π x10-7) N/A
or resistivity Ω-m
2
ρs surface charge density C/m
σ Ω m
-1 -1
electric conductivity
Q total charge C
f frequency 1/s
j -
imaginary number often viewed as
87
CFD-ACE_V2010.0_Modules_Manual_Part2
Spray Module
88
Modules
• User access subroutines to enable the user to specify properties, heat/mass transfer correlations,
initial conditions, and time step size, as well as evaluate parameters along trajectories for output.
Database
The Spray Module includes a database file (LIQPROP.DAT) that contains curve-fit coefficients for
evaluating properties of commonly used liquid species (fuels) for spray combustion problems.
LIQPROP.DAT uses the SI unit system. This database is independent of the property database for
gaseous species. From this database, the properties of liquids such as density, specific heat at constant
pressure, saturation pressure, and heat of vaporization can be calculated.
The species that are available currently in this database are: water (H2O), hydrogen (H2), oxygen (O2),
nitrogen (N2), methanol (CH4O), methane (CH4), ethanol (C2H6O), propane (C3H8), heptane (C7H16),
undecane (C11H22), and dodecane (C12H26). In order to get the properties correctly from the database,
the names (chemical formula) given in the brackets should be used when defining the spray liquid
species.
Spray-Theory
Spray Module
Theory-Introduction
In the Spray Module, discrete droplet/particle parcels are tracked through the computational domain by
solving the Lagrangian equations. For steady state calculations, a parcel is tracked through its lifetime
(until it leaves the domain or evaporates completely) and this procedure is repeated for all the parcels.
The local time step used for integrating the spray equations is estimated based on cell residence time and
maximum allowable (user defined) droplet temperature, velocity, and diameter change. For transient
sprays, the spray equations are integrated for all the parcels over a period of the global time step. Sub-
time steps are utilized if the internally calculated spray time step is smaller than the global time step..
The equation of motion for the droplet can be written as:
(13-1)
where m d is the mass of the droplet and υ = ui + vj + wk its velocity vector; u, v `and w are the Cartesian
velocity components; C D is the drag coefficient; ρ, U and p are the density, velocity and pressure of the
surrounding gas, respectively; A d is the droplet frontal area and V d is the droplet volume. For a spherical
droplet, A d = πd /4 and V d = πd / 6 where d is the droplet diameter. The gravity vector is represented by
2 3
g. Equation 13-1 accounts for the acceleration/deceleration of the droplet due to combined effects of drag
with gas and body forces such as gravity. Additional mass sources, Sm, can be added to this equation by
selecting the option for user-defined mass source in spray equation.
The drag coefficient, C D , for the droplet is a function of the local Reynolds number, which is evaluated as:
(13-2)
89
CFD-ACE_V2010.0_Modules_Manual_Part2
Spray Module
Theory-Drag Coefficient
The following correlations are used to calculate the drag coefficient:
Incompressible Flow
Compressible Flow
For compressible flows, two correlations, one for subsonic and one for supersonic flows, are used to
calculate the drag coefficient. Depending on the local relative Mach number (M), the appropriate
coorelation is used.
(13-3)
Subsonic Regime
(13-4)
(13-5)
(13-6)
(13-7)
90
Modules
(13-8)
Supersonic Regime
(13-9)
(13-10)
(13-11)
(13-12)
where is the droplet liquid density, and . The droplet locations are determined from the
velocities by numerically integrating the velocity-defining equation:
(13-13)
Spray Module
Theory-Droplet Heat and Mass Transfer
As the droplet moves through the surrounding medium, it exchanges heat and evaporates. The rate of
evaporation is modeled as (Abramzon and Sirignano, 1989):
(13-14)
where Sh is the Sherwood number, Γ m is the mass diffusion coefficient for the gas, and B m is known as
the Spalding number. Let and be the vapor mass fractions at the droplet surface and in the free
stream respectively. In this model is calculated by equating the surface partial pressure with the
liquid saturation pressure.
(13-15)
91
CFD-ACE_V2010.0_Modules_Manual_Part2
M d is the droplet molecular weight while M is the mixture molecular weight. The Spalding and Sherwood
numbers are obtained from the following expressions:
(13-16)
(13-17)
where Sc is the Schmidt number. This expression for Sh number (and Nu below) is known as the Ranz-
Marshall correlation (Ranz and Marshall, 1952). The mass transfer rate is affected by the evaporation
process since the presence of surface blowing increases the thickness of the laminar boundary layer. The
Sh number (and the Nu below) may be corrected for this affect according to Abramzon and Sirignano,
1989. It is appropriate to use this correction whenever evaporation is present. Since m d = ρ d πd / 6, the
3
mass conservation equation for the droplet is rewritten in terms of its diameter:
(13-18)
(13-19)
where is the sensible heat transferred to the droplet and T d is the droplet temperature. L is the latent
heat of vaporization for the droplet fluid and C d is the specific heat of the droplet. The sensible heat
transfer is calculated as:
(13-20)
where K and T g are the thermal conductivity and temperature of the gas, respectively. The Nusselt
number, Nu, is obtained from the following correlation:
(13-21)
where Pr is the Prandtl number. Substituting the above expressions into equation 13-19, the energy
equation is rewritten as:
(13-22)
where:
92
Modules
(13-23)
(13-24)
Spray Module
Theory-Droplet Source/Sink
As a droplet moves through a computational cell (in the Eulerian gas phase) and evaporates, it
contributes to the fuid-phase mass, momentum, and energy. Moreover, as a given cell may contain a
number of droplets, the new mass, momentum, and energy contributions should be computed by
summation. Thus, the solutions to the droplet equations provide source/sink terms to the gas phase
conservation equations.
At the end of each iteration, the source terms due to mass, momentum and energy transfer are calculated
and distributed to the neighboring cells. The source terms are interpolated to the neighboring cells so that
the source term is continuous as a function of the location of the droplet equations.
The net evaporation rate between the beginning and end of a time step is given as:
(13-25)
The subscripts named start and end refer to the beginning and end of an integration time step. The
number of drops in a given parcel, represented by , is calculated from the mass flow rate condition. The
source term for the gas-phase continuity equation is given by
(13-26)
where ∀ is the volume of the cell in which the mass addition takes place and is the accumulated mass
source term for the cell. This term appears on the RHS of equation 13-9 in addition to any other mass
sources.
If the evaporated droplet mass contributes to a species for which a mass fraction equation is being
solved, then it also appears as a source term in the mass fraction equation:
(13-27)
th th
where is the mass fraction of i species and is the evaporated mass of i species in the cell.
The preceding source term is added to the right hand side of equation 13-25 in addition to any reaction
rate source terms.
The momentum source term is evaluated as:
93
CFD-ACE_V2010.0_Modules_Manual_Part2
(13-28)
There are two contributions to the momentum source term: momentum transfer due to drag and
momentum of the evaporated mass. The source term for the gas-phase momentum conservation
equation is given by:
(13-29)
th
where the subscript j denotes the j coordinate direction with j = 1, 2, or 3. This term appears on the right
hand side of equation 13-2 in addition to any body force terms such as gravity.
The energy transfer between the droplet and gas is given as:
(13-30)
where h ο is the total enthalpy of the vapor at the vaporization temperature and h is the heat transfer
coefficient calculated from equation 13-20. The source term for the gas-phase enthalpy equation is given
by:
(13-31)
This term appears on the right hand side of either of the equation 13-4 or equation 13-7 in addition to any
other terms such as radiative source terms.
The source term calculated at the end of an integration time step is distributed to the neighboring
computational cells using a set of weighting factors. The weighting factor is inversely proportional to the
distance between the droplet location and the cell center. For 2D mesh, the source term is distributed to
four neighboring cells whereas for 3D mesh, it is distributed to the eight neighboring cells.
Spray Module
Theory-Drop Size Distributions
CFD-ACE+ provides the following options to specify droplet size distributions:
• User-specified discrete sizes
• Log-normal distribution
• Rosin-Rammler distribution
In the first option, you specify a discrete set of size groups as well as the fractions of injected mass
contained in the groups. The sum of the mass fractions for an injector must add up to unity. For log-
normal and Rosin-Rammler distributions, the mass fractions for each size group are calculated from the
distribution functions. You need to specify the volume mean diameter of the droplets, the standard
deviation, and the minimum and maximum size of the droplets. From the specified range of the droplet
size, a (user-specified) number of sizes are selected at equal intervals. The mass fraction for each size
group is calculated from the following equations.
Log-normal Distribution
94
Modules
where D is the volume mean diameter of the droplet cloud and is the standard deviation. And the
fraction of mass contained in size group D is calculated as follows:
Where the summation is over all the size groups of an injector and m is the total mass flow rate through
the injector.
Rosin-Rammler Distribution
Where q is a constant that lies between 1.5 and 4. The user needs to input the number of discrete size
bins in the Rosin-Rammler distribution. If the number of size bins is specified as N, the size range is given
by:
Then the mass fraction for each size group is calculated from the following equation:
95
CFD-ACE_V2010.0_Modules_Manual_Part2
Spray Module
Implementation-Grid Generation
You can apply the Spray Module to any geometric system (3D, 2D planar, or 2D axisymmetric).
Furthermore, all grid cell types are supported (quad, tri, hex, tet, prism, poly).
The general grid generation concerns apply, i.e., ensure that the grid density is sufficient to resolve
solution gradients, minimize skewness in the grid system, and locate computational boundaries in areas
where boundary values are well known. The only restriction is that the grid cell size should be much
larger than the droplet/particle size.
Spray Module
Implementation-Model Setup and Solution-Introduction
CFD-ACE+ provides the inputs required for the Spray Module. Model setup and solution requires data for
the following panels:
• Problem Type
• Model Options
• Spray Properties
• Spray Injectors
• Volume Conditions
• Boundary Conditions
• Initial Conditions
• Solver Control
Spray Module
Implementation-Model Setup and Solution-Problem Type
Click the Problem Type [PT] button to see the Problem Type Panel. See Control Panel-Problem Type for
details.
Select Spray to activate the Spray Module. If the problem involves evaporation, then heat transfer and
chemistry should also be selected.
Global
There are no settings under the Global tab that affect the Spray Module. However, activating gravity will
affect the spray particle trajectories.
Spray Module
Implementation-Model Setup and Solution-Model Options-Introduction
Click the Model Options [MO] tab to see the Model Options Panel. See Control Panel-Model Options for
details. All of the model options for the Spray Module are located under the Spray tab. The Spray Model
Options section includes:
• Model Options-Spray Models
• Model Options-Nusselt Number
96
Modules
Spray Module
Implementation-Model Setup and Solution-Model Options-Spray Models
The Spray Models section enables you to pick the type of exchange between the continuous phase
solution and the spray trajectory calculation. If the problem involves evaporation, then heat transfer and
chemistry should also be selected from the Problem Type tab. This will enable you to select three options
in the Spray tab:
• Drag - particle trajectory calculation is affected by drag between particle and fluid
• Heat Transfer - spray particles can exchange heat with the surrounding fluid
• Evaporation - spray particles can evaporate into gas
There are a few requirements for each of these models. The Flow Module must be activated in order to
use any of them. The Heat Transfer Module must be activated in order to use the heat transfer model
option. The Chemistry and Heat Transfer Module must be activated in order to use the evaporation model
option.
Spray Module
Implementation-Model Setup and Solution-Model Options-Nusselt Number
The Nusselt Number evaluation method is required for any spray simulation when heat transfer to/from
the drop is to be incorporated. There are three options available for calculating the Nusselt number: No
Convection Correction, With Convection Correction, and User Defined.
It is usually appropriate to utilize the convection correction option. See Theory-Droplet Heat and Mass
Transfer for details on the Nusselt number). The User Defined option instructs the solver to call the
spr_usr_nusl() user subroutine for your instructions on how to set the Nusselt number.
Spray Module
Implementation-Model Setup and Solution-Model Options-Sherwood Number
The Sherwood Number evaluation method is required for any evaporating spray simulation. There are
three options available for calculating the Sherwood number: No Convection Correction, With Convection
Correction, and User Defined. It is usually appropriate to utilize the convection correction option. See
Theory-Droplet Heat and Mass Transfer for details on the Sherwood number. The User Defined option
instructs the solver to call the spr_usr_sher() user subroutine for your instructions on how to set the
Sherwood number.
Spray Module
Implementation-Model Setup and Solution-Model Options-Spray Wall Boundary Treatment
This section enables you to select what happens to a spray particle when it strikes a wall boundary
condition:
• Bounce - Allows the particle to bounce (or reflect) off of the wall with a user supplied coefficient
of restitution (a coefficient of 0 implies that the particle will lose all momentum, a coefficient of 1.0
implies a perfect bounce).
97
CFD-ACE_V2010.0_Modules_Manual_Part2
• Stick - Can only be used when the Evaporating spray model is turned on. This option instructs
the solver to stop tracking the drop when it hits the wall and to evaporate the drop at the point of
impact.
• Vanish - Instructs the CFD-ACE-Solver to stop tracking the drop as soon as it hits the wall and to
neglect the mass of the drop that hits the wall.
The coefficient of restitution (CR) affects the Bounce option above. Normal and Parallel components of
the CR are specified separately. Setting both to the same vale has the same effect as a single standard
CR setting.
• Normal CR - This coefficient is applied only to the velocity component normal to the wall
• Parallel CR - This coefficient is applied only to the velocity component parallel (or tangent) to the
wall
Spray Module
Implementation-Model Setup and Solution-Model Options-Spray Boundary Options
Spray Module
Model Options-Turbulent Dispersion and Secondary Breakup
Turbulent Dispersion
Two turbulent dispersion models have been implemented in CFD-ACE+. The first is a stochastic model
that has been used successfully for many years, but which can require a fairly substantial computational
effort. The second model is a probability density function propagation (PDFP) model that can be very
efficient. The PDFP model was implemented through funding from the National Science Foundation
(NSF), but it is synergistic and relevant to the work for this NIST program, so it is briefly described below.
Stochastic
The stochastic turbulence dispersion model of Gosman and Ioannides (1981) was implemented for both
steady-state and transient. In this model, the computational parcel is tracked through a sequence of
turbulent eddies whose size and strength is estimated from local turbulent mean kinetic energy and
dissipation rate. The fluctuation gas velocity sampled from this model is used to displace the particle
position.
A test case in uniform turbulence was completed and compared to analytical results. The test case
consisted of uniform flow at 5 m/s with a turbulent viscosity of 0.001 kg/(m-s).
2 2 2 3
The kinetic energy was 1.0 m /s and the dissipation was 90.0 m /s . The particles and the fluid were
3
assumed to have a density of 1.0 kg/m and a size of 1 micron. The analytical solution assumes that the
particles are 'marked' particles that move with the fluid. A comparison with analytical results (Hinze,
1975, p. 428) was made at 1.0 m downstream of the release point. The calculation under predicted the
distribution to some extent, but the width of the distribution is similar.
PDFP
In many applications it is important to capture the effects of particle dispersion (in addition to the mean
particle information), which is a term used to describe the interaction of either solid or liquid particles with
the turbulent gas phase. This is often accounted for by perturbing the droplet velocities during the
98
Modules
trajectory based on the local turbulence quantities in a stochastic manner. This process results in an
expensive Monte Carlo type simulation to accurately describe the complete particle field. Several
thousand trajectories per particle size class and/or type usually need to be modeled. The innovative
Probability Density Function Propagation (PDFP) model implemented in CFD-ACE+ provides a complete
statistical representation of the condensed phase throughout the flow field with only one trajectory per
particle size class and/or type. This results in a one to two order magnitude decrease in the
computational time needed to characterize the particulates, making it significantly easier to model
problems containing large numbers of solid or liquid particles. The model can be used for either steady-
state or transient applications.
The Probability Density Function Propogation (PDFP) model is a Lagrangian particle model that tracks
the PDF of particle position. The PDF represents a group of particles having the same physical
properties and initial conditions. A description of the model can be found in Shirolkar (1996), and
Shirolkar and McQuay (1998). The most challenging aspecty of the model is to determine the spread in
the PDF along the trajectory. The assumption of a Gaussian PDF is usually made, although other
distribution functions could be used.
The following paragraphs provide a brief mathematical description of the PDFP model. If the particles
can be assumed to have a Gaussian distribution in space at any point in time, the three-dimensional PDF
can be written as:
To propogate the PDF shown in the above equation through space, expressions for the ensemble mean
particle location ( ) and the particle covariance matrix ( ) are needed as functions of time.
Expressions for these parameters (for a derivation see Shirolkar, 1996) are shown in the equations
below.
The ensemble mean particle velocity in the equation is obtained from the ensemble mean particle
momentum equation. This procedure is similar to that used to calculate individual particle trajectories
already present in CFD-ACE+. The solution of this equation is relatively straightforward, and with the
exception of the ensemble averaging, is the same as solving for an individual particle trajectory. The
significant challenge in the implementation of the PDFP model is to estimate the particle fluctuating
velocity correlations that appear in the equation. To estimate the velocity correlations, Shirolkar
defined a normalized Lagrangian particle correlation tensor, based on work by Taylor (1921), as shown in
the equation below:
Using this expression, the equation can be reduced to (no summation implied by the repeated
indices "i" and "j"):
99
CFD-ACE_V2010.0_Modules_Manual_Part2
If the particles are assumed to evolve in a random fashion, the normalized Lagrangian correlation tensor
shown in the above equation will depend on the particle residence time and the particle properties at the
present time. A common approximation for the correlation function is the Frenkiel function (1948).
In this equation is the particle Lagrangian time scale tensor, m is the negative loop parameter of the
functions, and is the residence time . To solve the above set of equations requires
estimation of the particle Lagrangian time scales and the particle variances. These parameters can be
determined from the information available from practical turbulence models, such as the standard
model, and by solving a set of coupled differential equations derived from the particle equation of motion.
Although not presented here, the details are given in Shirolkar et al. (1996).
The PDFP model was validated against two benchmark data sets for particle dispersion, Snyder and
Lumley (1971) and Weels and Stock (1983). Snyder and Lumley measured the transverse (relative to the
direction of particle motion) particle dispersion of 56.5 m hollow glass spheres, 87 m corn pollen
particles, 87 m solid glass spheres, and 46.5 m copper spheres. Wells and Stock measured the
longitudinal (in the direction of particle motion) particle dispersion for 6 m and 57 m glass beads at
several different drift velocities. Good agreement to both data sets was achieved.
Secondary Breakup
CFD-ACE+ provides 3 models for secondary drop breakup: the Rayleigh-Taylor model, the Taylor
Analogy Breakup (TAB) model, and the enhanced TAB model (ETAB).
Rayleigh-Taylor
At high Weber number, droplets can undergo secondary breakup. As the relative velocity between a drop
and the surrounding gas is increased, the drop distorts from its undisturbed spherical shape and
ultimately the drop becomes flattened, or disk shaped, normal to the flow of direction. The acceleration of
the flattened drop favors the development of Rayleigh-Taylor instabilities. Measured instability
wavelengths, λ, were found to agree with predictions from Rayleigh-Taylor wave theory (Taylor, 1950),
where is a model constant, = 1.1; is the droplet acceleration, which is obtained from:
The droplet is allowed to breakup only if the wavelength given by the Rayleigh-Taylor wave theory
equation is smaller than the droplet diameter, and the drop breakup time is reached. The breakup time
can be obtained by taking the reciprocal of the frequency of the fastest growing wave (Taylor, 1950;
Patterson and Reitz, 1998).
Taylor Analogy Breakup (TAB)
The TAB model is a classic method for calculating droplet breakup, which is applicable to many
engineering sprays. This method is based on analogy between an oscillating and distorting drop and a
spring-mass system. The external force acting on the mass, the restoring force of the spring, and the
damping force are analogous to the drag force, the liquid surface tension force, and the liquid viscosity
force, respectively. Thus, the distortion parameter, X=2y/r (y is the displacement of the equator of the
drop from its equilibrium position) is calculated by solving the spring-mass analogy equation:
100
Modules
where
The droplet is assumed to break up if the distortion grows to a critical ration of the droplet radius. The
TAB model first determines the amplitude for an un-damped oscillation ( ) for each droplet at a
certain time step:
Equation 1
and according to equation 1, breakup is possible only if the following conditions are satisfied:
or
This is the limiting case because damping will only reduce the chance of breakup. If a droplet fails the
above criterion, breakup doesn't occur, but the calculation of X and its derivative should be updated.
If the above criteria are met, then breakup is possible and the breakup time must be determined to see if
breakup occurs within the current time step. The breakup time is set to the time required for oscillations
to grow sufficiently large that the magnitude of the droplet distortion, X, is equal to unity. From equation
1, we have:
where is the breakup time of droplet. If ( is current time step), breakup occurs. And the
Sauter mean radius of the resulted droplets is calculated by:
Once the size of the child droplets is determined, the number of child droplets can be determined by
mass conservation.
101
CFD-ACE_V2010.0_Modules_Manual_Part2
ETAB
The criterion of the ETAB model is the same as that of the TAB model. The only difference is the
calculation of Sauter mean radius. In the TAB model, the calculation of SMD is based on the assumption
that the combined energy of child droplets is equal to the energy of the parent droplet, while in the ETAB
model, the SMD calculation is based on the following empirical equation:
where
Spray Module
Implementation-Model Setup and Solution-Model Options-User Defined Spray Equation Sources
Activating the User Defined Spray Equation option instructs CFD-ACE-Solver to make calls to certain
user subroutines where you may add sources to the spray equations.
• Momentum - forces a call to spr_usr_mom().
• Mass - forces a call to spr_usr_mass() and should only be activated if the Evaporation spray
model is turned on. The Energy option forces a call to spr_usr_engy() and should only be
activated if the Heat Transfer spray model is turned on.
Spray Module
Implementation-Model Setup and Solution-Model Options-User Access
The User Access options access user subroutines at the start of every spray calculation and at the end of
every spray calculation.
• At Spray Start - Forces a call to the spr_usr_start() user subroutine.
• At Spray End - Forces a call to the spr_usr_end() user subroutine.
Spray Module
Model Options-Advanced/Miscellaneous Forces
The Miscellaneous Forces section enables you to choose the following options:
Flash Evaporation
The flash evaporation is modeled by the following equations
102
Modules
Pressure Gradient
The pressure gradient force on a spray droplet is given by
103
CFD-ACE_V2010.0_Modules_Manual_Part2
T Saffman lift force (or shear-induced lift force) arises due to the surface pressure distribution on a droplet
in the presence of a velocity gradient in the flow field and plays an important role in determining droplet
deposition patterns in spray simulations. Generalized Saffman lift force can be expressed as (Crowe et
al, 1998):
where , and are fluid and particle velocity vectors respectively, is the Saffman lift
coefficient, is the droplet diameter, and and are the dynamic viscosity and density of the fluid
respectively.
Brownian Motion
The Brownian force experienced by droplets results from the impact of carrier fluid molecules on the
particles and is significant for sub-micron particles. Brownian force can be evaluated as (Li & Ahmadi,
1992)
where is the Cunningham correction factor, and are the dynamic viscosity and density of the
carrier fluid respectively, is the Boltzmann constant, T is the absolute temperature, is the time step,
and R is a Gaussian random number bounded by -1 and +1.
Thermophoretic Force
Thermophoretic force on the particles is given by (Crowe et al, 1998)
where = 1.17 is the temperature slip, = 2.18 and = 1.14 are the thermal and momentum
exchange coefficients respectively, Kn is the Knudsen number, , v, and are dynamic viscosity,
kinematic viscosity, and thermal conductivity of the carrier fluid respectively, is particle thermal
conductivity, is the particle diameter, is the local temperature gradient, and T is the local absolute
temperature
Spray Module
Implementation-Model Setup and Solution-Spray Properties
All of the properties for the spray particles are taken from the Database Manager when you specify the
spray Material Name under the 'Inj/Spray' panel. The database option is only available for the liquid
properties and the entries can be found in the LIQPROP.DAT file (this file uses the SI unit system). The
User Defined option calls the appropriate user subroutine to enable you to specify these properties. See
the Database Manager - Material Properties - Liquids section of the User Manual for more information on
the available options for these properties.
Spray Module
Introduction to Spray Injectors
The Spray INJ panel defines the spray injector settings including spray location and type of spray
injection. When you click this tab the panel opens with the message: Select or add an injector in
Geometry list in left bottom.
104
Modules
Select an injector and fill in the appropriate information to define which liquid/solid is being injected, the
mass flow rate, velocities, size distribution, and location of the injector.
Spray Module
Injector Types - Pressure Swirl Atomizer
This type of atomizer accelerates the liquid through nozzles known as swirl parts into a central swirl
chamber. The swirling liquid pushes against the walls of the swirl chamber and develops a hollow air
core. It then emerges from the orifice as a thinning sheet, which is unstable, breaking up into ligaments
and droplets. The pressure-swirl atomizer is widely used for liquid-fuel combustion in gas turbine, oil
furnaces, and direct-injection spark-ignited automobile engines. The whole process from injector flow to
fully developed spray can be divided into three steps: film formation, sheet breakup, and atomization.
Our focus will be on the last two steps. In the following sessions, a brief introduction to the linear
analysis is presented on the liquid sheet stability, and the breakup mechanism is introduced.
Linear Stability Analysis
The present stability analysis for a liquid sheet, which includes the effects of the surrounding gas a liquid
velocity, is formulated from the equations of motion and follows the approach of Sterling and Sleicher
(1975), Levich (1962), and Reitz and Bracco (1986) for the analysis of the stability of cylinder liquid jet.
Consider a two-dimensional, viscous, incompressible liquid sheet of thickness 2h moves with velocity U
through a quiescent, inviscid, incompressible gas medium. The liquid and gas have densities of and
, respectively, and the viscosity of the liquid is . A coordinate system is used that moves with the
sheet, and a spectrum of infinitesimal disturbances of the form
(1)
is imposed on the initially steady motion, which produces fluctuating velocities and pressures , , and
for the liquid and , , and the gas. In equation (1), is the initial wave amplitude, is
the wave number, and is the complex growth rate. The most unstable disturbance has the
largest value of , denoted by ,is assumed to be responsible for breakup. The resulting ligament size
is related to the maximum unstable wavelength , where is the wave number
corresponding to the maximum growth rate . Thus, it is desired to obtain a dispersion relation
from which the most unstable disturbance can be deduced.
To obtain the dispersion relation, the linearized continuity and momentum equations are solved, and the
following relation between and k is obtained for the case of sinuous wave modes (Senecal et al, 1999):
(2)
where
(2b)
105
CFD-ACE_V2010.0_Modules_Manual_Part2
speed of modern fuel injection is high enough that the film Weber number is often well above the critical
limit.
Li and Tankin (1991) derived a dispersion relation similar to equation (2) for a viscous sheet in the
sinuous mode. Their equation, which was obtained from a linear analysis with a stationary coordinate
system, can also be used to obtain equation (2).
To provide a simplified viscous dispersion relation, equation (2) for sinuous waves is further analyzed by
Senecal et al. [1999]. First of all, an order of magnitude analysis using typical values of and from
the inviscid solution shows that the terms of second order in viscosity can be neglected in comparison to
the other terms in equation (2). In addition, the density ration is on the order of in typical application
and hence it is assumed that . The resulting growth rate is given by:
(3)
(4)
Furthermore, if short waves are assumed for high speed sheets and , equation (3) reduces to
(5)
Breakup Mechanism
The physical mechanism of sheet disintegration proposed by Dombrowski and Johns (1963) is adopted in
order to predict the drop size produced from the primary breakup process. Disintegration occurs due to
the growth of waves on the surfaces caused by the aerodynamic forces acting on the sheet. Once the
waves reach critical amplitude, fragments of liquid are broken off which contract to form cylindrical
ligaments that are believed to move normal to the ligament axis. As a result, capillary forces cause the
unstable ligaments to break into drops.
Equations (4) and (5) indicate that while the growth rate of long waves depends on the sheet thickness,
the growth rate of short waves is independent of the thickness of the sheet. Thus, in order to predict the
onset of ligament formation for short wave growth ( > 27/16), a sheet breakup length is based on the
analogy of the prediction of the breakup length of cylindrical liquid jets (Reitz and Bracco, 1986). If the
surface disturbance has reached a value of at breakup, a breakup time can be evaluated via:
(6)
where is the maximum growth rate obtained from equation (5). Thus, the sheet will breakup at length
given by:
106
Modules
(7)
where the quantity ln( / ) is an empirical sheet constant from 3 to 12 based on the work of Dombrowski
and Hooper (1962), a default value of 12 is specified in CFD-ACE+. It is important to note that the
quantity of V is the absolute velocity of the liquid sheet while the velocity of U is the relative velocity
between the liquid and gas.
The diameter of ligament formed at the point of breakup can be obtained from a mass balance. If it is
assumed that the ligaments are formed from tears in the sheet once per wavelength, the resulting
diameter is given by:
(8)
where is the wave number corresponding to the maximum growth rate. The ligament diameter
depends on the sheet thickness, which is a function of breakup length. The film thickness is calculated
from the breakup length and the radial distance from the centerline to the mid-line of the sheet at the
atomizer exit, :
(9)
The above mechanism is not logical for short waves. For short waves, the determination of ligament
diameter is simpler. The value of is assumed to be linearly proportional to the wavelength that
breakup the sheet:
(10)
(11)
(12)
107
CFD-ACE_V2010.0_Modules_Manual_Part2
which is Weber's (1931) result for the wave number corresponding to the maximum growth rate for the
breakup of a cylindrical, viscous liquid column. Substitution of equation (12) into equation (11) gives:
(13)
(14)
(15)
(16)
The initial sheet thickness is related to mass flow rate and u by conservation of mass:
(17)
Spray Module
Injector Types - Plain Jet Atomizer
In a plain jet, a uniform, homogeneous phase is injected from an orifice into the ambient surrounding
phase. The injected phase is typically a liquid, which is usually accelerated by passage through a nozzle
before issuing from the injector. The cross-sectional shape of the jet is partly determined by that of the
orifice, and is here taken to be circular. The primary atomization process for a plain jet injector is therefore
driven by the instabilities of a moving column of liquid and its interactions with the surrounding gas. CFD-
ACE+ has two different primary atomization models for the plain jet: (1) the "blob" model; and (2) the
Boundary Layer Stripping (BLS) model. These two models are further described below.
BLOB Model
Jet stability analysis
The jet breakup theory considers a cylinder liquid jet issuing from a circular orifice into an incompressible
gas. The stability analysis of the liquid surface is based on the approach of Sterling and Sleicher (1975),
Levich (1962), and Reitz and Bracco (1986) for the analysis of the stability of cylinder liquid jet. The
axisymmetric surface displacement can be written as:
(1)
then the solutions to the linearized momentum equations can be expressed as:
108
Modules
(2)
(3)
where and are the velocity potential and stream function, respectively; C1 and C2 are integration
constants; and are modified Bessel functions of the first kind; . In addition, the
fluctuating gas pressure at the liquid-gas interface is obtained by solving the inviscid gas momentum
equation:
(4)
where a is the radial position of liquid surface; K0 and K1 are modified Bessel functions of the second
kind. The linearized boundary conditions of liquid jet are:
(5)
(6)
and
(7)
with above boundary conditions, the constants C1 and C2 can be eliminated and a dispersion equation is
obtained:
(8)
Reitz (1987) showed that equation (8) predicts a maximum growth rate exists for a given set of flow
conditions. Further simplified equations to obtain the maximum growth rate and corresponding
wavelength are proposed by Reitz (1987) through curve fits of numerical solutions to equation (8):
109
CFD-ACE_V2010.0_Modules_Manual_Part2
(9)
(10)
(11)
where .
The mass of new droplets formed due to breakup is subtracted from the parent drops as the breakup
proceeds. The change of the radius of a parent drop is assumed to follow the rate equation:
(12)
(13)
Fragments are stripped from the liquid column if the satisfies the following criteria:
110
Modules
(14)
where
(15)
(16)
and here are based on the gas velocity component normal to the liquid jet direction instead of
the relative velocity. If the column stripping does occur, the amount of mass removed from the column is
calculated as:
(17)
where
(18)
(19)
(20)
and is the lifetime of the liquid column drop. The factor causes the shedding rate to increase
linearly with distance away from the injection location. The shed drop size SMD is modified from Hsiang
and Faeth (1992) as:
111
CFD-ACE_V2010.0_Modules_Manual_Part2
(21)
The resulted drop velocities are set loosely based on Chou et al. (1997):
(22)
(23)
(24)
where Rnd represents any random number between 0 and 1; , are normal velocity component and
the cross flow velocity component, respectively; is the lateral velocity component. When the column
breakup time is exceeded, parcels created through the column stripping mechanism are considered to be
fragments, which may undergo further breakup. The liquid jet column breakup time is calculated as:
(25)
which is modified from Mazallon et al. (1999). When , the column is broken into new parcels
immediately with . The resulted velocity components are the same as equations (22)-
(24) except:
(26)
Those parcels formed from column breakup are still relatively large and are designated as fragments.
These fragments are further broken into small drops according to a modified version of BLS model for
drops (Delplanque et al., 1994):
(27)
where
(28)
(29)
The mass shed from a fragment in a time step and the SMD are calculated as:
112
Modules
(30)
(31)
A fragment can continue to break up until it no longer meets the criteria of equation (27) or until its size is
less than the newly created drops. Once the fragment breakup process is complete, drops may break up
further based on a Rayleigh-Taylor secondary breakup method.
Spray Module
Implementation-Model Setup and Solution-Volume Conditions
Spray Module
Implementation-Model Setup and Solution-Boundary Conditions
To assign boundary conditions and activate additional panel options, select an entity from the viewer
window or the BC Explorer. The boundary condition treatments for droplets are given below:
• At the inlets and outlets, the droplet/particle leaves the domain and is no longer tracked. The
exception to this is for inlets if "Treat Inlets as Wall" has been set.
• At symmetry boundaries, specular (coefficient of restitution of unity) reflection of droplet/particle
position and velocities are imposed.
• At the wall boundaries, several options are available controlled by the appropriate setting on the
Model Options page. See Model Options-Spray Wall Boundary Treatment.
• The droplet is allowed to stick to the wall.
• The droplet is allowed to evaporate at the wall (for example, in the case of evaporation).
• The droplet/particle is specified to bounce. (A proper value for coefficient of restitution should be
specified.) If the coefficient of restitution is specified as 0 (default), then the droplet/particle slides
along the wall without any friction or heat transfer with the wall.
Spray Module
Implementation-Model Setup and Solution-Initial Conditions
The initial conditions for the discrete phase, i.e. the droplet/particle locations, velocities, and mass flow
rate, could be calculated from an atomization model (not included) for applications in which an atomizer is
used to create the droplets. Since a reliable atomization model is not available, it could be specified from
experimental data taken at a location very close to the atomizer. If experimental data is not available, then
reasonable values could be guessed.
For example, if an atomizer produces a hollow cone spray, and if the cone angle and liquid pressure are
known, then the droplet velocities could be calculated within a reasonable uncertainty. Often times, the
locations of the injectors are specified at the center of the atomizer (little downstream). But if the intact
length of the liquid sheet could be estimated, then droplets could be placed at the end of the liquid sheet.
The mass flow rate of all the injectors should be equal to the total flow rate through the atomizer. It is
important to note that the injectors should be distributed spatially such that the mass flux distribution
matches with experimental data or that which is expected from the atomizer.
113
CFD-ACE_V2010.0_Modules_Manual_Part2
Spray Module
Implementation-Model Setup and Solution-Solver Control-Introduction
For droplet/particle transport problems without heat and mass transfer interactions, the default values for
solver controls are generally adequate. More relaxation than the default values may be needed for cases
in which the mass loading is relatively high. Since the spray sources in the gas phase equations introduce
stiffness, heavy inertial relaxation for dependent variables and linear under relaxation for auxiliary
variables may have to be used for cases such as spray combustion.
The Solver Control Spray tab contains the following sections that enable you to change the specifications
for your problem. For details on iterations, spatial, solvers, relax, limits, and advanced tabs see User
Manual-Overview-Control Panel-Solver Controls.
• Solver Control-Spray Calculation Control
• Solver Control-Spray Time Step Control
• Solver Control-Max Changes in a Spray Time Step
• Solver Control-Spray Fluid Coupling
• Solver Control-Coupling Source Linear Relaxation
• Solver Control-Artificial Spray Dispersion
• Solver Control-Ramp Coupling Source
Spray Module
Implementation-Model Setup and Solution-Solver Control-Spray Calculation Control
Spray Calculation controls spray trajectories processing. The controls available in this section depend on
whether the problem is steady or transient. This section enables you to change the following parameters:
• Starting Iteration - For Steady-state cases, this value tells the solver which iteration to start
calculating spray trajectories. This is useful if you want the flow field to setup before initiating
spray. For Transient cases, this value tells the solver the first time step to begin injecting droplet
parcels.
• Iteration Frequency - Instructs how often (in solver iterations) that a new spray calculation
should occur. Increasing this value may speed up the solution process since the spray trajectory
calculation can sometimes take longer than a solver iteration. However, increasing this value will
also typically hurt solution convergence.
• Starting Time Step - For transient cases, this value tells the solver the first time step (inclusive)
to inject droplet parcels.
• Ending Time Step - For transient cases, this value tells the solver the last time step (inclusive) to
inject droplet parcels.
• Injection Frequency - Instructs how often (in time steps) to inject new droplet parcels. Note that
the mass associated with each droplet parcel is adjusted based on the Injection Frequency so
that the mass flow rate is maintained.
• Minimum Particle Size - Determines the minimum size of a particle to track. When the particle
size becomes smaller than this value it will no longer be tracked and it will be immediately
evaporated.
Spray Module
Implementation-Model Setup and Solution-Solver Control-Spray Time Step Control
114
Modules
Spray Time Step Control provides inputs that determine the time step size used for the spray trajectory.
Because the Spray Module adaptively sets the time step size, these options provide limits for determining
the time step size:
• User Sub(spr_usr_tstep) - The solver calls the spr_usr_tstep() user subroutine for your
instructions on how to set the spray trajectory time step.
• Minimum Time Step - Constrain the spray time step size to stay within the minimum specified
limits.
Maximum Time Step - Constrain the spray time step size to stay within the maximum specified limits.
• Maximum Integrations - For steady cases, this limits the number of time steps that will be
performed for each trajectory. This is useful if the spray particle becomes entrained in a re-
circulation zone or is sliding slowly along a wall. The particle is not evaporated when the
trajectory calculation is stopped in this manner.
• Steps Per Cell - provides control over the internal calculation of the spray time step by adjusting
the time step size so that you will get approximately the specified value of time steps while the
spray particle is in a single cell. This means that smaller cells will force smaller spray time steps.
Spray Module
Implementation-Model Setup and Solution-Solver Control-Max Changes in a Spray Time Step
The Max Changes in a Spray Time Step provides control over the internal calculation of the spray time
step by limiting the maximum change of droplet properties. These properties are velocity, temperature (for
heat transfer cases), and diameter (for evaporation cases). If any of these properties change by more
than the specified amount, the time step size will be reduced accordingly.
Spray Module
Implementation-Model Setup and Solution-Solver Control-Spray Fluid Coupling
By default the Spray Fluid Coupling options, Fluid Phase Source Terms and Momentum Source Terms
are activated. This means that the spray solution will affect the fluid solution. This is required for
evaporating sprays. If you think that the spray has a negligible effect on the fluid/gas phase solution then
you can turn these options off. If this option is off for steady cases, the spray will be solved only on the
last iteration since it has no effect on the fluid phase solution.
Spray Module
Implementation-Model Setup and Solution-Solver Control-Coupling Source Linear Relaxation
If you activate the Gas Phase Source Terms option, you may control the coupling of the source term by
introducing a linear relaxation factor.
Spray Module
Implementation-Model Setup and Solution-Solver Control-Artificial Spray Dispersion
If you have activated the Gas Phase Source Terms option, you may add artificial dispersion by activating
Artificial Spray Dispersion. By default, this option is OFF so that the spray source terms are transferred to
the continuous phase by interpolating all of the source terms to the cell that is occupied by the spray drop
and its immediate neighbors in a continuous fashion. If you turn this option on, then you may specify the
number of levels to perform an artificial dispersion operation which will further spread the source term to
more neighboring cells. A higher value increases the spreading of the source term. Although this is an
arbitrary mechanism, it can significantly improve solution convergence for highly loaded spray cases.
Spray Module
Implementation-Model Setup and Solution-Solver Control-Ramp Coupling Source
115
CFD-ACE_V2010.0_Modules_Manual_Part2
If you have activated the Gas Phase Source Terms option, you may ramp the source terms during the
start-up of the solution process. If the Ramp Coupling Source option is turned on, then you will be able to
choose the number of iterations over which the source term should be linearly ramped up. On the first
iteration, no source term will be applied, on the iteration specified (and every iteration thereafter), the full
source term will be applied.
Spray Module
Implementation-Post Processing
CFD-VIEW can post-process the solutions. The spray data is written to a separate ASCII file with the
name: model_name.SPY.
116
Modules
It should be noted that the solver calculates several time steps based on mass/momentum/heat transfer
rates and the final time step selected will be the smallest of these including the user defined time step.
How do I specify the droplet initial conditions?
The initial conditions for the discrete phase, i.e. the droplet/particle locations, velocities, and mass flow
rate, can be calculated from an atomization models for some applications in which an atomizer is used to
create the droplets. Or, initial conditions can be specified from experimental data taken at a location very
close to the atomizer. If experimental data is not available, then reasonable values can be estimated. For
example, if an atomizer produces a hollow cone spray, then if the cone angle and liquid pressure are
known, then the droplet velocities can be calculated within a reasonable accuracy.
Often times, the locations of the injectors are specified at the center of the atomizer (little downstream).
But if the intact length of the liquid sheet could be estimated, then droplets could be placed at the end of
the liquid sheet. The mass flow rate of all the injectors should be equal to the total flow rate through the
atomizer. It is important to note that the injectors should be distributed spatially such that the mass flux
distribution matches with experimental data or that which is expected from the atomizer.
How can I read the spray data in CFD-VIEW?
The spray data is written to a separate file with the name: model_name.SPY. This is a formatted file that
was used for the structured solver, CFD-ACE. In order to read this spray data file, you must first read the
DTF file for the simulation. Then click on import Plot3d data submenu on the File menu in CFD-VIEW. In
the file open panel, select the spray, read additional data, PLOT3D, formatted buttons. The files names
that end with SPY will automatically appear. Select the appropriate spray file.
What are the optimum solver parameters for convergence of spray problems?
For many spray combustion problems, solver convergence can pose difficulties due to the strong coupling
between the phases. In general four orders of magnitude in convergence is acceptable. In order to
achieve this, heavy inertial relaxation for dependent variables (~ 0.5) and heavy under-relaxation (~ 0.2)
for auxiliary variables can be used. However, the inertial relaxation for the enthalpy equation should be
small (~ 0.05).
How can I manage spray source/sinks for getting convergence?
Spray problems can pose difficulty in convergence because of the stiffness introduced by the interaction
of the source/sink terms, especially for problems in which the mass loading is high locally (such as fast
evaporation in a region). There are several options available to manage the spray source/sink terms in
order to obtain better convergence:
1. Start the spray calculation after the gas phase equations have converged to a reasonable extent.
Delayed spray calculation can be accomplished by selecting a start iteration greater than 1 in
Solution Control/Spray (only for steady state cases). For unsteady cases, try to start from a
realistic initial condition such as a converged steady case.
2. Ramp the spray sources/sinks by selecting the ramp source option in the MD panel. Specify the
starting iteration (Io) and the number of iterations (N) over which the sources would be ramped
linearly. After (Io + N) iterations, the total sources/sinks will be added at each iteration.
3. The spray sources/sinks could be distributed to neighboring cells using an artificial dispersion
technique. This is justifiable because the Spray Module does not track all the droplets/particles,
but parcels. It must be noted that this dispersion technique is fully conservative so that mass is
conserved.
4. Utilize the artificial dispersion parameter.
Why do transient spray problems take more computational time?
To satisfy the spray mass flow rate condition, new parcels are introduced at each time step (note that the
time step interval, or injection frequency, can be set by the user in Solution Control/Spray). As the
simulation marches in time, the number of parcels tracked increases as NT*NP where NT is the number
of time steps, and NP is the number of parcels introduced at each time step.
117
CFD-ACE_V2010.0_Modules_Manual_Part2
Example:
# Point1 1 0
0.1 0.1 0.1 0.1 0.1 0.1 20 1000 1
# Point3 1 0
0.2 0.2 0.2 0.0 0.0 0.0 20 1000 1
# Point7 1 0
0.4 0.4 0.4 0.0 0.0 0.0 20 1000 1
Notes: The last value in the first line will be ignored if you are not running a dielectrophoresis case. Also,
for non DEP cases you do not have to provide an electrical permittivity and conductivity.
References:
Abramzon, B. and Sirignano, W.A., "Droplet Vaporization Model for Spray Combustion Calculations,” Int.
J. Heat Mass Transfer, Vol. 32, No. 9, pp. 1605-1618, 1989.
Ranz, W.E. and Marshall, W.R., "Evaporation from Drops - I and II,” Chem. Engr. Prog. 48, 141 and 173,
1952.
Shiller, L. and Naumann, A. "Uber die grundlegenden Berechungen bei der Schwerkraftaufbereitung, Ver.
Duet. Ing., 77, 318, 1933.
Tanner, F.X., "Liquid Jet Atomization and Droplet Breakup Modeling of Non-Evaporating Diesel Fuel
Sprays,” J. of Engines, Vol. 106, Sec. 3, pp.127-140, 1998.
118
Modules
different approaches have been developed to describe such flows. Trajectory-based (Lagrangian-
Eulerian) methods, where the fluid phase is solved in Eulerian frame of reference whereas the particles
are tracked as discrete entities in a Lagrangian reference frame, are well-established. However,
traditional Lagrangian methods are largely applicable to situations where the particle size is 'small’ (point
particle approximation) compared to the flow length scales of interest. For 'large’ particles, where the
particle dimensions become comparable to the flow length scales, the discrete phase can no longer be
treated as point particles. These particles affect the flow of the surrounding fluid substantially and such
influence has to be explicitly modeled (via the resolution of boundary layers, for instance). When coupling
the dynamics of such particles with other field effects (electric, magnetic, optic, etc.), the local details in
the vicinity of the particles become important. In terms of computational modeling of large particle
systems, one now needs:
1. Ability to correctly represent disperse phase particles that can occupy several computational cells
(large particles or macroparticles) and account for the effects of the blockage of these cells on the
surrounding flow.
2. Include the dynamics of the particle under the action of fluid drag as well as other forces such as
body forces. This would typically include a 6 Degrees Of Freedom (6-DOF) model to describe the
translational and rotational motion of the particles.
3. Ability to 'move’ the particles in the surrounding flow grid, and keep track of the interface between
the particle and the fluid.
The Macroparticle module in CFD-ACE+ was developed to address this need to model large particles in
fluids. In the Macroparticle module, the discrete phase tracking is achieved by solving the governing
momentum conservation equations in a Lagrangian frame of reference. The effect of the particles on the
surrounding fluid (two-way coupling) is explicitly accounted for. The Macroparticle Module documentation
includes:
• Applications
• Features and Limitations
• Theory
• Implementation
• References
Macro Particle Applications
The Macroparticle Module can be used for any simulations involving moving body calculations. Typical
applications include:
• Cellular- or particulate-based microfluidic applications
• Cell or microsphere surface binding in bioanalytical systems
Macro Particle Features and Limitations
Features
The Macro Particle module should be used when the size of the particles is rather large in comparison to
the flow path and is larger than the cell size. Some of the features of the Macro Particle module are:
• Multiple particles may be used
• Particle - particle interactions
• Coupled with the following modules
o Flow
o Turbulence (k-ε only)
119
CFD-ACE_V2010.0_Modules_Manual_Part2
o User Scalar
o Chemistry (only Biochemistry surface reactions)
Limitations
The following limitations apply when using the Macro Particle module:
• Cannot be run in parallel
• Cannot cross an arbitrary interface
• Can be run with all modules, but is only coupled with the modules listed in the Features
section.
Macro Particle Theory
Macro Particle
Theory
Multiphase flow modeling techniques (Crowe et al., 1998) are broadly classified into: (1) Two-fluid
approach (Eulerian), and (2) Trajectory approach (Lagrangian). The two-fluid approach is largely
applicable to multiphase modeling where different phases are evenly distributed. For modeling of discrete
particles, this approach is fraught with challenges in physical (continuum description of particles,
boundary condition specification, etc.) and numerical (artificial dispersion, etc.) modeling. Traditional
Lagrangian modeling (McLaughlin, 1994, Crowe et al., 1996) of particulate systems has largely been
carried out using a point-particle approximation (Maxey & Riley, 1983), similar to the one used in CFD-
ACE+ Spray module.
In recent years, several researchers have proposed novel approaches to accurately model motion of
large particles in liquids. These can be broadly classified into multi-domain and single domain methods. In
multi-domain techniques (called overset mesh or chimera, Steger et al., 1983), a unique computational
domain is prescribed to an individual particle and information is exchanged between the background and
the overset meshes. While these methods are very accurate in that no additional physical models are
required, they are ill suited for multi-body computations in close proximity to each other or to the
boundaries. Single domain methods have been successfully employed by several researchers for
simulation of large-scale particle-fluid systems. Body fitted, unstructured, Arbitrary Eulerian-Lagrangian
(ALE) methods have been reported by Hu et al. (2001), among others. Structured approaches include the
fictitious domain method (Glowinski et al., 2001) and the Immersed Boundary Method (Peskin, 2002).
Other domain embedding techniques include the approximate Eulerian-Lagrangian methods (Hofler &
Schwarzer, 2000), where a volumetric force is used to emulate rigid body motion of the particle, in a
manner similar to the force coupling methods introduced by Maxey & Patel (1997).
The Macroparticle module in CFD-ACE+ uses a single domain formulation termed the Surface Marker
Point Approach, which is similar to the Immersed Boundary Method. In this approach, an individual
particle is represented by a center of mass location and a pre-specified number of marker points and
tracked in a Lagrangian-Eulerian fashion. The region occupied by the particle is blocked off to the fluid
using a porosity approach and tangential no-slip and zero normal flux boundary conditions are imposed
on the surface of the particle. A six-DOF model is used to calculate the force balance (including fluidic, as
well as body forces) on the particle, which in turn is used to compute the motion of the particle. It should
be noted that the accuracy of this approach depends on the resolution of the macroparticle, which in turn
depends on the computational grids being used in the simulations. Intuitively, such an approach may not
be feasible for flow domains with computational cells much larger than the particle.
120
Modules
Macro Particle
Fluid Flow
The transport in the bulk fluid occurs due to convective and diffusive mechanisms and is described by
Naviér-Stokes equations. For incompressible laminar flow, these can be written as:
(1)
(2)
where , , and are the fluid velocity, density, pressure and gravity respectively. Note that the
conservation equations are modified to account for , the fluid volume fraction. Fluid stresses ( ) are
expressed in terms of the basic variables using constitutive relations.
Macro Particle
Particle Dynamics
As mentioned previously, in the Surface Marker Point approach, the particle is represented by a center of
mass location and a pre-specified number of marker points. The region occupied by the particle is
blocked off to the fluid using a porosity approach and tangential no-slip and zero normal flux boundary
conditions are imposed on the surface of the particle. 6-DOF motion of the rigid body is solved in a
Lagrangian fashion (Corben et al., 1950):
(1)
(2)
121
CFD-ACE_V2010.0_Modules_Manual_Part2
where is the momentum (angular momentum) defined as , is the mass of the body, is the
linear velocity, is the angular velocity, is the mass moment of inertia tensor, is the resultant force
vector acting on the body, and is the moment vector about the body center of mass. The forces acting
on the body include hydrodynamic as well as external body forces (if any) and impulse force imparted due
to wall interactions and/or collisions for multiple body calculations. Hydrodynamic forces on the particle
are computed by explicit numerical integration along the surface:
(3)
where is the pressure at marker point cell , is the stress tensor at marker point cell , is the
outward normal at marker point , is the surface area associated with marker point , and is the
total number of marker points.
Macro Particle
Particle Collisions
Particle direct interactions in the Macro Particle module are implemented for smooth collisions, i.e., if two
rigid bodies collide, the velocities coincide at the point of contact. It should be mentioned that in reality,
collisions in liquid-solid systems could only take place if the incidental energy is sufficient enough to
overcome lubrication effects. If the rupture of the interstitial liquid film is not taken into account, smooth
particles should not collide, as the repulsive lubrication forces become increasingly large as the film
becomes smaller. The primary purpose of the particle-particle collision models in the surface marker
approach is to prevent interpenetration. The models described below are intended to attain this objective
without impacting the overall behavior of the particles significantly. Two different approaches are
implemented in CFD-ACE+ Macro Particle module to model rigid body collisions.
In the hard sphere approach, frictionless rigid body collisions are enacted based on classical dynamics
theory. An impulsive force and torque are applied at the moment of impact and post-collisional
translational and rotational velocities are updated in accordance with conservation of energy and
momentum. The collision impulse is defined as:
(1)
where is the collision force vector and is time over which the collisional force acts. The change in
linear and rotational velocities is computed from
(2)
(3)
where is the mass of the rigid body, is the moment of inertia tensor, and is the position vector of
contact point with respect to the center of mass of the rigid body. Note that kinetic energy losses during
122
Modules
the collision process can be modeled using a coefficient of restitution approach. The impulse calculation
is described in more detail below.
Impulse Calculation
The collision process between two rigid bodies A and B is schematically shown in the figure below. The
particles undergo collision at , the Moment Of Impact (MOI). The global position vectors of the contact
points on the particles are denoted by and respectively, and is the outward normal vector (based
on particle B) at MOI.
At the MOI , the velocities of the contact points, and , on body A and B are given by
(4)
(5)
where and are velocities of center of mass of body A and B respectively, and are
rotational velocities of particles A and B respectively, and are the position vectors (global
coordinate system) of center of mass of body A and B respectively. The relative normal velocity between
contact points and is defined as
(6)
It is immediately obvious that the collision will take place only if the projection of relative normal velocity to
normal direction ( ) is negative.
123
CFD-ACE_V2010.0_Modules_Manual_Part2
(7)
The relative normal velocities right before (denoted with a '-’ sign superscript) and right after (denoted with
a '+’ sign superscript) the collision are computed as
(8)
(9)
The coefficient of restitution or recovery coefficient, , is defined as the ratio of post- and pre- collision
relative velocities
(10)
Applying the collision impulse to each body we obtain the change in translational and rotational velocity of
the center of mass. For instance, the post-collisional velocities for body A are given by
(11)
(12)
where is the mass of body A, is the velocity vector of center of mass of body A, and is the
position vector of contact point with respect to the center of mass (local coordinate system). The velocity
of contact point after the collision is given by
124
Modules
(13)
(14)
Similarly for body B, the post-collisional translational velocity of point can be obtained by substituting
with the opposing impulse ( ):
(15)
Thus the relative velocity at contact point between body A and B is obtained as
(16)
The relative normal velocity at contact point after the collision is therefore obtained by projecting the
relative velocity vector along the normal vector
(17)
Substituting this equation in the definition of the coefficient of restitution given by equation 10, we have
(18)
(19)
Note that if we let and to be infinite, the two body collision reduces to rigid body-wall interaction.
For particle-wall interactions, where 'Bounce with Impulse’ treatment is specified, this approach is used to
evaluate post collisional state of the particle.
125
CFD-ACE_V2010.0_Modules_Manual_Part2
In the repulsive force approach (termed soft collisions), short-range forces are used to prevent rigid
bodies from penetrating each other or the domain boundaries. This approach is similar to the ones used
by Glowinski et al. (2001) and Pan et al. (2002) and is derived from Molecular Dynamics (MD)
methodology. In order to treat particle direct interactions using soft collisions, repulsion forces derived by
truncation of the Lennard-Jones potentials are used to impose non-penetration constraints. In contrast to
hard sphere collisions, where the concept of impulse is employed to modify particle velocities, particle
forces in the repulsive force method are computed such that the particle "relaxes” to a post-collisional
state. It should be mentioned that this approach is best suited for near close-packed situations, even
though the nature of the interactions force limits the ability of the system to reach a fully close-packed
state.
In the repulsive force method, the interaction force on particle due to particle is given by
(20)
where is the separation vector between and , is a characteristic region of influence, is a force
scaling parameter, and is the force range. The total net force acting on particle is obtained by
summing up the contributions from all the other particles within the computational domain. The
characteristics of the force are tailored such that it acts only over a short range to prevent particle overlap
by acting continuously over time, which enhances model stability and robustness. Particle-wall collisions
in this approach are enacted using a simple reflection procedure, when 'Bounce’ treatment is specified.
Macro Particle
User Scalar or Analyte Transport
User scalar or analyte (for biochemistry applications) transport in the bulk is solved using a convective-
diffusive framework
126
Modules
(1)
where , , and are the scalar or analyte concentration, diffusivity and source/sink term (due to
surface binding, for instance), respectively.
Macro Particle
Surface Biochemistry
The figure below is a simplified schematic of the processes occurring within the flow domain and the
particle surface. Surface biochemical reactions are used to compute the source/sink term to be used in
the bulk transport . This is accomplished by imposing mass balance constraints on the rigid body surface.
Similar to regular surface biochemistry, it is assumed that there are two types of species present on the
surface of the particle - analytes (bulk phase species) and receptors (surface phase species). Each
species must satisfy a balance equation at a reactive surface depending on the type of species. For
analyte species the following balance equation is used:
(1)
which balances the diffusive flux of the analyte to/from the particle surface with the reaction
(consumption/generation) flux at the surface. For surface phase species, there is no diffusive transport
term and only the evolution of the receptor species is tracked at the surface.
(2)
(3)
127
CFD-ACE_V2010.0_Modules_Manual_Part2
The forward and reverse rate coefficients can be constants, or have an Arrhenius type dependence. After
computing the rate expressions for each step, the individual species source/sink term for the species
can be computed by summing over the individual step rates:
(4)
where represents the number of steps in a particular reaction mechanism. The function is zero
if species does not participate in step , whereas is 1 or -1 if species is a product or a reactant in
the reaction step.
Macro Particle Model Setup
Macro Particle
Implementation Introduction
CFD-ACE+ provides the inputs required for the Free Surface Module. Model setup and
solution requires data for the following panels:
• Problem Type
• Model Options
• MacP (Point Conditions)
• Solver Control
• Output
Macro Particle
Implementation and Grid Generation
The Implementation section gives details about how to setup a model for simulation using the Macro
Particle Module. The Macro Particle Module Implementation section includes:
• Model Setup and Solution - Describes the Macro Module related inputs to CFD-ACE-SOLVER
• Post Processing - Provides tips on what to look for in the solution output
The following geometric systems are supported by the Macro Particle Module: 3D, 2D Planar, 2D
Axisymmetric. All grid cell types are supported (quad, tri, hex, tet, prism, poly).
The general grid generation concerns apply, i.e., ensure that the grid density is sufficient to resolve
solution gradients, minimize skewness in the grid system, and locate computational boundaries in areas
where boundary values are well known. Also, you must ensure that the particle is larger than the smallest
cell. If the particles are smaller than the cells, you will need to refine your grid or use the Spray Module.
Macro Particle
Problem Type
Click the Problem Type [PT] tab to see the Problem Type Panel. See Control Panel-Problem Type for
details.
Select Macro Particle to activate the Macro Particle Module. The Macro Particle module should be used
when you have particles that are larger than the cell size, i.e. the particle completely blocks some cells. If
your particles are smaller than the cell size, the Spray Module should be used or you should refine your
grid.
128
Modules
Macro Particle
Model Options - MacP Tab
Motion Options
Allow Translation
The Allow Translation option will allow the particle to translate in any direction (X, Y, X). The particle
orientation will remain constant (i.e. not allowed to rotate). If the particle being modeled is stationary, do
not check this option.
Allow Rotation
The Allow Rotation option will allow the particle to rotate about the center of gravity of the particle. If this
option is not checked, then the particle will not be allowed to rotate and will have a constant orientation
(which is specified on the MacP tab).
Biochemistry Options
This option is only available when the Chemistry −> Liquid Phase −> Biochemistry is activated. If the
Allow Biochemistry Binding option is activated, the surface reaction mechanism should be supplied on the
MacP tab for each particle.
Force Options
Pressure Forces
Activating the Pressure Forces option will calculate the pressure forces on the particle. The calculated
pressure forces will then cause the particle to move if the Allow Translation and/or the Allow Rotation
options are activated.
Viscous Forces
Activating the Viscous Forces option will calculate the viscous forces on the particle.
Body Forces
Additional body forces, other than gravity, can be added to the particle. This is available through a user
subroutin (upblock_bforce). A user specified body force may be added to the particle through the user
subroutine.
If there are multiple particles in the model and the interactions between the particles need to be detected,
then the Detect Collisions option should be selected. If this option is not selected when multiple particles
are used, then there will be no particle-particle interactions, i.e. one particle could pass through another
particle without interacting.
Hard Collisions
This options allows for the specification of the coefficient of restitution between the particles. A coefficient
of 0 means the particle will loose all normal momentum and a coefficient of 1 means implies a perfect
129
CFD-ACE_V2010.0_Modules_Manual_Part2
bounce between the particles. For values between 0 and 1, there will be momentum transfer between the
two particles.
Soft Collisions
Bounce
This options allows for the specification of the normal and tangential coefficient of restitution between
particles and walls. The normal coefficient refers to the normal component between the particle and the
wall. For a normal coefficient of 0, all normal momentum of the particle will be lost. A normal coefficient
of 1 implies a perfect bounce of the particle. The tangential component refers to the tangential
component between the particle and the wall. A coefficient of 0 means the particle will loose all tangential
momentum and a coefficient of 1 means implies a perfect bounce. For values between 0 and 1, the
particle will loose some momentum.
This option only allows for momentum transfer in the normal direction between the particle and the wall.
Stick
For this option, any particle that comes in contact with a wall will stick to the wall.
130
Modules
Macro Particle
Model Options - Adv Tab
There is one setting on the Model Options −> Advanced tab that applies to macro particles. You can
either allow the solver to calculate which cells should be blocked by the particle, or you can write a user
subroutine (UPBLOCKAGE) to determine which cells should be blocked. In general, the solver will block
any cell which is entirely or partially inside the particle.
Macro Particle
Macro Particle Settings
To add a macro particle, select the MacP tab and select Add in the Model Explorer, as shown in Figure 2.
131
CFD-ACE_V2010.0_Modules_Manual_Part2
Properties
Once a particle has been added, the properties to be used for the particle need to be specified. Only
solid materials may be specified for the particle. Currently, the only property used is the density of the
solid (other properties may be specified in the Database Manager, but are ignored by the solver). New
solids may be created in the Database Manager.
Biochemistry Binding
If the Chemistry module is activated and the Biochemistry option has been chosen, then binding on the
surface of particles is allowed. A binding mechanism should be specified for each particle in the
simulation. If certain particles should have no binding, the mechanism must be specified such that the
surface density of the receptor groups is zero.
Particle Shape
There are several options for specifying the shape of the particle. The different options available are:
2D 3D
Circle Sphere
Ellipse Spheroid
Square Cube
Rectangle Cuboid
Capsule Cap Cylinder (Capsule)
- Cylinder
Arbitrary Arbitrary
Circle/Sphere
This option allows you to create a circle in 2D cases and sphere in 3D cases.
Inputs: Center point and circle/sphere diameter
132
Modules
Ellipse/Spheroid
This option allows you to create an ellipse in 2D cases and a spheroid in 3D cases.
Inputs: Major and minor diameter, Orientation
Square/Cube
This option allows you to create a square in 2D cases and a cube in 3D cases.
Inputs: Length, Orientation
Rectangle/Cuboid
This option allows you to create a rectangle in 2D cases and a cuboid in 3D cases.
Inputs: Length, Width, Height (3D), Orientation
Capsule/Cap Cylinder
Cylinder
Arbitrary
This option allows you to create a particle of arbitrary shape using an input file or user subroutine
(UPOINT_INIT) for marker point specification. The particle can be closed or open. An example of an
open particle would be a string or particles where there are gaps through which flow may pass. The file
format is the following
Each line represents the position, initial velocity, and diameter of a marker point. The more marker points
that are created, the better resolved the shape will be. If the arbitrary shape is best defined by an
equation, the UPOINT_INIT user subroutine may be the best option.
133
CFD-ACE_V2010.0_Modules_Manual_Part2
Particle Temperature
This options allows you to specify the temperature of the particle. Although the Macro Particle module is
not coupled to the Heat Transfer module (i.e. no conjugate heat transfer), the particle can heat the
surrounding fluid. An example would be flow over a cylinder where the cylinder is heated, where the
isothermal temperature of the walls could be specified.
134
Modules
Crowe, C.T., Chung, J.N. & Troutt, T.R., 1996, "Numerical Models for Two-Phase Turbulent Flows,” Ann.
Rev. Fluid Mech., 28, 11-43.
Crowe, C.T., Sommerfeld, M., & Tsuji, Y., 1998, Multiphase Flows with Droplets and Particles, CRC
Press, Boca Raton, FL.
Glowinski, R, Pan, T.-W., Hesla, T, Joseph, D. D., & Périaux, J., 2001, "A Fictitious Domain Approach to
the Direct Numerical Simulation of Incompressible Viscous Flow past Moving Rigid Bodies:
Application to Particulate Flow,” J. Comput. Phys., 169, 363-426.
Hofler, K. & Schwarzer, S., 2000, "Naviér-Stokes Simulation with Constraint Forces: Finite-difference
Method for Particle-laden Flows and Complex Geometries,” Phys. Rev. E, 61, 6, 7146-7160.
Hu, H.H, Patankar, N.A. & Zhu, M.Y., 2001, "Direct Numerical Simulations of Fluid-solid Systems using
the Arbitrary Langrangian-Eulerian Technique,” J. Comput. Phys., 169(2), 427-462.
Kalthoff, W., Schwarzer, S. & Herrmann, H., 1997, "An Algorithm for the Simulation of Particulate
Suspensions with Inertia Effects,” Phys. Rev. E, 56(2), 2234-2242.
Maxey, M. R. & Patel, B. K., 1997, "Force-coupled Simulations of Particle Suspensions at Zero and Finite
Reynolds Numbers,” ASME FEDSM 97-3188.
Maxey, M.R. & Riley, J.J., 1983, "Equation of Motion for a Small Rigid Sphere in a Nonuniform Flow,"
Phys. Fluids, 26, 883-889.
McLaughlin, J. B., 1994, "Inertial Computation of Particle Turbulent Interaction,” Intl J. Multiphase Flow,
20, 211-232.
Peskin, C.S., 2002, "The Immersed Boundary Method,” Acta. Numerica, 11, 479-517.
Steger, J.L., Dougherty, F.C. & Benek, J.A., 1983, "A Chimera Grid Scheme,” Advances in Grid
Generation, (ed. Ghia, K.N. & Ghia, U.), ASME FED, 5.
135
CFD-ACE_V2010.0_Modules_Manual_Part2
136
Modules
(14-1)
where F is the liquid volume fraction, t is time, ∇ is the standard spatial grad operator, and is the
velocity vector. This equation must be solved together with the fundamental equations of conservation of
mass and momentum (and energy when activated) in CFD-ACE+ to achieve computational coupling
between the velocity field solution and the liquid distribution.
The solution of equation 14-1 in conjunction with the standard conservation equations in CFD-ACE+
requires three related actions:
• Compute mixture properties
• Reconstruct the fluid-fluid interface in each cell
• Determine the contribution of the secondary fluid flux to the equations for conservation of mass,
momentum, energy and volume fraction
(14-2)
where is the volume-averaged quantity, φ 2 is the value of the property for fluid two (water), φ 1 is the
value of the quantity for fluid one (air). For a mass-specific quantity, equation 14-2 must be extended to
include the effect of density, ρ:
(14-3)
Fluids 1 and 2 can be interchanged. In principle, interchanging the fluids should have no effect on the
results; in practice, it has a slight effect, which is more pronounced for problems with surface tension.
137
CFD-ACE_V2010.0_Modules_Manual_Part2
138
Modules
therefore generally have the shape of an arbitrary polygonal face. The facet in a cell is fully defined by
specifying: (i) the spatial orientation of the infinite plane that contains the facet; and, (ii) the location of a
point within the cell through which the infinite plane passes. The orientation is specified by specifying the
outward-pointing unit normal of the infinite plane, and the sense of the normal is here arbitrarily chosen so
that it points out of the liquid phase and into the gas phase. The unit normal of the plane is determined by
assuming that it is parallel to the gradient vector of F. The gradient of F is determined from the local
distribution of F in a set of cells which includes the target cell and all its immediate neighbors. The
location of the anchor point is determined by finding the infinite cutting plane perpendicular to the unit
normal of the infinite plane that truncates the correct liquid volume from the cell, in that it satisfies the
condition:
V cut = F V c (14-4)
where V cut is the volume of the cell truncated by the cutting plane, V c is the volume of the whole cell, and
F is as defined above.
In the PLIC scheme, each cell has a unique surface normal that can be used to compute the surface
curvature from cell to cell. This calculates and adds surface tension forces for the free surfaces.
In VOF module, the volume fraction is computed by solving scalar convection equation
(1)
The value of f is zero when the cell does not contain the r field-associated fluid, and one when the cell is
totally filled with that fluid. Cells located at the interface are filled with two fluids, thus the f fields at these
locations have values between zero and one. The VOF method is capable of modeling flows with
complex free-surface geometries, including flows in which fluid volumes separate and reattach; yet it is
remarkably economical in computational terms, requiring only a mesh-sized array for storing the f field
and an algorithm to advect the f field at each time step. Because the f fields represent averaged volume
fractions of fluids within each cell of the computational domain, information about interfaces is not readily
available and as the fluids flow through the fixed grid, the fluid-fluid interfaces may cut through
computational cells. In this case, extreme care should be taken in advecting the f fields so as to preserve
the sharpness of interfaces. Over the years, a number of advection schemes have been developed,
which, for Eulerian meshes, can be classified under two categories denoted in the literature as interface-
tracking methods and interface-capturing methods.
In interface-tracking methods, the interface is explicitly reconstructed and used in the evaluation of the
advection scheme, i.e., the advected f fluxes depend explicitly on the position of the interface within the
individual computational cell. Hence the accuracy of the reconstructed interface plays a critical role in the
performance of the advection scheme. Examples of interface-tracking methods include the well-known
SLIC [1-4] and PLIC algorithms and their many variations (e.g., PROST [5], DDR [6], etc.).
The main drawback of interface-tracking methods is the algorithmic complexity involved in reconstructing
the interface in a continuous manner across the computational domain, with this difficulty being
compounded in three-dimensional problems. In interface-capturing methods, the f value at a control-
volume face is formulated algebraically, without reconstructing the interface. Generally, in interface-
capturing methods a compressive scheme is used to avoid smearing of the interface. However, this has
been found to lead to stepping of the interface (i.e., loss of curvature) whenever the flow is not aligned
139
CFD-ACE_V2010.0_Modules_Manual_Part2
with the computational grid. Workers have remedied this problem by adopting a switching strategy that
toggles between a compressive and a noncompressive scheme depending on some criterion related to
the f field. Many of these schemes base the switching criterion on a function of the angle formed between
the interface normal direction, readily obtained using the gradient of the f field, and the grid orientation.
In interface-capturing methods, the discretization of the f equation in both the time and spatial domains
has to be accurate enough to prevent the smearing associated with numerical diffusion. Generally the
base scheme is the upwind scheme, but other higher-order schemes could also be used. For upwind
scheme, it is computationally robust and efficient, but suffers from substantial numerical diffusion. The
second-order Crank-Nicholson and the second-order Euler schemes are better behaved in that respect
but can still lead to over-undershoots with large time steps, as they are not bounded. To avoid those
drawbacks, some new schemes are proposed, e.g., CICSAM[7], HRIC [8], and STACS[9].
A free-surface solver based on a homogeneous multiphase model is available in V2008 of ACE+. The
free-surface solver is of the volume of fluid type (volume fraction) with additional numerics to enhance its
robustness and accuracy. The solver relies on an interface capturing method to advect the inter-fluid
interfaces, furthermore no restriction is placed on the type of mesh used within the computational domain,
and the fluids can be compressible and/or incompressible.
The best approach for blending is to have a continuous switching function whereby the values of a
compressive and a HR advection scheme are blended together, with the blending factor depending on
the angle between the flow direction and the grid lines. The angle can be determined using the grid
orientation at the integration face and the gradient of the f field, whose unit vector represents the direction
normal to the interface (see Figure 1). This general approach has been utilized in the CICSAM, the HRIC
and the STACS schemes, even though different blending functions are used in these schemes, as will be
described later. From the above it is clear that an "interface-capturing’’ scheme based on the switching
strategy should possess the following attributes:
1. It should be based on a combination of compressive and high-resolution schemes.
2. Its blending function should be based on the angle between the interface direction and the grid
orientation, preferably in a continuous fashion.
The reasoning followed in defining the blending function is illustrated in Figure 2. If the cell has started to
be filled with fluid from the upwind side of the interface and the interface is parallel to the cell face (Figure
2a), then only fluid present at the downstream cell should be convected through the cell face. In this case
a compressive scheme should be used. However, if the interface is perpendicular to the cell face (Figure
2b), then the convected fluid is expected to be of the same composition as the upwind cell; in this case a
HR scheme would be appropriate. When the fluid-fluid interface is parallel to the cell face but most of the
cell is filled with fluid from the upwind side of the interface (Figure 2c), then either scheme could be used.
The above-mentioned situations represent extreme cases in which the fluid-fluid interface is either parallel
or perpendicular to the control-volume face. In general, the angle between the interface and cell face is
between these two extremes (i.e., the angle usually varies between 0 and , Figure 2d and the value
of f at the interface should be obtained by blending the advection schemes of the extreme cases, with the
normalized value of f at the control-volume face given by
(2)
Where is the blending function that varies between 0 and 1, is the angle between the interface and
cell face. The normalized volume fraction is defined as
140
Modules
(3)
with the subscripts U and D referring to values at the upstream and downstream locations.
The high-resolution interface-capturing scheme (HRIC) is based on a blending of the bounded downwind
(BD) and upwind differencing schemes (UD), with the aim of combining the compressive property of the
BD scheme, which can be viewed as a steady-state version of the HYPER-C scheme, with the stability of
the UD scheme. The normalized functional relationship of the BD scheme is given by
(1)
141
CFD-ACE_V2010.0_Modules_Manual_Part2
The functional relationship of the HRIC scheme is also function of the angle between the normal to the
interface (defined by the gradient of the f field) and the normal to the cell face. For an interface aligned
with the cell face ( = 0) the bounded downwind scheme is used, while for an interface perpendicular to the
cell face the upwind scheme is used. For an interface with between these two limits, the blending
function is defined as
(2)
(3)
With this formulation, the blending of the UPWIND and DOWNWIND schemes is dynamic and accounts
for the spatial distribution of the f field. Muzaferija further modifies the value of to account for the local
Courant number (Co)
(4)
142
Modules
Normalized variable diagrams (NVDs) of (a) HRIC, (b) CICSAM, (c) SUPERBEE, (d) STOIC, and (e)
STACS scheme
For a Courant number below 0.3 the scheme is not modified, while for a Courant number above 0.7 the
upwind scheme is used. For Co values between 0.3 and 0.7, the interface value is computed by
(5)
143
CFD-ACE_V2010.0_Modules_Manual_Part2
The normalized variable diagrams (NVDs) of the HRIC scheme for Co values in the various regimes are
depicted in Figure 3a. It is clear from the NVD diagram that for Courant number values above 0.7, the
HRIC scheme basically reverts to the very diffusive UPWIND scheme, and even for moderate values of
the Courant number, the scheme would still be very diffusive.
It has been shown that predictions generated using the above schemes deteriorate with increasing values
of the Courant number as these schemes blend with the UPWIND scheme and become identical to it at a
Courant number of 0.7 for HRIC and 1 for CICSAM. This behavior is a result of the temporal bounding
used, which was originally designed by Leonard [10] for the explicit QUICKEST scheme. While this is
needed for explicit transient schemes, its use in an implicit method increases numerical diffusion.
Marwan Darwish [9] suggested Switching Technique for Advection and Capturing of Surfaces (STACS)
scheme, in which the selected compressive scheme is SUPERBEE, a bounded version of the downwind
scheme, while the high-resolution scheme is STOIC. Moreover, because of the use of an implicit transient
discretization, no transient bounding is applied. Furthermore, in order to minimize the stepping behavior
of the highly compressive SUPERBEE scheme, the blending between the two schemes is performed
using Equation (6) with set to , which enables a rapid but smooth switching away from the
compressive scheme for the case where the normal to the free-surface face is not along the grid
direction. The normalized variable diagrams of the SUPERBEE, STOIC, and STACS schemes are
displayed in Figures 3c, 3d, and 3e, respectively.
The normalized variables relationship for the STACS scheme is given by
(1)
(2)
(3)
144
Modules
The CICSAM (Compressive Interface Capturing Scheme for Arbitrary Meshes) scheme is also an
interface-capturing scheme based on the blending strategy. However, rather than choosing the
DOWNWIND and UPWIND schemes as base schemes, it uses, respectively, the HYPER-C scheme and
the ULTIMATE-QUICKEST (UQ) scheme, with HYPER-C being utilized when the cell face is
perpendicular to the interface normal vector and the UQ employed when the normal vector to the face is
aligned with the normal to the interface. The HYPER-C scheme is a bounded downwind scheme that is
constructed by enforcing the transient convection boundedness criteria (CBC) onto the DOWNWIND
scheme and is expressed as
(1)
(2)
(3)
(4)
(5)
The blending function is based on the angle between the gradient of the volume fraction at the
interface and the normal to the cell face. For an angle =90, i.e., when the interface normal is
perpendicular to the cell face normal, is zero and the UQ scheme is used, and for =0; i.e., when the
flow interface is aligned with the face normal, the HYPER-C scheme is used. The NVD of the CICSAM
scheme, depicted in Figure 3b, reveals that with increasing Courant number the scheme becomes more
and more diffusive as its NVF function reverts to that of the UPWIND scheme.
145
CFD-ACE_V2010.0_Modules_Manual_Part2
Computation of the Secondary Fluid Flux in the PLIC Method by Back Projection from a Cell Face
Two important implications of back projection approach are that the secondary flux:
• is computed explicitly based on the old liquid surface orientation
• and depends on information from the upwind cell only
As such, the surface advection calculation is explicit and has an inherent stability limit on the allowable
timestep magnitude. Within each computational cell, the stability limit is given by the so-called Courant
Condition:
(14-5)
where ∆t c is the maximum timestep that can be taken in cell c, and and d c are respectively the local
velocity vector and the local cell dimension (or length-scale).
The CFL Number, defined by:
(14-6)
specifies the distance crossed by any phenomenon, effect, or wave traveling at velocity v c , as a fraction
of the local length-scale of the cell. Thus, for example, a CFL value of 0.2 in a VOF computation would
allow the fluid or the interface to cross 20% of the width a cell during each time-step, regardless of the
actual velocity or cell dimension. For stability of the solution algorithm used to integrate equation 14-1, the
CFL Number must satisfy the condition CFL < 1, but in practice, values as low as 0.1 are often
required.
146
Modules
In practice, this means that the timestep size for mixed cells must be restricted to ensure that the free
surface crosses less than a cell during that timestep. This may be accomplished automatically in CFD-
ACE+ via an automatic timestep option that controls to the Courant Condition.
Multi-to-One Cell
(14-7)
,
where σ is the surface tension between the two fluids. This leads to:
This formulation requires no information about the global interface shape. Instead, only the geometry of
the edges of the interfaces within the individual computational cells are needed. When applied at the
discrete level, the computation is conservative, in that when the interface is broken into many small
pieces, the total force is equal to summation of the forces over all the constitutive pieces. Mathematically,
due to the opposite traversal direction along each edge separating two cells, the contour integrals cancel
except for the contributions along the outer boundary of the entire surface.
The model is implemented as:
147
CFD-ACE_V2010.0_Modules_Manual_Part2
Marangoni Effect
Equation 14.7 above and the discussion following it assume that the surface tension coefficient, σ, is
constant. In that case, as mentioned above, the contour integral of the surface tension force along the
closed curve bounding any interface element (which is expressed in the right-hand side of Equation 14.7)
has a resultant that is directed perpendicular to the interface element, and therefore can be balanced in
static equilibrium exclusively by a pressure difference across the interface, as expressed in Equation
14.7. If, however, the surface tension coefficient is not constant, the resultant of the force contour integral
defined above will generally have a non-vanishing component in a direction tangent to the interface
element, typically leading to flow along the interface. This phenomenon (of motion along the interface and
in the two fluids attached to it due to variations or gradients in the surface-tension coefficient) is known as
the Marangoni Effect. Variations in the surface tension coefficient can arise through dependence of the
surface tension coefficient on solution variables or material properties. For example, the surface tension
coefficient may depend on temperature, species concentration(s), electric-field intensity, or even the local
curvature of the interface itself. ACE+ enables such dependence to be modeled either by specification
through the GUI or through User Subroutines.
The Marangoni force in ACE+ is computed by evaluating the right-hand side of Equation 14.7 for each
interface “fragment” in each computational cell, computing the tangential component of this force, and
then applying that component as a source term in the discrete momentum equation for the cell in which
the interface fragment lies. These tangential forces are computed in addition to the normal forces
described above.
148
Modules
velocity can be appreciable. For example, the contact angle may vary by almost 180 degrees for very
large ranges of velocity, and often varies by 20 degrees or more over the velocity ranges encountered in
a typical micro-fluidic problem. For convenience, only the most realistic and commonly-used contact-
angle functions have been provided in the set of Dynamic Methods in CFD-ACE+. As mentioned above,
for functions other than these, or for different or more general dependencies, such as those involving the
temperature or the location on a surface, the user-subroutine facility must be used instead.
One concern that should be borne in mind when using a velocity-dependent contact-angle model is its
potential adverse effect on the convergence rate of the numerical solution. This complication arises from
the strong coupling between (and the circular dependence on each other) of the local velocity around a
contact line or point, and the location of the contact line or point. At best, the effect of this coupling would
be an increase in the number of iterations required to achieve a converged solution (and hence an
increase in the total solution time). This effect can be reduced by avoiding contact angle functions with
steep gradients.
You must select or specify any desired functional dependence of the contact angle as an explicit, closed-
form function. The CFD-ACE-Solver does not have any capabilities for making automatic predictions of
the functional dependence of the contact angle, whether from the thermodynamic or physical properties
of the materials involved, or from any other ab-initio physical models. The required functional dependence
is typically obtained independently, by curve-fitting of appropriate experimental measurements of the
behavior of the contact angle.
where θ is the evaluated contact angle between the liquid-gas interface and the wall; w is the unit wall
normal directed into the wall, and w is unit tangent to the wall, directed into the liquid phase, as shown
below. It shows a contact point, and the associated vector directions, including that of the interface
normal n. The contact point is marked by the symbol x w .
149
CFD-ACE_V2010.0_Modules_Manual_Part2
Mass Diffusion
Considering a control volume such as the one in Figure 1, the mass conservation equation for the vapor
can be written as:
(1-1)
where is the evaporation rate and f is the mass fraction of the vapor. The diffusion rate is then given
by
(1-2)
where the subscript "0" is the position of the liquid-vapor interface. Thus, the evaporation rate can be re-
written as
(1-3)
In the boundary layer, let define the value of f in the free stream, which leads to the following
150
Modules
(1-4)
(1-5)
The conductance is independent of for near zero, but may become a strong function of as
becomes large. However, this does not invalidate the definition of the mass conductance.
Figure 1. Control volume at the phase interface for the mass diffusion equation
With the definition of the mass conductance in equation (1-5), we then have the following for the
evaporation rate
(1-6)
The mass fraction of can be obtained from the equilibrium assumption at the interface, and it is a
function of temperature.
If the Lewis number (Le), a ratio of the thermal diffusivity to the mass diffusivity, is assumed to be unity in
the boundary layer, we have
(1-7)
(1-8)
151
CFD-ACE_V2010.0_Modules_Manual_Part2
which leads to
(1-9)
152
Modules
CFD-ACE+ provides the inputs required for the Free Surface Module. Model setup and solution requires
data for the following panels:
• Problem Type
• Model Options
• Volume Conditions
• Boundary Conditions
• Initial Conditions
• Solver Control
• Output
Click the Problem Type [PT] button to see the Problem Type Panel. See Control Panel-Problem Type for
details.
Select Free Surface to activate the Free Surface Module.
Click the Model Options [MO] tab to see the Model Options Panel. See Control Panel-Model Options for
details.
Shared
Select the Standard or Automatic Transient Option. For the Automatic timestep option, input a target CFL
number to limit the advection of the surface relative to the grid. For example, a CFL target of 0.2
(recommended) will limit the surface motion to 20 percent of the cell volume per timestep.
VOF
CFD-ACE+ provides two options in the VOF module: Surface Reconstruction and Time-Integration
scheme.
Surface Reconstruction
Surface Reconstruction
There are six options available under Surface Reconstruction: 2nd Order (PLIC), 1st Order (SLIC), 0th
Order, HRIC, STACS, and CICSAM. When the PLIC option is selected, the Surface Tension and
Pressure calculation options become are available.
Surface Tension
Select the Surface Tension option to incorporate surface tension effects. The HRIC, STACS, and
CICSAM schemes may be used with unstructured grid systems, but only without surface tension
modeling. Surface tension capabilities will be added for these schemes in a later release. If the Surface
Tension check box is activated, an additional check box for Marangoni Effects appears in the GUI and
can also be activated if desired. As explained above, selecting the Marangoni Effects option will include
the effects of tangential surface tension forces (in addition to the normal surface tension forces) that are
153
CFD-ACE_V2010.0_Modules_Manual_Part2
acting on the interface. If the Marangoni Effects option is not activated, then the Maragoni forces will not
be calculated, even if the surface tension coefficient is not constant. If the surface tension coefficient is
constant, or if the tangential (Marangoni) forces are known to be unimportant in the simulation being
prepared, there is no reason to activate the Marangoni Effects model. In fact, in such cases, it is
recommended to not activate the Marangoni Effects model to avoid the introduction of additional errors in
the solution. For example, if the surface tension coefficient is constant then theoretically, the tangential
forces along interfaces should cancel, but in practice, the discretization errors in the solution, including
those from reconstructing the geometry of the interface and calculating its normals and curvatures, will
introduce errors that will show up as spurious tangential forces in the solution.
Selection of the Time-Integration Scheme for the F Equation
Two options are available for specifying the time-integration scheme used to evolve equation 15-1:
explicit and implicit. The explicit option corresponds to a Forward-Euler Scheme, in which the fluxes are
computed from the F distribution at the previous time-level. The implicit option corresponds to a
Backward-Euler Scheme, in which the fluxes are computed from the F distribution at the current time-
level.
Compared to the implicit option, the explicit option exhibits greater stability and better convergence, but
slightly lower accuracy. The effects on stability and convergence are especially pronounced for problems
in which the source terms that drive the flow depend on the F distribution. The prime example of such a
problem is a surface-tension-dominated flow problem. The reason for the greater stability and improved
convergence for such problems is that the source term does not change throughout the outer-iteration
loop. The explicit option is also slightly faster than the implicit option.
Phase Change
In order to select the Evaporation/Phase Change feature with VOF, the additional modules of Chemistry
and Heat Transfer must be selected on the PT tab. Then, the option for evaporation will be available on
the MO tab, under VOF. When running an evaporation case, the species that will be evaporating must be
specified. The saturation pressure of the evaporating species is defined using the evaluation method:
Piecewise Linear in T. For this method, data pairs of the saturation pressure and temperature must be
input.
Click the Volume Conditions [VC] tab to see the Volume Conditions Panel. See Control Panel-Volume
Conditions for details. Before any volume condition information can be assigned, one or more volume
condition entities must be made active by picking valid entities from either the Viewer Window or the VC
Explorer. The material selection for both fluids must be specified for each volume condition, regardless of
the initial distribution of the two.
Properties
A second set of properties is required, corresponding to the properties of the second fluid. The following
property options are available:
• Density Options
• Constant
• Polynomial Function of Temperature
• Conductivity (when Heat Transfer is activated)
• Constant
• Specific Heat (when Heat Transfer is activated)
• Constant
154
Modules
In addition to the standard initializations required for the velocity field (and all other universal field
variables present), the Free Surface Module requires the specification of the initial distribution of the
volume fraction of the secondary fluid, F, throughout the solution domain. This initial condition
specification may be applied through five different options that the user can select. These five different
options are as follows:
1. All Fluid 1, that is, F = 0 throughout the domain or current volume condition.
2. All Fluid 2, that is, F = 1 throughout the domain or current volume condition.
3. Foam, that is 0 < F < 1 throughout the domain or current volume condition.
4. With the use of one or more of the following geometric shapes:
• Planes, with Fluid 1 on the side of the plane that is in the direction of the normal vector
specified by the user.
• Spheres, with Fluid 1 or Fluid 2 inside the sphere, as specified by the user
• Cubes, with Fluid 1 or Fluid 2 inside the cube, as specified by the user.
• Oher inputs that define the geometry of the individual shapes are required. The user may
select any multiples and combinations of the above shapes, with overlaying of different
shapes allowed, to specify different initialization regions and to create composite filling
patterns in the solution domain.
5. With the use of a set of points. This initialization method is also called the “Initial Data” approach,
and is available only with the Global IC initialization option. The Initial Data approach requires the
user to specify a set of points in a data file according to a pre-determined format. When this
option is selected, the user must specify the name of the file, and the directory from which the file
can be retrieved by the solver at the start of execution. With this option, the entire domain is first
initialized with primary fluid, that is, F = 0 is set everywhere. Then, for each point in the data file,
every cell whose centroid falls within a specified distance from that point is re-initialized with the
secondary fluid, that is, with F = 1. The user can choose the distance (which will be the same for
all the points), and can choose whether the distance is interpreted in the sense of an “inclusion
sphere” centered at the point or in the sense of an “inclusion cube” centered at the point. No
intersection calculations are performed between the spheres or cubes defined in the data file and
the cells of the grid: each cell is either assigned a value of F = 0 or F = 1, depending entirely on
whether it falls inside at least one inclusion sphere or inclusion cube. The pre-determined format
of the data file is as follows:
n d option
x 1 y1z 1
...
155
CFD-ACE_V2010.0_Modules_Manual_Part2
...
...
xnynz n
where:
n denotes the number of points in the data file
d denotes the radius of the inclusion sphere or the half width of the inclusion
cube that are to be centered at each point
option denotes the metric definition to be applied to d. If option is set to 1,
then the distance is measured in the sense of an inclusion sphere in 3-D or an
inclusion circle in 2-D. If option is set to any integer other than 1, then
distance is measured in the sense of an inclusion cube in 3-D or an inclusion
square in 2-D
x, y, z denote point coordinates
The figure below shows a schematic of the initialization process described above in a 2-D domain for both
the "sphere" and the "cube" options.
Schematic showing the Initial Data initialization procedure for the F function.
In order to enable the user to obtain smoother initializations with the Initial Data approach, the solver
actually creates some additional linearly-interpolated points between each two points given in the data
file, and then applies the distance checking procedure to all points (including the original points and the
internally-created, interpolated points).
Default solver control settings are recommended with approximately 30 iterations per timestep. See
Control Panel-Solver Control for details.
Advanced
CFD-ACE+ provides two options under the Advanced tab in the Solution Control tab: Removal of Flotsam
and Jetsam and Surface-Tension or Capillary-Wave Damping. The Advanced Tab includes:
156
Modules
One of the well-known defects of the VOF methodology, especially if a lower-order surface-reconstruction
method such as SLIC is used [4], is the generation of tiny isolated droplets of liquid in gas regions, and of
tiny isolated bubbles of gas in liquid regions. These droplets and bubbles are collectively called flotsam
and jetsam. This effect becomes more pronounced with an increase in any of the following:
• Poor convergence, perhaps due to an insufficient number of iterations for each time-step
• Too large a setting for the CFL Number
• Excessive skewedness in the grid, or other contributors to poor grid quality.
The formation of flotsam and jetsam should be minimal with the PLIC surface reconstruction (compared to
the SLIC surface reconstruction), but cannot be completely eliminated, especially if either the
convergence or the grid quality is poor.
The flotsam and jetsam filter attempts to eliminate flotsam and jetsam at the earliest stage of their
formation, before they begin to adversely affect the solution, and before they grow in size. However, this
filter should be regarded as a means of treating the symptom, and not the cause. Nevertheless, if the
cause cannot be treated, then using this filter will at least prevent some of the negative side effects of
flotsam and jetsam, not least of which is a reduction in the time-step size.
The flotsam and jetsam filter can be turned on or off under the Advanced Tab in the Solver Control menu.
The frequency, that must be supplied if the option is activated, specifies the number of time-steps
between each invocation of the filter during a computation. Unless there is good reason to do otherwise,
this number should be set to 1.
The capillary-wave damping option increases the viscosity in the vicinity of the interface, in an attempt to
damp the capillary waves that are invariably generated in the interface under surface-tension forces.
These waves are of no interest in any equilibrium problem, or in any problem in which only the gross
propagation of the interface is being studied. However, the magnitude of the tangential velocities that they
create is often 10-100 times greater than the magnitude of the normal velocity of the interface. Since the
size of the time-step is controlled by the CFL Number, these large tangential velocities can in turn
unnecessarily reduce the time-step by one or two orders of magnitude. Increasing the viscosity around
the interface, however, could adversely affect the final solution, and must therefore be done very
carefully.
The general guidelines for using capillary-wave damping are:
• Activate the capillary-wave damping option only if the interface motion is exhibiting symptoms of
157
CFD-ACE_V2010.0_Modules_Manual_Part2
provided for up to three layers (or levels) of cells around the interface. Level 1 refers to the interface cells
(i.e., to the cells which contain an interface segment); Level 2 refers to the layer of cells next to those of
Layer 1; and, Layer 3 refers to the layer of cells next to those of Layer 2. The Basic Input option allows all
three layers to be given the same value simultaneously, while the Advanced Input option allows the
values to be independently specified for each of the three layers. The "wall” layer can be used to reduce
the viscosity near walls, from the final value after adding the damping, and should not be increased above
1.
Typical suitable ranges for the increase in the viscosity are 2-10 for the liquid, and 200-1000 for the gas
(which is typically 100 times less viscous than the liquid). A general guideline is that the liquid and gas
viscosities should be increased so that they have the same final values. For problems where one of the
fluids has a very high viscosity, no damping should be needed.
The necessary increase in viscosity needed to reduce the capillary waves to a tolerable level depends on
the relative strength of the surface-tension forces, and on the three-dimensionality of the interface. For
interfaces in thin cavities, especially with the Hele-Shaw model, the needed increase tends to be as small
as 2-5 in the liquid, and 200-500 in the gas. For three-dimensional interfaces with relatively very high
curvatures and very strong surface-tension forces, the necessary increase in the liquid viscosity could be
as large as 50. In all cases, the lowest value of the viscosity increases can be arrived at by trial and error,
and by monitoring the response of the time-step size to changes in the damping.
Although increasing the viscosity in the interface region by a factor of 10 or more may seem likely to
significantly affect the final solution, numerical experiments confirm that doing so under the right
circumstances can have little or no detectable effects on the final solution. It should also be noted that the
finer the grid around the interface, the more effective the capillary-wave damping, and the less effect it
has on the final solution.
Several additional outputs are available when the Free Surface Module is active. The most common one
is the output of the volume fraction F, which can be used to reconstruct the interface position. A key
assumption in the Free Surface model is that, within a given numerical control cell, the two fluids are
assumed to exhibit no slip, i.e. they have the same local velocity. In this respect, the Free Surface Module
is radically different than the Two Fluid Module, which solves separate momentum equations for each
component and does allow slip. Output specific to free surfaces is not available.
Summary Output
Under the Summaries tab, the user can activate the output of Mass Balance Summary data and the
output of the Force and Moment Summary Data. The activation of these summary outputs as well as the
data that is output for these summaries are explained further in the Flow Module, and the outputs have
the same meaning regardless of whether the VOF Module is active or not, since these summaries are
determined entirely by the flow field and the fluid properties.
Graphical Output
Under the Graphics tab, you can select the variables to output to the graphics file (modelname.DTF).
These variables will then be available for visualization and analysis in CFD-VIEW. Activating the Free
Surfaces (VOF) Module allows output of these variables:
158
Modules
Variable Units
3
Fluid1 Density kg/m
3
Fluid2 Density kg/m
Fluid2 Enthalpy
J/kg
(only if Heat Transfer is activated)
Surface Curvature
1/m
(only if Surface Tension is activated)
159
CFD-ACE_V2010.0_Modules_Manual_Part2
160
Modules
should probably avoid performing any serial calculation with the VOF module using a grid with 100,000 or
more cells, and should make every effort to prevent the number of cells from increasing above 10,000.
Keeping the number of cells small is not always feasible nor easy for a given, built-up, complex system or
complex geometry, especially because there are constraints on the minimum grid density required to
resolve interface shapes, as discussed in more detail below. Because of this, you should consider a
component-by-component-modeling approach for studying a complex system, and should consider the
applicability of the available reduced models (such as the Hele-Shaw model, for example) whenever
possible. As always, in making simplifications, judgement, and a good understanding of the problem are
needed to avoid making a simplification that significantly changes the solution from the true one.
Another implication of using the geometrically-based flux-calculation approaches described above is the
very high susceptibility of the solution quality to the grid-quality in general, and to the following
aspects of it in particular: (i) orthogonality; (ii) smoothness; and (iii) aspect ratios above 10. These
influences are especially important in any region through which the interface will pass during a
computation. Poor orthogonality most directly affects the accuracy of the flux calculations, while poor
smoothness and high aspect ratios most directly affect the accuracy of the surface-reconstruction. The
most ideal grid for a 2D VOF calculation is a square domain with square cells; the most ideal grid for a 3D
VOF calculation is a cubic domain with cubic cells. Grids should be designed and built to approach this
ideal as closely as possible. Using the multi-block grid-generation technique, it is usually possible to
create a grid of acceptable quality even for a complex geometry.
Another important consideration affecting the design of a grid for a VOF computation is the number of
cells required to resolve the shape of an interface. A minimum of 4-5 cells are required across a gap
(such as the gap between two flat plates, or the gap across the width of a channel) to provide adequate
resolution of the shape of the interface in that gap. Using 8 cells across a gap provides ample resolution.
If the necessary resolution is not available, the accuracy of the curvature calculation will be adversely
affected.
1 / Weber
(relative to momentum force)
Number:
Froude/Weber #
(relative to gravity forces)
:
Reynolds
(relative to viscous forces)
/Weber # :
where σ is the coefficient of surface tension, ρ is the density, r is a characteristic radius of curvature, U is
a characteristic velocity, g is the gravitational constant, L is a characteristic length-scale for the viscous
effects, and µ is the viscosity.
Surface-tension-driven flows are source-term dominated, and therefore more difficult to converge. If
instabilities or convergence difficulties are encountered when activating the surface-tension model, the
161
CFD-ACE_V2010.0_Modules_Manual_Part2
problem may be re-tested with a reduced value of the surface-tension coefficient and even with the
surface-tension coefficient brought down to zero), just to see if the problem is specifically caused by a
very dominant surface-tension effect. However, if surface tension is important, the full value of the surface
tension coefficient must be eventually used in the computation to obtain the correct solution. This
treatment is analogous to the testing of a natural-convection problem with a reduced gravitational
constant.
.
Reducing the CFL Number generally increases the accuracy and stability of a computation, but
also increases the required number of time-steps for a given total integration time.
The optimal value of the CFL Number depends on whether the surface-tension model is activated or not:
if it is not, then a value of 0.2 is usually the most appropriate; if it is, then a value of 0.1 is usually the most
appropriate. For all computations, the use of the automatic time-stepping (as opposed to fixed time-steps)
is strongly recommended. This allows the code to automatically determine the appropriate time-step size
from the given CFL Number and the velocity field. If the automatic option is selected, then the maximum
and minimum time-step sizes must also be appropriately chosen to not interfere with the automatic
determination of the time-step. Establishing suitable values may require several trial runs to be made for
the problem, but it is often best to set the minimum and maximum time-step sizes to respectively 0 and 1
(where the latter is a very large number for most surface-tension-driven flows). The initial CFL Number
must also be sufficiently low to prevent failure of convergence during the first time-step, but not too low
that the code wastes several time-steps increasing it to an appropriate value. The appropriate initial value
clearly depends strongly on the problem. Values as low as 1.0E-10, and as high as 1.0E-02 are used on
a regular basis. Again, the most suitable value for the initial time-step size can easily be found by trial and
error.
If the interface appears to become stuck or if it seems to move with a displacement far less than given by
the value of CFL * Characteristic Cell Length, then there must be something holding the time-step back.
The possible causes for this, and the corresponding solutions are:
1. A mixed cell with a minute fraction of one of the two fluids is located in a high-velocity region. This
is typically the result of the creation of flotsam and jetsam (see above) in that cell, where fluid is
being stripped from the interface, at least partly as a consequence of a numerical artifact.
• Solution A: Use a smaller CFL Number, and start from an initial condition without flotsam and
jetsam
• Solution B: Use capillary-wave damping
• Solution C: Use a better-quality grid
2. There is a high velocity tangential to the surface.
• Solution A: Determine if the high tangential velocities are physical and relevant to the problem. If
they are, there is little that can be done. If they are not, then:
o Use capillary-wave or viscous damping or
162
Modules
o Examine and modify the boundary conditions that may be generating the high winds
Free Surface Module Special Discussion
This section describes the form and implementation of the various Newtonian and non-Newtonian
rheology models that were incorporated into the VOF scheme.
A viscosity model is considered Newtonian if the local viscosity depends only on the instantaneous, local
values of any primitive variable or combination of primitive variables. These variables could include
temperature, composition, extent of relaxation toward an equilibrium state, or pressure, for example. If the
viscosity depends on any quantity that is obtained from spatial gradients or integrals of the primitive
variables, such as the strain-rate tensor or any of its invariants, then the viscosity model is considered
non-Newtonian. A viscosity model is also considered non-Newtonian if the viscosity depends on historical
values or on temporal rates of change of any combination of primitive variables or on temporal rates of
change of the spatial derivatives or integrals of any combination of primitive variables.
A commonly used sub-category of non-Newtonian viscosity models is that of the so-called Generalized
Newtonian Fluid. The characterizing feature of this sub-category is that the value of the viscosity depends
on the current values of only the primitive variables and possibly their spatial derivatives and integrals,
with no dependence on past values or temporal rates of change. This extension enables the Generalized
Newtonian Fluid to represent, in addition to all the viscous effects of a Newtonian Fluid, the effects of
shear-rate. Even this extension, however, remains quite limited in its ability to represent the range of
physical behaviors typically associated with generic non-Newtonian and polymeric materials, such as rod-
climbing, jet-swell, recoil, or any other viscoelastic or time-dependent effects, or effects that are caused
by non-zero values for the normal components of the stress tensor.
Using the above definitions, all viscosity models implemented in CFD-ACE+ are either for Generalized
Newtonian Fluids or for Newtonian Fluids. In particular, the Cross, the Power Law, and the Carreau
models are all for GNFs, while the Sutherland, the Modified Cross, and the Square-Root of Temperature
Law models are for Newtonian Fluids. The two constant viscosity options are also for Newtonian Fluids.
With viscosity denoted by µ, the specific formulations for each of the viscosity models implemented in the
VOF methodology are:
Constant Kinematic Viscosity Model:
µ = ρν (14-8)
where ν is the constant kinematic viscosity, and ρ is the density. If ρ varies with space or time, µ will also
vary accordingly.
Constant Dynamic Viscosity Model:
µ=κ (14-9)
where k is a constant. Thus, the viscosity remains constant in space and time, regardless of any other
variations in properties or variables.
Sutherland's Law Model:
(14-10)
163
CFD-ACE_V2010.0_Modules_Manual_Part2
where K, n 1 , n 2 , and T* are constants, is the second invariant of the strain-rate tensor, and T is the
temperature in K.
Carreau Law Model:
(14-13)
where µ ο and µ ∞ are respectively the zero shear-rate and infinite shear-rate viscosities, K, a, and n are
model constants, and is the local shear rate.
Free Surface Module References
Darwish, M. and Moukalled, F., Convective schemes for capturing interfaces of free surface flows on
unstructured grids, Numer. Heat Transfer B, vol. 49, pp. 19-42, 2006.
Harvie, D. J. E. and Fletcher, D. F., A New Volume of Fluid Advection Algorithm: The Defined Donating
Region Scheme, Int. J. Numer. Meth. Fluids, vol. 38, pp. 151-172, 2001.
Hirt, C.W. and Nichols, B.D., "Volume of Fluid (VOF) Method for the Dynamics of Free Boundaries."
Journal of Computational Physics, Vol. 39, pp.201-225, 1981,
Kothe, D.B., Rider, W.J., Mosso, S.J., and Brock, J.S., "Volume Tracking of Interfaces Having Surface
Tension in Two and Three Dimensions." AIAA Paper, 96-0859, 1996.
Leonard, B. P., The ULTIMATE Conservative Difference Scheme Applied to Unsteady One-Dimensional
Advection, Comput. Meth. Appl. Mech. Eng., vol. 88, pp. 17-74, 1991.
Moukalled, F. and Darwish, M., A New Bounded-Skew Central Difference Scheme-Part I: Formulation
and Testing, Numer. Heat Transfer B, vol. 31, pp. 91-110, 1996.
Noh, W.F. and Woodward, P.R., "SLIC Simple Line Interface Method)." In A.I. van de Vooren and P.J.
Zandbergen, editors, Lecture Notes in Physics 59, pp. 330-340 1976.
Noh, W. F. and Woodward, P., SLIC (Simple Line Interface Calculations), Lecture Notes Phys., vol. 59,
pp. 330-340, 1976.
Puckett, E. G., A Volume of Fluid Interface Tracking Algorithm with Applications to Computing Shock
Wave Rarefraction, Proc. Fourth Int. Symp. on Computational Fluid Dynamics, University of California
at Davis, Davis, CA, 9-12, September, 1991, pp. 933-938.
Rider, W.J., Kothe, D.B., Mosso, S.J., Cerrutti, J.H., and Hochstein, J.I., "Accurate Solution Algorithms for
Incompressible Multiphase Fluid Flows." AIAA Paper, 95-0699.1995.
Rudman, M., Volume Tracking Methods for Interfacial Flow Calculations, Int. J. Numer. Meth. Fluids, vol.
24, pp. 671-691, 1997.
Ubbink, O. and Issa, R., A Method for Capturing Sharp Fluid Interfaces on Arbitrary Meshes, J. Comput.
Phys., vol. 153, pp. 26-50, 1999.
Unverdi, S. O. and Tryggvason, G., A Front Tracking Method for Viscous, Incompressible Multi-fluid Flow,
J. Comput. Phys., vol. 100, pp. 25-37, 1992.
Young, T., "Philosophical Transactions of the Royal Society of London." Volume 95, Page 65, 1805.
Youngs, D. L., An Interface Tracking Method for a 3D Eulerian Hydrodynamics Code, Tech. Rep.
44=92=35, Atomic Weapons Research Establishment, Harwell, UK, 1984.
164
Modules
Plasma Module
165
CFD-ACE_V2010.0_Modules_Manual_Part2
166
Modules
167
CFD-ACE_V2010.0_Modules_Manual_Part2
capability remains supported for backward compatibility and advanced users. This feature can be
activated under MO -> Advanced -> Plasma.
The fixed setting of different parameters is used to automatically generate LUTs as shown in figure 1.
These fixed settings for the automatic Lookup Table generation process should be applicable to a large
class of problems and runs a one-cell case (similar to the manual process). Note that the inputs for
electric field excitation frequency and the range and number of points for Ne and E0 specified in MO -->
Adv in the user’s case will overwrite the corresponding settings in figure 1.
168
Modules
(Note that the settings for electric field frequency and the range and bins for Ne and E0 are
overwritten from the user case inputs in MO --> Adv.)
When this automatic LUT generation feature is used, a subdirectory casename_LUT.LUT is created in the
user’s current case directory and the one-cell problem is solved using the settings described earlier to
generate Lookup Tables files with the extension "IN", similar to the existing manual procedure. Based on
the user’s selection of table in the LUT manager window (see Plasma>Model Settings>Run), the
appropriate lookup table information is automatically updated into the user case (DTF) file. Once a lookup
table is "applied" to the original user’s case case.DTF file, reactions in the volumetric mechanism that had
cross-section data will have their cross-section data supplanted with lookup table data. A copy of the
original case.DTF called case.backup.DTF file is created when lookup tables are generated, in case the
user needs to recover or reuse the original file. Once the automatic update occurs, the pop-up LUT run
manager window disappears and the solver run begins as usual. A couple of .DTF files, one each of
.OUT, .RSL and .DAT files are written as well in the subdirectory but are not used.
Advanced users who need settings other than those set in Figure 1 should follow the existing manual 2-
step procedure instead of this automatic generation feature.
Limitations
1. The total pressure used in automatic LUT generation is Reference Pressure + pressure specified
in the IC tab - it is assumed all the plasma VCs have the same initial pressure.
2. Lookup Tables are generated at the beginning of the run only; currently the feature does not
support recalculating these tables during a run if the mixture composition changes significantly
due to chemical kinetics such as extensive fragmentation of the feedstock gases.
3. Parallel run is not supported currently using this feature.
See Also:
Plasma-> Implementation-> Model Setup and Solution -> Model Options->Advanced
Plasma-> Implementation-> Model Setup and Solution -> RUN
Plasma-Theory
Plasma Module
Theory-Introduction
Plasma is often called the fourth state of matter. It is a quasi-neutral mixture of charged species in which
the density of negatively charged species is well balanced by the density of positively charged species.
An equilibrium plasma is characterized by its temperature and pressure. In such a plasma, the
temperatures of all species are equal and the densities of species are determined by Saha’s equation.
Some industrial plasmas, including DC arcs and RF plasma torches operating at atmospheric pressure,
are close to thermodynamic equilibrium. Several processes occurring in the plasma lead to deviation from
the thermodynamic equilibrium.
In gaseous plasmas, the charged species are created during gas breakdown. During the breakdown, the
primary electrons, which are always present in a small quantity in any gas, are accelerated by external
forces (usually applied electric fields). These energetic electrons collide with the heavy species. This
brings about ionization of the neutral particles, producing new electrons and ions. Since the ionization
fraction of most gases is less than a few percent (except in the case of fusion plasmas), electrons
experience collisions mostly with neutral gas species. The very large disparity between the mass of an
electron and the heavy species makes the energy transfer during elastic collisions between electrons and
heavy particles very inefficient. At low gas pressures, the temperature of the heavy species (ions, and
neutral atoms) is close to room temperature. Such plasma is called cold plasma to distinguish it from the
hot plasma of thermonuclear fusion. However, the electron temperature in cold plasma can exceed the
temperature of heavy particles by two orders of magnitude and the plasma represents a system that is far
from thermodynamic equilibrium.
169
CFD-ACE_V2010.0_Modules_Manual_Part2
The non-equilibrium nature of gaseous plasma manifests itself in several ways. At low gas pressure, the
plasma chemistry, in most part, is dominated by electron induced reactions. Since electron temperature,
Te, can exceed the gas temperature, T, by two orders of magnitude (see figure below). These two
temperatures gradually merge together with increasing gas pressure and the plasma approaches the
local thermodynamic equilibrium (LTE). The pressure at which T approaches Te depends on the gas
composition. There may be compositions where Te and T merge together above 760 torr (1 atm).
Electron and gas temperatures in a DC discharge of mercury and rare gas mixture at different
gas pressures and same current.
The walls of the discharge chamber and electrodes serve as sinks of charged particles; electrons are
absorbed at the walls, and ions are converted into neutrals. As a result, at low gas pressures, plasma
density is spatially non-uniform exhibiting a maximum at the center and a minimum towards the walls.
At low gas pressures, many characteristics of the plasma depend on the product pL, where p is the gas
pressure and L is a characteristic size of the device chamber. At pL=10-3-10 Torr cm, the charged
particle generation in the bulk is balanced by the loss of charged particles to walls of the discharge
chamber due to diffusion. At steady state, each electron needs to generate a new electron-ion pair before
reaching the walls in order to maintain the dynamic balance. Since electrons tend to diffuse faster than
ions, a small net positive charge and an electrostatic field build up in the plasma. This process balances
the fluxes of charged particles and maintains the quasi-neutrality. At high pressures, there might be a
local balance of particle production and loss at each point of discharge volume.
The quasi-neutrality of plasma is violated near boundaries where a space-charge sheath is created to
ensure the balance of currents. As a result, the bulk plasma region with a weak electrostatic field is
separated from the walls/electrodes by thin sheaths with strong electric fields.
The theory on CFD-ACE+ Plasma Module is described in the following sections:
• Theory-Types of Discharges
• Theory-Basic Equations
• Theory-Solution Method
• Theory-Plasma Chemistry
Plasma Module
Theory-Types Of Discharges
170
Modules
Depending on the way electromagnetic power is coupled to the plasma, one can distinguish between
various kinds of discharges according to the frequency range of the electromagnetic sources:
• Direct Current (DC) glow discharges
• Radio Frequency (RF) discharges which can further be subdivided into capacitively coupled and
inductively coupled
• Microwave discharges operating in the GHz frequency range including electron cyclotron
resonance (ECR)
Capacitively coupled plasmas are created by applying RF voltage to electrodes that can be immersed into
plasma or separated from the plasma by dielectric layers. The distinguishing feature of CCP (sometimes
called plasma diodes) is the presence of oscillating sheaths near electrodes where displacement current
dominates conduction current.
The ICP is created by an RF current passed through a coil that generates an RF magnetic field. This
magnetic field induces a circular electric field whose lines of force are concentric with the primary coil
current. The inductive discharges are, in principle, electrodeless.
In practice, different mechanisms of plasma generation can co-exist in one reactor. Frequently, the
plasma is maintained by inductive power coupling, whereas a small capacitive coupling exists near a
biased electrode immersed into the plasma to extract ions.
Approximately, the plasma density is proportional to the square root of absorbed power for CCP and is
linear with power for ICP. Thus, ICPs are more efficient at higher plasma densities. Conventional non-
-3
magnetized CCPs are limited to a density of 1010 cm whereas the density limitations in ICP are of the
-3
order of 1012 cm . The upper density limit in ICP is dictated by power absorption in the skin layer whose
-1/2
thickness scales electron density as ne .
Currently CFD-ACE+ can be applied to simulate the discharges in the frequency ranges of DC, ICP and
CCP. Further development is under way for microwave type of discharges.
Theory-Basic Equations
Plasma Module
Theory-Basic Equations-Introduction
Electrons are the key component of plasmas. The Plasma Module solves equations for electron number
density, drift velocity and temperature. The Chemistry, Heat, Flow, Electric and Magnetic Modules in
CFD-ACE+ are enhanced to accommodate features required for plasma simulations. The Plasma Theory
Basic Equations section includes:
• Basic Equations-Electrons
• Basic Equations-Electromagnetics
• Basic Equations-ICP Simplification
• Basic Equations-Ions and Neutrals
• Basic Equations-Gas Flow
• Basic Equations-Heat Transfer
• Basic Equations-Reaction
Plasma Module
Theory-Basic Equations-Electrons
171
CFD-ACE_V2010.0_Modules_Manual_Part2
(15-1)
Here n e is the electron number density, Γ e is the electron density flux, and S is the source of electrons
produced or consumed in chemical reactions. The drift diffusion approximation is used for the electron
density flux:
(15-2)
where µ e is the electron mobility, D e is the diffusion coefficient and ϕ is the electrostatic potential. The
electron temperature T e is found from the electron energy balance:
(15-3)
where the energy transfer is due to convective flux and thermal diffusion, with coefficient
The last term on the RHS of Eq. (15-3) accounts for the electron energy loss due to the electron induced
reactions. The power density, P, stands for the energy absorbed by electrons, such as Joule heating,
inductive heating and external heading, P ext . External heating can be used, for instance, to simulate
collision-less heating.
(15-4)
(15-5)
(15-6)
(15-7)
with ε 0 , ν m , ω ep , ω as permittivity of free space, electron collision frequency, electron plasma frequency
( ) and angular frequency of the induction power source. The electrostatic potential,
φ, and vector magnetic potential, , are provided by the Electric and Magnetic modules in CFD-ACE+
A new electron stochastic heating model has been implemented especially for electrons in low pressure
and high RF frequency conditions. A novel "self-aware" approach is implemented, which allows ACE+ to
automatically detect regions in the plasma where stochastic sheath heating occurs. The heating power is
defined as:
172
Modules
(15-8)
where is the integrated electric field (or the electric potential difference) from the local plasma cell to the
powered electrode, m is the electron mass, e is the electron charge, ω is the highest angular driving
frequency at the powered electrode, and
is the electron temperature.
Note: In V2010.0 and later, the new electron stochastic heating model described above is activated by
default. The old heating model was defined as:
(15-9)
(15-10)
where f 0 (ε) is the 0 -order term of the electron energy distribution function and the electron density factor
th
(15-11)
The electron diffusion coefficient is , and mean free path . The electron
velocity can be expressed in terms of kinetic energy, ε (in Volts), as
(15-12)
173
CFD-ACE_V2010.0_Modules_Manual_Part2
(15-13)
where n s , , and are the density, elastic electron momentum transfer cross section and
inelastic cross section for neutral species s, respectively. is the electron-ion momentum transfer
(Coulomb) cross section
(15-
14)
ε 0 is the permittivity of free space and γ E is the ration of conductivity in a fully ionized plasma to that in a
Lorentz gas. For a gas containing only singly charged ions, γ E = 0.582.
Remark: In the CFD-ACE+ implementation (prior to v2006) of Eq. (15-13), only the electron momentum
transfer (i.e., ) is considered. The inelastic cross-section is in general much smaller than that of
elastic momentum transfer and thus can be neglected. However, the Coulomb collision term (i.e., the last
term on the RHS of Eq.(15-13)) may become substantial when electron density is high. The new
formulation for the collision frequency in Eq.(15-13) will affect diffusion coefficient for electron kinetics,
and thus Eqs.(15-9&10) for mobility and diffusivity. Starting from v2006, the inelastic and Coulomb
contributions are taken into consideration by default. To de-select the inelastic collision contribution, use
script command (under Tools/Special DTF Update in GUI): "ecoll_freq_inelastic int 1 0”. To
de-select the Coulomb collision contribution, the script command is "ecoll_freq_eion_Coulomb int
1 0”.
When electron kinetic model is activated, the electron mobility and diffusivity are calculated as Eqs.(15-9)
and (15-10), respectively for plasma region unconditionally (overtaking the GUI selections on the
"Electron Collision Frequency” and "Electron Diffusion Coefficient”). At least one step of electron elastic
(momentum transfer) reaction with cross-section needs to be included in the reaction mechanism.
Maxwellian Distribution: : In the case of plasma fluid model (electron kinetic model is NOT activated),
the electron energy distribution is assumed as Maxwellian , in which energy and
mean temperature are all in Volts. The Eqs (15-9 and 10) now become
(15-15)
Notice that implemented prior to v2006, the electron mobility and diffusivity were calculated as
(15-16)
(15-17)
The electron mobility and diffusivity by Eq.(15-15) will be noticeably different from those by Eq.(15-16).
Since v2006, Eq.(15-15) is used as default (see GUI selection of "Function of Te” for electron
collision/mobility). To switch back to the old Eq.(15-116), use script command "mobe_De_MAXWELL_edf
int 1 0”.
174
Modules
A simple alternative to estimate the electron collision frequency is also provided in CFD-ACE+:
(15-18)
where constant value ν 1 is specified in CFD-ACE-GUI (2E+07 as default). P and T are gas
thermodynamic pressure (in Pa) and temperature (in Kelvin), respectively.
Boundary Conditions for Electrons
Assuming no reflection, the electron flux normal (subscript n) to the electrodes or walls is given by:
(15-19)
where the electron thermal velocity and γ is the is the secondary emission
coefficient. q i , Γ i,n are (ion) species charge unit and number flux. The electron energy flux to the
electrodes and walls is given as:
(15-20)
Plasma Module
Theory-Basic Equations-Electromagnetics
(15-21)
While Dirichlet and Neumann conditions are normally applied at the boundaries to φ, dielectric and
external circuit boundary conditions are added for plasma (CCP option).
Boundary Conditions for the Dielectric
At dielectric surfaces, the net surface charge σ s is obtained from the particle fluxes as:
(15-22)
The boundary condition for the electric potential on a dielectric surface becomes:
(15-23)
where is the normal component of the field on the surface. Subscript 2 represents the solid
dielectric material. The norman , n, is pointing from the plasma to the dielectric regions.
Boundary Conditions for the External Circuit
175
CFD-ACE_V2010.0_Modules_Manual_Part2
At the boundary that is connected to the external circuit, the current is obtained from the plasma solution
and potential is returned from the external circuit solution. Consider a simple external circuit consisting of
the power source and one capacitor:
(15-24)
(15-
25)
Plasma Module
Theory-Basic Equations-ICP Simplification
When the ICP option is chosen, the quasi-neutrality condition can be used instead of solving Eq. (15-1).
The electron number density is obtained from the algebraic relation:
(15-26)
where n i denote the ion number densities. Also, in this case, instead of solving the Poisson equation, the
electrostatic potential is calculated from charge conservation using quasi-neutrality constraint:
(15-27)
where is the total conduction current density due to both electrons and ions. In many cases, where no
net current is present, Eq. (15-27) ca be reduced to . Neglecting the current contribution from ions,
this ambipolarity assumption results (using electron drift-diffusion) in the electrostatic field:
176
Modules
(15-28)
which implies that electrons are in electrostatic equilibrium with the field E, if net conduction current is
zero.
Plasma Module
Theory-Basic Equations-Ions and Neutrals
The heavy particles (ions and neutrals) in plasma are solved from the species transport equations
implemented in the Chemistry module. For species transport in plasma, ion drift effect needs to be
considered. The mass flux of each species can be expressed as follows:
(15-29)
where Y i is the species mass fraction, ρ is mass density, D i is the diffusion coefficient. The first term is
due to concentration gradient, the second term by the drift velocity, and the third term is the correction to
ensure species flux conservation.
Ion Drift-Diffusion Approximation
The drift velocity is simplified as the linear function of ion mobility and electric field, and can be expressed
in following way to ensure the drift flux conservation:
(15-30)
(15-
31)
where the ion collision frequency and ion mobility is related as and the drift velocity is then
expressed as:
(15-32)
Ion Mobility
Transport properties of ions are calculated in two different ways. For a rough estimate, the diffusion
coefficient of an ion is assumed to coincide with the diffusion coefficient of corresponding neutral, and the
mobility is calculated using Einstein’s relation.
(15-33)
177
CFD-ACE_V2010.0_Modules_Manual_Part2
(15-34)
where α j is the polarizabilty measured in and m r is the reduced mass of the ions and neutrals. For
standard mobility in a parent gas:
(15-35)
where σ ex,j is the charge exchange cross-section measure in . Blanc's las is used to calculate ion
mobility in a gas mixture
(15-36)
where X j is the mole fraction of species j. Using the same Einstein relation of Eq. (15-33), the ion
diffusivity can also be derived from the ion mobility.
Surface Reaction Boundary Conditions
The fluxes of gas-phase species at reacting surfaces are determined by the flux balance equation. For ion
surface recombination , the flux balance at the surface becomes:
(15-37)
where J i,n and are species normal mass flux and surface production rate, respectively. The surface
reaction rate is typically expressed using the sticking coefficient model, with sticking probability on the
characteristic flux being:
(15-38)
w
where C i is the surface value of species concentration to be solved from the flux balance equation. The
characteristic velocity can be expressed for neutral and ion species according to how sheath is modeled:
(15-
39)
In Eq. (15-39), the characteristic velocity is thermal velocity for neutral. For ion, the characteristic velocity
is the Bohm velocity at the pre-sheath if sheath is not resolved (using ICP option). But if sheath is
resolved (using CCP option) actual ion drift velocity is used to ensure the supersonic (free exit) condition.
178
Modules
Plasma Module
Theory-Basic Equations-Gas Flow
The presence of plasma also modifies the flow module which solves the momentum equation for the
mass averaged velocity. The static pressure, p, is a sum of the partial pressures of all species, including
the contribution of electron pressure:
(15-40)
where k is the Boltzmann constant. For weakly-ionized plasmas, the ion contribution to the total pressure
-3
can be neglected since the density ratio of ions to neutrals does not exceed 10 . However, the electron
2
pressure may be substantial since the ratio T e /T is in the order of 10
Plasma Module
Theory-Basic Equations-Heat Transfer
The Heat module is enhanced to accommodate the heat sources related to Plasma. The energy equation
for heavy particles can be expressed in terms of gas temperature as:
(15-41)
The gas heating sources are added to account for the following plasma phenomena: 1) electron elastic
collisions; 2) ion Ohmic heating; 3) ion surface recombination and kinetic impact.
Electron Elastic Collisions
The last two terms on the RHS of Eq. (15-41) account for the heat release due to non-electronic
reactions, as well as electron momentum transfer steps (electron elastic collisions). The heat of reaction
for step r is calculated as:
(15-42)
where m (r) is the molecular mass of neutral species associated with momentum transfer step r. and
represent the stoichiometric coefficients of the products and reactants, respectively. The associated
mean electron energy (in volts) can be expressed as:
(15-43)
where reaction rate constant is given by Eq.(15-51). For Maxwellian f 0 , the mean can be expressed as
. Prior to version 2006, the energy transfer for electron elastic collision was calculated as
(15-44)
179
CFD-ACE_V2010.0_Modules_Manual_Part2
in which apparently the factor of 3/2 was missing as compared with Eq.(15-42). If the electron kinetic
module is NOT activated, it is allowed to switch back to Eq.(15-44) by updating the DTF file using the
following string "mom_transf_MAXWELL_edf int 1 0”.
Ion Ohmic Heating
The gas heating source term Q ion due to the ion-neutral collision can be expressed as ion Ohmic heating:
(15-45a)
where V i are µ i ion velocity and ion mobility, respectively. If the ion velocity is simplified as drift velocity,
then
(15-45b)
(15-46)
where J i,n and h i are species flux and enthalpy, respectively. The first term already considered in v2004,
accounts for the chemical heat release due to the surface reaction (including ion surface recombination).
The second term is due to ion kinetic energy release on surface impact and
(15-
47)
where is the ion potential gain (calculated using the sheath model) across the sheath.
Stochastic Heating
The physics of "stochastic heating" plays an important role in CCP discharges, especially when lower
pressures and/or higher frequencies are used. Electrons are "heated" via 2 major mechanisms in
capacitively coupled discharges - Ohmic heating, which is a collisional mechanism; and Stochastic
heating, which is a collisionless mechanism. Momentum transfer from high-voltage moving sheaths leads
to stochastic heating. Stochastic heating is localized in the sheath area.
In V2010.0, a new model for stochastic heating in the Capacitively Coupled Plasma model has been
developed. In this new model, an analytical formula published by Lieberman, et al, is used and a novel
"self-aware" approach is implemented, which allows ACE+ to automatically detect regions in the plasma
where stochastic sheath heating occurs. The new implementation allows ACE+ to detect the sheath
regions adjoining the sinusoidal powered surfaces. The stochastic heating power is then computed using
the equation 15-8.
This can be activated under the MO -> Advanced tab when CCP plasma is selected. For mor details, see
the Model Setup and Solutions > Model Options > Advanced page. The output of "Stochastic Heating" in
-3
units of W-m will be written out to the DTF file and can be seen in CFD-VIEW. This occurs only when the
option for electron stochastic heating is turned on in the MO -> Advanced tab. The user can get a better
idea of how power is going into stochastic heating, and what regions of the discharge are affected.
180
Modules
Plasma Module
Theory-Basic Equations-Reaction
Electron induced gas phase reactions are introduced in the Chemistry module to accommodate plasma
reactions. The rates of electron induced reactions depend on electron temperature in eV, while the rates
of reactions among heavy species depend on gas temperature in Kelvin.
Three types of electron-induced reactions can be specified. The first type is of the form:
(15-48)
where A and B are different species (neutral, ion, excited). In CFD-ACE-GUI Database Manager, the
properties of these species are specified (in Gas Species), and the electron energy loss specified (in
Volume Reaction) should be set to zero. These reactions define the source term in the Boltzmann
equation solved by the Kinetic Module, or the species source terms in the Chemistry Module and energy
loss term in the electron energy equation in the Plasma Module.
The second type of reactions is of the form:
(15-49)
where A denotes the same species. These reactions describe quasi-elastic collisions such as excitation
of molecular vibration and rotations. You must input the energy loss for these reactions in Volume
Reaction panel Database Manager. This type of reactions is used when you choose not to solve for the
transport of these individual states (species) in the chemistry module, but have to include the important
electron energy loss in these processes to obtain realistic electron temperature. These reactions produce
a contribution to the energy convection term in the Boltzmann equation, or to the electron energy
(temperature) balance in the plasma module.
The third type of reaction is elastic collisions. These reactions are of the same type as Eq. (15-49) but the
energy loss is set to zero in the Volume Reaction panel Database Manager. In this case, CFD-ACE+
calculates the energy loss assuming elastic collisions from the mass ratio of colliding particles.
If you activate the Kinetic Module, you have to specify the cross-section for each electron induced
reaction.
Electronic Reaction Rate: In the Gas Phase Plasma Chemistry, the electron induced reaction rate can be
written as:
(15-50)
where n (j) is the neutral species density associated with reaction step j. The rate constant is
(15-51)
where the electron energy distribution function f 0 can be obtained from Kinetic solution or simply assumed
as Maxwellian, and σ j (ε) is the cross-section. In the Reaction Manager, the following options are available
to express the reaction rate constant K:
• One can define K in Arrhenius form as function of the electron temperature Te (expressed in eV)
• By selecting Cross-Section option, a window will pop up where one can import collision cross
sections by reading an ASCII file containing pairs of numbers for energy ε (in eV) and σ(ε) (in A ).
2
The C-S information for this reaction step is then saved by the Reaction Manager.
181
CFD-ACE_V2010.0_Modules_Manual_Part2
• If the Kinetic module is activated, the electron energy distribution function f0 is calculated by
solving the Boltzmann equation (otherwise as Maxwellian by default) and the electron induced
reaction rates are calculated according to Eq. (15-50).
Surface Plasma Chemistry
Surface reaction is a very important aspect of plasma process, since it involves phenomena such as
material etching, deposition, as well as simple ion surface recombination and other neutral reactions. The
typical reaction can be represented by the sticking model, in which the surface reaction rate is expressed
as the limiting flux with the sticking probability. The details are discussed in Surface Reaction Boundary
Conditions.
Another type of plasma surface reaction has ion energy dependency for the reaction rate and yield.
Consider ion induced surface reaction:
+
in which ion species A recombines to become neutral species A and reacts with species B i . The sign #
indicates ion energy enhanced yield factor. The ion energy dependent rate/yield can be expressed as:
+
where E th , f, g, y, z, and h are parameters from the mechanism, and the mean ion energy for species A
under RF bias can be calculated from Ion Energy Distribution Function (IEDF) by sheath model:
Plasma Module
Theory-Solution Method
The solution methodology of the plasma governing equations depends on the selection of the ICP or the
CCP option.
ICP
If the ICP option is chosen, the Plasma module solves for the electron energy balance equation and uses
the quasi-neutrality condition to calculate the electron density and the electrostatic field. Normally steady-
state equations are solved. ICP option is intended for the simulation of the inductively coupled plasma, in
which the electron is heated by the inductive heating power as described in Eq.(15-6). The magnetic field
and the induced electric field are solved in the Magnetic Module.
CCP
Two distinguishing features of CCP consist of a) the necessity of a Poisson solver due to the large
thickness of the space charge sheath, and b) the necessity of transient simulations because electrons
respond to oscillations of the RF electric field. Thus, CCP simulations deal with the enormous disparity in
the spatial and temporal scales governing the gas flow, chemistry and transport of electrons.
182
Modules
183
CFD-ACE_V2010.0_Modules_Manual_Part2
The Implementation section describes how to setup a model for simulation using the Plasma Module. The
Plasma Implementation section includes:
• Grid Generation - Describes the types of grids that are allowed and general gridding guidelines
• Model Setup and Solution - Describes the Electroplating Module related inputs to the CFD-ACE-
Solver
• Post Processing - Provides tips on what to look for in the solution output
Plasma Module
Implementation-Grid Generation
Use CFD-GEOM to create geometry and generate structured or unstructured grids. For CCP option, due
to the exponential scheme, structured grid is highly preferred, and grid needs to be made orthogonal as
much as the geometry allows. Since sheath is included in the solution domain for CCP, this thin sheath
layer needs to be proper resolved (~ 3 to 5 cells). For ICP option, both structure and unstructured grids
are allowed, and there is no need to resolve the sheath (since it is not included).
Plasma Module
Implementation-Model Setup and Solution-Introduction
CFD-ACE+ provides the inputs required for the Plasma Module. Model setup and solution requires data
for the following panels:
• Problem Type
• Model Options
• Volume Conditions
• Boundary Conditions
• Initial Conditions
• Solver Control
• Output
Plasma Module
Implementation-Model Setup and Solution-Problem Type
The Problem Type [PT] menu provides options for specifying the nature of the problem, setting the
model’s time dependency and other global parameters. See Control Panel-Problem Type for details.
Select Plasma to activate the Plasma Module. Plasma enables a solution for electron density and
temperature, and activates automatically the Chemistry option enabling transport/chemistry of ions and
neutrals, and the Flow option enabling gas flow simulations. Plasma also requires using one (or both) of
the two electromagnetic field options. If ICP option is selected for inductively coupled plasma, the
Magnetic Module must be activated while Electric (conduction problem) Module is optional. If CCP option
is selected, the Electric Module (electrostatic problem) must be activated. Please notice that the plasma
option ICP really means the quasineutrality assumption, and this option is not limited to the inductively
coupled plasma. Likewise, the option CCP means a more general plasma fluid model, and it can be used
not only for capacitively coupled plasma, but many more discharges such as DC, ICP, etc.
The Plasma Module can be used with the following modules: Heat, Turbulence, User Scalar, Radiation,
Grid Deformation, and Stress.
Plasma does not support use with Spray, Free Surface, Two Fluid, Cavitation, Electroplating, and
Biochemistry.
184
Modules
Plasma Module
Implementation-Model Setup and Solution-Model Options-Introduction
Click the MO tab to activate the Model Options Panel. See Control Panel-Model Options for details. For
the Plasma module, the following tabs should be active: Shared, Flow, Chem, Plasma, and either Magnet
or Electric. The Plasma Model Options section includes:
• Model Options-Positive Column
• Model Options-Global Model
• Model Options-Shared
• Model Options-Plasma
• Model Options-Magnet
• Model Options-Electric
Plasma Module
Implementation-Model Setup and Solution-Model Options-Shared
The Shared tab enables you to enter data for inductively or capacitively coupled plasma.
ICP
ICP may be solved as either a transient or steady state model. The Steady State mode is recommended
for an ICP reactor maintained at a high frequency in a continuum power regime. Use the Transient option
for low frequency ICP and pulsed power operating regimes. For simulations of ICPs driven by multiple
coils with different frequencies, you must also use the Transient option.
The Transient option requires selecting a method for calculating the time step and defining a number of
time steps for the simulations.
CCP/Global
Select the Transient option for CCP/Global. For "Transient Time Step”, specify the time step size ∆t shared
and the step number. For a typical RF CCP discharge, it may require several thousand RF cycles to
reach the periodic steady-state. Therefore the step number will be the product of cycle# and time-
step#/cycle.
Plasma Module
Implementation-Model Setup and Solution-Model Options-Global Model
Species mass fractions are solved for heavy particles such as ions and neutrals. The Gas Phase
Reaction needs to be activated for plasma volume reaction source. The Database Manager has a built-in
library under the Volume Reactions folder on plasma reaction mechanisms for some simple gases. These
libraries are suggested as starting point for design of user specific reaction mechanisms.
3
For plasma reactions, the units of gas phase reaction rate are #/m -s. The electron induced reactions are
distinguished by the presence of electrons (denoted by e or E) in the reaction mechanism. For instance,
to specify direct ionization of argon atom by electron impact, enter in the Database Manager
AR + e −> AR + 2e
The rate of electron induced reactions depends on the electron temperature which is measured in eV.
Gas temperature is measured in K.
Please notice that the CHEMKIN option cannot be used with the Plasma module.
185
CFD-ACE_V2010.0_Modules_Manual_Part2
Plasma Module
Implementation-Model Setup and Solution-Model Options-Plasma
The Plasma tab enables you to enter data for Inductively Coupled Plasma (ICP), Capacitively Coupled
Plasma (CCP), and DC.
ICP Options
Under the ICP tab, select one of the following methods for specifying the Power Deposition:
Coil Current - This option ensures that the electric current in each coil will be specified explicitly in the
Volume Condition (VC) menu.
Absorbed Power - This option enables you to specify the total power absorbed in the plasma. For this
option, the initial current in the coils will be adjusted proportionally during the simulation to match the
computed power absorbed in the plasma with the value entered here in the Power input field.
Pulsed Power - This option is used for transient simulations of pulsed power ICPs. After selecting the
Pulsed Power option, you are prompted to specify the Pulse Period (in sec) and Duty Cycle in %.
CCP Options
CCP Time Step: Under the CCP option, one can specify (independently of the main time step) the CCP
time step size to resolve fast electron and electrostatic field oscillations. There are two equivalent
choices: the number of Time Steps per Cycle, or Explicit Time Step (in sec). If "Time Steps per Cycle” is
chosen and multiple cycles are involved, the "Cycle” by default means the fastest one. But this default
can be overwritten by activating "Cycle Freq Specified”. To properly resolve the field oscillations, it
is recommended to use at least twenty steps per smallest RF period.
External Circuit: For CCP discharge driven by RF source, normally an external circuit is placed between
the RF voltage source and the (metal) electrode (see Fig. 15-4 and Eqs. 15-24 & 25). It is required to turn
on the single capacitor external circuit, the only available option for current version of CFD-ACE+.
The capacitance is defined as per total areas of the electrode at which the same RF voltage source
(Sinusoidal BC of electric) is prescribed. For 2-D axisymmetric, the value of area corresponds to the
whole 360o of electrode. For 3-D, the area is for the actual electrode surface as defined (scale to the pie-
sector if that is the case).
CCP Power: For RF driven CCP discharge, in many cases the source voltage is not known a prior. In
stead, the power deposition in the plasma is a more appropriate parameter to specify. To control the
power deposition, check the box "Fixed CCP Power”, and enter the corresponding parameters. The total
CCP power is calculated as:
which is combined electron and ions Ohmic heating summed over whole reactor and averaged over one
CCP cycle. The Vrf of the RF source voltage (initially specified as sinusoidal BC for electric potential) will
be adjusted every cycle to meet the prescribed CCP power target as the CCP solution reaches the
periodic steady-state.
Power Target (in Watts): For 2-D axisymmetric, the value corresponds to whole volume of the reactor of
360o. For 3-D, the value is for the actual reactor geometry as defined (scale to the pie-sector if that is the
case).
Adjusting Vrf after # Cycles: The Vrf of the applied RF (ac) voltage source (at the sinusoidal BC for
electric potential) will be adjusted after the specified number of CCP cycles if this is a new run. For restart,
the adjustment starts after the first cycle.
186
Modules
Max dVrf per Cycle: The maximum change in Vrf allowed for each cycle. Use value not larger than one
volt for solution stability.
Relax Vrf: The relaxation used for Vrf update. Use value 0.1 or less for solution stability.
Ion Momentum: By default the drift-diffusion approximation is used for ions. But when ion collisions
become less dominant, it is desirable to solve ion momentum equation to include ion inertial effect. This is
realized by checking the box "Ion Momentum”.
Gas Heat: When Heat Transfer (Heat) is activated, by default the gas "Temperature Equation” (see Eq.
15-41) is solved for better convergence. Also the plasma heating mechanisms of "Joule Heating” by ions
(see Eq. 15-45) and "Surface Heating” by ion impact (see Eq. 15-46) are included as default. The Gas
Heat selections also apply to Plasma Model option ICP.
Inductive Power: When CCP option is used for the inductively coupled plasma, the inductive power
selections are the same as described above under "ICP options”. Since ICP discharge are generally at
very low pressure (~ 1 mTorr), ion momentum equation needs to be activated to better model the ion
transport across the (collisionless) sheath.
Stochastic Heating: When CCP option is used for low pressure high RF frequency discharges, electron
stochastic heating (described in Eq. 15-8) can be included by tab Adv/Plasma. To activate, check box
"Stochastic Heating” and enter the proper values for "Aptitude Co.” and "Electrode Gap”.
DC Options
For the DC model, specify the applied direct current and the initial value of the applied electric field.
Select explicit time step, with the time step size to resolve proper time scale.
Plasma Module
Implementation-Model Setup and Solution-Model Options-Positive Column
Using the CCP option, a special one-dimensional (using one cell in axial direction) Positive Column model
can be activated. In this model, the plasma is assumed to be axially uniform and is driven by the axial
current. Thus the governing equations (including both the electron density and energy equations) under
CCP option are reduced to one-dimensional in radial direction. To active the one-dimensional Positive
Column model, select Plasma Model as CCP from the MO panel, and check the box for Positive Column.
Then specify the applied axial current, initial value for the axial electric field and the relaxation parameter.
The electric field will be adjusted in the course of simulation according to:
where E* 0 is the value of E 0 at the current iteration, I 0 and I are the target and current values of the
discharge current, and ϑ is the relaxation parameters.
Plasma Module
Implementation-Model Setup and Solution-Model Options-Global Model
The Global models have been used to analyze low-pressure plasma reactors used in semiconductor
manufacturing applications. The global model assumes that the loss of charged particles and active
radicals in the reactor is controlled by their diffusion to the walls. By introducing characteristic diffusion
length Λ= for a cylindrical reactor of radius R and length L, the loss rate of positively charged ions is
expressed through the characteristic diffusion time , where D is the ambipolar diffusion coefficient.
The electron distribution function is calculated by solving the Boltzmann equation using the so-called non-
local approach. This approach assumes that the EEDF depends on the total electron energy and that the
energy dependence of the EEDF is obtained by spatial averaging of the electron heating sources. When
you select the Global model, you must specify the total power adsorbed in plasma.
187
CFD-ACE_V2010.0_Modules_Manual_Part2
The Global Kinetic model is activated by selecting Global in the MO Panel. It can be used with and
without Kinetic module. For details, contact ESI CFD Customer Support to obtain a tutorial that describes
how to use the Global Kinetic model and the procedure of coupling kinetic, plasma, and chemistry
modules. Use the Global Kinetic model for quickly simulating low-pressure chemically reactive plasma,
analyzing the sensitivity of chemistry models, and analyzing electron kinetic effects on plasma
composition.
Plasma Module
Implementation-Model Setup and Solution-Model Options-Magnet
The following are the ICP related options under the Magnet tab. The Magnetic Module must be activated
for ICP. Under the Magnetic Field option, select one of the following for ICP:
• AC Single Frequency - For an ICP reactor operating at a single frequency, the AC Single
Frequency option is recommended. Under this option, harmonic time variation of the
electromagnetic fields is assumed with the Field Frequency you have specified.
• DC+AC Single Freq. - This enables you to calculate static magnetic field produced by DC
sources or permanent magnets in addition to the electromagnetic fields created by AC sources.
• Transient - Maxwell equations are solved in time domain enabling arbitrary time dependence of
the fields for transient and multi-frequency processes.
• Displacement Current - This button can be activated for AC and Transient Problems to account
for the displacement current in the Maxwell equations. The displacement current is normally
needed only for very high frequency (> GHz) situation. See the Magnetic module for details.
Plasma Module
Implementation-Model Setup and Solution-Model Options-Electric
The Electric Tab enables you to access the CCP related options. The Electric Module must be activated
for CCP.
Electric Field Options
Under the Electric Field Option, you can only select the Electrostatic option when CCP has been
selected. The only Electric Solver Option for Plasma is the Finite Volume Method (FVM).
Model Options-Advanced
Plasma Module
Implementation-Model Setup and Solution-Model Options-Advanced
188
Modules
Theory
In the volume of a partially conducting dielectric material, the charge conservation equation (equation 10-
6) is solved for, in addition to the Poisson’s equation for the electrostatic potential. The current density is
assumed to be only due to that of conduction,
(10-16)
where is the conductivity of the material as specified by the user and E is the calculated electrostatic
field.
At the surface of the dielectric bordering the plasma, the surface charge density calculation allows for
charge leakage into the volume of the dielectric, in addition to the flux ( ) of all charged particles i,
(10-17)
In the volume of a partially conductive dielectric, the volumetric charge is solved at each cell explicitly at
CCP time step. The updated charge is used in the solution of Poisson’s equation.
Model Option
In order to activate this feature, check "Lossy Dielectric" option on the Advanced tab (on the Model
Options tab) as shown below.
Boundary Condition
At metallic boundaries, zero charge and the applied electric potential are enforced. The dielectric
boundary condition should be used for Dielectric material and lossy dielectric material.
Volume Condition (Properties)
The relevant property for this feature is the electrical conductivity in the lossy dielectric specified in units
of 1/Ohm-m as shown below. The option is only available upon checking the Lossy Dielectric box when
the Volume is specified as a Dielectric as shown below.
189
CFD-ACE_V2010.0_Modules_Manual_Part2
Output
Outputs that are updated as a result of this feature are the volumetric charge (Qvol) and electric potential
(elpot) within the volume of lossy dielectric materials.
User Tip
For a given electric conductivity of dielectric material, select such a way that the dielectric
relaxation criterion is not violated to prevent divergence , i.e., . For instance, with a
of 1 ns, the higher limit for would be 8.85e-3 (1/Ohm-m) and so on.
190
Modules
See Also:
Plasma-> Feature->Automatic LUT Generation
Plasma-> Implementation-> Model Setup and Solution -> RUN
See Also
Electric Conduction (Currents in Conductors and Lossy Dielectrics)
Electric Conduction-DC Conduction
Electric Conduction-AC Conduction (One Frequency)
Electric Conduction-Transient Conduction (Time Domain)
Plasma Module
Implementation-Model Setup and Solution-Model Options-Advanced
191
CFD-ACE_V2010.0_Modules_Manual_Part2
1. the user needs to specify initial pressure (Gauge pressure in Pa). Total of Gauge +
Reference pressure will be used for LUT generation calculation. If Gauge pressure is
zero, the reference pressure (specified under MO/Flow) will be used for calculation of
LUT.
2. the user needs to specify Initial Mixture.
Note: The user must check this option if user applied initial condition settings of volume by volume.
The total pressure (Reference Pressure + pressure specified) & mixture used in automatic LUT
generation is in the IC tab - it is assumed all the plasma VCs have the same initial pressure &
mixture. So if the user is using volume by volume initial condition settings, he must select "Override
IC settings" option here (MO/Advanced -> Plasma) and specify single initial pressure and mixture
for LUT calculation.
If initial condition setting is "For All Volumes" then users do not have to check this option. Initial
pressure and mixture will be used from IC settings for LUT calculation.
192
Modules
When the CCP option is used for low pressure, high RF frequency discharges, electron stochastic heating
(described in equation 15-8) can be included on the Adv/Plasma tab.
The new "Stochastic Heating" model is implemented in V2010.0. In this new model, an analytical formula
published by Lieberman, et al, is used, and a novel "self-aware" approach is implemented. This allows
ACE+ to automatically detect regions in the plasma where stochastic sheath heating occurs. To activate,
check the "Stochastic Heating" box. No other user input is required for the new model.
193
CFD-ACE_V2010.0_Modules_Manual_Part2
Plasma Module
Implementation-Model Setup and Solution-Model Options-Advanced
Circuit Model
For multi-dimensional reactor scale models to inform real time control for plasma processing (such as to
determine settings like electrode voltage) requires accurate coupling between circuit parameters and
plasma characteristics. The new electrical circuit coupling to SPICE is developed to address these types
of needs as well as to enable a more general investigation of the relationship between the external circuit
and plasma properties in radio frequency discharges. The ACE+ solver itself has been enhanced to
include a legacy SPICE code from the University of California, Berkeley. A newly developed interface
allows for mutual exchange of information between the reactor and the circuit. This ‘loose coupling’
methodology allows the ACE+ Plasma solver to account for external circuit interactions to a greater
degree of generality. Proposed external circuit coupling to SPICE will enable generic circuit
representations at several reactor surfaces as shown in Figure 1below.
194
Modules
Figure 2: The plasma reactor represented as a set of interconnected resistors "Rpi" and sheaths
"Sheath i" connected to external circuit elements.
With a multidimensional model such as ACE+, the currents flowing into the electrode surfaces (P1, P2
etc.) can directly be computed from known plasma characteristics. Therefore, the approach employed in
this methodology is to use currents computed from the Plasma model as inputs to a SPICE model
describing only the external circuit details.
Source code for the legacy SPICE solver version 2g6 is directly downloadable from the website at the
University of California, Berkeley. This code was developed more than 25 years ago for simulations of
integrated circuits and was used as a codebase for several commercial flavors of SPICE solvers in the
electronics design industry. External circuits used to power plasma processing reactors involve elements
such as resistors, capacitors, inductors and voltage sources and all of these are fully supported in the
SPICE version 2g6.
ACE+ and SPICE Coupling
A schematic representation of the loose coupling methodology is shown in Figure 3.
195
CFD-ACE_V2010.0_Modules_Manual_Part2
Figure 3: Schematic of the coupling of ACE+ and SPICE. The time step for a single call to the
SPICE solver corresponds to the time step in the CCP module.
User Input
The newly developed coupled SPICE-ACE+ option is available in the MO > Adv tab for CCP Plasma only
as shown in Figure 4.
Figure 4: The SPICE-ACE+ coupled model option is in the MO > Adv tab.
On activating a circuit model under MO > Adv, the user also needs to select netlist input file for the circuit.
Note: Make sure that you provide correct details in the netlist input file. There are a few rules as noted in
the later section of this document corresponding to the netlist input file name and format, sinusoidal
voltage sources, current legs and shunt resistance.
Once the SPICE-ACE+ coupled option is chosen (in other words, the circuit model is selected), the
following settings are needed by the user:
1. Specify source frequency under MO > Plasma when a sinusoidal source is used, as shown in
figure 5. The user needs to check the “Cycle Frequency” check box and specify the frequency. In
case of dual frequency, make sure to specify a higher frequency.
196
Modules
Figure 6: At Electric BCs, a new "External Circuit" option is available for the Fixed Potential type.
Once selected, a node number for this BC needs to be specified.
Output
An output file casename.circuit.out is written out every time step with the results of the SPICE model.
While the data contained is only pertinent to the last time step and is useful for the ACE+ solver, the user
gets no additional information over what is written to the DTF file. Similarly, casename.circuit.in has
information for SPICE but the user gets no additional information. There is no change in the contours
viewable in CFD-VIEW. The updated results from SPICE are reflected in the electric potential variables
el_pot and Pt_avg.
Spice Input Netlist Notes
1. Files named casename.circuit.in and casename.circuit.out are reserved for the coupled solver to
work with. The input circuit model file chosen in the GUI should NOT be named
casename.circuit.in where casename is the actual DTF file name.
2. The SPICE input file has a prescribed format for sinusoidal voltage inputs. This needs to be
strictly adhered to, so the solver picks up the information correctly.
1. The SPICE-ACE+ interface converts sinusoidal voltage inputs provided by the user in the
netlist file into PieceWiseLinear or PWL inputs to SPICE with the values corresponding to
the particular time in the simulation. This part occurs “behind-the scenes” and is provided
as information just in case the user looks at the casename.circuit.in file and finds a
“different looking” voltage input!
3. A constant current source from ground needs to be specified for each of the reactor boundary
circuit nodes. This value is overwritten from the solver as the simulation progresses.
4. The RSHUNT option is unavailable in the SPICEv2g6 used here. A high resistance on the order
of GigaOhms needs to be manually specified for a dc path to ground from each node not
connected to sources.
5. At this time, external circuit current sources have not been tested and should not be used in the
netlist input file.
Example SPICE netlist Input:
Lets consider the example of a capacitor of 0.6 nF in series with the external voltage source of 13.56 MHz
and 100 V amplitude to the powered reactor boundary surface as shown in figure 7.
197
CFD-ACE_V2010.0_Modules_Manual_Part2
198
Modules
j. The 44th to the 53rd column should contain the value of the excitation frequency in Hz.
k. The 54th column should be empty.
l. The 55th to 64rd column is for the time-delay - the time in seconds from the start of the simulation
after which this source is turned on.
m. The 65th column should be “)”.
The underscore “_” fields in the format descriptor indicate the empty spaces.
Every line after the first contains a circuit element - capacitor, resistor, inductor or a voltage source. The
circuit elements are recognized by the first letter - eg. C for capacitor, R for resistor, V for voltage source.
The line starting with I specifies a dummy current of 1mA coming in from the ground to node 2.
The resistor from node 2 to ground having a large value of 1e9 is necessary to provide SPICE with a DC
path to ground. In general, such a line in the input netlist is necessary for all node numbers not connected
to voltage sources.
The lines beginning with .TRAN to .END must not be changed except in the following way:
a. The .ic line should have V(1)=0 V(2)=0 …and so on for all node numbers used in this netlist file.
b. The .PRINT TRAN line should include V(1) V(2)..and so on for all node numbers used in this
netlist file.
Similarly, the circuit netlist file can be generalized to include other surfaces, or include more circuit
elements, etc. Care should be taken to appropriately “hookup” other surfaces with their node numbers in
the BC tab in the GUI.
Other Notes and Limitations
1. The coupled SPICE-ACE+ option works only with Plasma model CCP.
2. The SPICEv2g6 User Manual as obtained online is provided as a separate pdf document for the
users’ reference.
Backward Compatibility
1. To revert to the previous one-capacitor circuit model, a special DTF update with the following 2
lines is necessary:
V_float int 1 1
Cb_per_unit_area double 1 XX
where XX is the single capacitance in F/m2 that would otherwise be specified in the GUI.
To revert to the previous one-capacitor circuit model for multiple electrodes with different
capacitors at different electrodes, similar special DTF update is necessary:
V_float int 1 1
RF_Vsin_Cb_Num int 1 N
RF_Vsin_Src_Freq double N f1 f2 ....fN
RF_Vsin_Cb_per_Area double N C1 C2 ....CN
2. The SPICE-ACE+ option and the one-capacitor circuit model will not work together. If the SPICE-
ACE+ checkbox option is selected in the GUI, then the one-capacitor circuit model will be
ignored.
References
[1] S. Rauf and M. J. Kushner, Journal of Applied Physics, 83, 5087 (1998)
[2] http://embedded.eecs.berkeley.edu/pubs/downloads/spice/index.htm
[3] S. Rauf and M. J. Kushner, IEEE Transactions on Plasma Science, 27, 1329 (1999).
199
CFD-ACE_V2010.0_Modules_Manual_Part2
See Also:
Plasma-> Feature->Automatic LUT Generation
Plasma-> Implementation-> Model Setup and Solution -> RUN
See Also
Electric Conduction (Currents in Conductors and Lossy Dielectrics)
Electric Conduction-DC Conduction
Electric Conduction-AC Conduction (One Frequency)
Electric Conduction-Transient Conduction (Time Domain)
Plasma Module
Implementation-Model Setup and Solution-Volume Conditions-Introduction
Click the VC tab to see the Volume Conditions Panel. See Control Panel-Volume Conditions for details.
For Plasma, set the Plasma related Properties. For ICP, set the electromagnetic sources. The Volume
Conditions section includes:
• Volume Conditions-Properties
• Volume Conditions-Magnet
Volume Conditions-Properties
Plasma Module
Implementation-Model Setup and Solution-Volume Conditions-Properties-Introduction
Select Properties in the Setting Mode list. Next, select the Volume from the graphics window or the
BC/VC Explorer (below the Graphics window). Plasma properties apply to the fluid region, where you
must set the electron collision frequency and the electron diffusion coefficient. The Plasma Volume
Condition Properties include:
• Properties-Electron Collision Frequency
• Properties-Electron Diffusion Coefficient
• Properties - Electromagnetic
Plasma Module
Implementation-Model Setup and Solution-Volume Conditions-Properties-Electron Collision
Frequency/Mobility
When Kinetic Module is activated for electron kinetics, the electron collision frequency, mobility and
diffusivity will be calculated by Eqs. (15-17), (15-9) and (15-10), respectively. The selections of Constant,
Function of Te, and User Subroutine evaluation methods are available for specifying the collision
frequency when Kinetic Module is not activated.
Constant
A constant value is specified in the field. Based on this constant value, the code calculates a cell-by-cell
electron transport collision frequency using Eq. (15-18). And the electron mobility is derived using Eq.
(15-16).
Function of Te
The electron collision frequency is calculated on the basis of electron collision cross-sections specified in
the Database Manager. One must choose "Collision Cross Section” as the Step Type for the elastic
collision (or momentum transfer) steps in the plasma reaction, and specify the collision cross section as a
200
Modules
function of electron energy. The Eq.(15-17) is used to calculate the electron collision frequency. Also the
electron mobility and diffusivity are calculated according to Eq. (15-16) with Maxwellian distribution when
electron Kinetics is not solved.
User Subroutine
The subroutine uecoll_frequency calculates the transport collision frequency as a function of the electric
field and other plasma parameters.
Plasma Module
Implementation-Model Setup and Solution-Volume Conditions-Properties-Electron Diffusion Coefficient
If the Kinetic Module is not activated for electron kinetics, the Maxwellian distribution is assumed and
Einstein’s Equation becomes the only evaluation method available for specifying the Electron
Diffusion Coefficient.
Plasma Module
Implementation-Model Setup and Solution-Volume Conditions-Properties-Electron Collision
Frequency/Mobility
Under the "E/M” tab, when Magnetic Module is activated (for inductive coupling), the Electric Conductivity
in plasma is needed and calculated by Eq. (15-7). In solid region, the Electric Conductivity should be
chosen as "Perfect Insulator” since only non-conducting dielectric materials are allowed for Plasma
Module.
The Relative Permittivity and Relative Permeability can be specified as allowed by GUI.
The Space Charge in plasma is automatically calculated (see Poisson Eq. 15-21), and zero value needs
to be entered for additional charge.
Volume Conditions-Magnet
Plasma Module
Implementation-Model Setup and Solution-Volume Conditions-Magnet-Introduction
For ICP applications, you will need to set sources of the electromagnetic fields. Typically, ICP reactors
use AC coils to create an alternating magnetic which induced an electric field that sustains the plasma.
Static magnetics fields may also be created using either DC currents or permanent magnets (although
steady fields, alone, will not generate plasma).
201
CFD-ACE_V2010.0_Modules_Manual_Part2
The AC Source option, enables you to add an AC electrical current source to the control volume.
Current
If you have selected the AC Single Frequency Magnetic Field Option in the Magnet tab of the Model
Options panel, then the Current amplitude (in Ampere) is input for two-dimensional geometries. Please
notice that the value of the current is for the surface patch as selected. For example, if the coil current is
40 A, but the coil cross section area is divided in 4 equal patches, the value to be entered for each patch
becomes 10 A. For 3D geometries, you must specify the Current Density in units of amps per square
meter.
If you have selected Total Power under the MD/Plasma Power Specification option, for AC Single
Frequency, the current of each electromagnetic source will be automatically and proportionally adjusted
to achieve the specified power absorption in the plasma.
Frequency
If you have selected the AC Single Frequency Magnetic Field Option in the MD/Magnet panel, then the
Frequency of all Electromagnetic Sources is fixed to the value specified under the MD/Magnet Field
Frequency window and cannot be modified under VC’s.
If you have selected the Transient Magnetic Field Option in the MD / Magnet panel, then you can set a
different frequency for each electromagnetic source.
Current Direction
For 2D, axisymmetric geometries, it is assumed that the direction of the electric current is axisymmetric.
For two-dimensional, plane geometries, it is assumed that the direction of the electric current is in the
positive z direction.
For three-dimensional geometries, six input fields (x 1 ,y 1 , z 1 , and x 2 ,y 2 , z 2 ) under Axial Direction option
will appear for specifying two points that define the axis of symmetry for the current.
Inherent in this method is the assumption that the Coil and its associated current are axisymmetric. For
non-axisymmetrical coils, the user should use the Electric Module to solve for the current in the coil as a
function of voltage drop between the coil terminals.
Plasma Module
Implementation-Model Setup and Solution-Volume Conditions-Magnet-DC Source
The option of introducing a DC source into a Coil volume is available when the magnetic field is being
solved as a Static field. It is also available when you select the AC+DC option in the MD/Magnet panel.
For a DC source, you must specify the coil current magnitude and, for 3D, the axis of symmetry an AC
source. The procedure for setting these values for CD source is the same as for AC Source.
Plasma Module
Implementation-Model Setup and Solution-Boundary Conditions
Plasma
Click the BC tab in the Control Panel, then select Plasma tab to specify Plasma related Boundary
Conditions. To assign boundary conditions and activate additional panel options, select an entity from the
viewer window or the BC Explorer.
Under Plasma options ICP and CCP, one must select from the following options for electron temperature
boundary condition at Inlet, Outlet, Rotating Walls, Wall, and (material: solid-fluid) Interface boundaries:
• Thermal Flux Balance - This default option does not require any additional input.
• Fixed Gradient - Input the variable dTe/dn which specifies the value of the normal derivative for
the electron temperature on a surface.
• Fixed Te - Specifies the value of Te on a surface in eV.
202
Modules
Under Plasma options CCP, there are only two options for electron number density boundary condition at
Inlet, Outlet, Rotating Walls, Wall, and (material: solid-fluid) Interface boundaries:
• Thermal Flux Balance - Under this default option, it allows the secondary electron emission with
the coefficient as constant or user-defined.
• Zero Gradient - The electron number density at the boundary is calculated as the same as the
adjacent cell value.
Under Plasma options ICP, the BC recommended is the Fixed Gradient at zero value, which means that
no electron energy is lost at the boundary. User can also specify the Thermal Flux Balance BC, which
allows the energy loss limited by thermal flux.
Under Plasma options CCP, the BC recommended is the Thermal Flux Balance for both Te and Ne.
Electron Secondary emission needs to be activated at surfaces like Cathode. At Outlet boundary, since
the normal electric field is zero, Fixed Gradient at zero for Te and Zero Gradient for Ne are more
appropriate.
Chem
Continue on the BC tab in the Control Panel, and select Chem tab to specify Plasma related chemistry
Boundary Conditions.
Under Plasma options ICP and CCP, the following options for chemsitry boundary condition at Inlet,
Outlet, Rotating Walls, Wall, and (material: solid-fluid) Interface boundaries:
• Surface Reaction - This BC needs to be specified for all Wall surfaces, including solid-fluid
material interfaces. Proper surface reaction mechanism needs to be input.
• Mixture Definition - For Inlet / Outlet boundaries, proper mixtures need to be defined and the
mixture name needs to be input.
Electric
Continue on the BC tab in the Control Panel, and select Electr tab to specify Plasma/CCP related electric
Boundary Conditions for all BC types:
• Fixed Potential - Specify the electric potential using Constant, Sinusoidal, User-Sub and
Parametric. The Sinusoidal is a special type of electric BC to be used to provide the information
on the sinusoidal function (e.g., RF frequency) and the connection to the external circuit.
• Dielectric -his BC needs to be specified for the dielectric surface, where the surface charge will
be calculated in CFD-ACE+ solver.
• Surface Charge -Equivalently, this BC is used to specify the normal electric field at outer
surfaces. For example at Outlet, the normal electric field becomes zero if the surface charge
density is specified as zero.
Plasma Module
Implementation-Model Setup and Solution-Initial Conditions
Click the IC tab, then select the Plasma tab to specify Plasma related Initial Conditions.
• Electron Temperature (Te) -Electron Temperature (T e ). The only evaluation method available is
Constant. Here you must specify the electron temperature in eV.
• Electron Concentration (Ne) - Evaluation method Constant is available and you may specify the
3
electron number density in #/m , or using quasi neutrality.
For species Mixture Specification under Chem tab of IC, the mixture needs to be defined to contain
proper amount of net ion charge, and its Mixture Name needs to be input. For CCP the initial net ion
15~16 3 17~18 3
charge density is typical about 10 #/m , and 10 #/m , for ICP.
203
CFD-ACE_V2010.0_Modules_Manual_Part2
Plasma Module
Implementation-Model Setup and Solution-Solver Control
Click the Solver Control [SC] tab to see the Solver Control Panel. See Control Panel-Solver Control for
details. Next, select the following tab folders to specify Plasma related Solution Control Parameters.
Spatial Differencing
For CCP, exponential scheme is used to compute electron temperature (T e ) and electron number density
(N e ).
Solver
As with the other variables, you can pick one of the many solvers available for T e and N e .
Relax
Different relaxation parameters are required for ICP and CCP solvers, since the governing equations and
solution approaches are different. For ICP, where the steady-state equations are solved, it is important to
optimize the relaxation parameters for a good convergence. Since the transport rate of ions is typically
two orders higher than that of neutral species due to the drift flux, both inertial and linear relax on ions are
(some two-order) heavier than neutrals. Whereas for CCP, where time scales of all the variables are
resolved by the transient solution, the relaxation is mostly unnecessary. When gas heat is solved, the
inertial relax on enthalpy is light, but the linear relax on temperature is heavy. The following table lists the
typical values of relaxation parameters for all variables involved in ICP and CCP solution.
Limits
The Limits option lets you set minimum and maximum values for certain dependent and auxiliary
variables. Limits on T e and N e can be set in this panel.
Adv
• Chemistry (Species Mass Fractions) - Cut Diffusion - Enables the species mass flow rate as
specified via convection at Inlet BC. Also for CCP, the ions are allowed to recombine into its
neutral at the Inlet.
• Plasma (CCP) - CCP Iterations - Enables additional control of sub-iterations for electron number
density, electron temperature and electrostatics potential solution. We recommend that you
204
Modules
increase the sub-iteration number (up to 20) when system pressure decreases (below 0 (100)
mTorr), but decrease (to one) when system pressure becomes high (atmospheric).
• Plasma (CCP) - Ramping Electron Heating - For high (RF) frequency CCP discharge, the
electron Joule heating is ramped for the first 400 time steps for NEW runs in order to minimize the
disturbance due to the initial guess. For restart runs, the ramping is not used no matter if the box
is checked or not. For some low frequency (~ KHz) discharges (e.g., atmospheric pressure DBD),
the ramping may not be desirable and thus be unchecked.
• Monte Carlo - Monte Carlo Transport - The Monte Carlo Transport option enables the post-
processing of the heavy particle (hot neutrals) and generates the input information for CFD-
TOPO, the feature scale simulation tool of ESI CFD.
Plasma Module
Implementation-Model Setup and Solution-Output
Summary Output
Under the Summaries tab, select the summary information to be output by the solver. . See Control
Panel-Summary Output for more information on the general summary output options, including boundary
integral output, diagnostics, and monitor point output. The plasma related print out are:
CCP Monitor Points
For Plasma/CCP option, monitor points are added to print out (in file_name.CCP_bc.##.MON) the source
voltage, electrode surface voltage, currents (electron, ions, displacement and total), and ion energy
distribution function (IEDF) at electrode boundaries. To activate the CCP Monitor, click
Out/Print/Plasma(CCP) and check the box "Monitor V-J at Boundary Points”. Enter the x,y,z locations (will
be used to find the closest boundary faces) to specify the monitor points.
CCP Cycle Average Summary
For Plasma/CCP option, CCP Cycle averaged quantities are summarized in file_name.CCPavg. This is a
useful means beside the monitor points, to check if the RF CCP solution reaches the periodic steady-
state.
Plasma Heating Balance
To output the interested volume heating sources and surface heating fluxes. The print out of the following
items is enabled:
1. The volumetric heat release of each gas-phase reaction step, excluding the steps involving
electron is calculated as the reaction rate times the heat of reaction, and printed as the
summation over all volumes per those reaction steps (tagged by the reaction equation). The
radiation heat release can then be found according to the reaction equations.
2. The volumetric heat release from electron elastic (momentum transfer) reaction with neutrals is
calculated as defined by Eq.(15-42), and printed as the summation over all volumes per those
reaction steps (tagged by the reaction equation).
3. Ion Ohmic heating by Eq.(15-45) is printed as the summation over all volumes and all ions.
4. The surface heat fluxes (see Eq. 15-46) due to surface reaction are printed as the summation of
surface reaction heat release over each surface patch as well as over all surfaces, per each
surface reaction step. The surface heat fluxes due to ion kinetic energy impact are also printed as
the summation of surface all ions kinetic heat release over each surface patch as well as over all
surfaces.
5. Currently the Joule heating on electron and ions, and electron energy loss due to gas-phase
reaction (for those steps involving electron) are printed out as summation over volumes.
205
CFD-ACE_V2010.0_Modules_Manual_Part2
Graphical Output
For CCP, you may specify how often to output the results for a complete CCP cycle. This Plasma (CCP)
Results output flag is expressed in terms of the CCP Cycle Frequency. For example, if you set the
Timestep Output Frequency value to 1000, a CCP Cycle Frequency of 500, and has selected 20
timesteps per cycle for CCP timestep size in the MO/Plasma menu (see Model Setup and Solution-Model
Options) then the results for a full cycle (20 timesteps in this case) will be output starting at the timestep
corresponding to the beginning of the 500th CCP cycle to the end of that cycle. (That is, timesteps 9981->
10000, 19981->20000, 29981-30000, etc. will be output in addition to the timesteps corresponding to a
Timestep Output Frequency value of a 1000).
The complete set of DTF files output would correspond to the following timesteps - 1000, 2000, 3000,...,
8000, 9000, 9981-10000, 11000, 12000,..., 18000, 19000, 19981-20000, 21000, 22000,..., 28000, 29000,
29981-30000,... etc.
Under the Graphics tab, you can select the variables to output to the graphics file (modelname.DTF).
These variables will then be available for visualization and analysis in CFD-VIEW. Activating the Plasma
Module allows output of the variables listed:
3
Electron Density 1/m
Electron Temperature eV
Electron Mobility V
2
1/m -
Electron Flux CCP only
s
3
Average Electron Density 1/m CCP only
Average Electron
eV CCP only
Temperature
Plasma Module
Implementation-Model Setup and Solution-Run
206
Modules
When the "Offline Boltzmann Lookup Table" option is selected under MO/Advanced/Plasma, the Run tab
gets an additional button as shown below, guiding the user to generate new Lookup Tables (or apply
existing ones).
207
CFD-ACE_V2010.0_Modules_Manual_Part2
208
Modules
The user can either (a) generate and then apply LUTs or (b) apply a previously generated LUT
When the "Automatic Run" option is selected, the user is required to select one of the look-up Table
options as shown in the figure above.
1. Te: If this table is used, the rate coefficient is from the lookup table based on the E field in the
plasma, while the mobility and diffusion coefficient is calculated based on Maxwellian EDF.
2. Te-md: This table contains ( , )-based lookup table for reaction rate, as well as mobility and
diffusion data.
3. E: contains ( , E)-based lookup table for reaction rate, as well as mobility and diffusion data.
4. E/p: contains ( , E/p)-based lookup table for reaction rate, as well as mobility and diffusion data.
The solver generates LUTs using the user’s case conditions, in particular - the total pressure (including
the Reference pressure) and the gas composition specified in the IC tab, the volumetric reaction
mechanism, the electric field excitation frequency and the range of Ne and E0 information specified in MO
-> Adv are used.
The user can also apply existing LUT. To apply an existing LUT, deselect "Automatic Run". By default,
"other" is selected and the user can browse to the file. Caution should be exercised since LUTs are
specific to the user’s case conditions and inputs noted in the paragraph above. After applying the
selected LUT, the user will have to submit the DTF file to the solver by clicking "Submit to Solver" on the
Run tab.
See Also:
Plasma -> Feature -> Automatic LUT Generation
Plasma -> Implementation -> Model Setup and Solution -> Model Options -> Advanced
Plasma Module
Implementation-Post Processing
CFD-VIEW can post-process solution results. A complete list of output variables available as a result of
using the Plasma Module is shown in the table.
Te Electron temperature eV
209
CFD-ACE_V2010.0_Modules_Manual_Part2
Theory
CFD-ACE+ kinetic module solves the Boltzmann equation to obtain the electron energy distribution
function (EEDF) f0. The reaction rate constant R, diffusion coefficient Dr, and mobility of electrons are
then calculated from f0 as follows:
Here u is the kinetic energy, e is the unit charge, v is the momentum collision frequency, and is the
mass of electron.
210
Modules
CFD-ACE+ kinetic module can generate three types of 2-D lookup tables. They are LUT( , ), LUT( ,
E), and LUT( , E/P) respectively, with the electron density, the electron temperature, E the
electric field strength, and p the pressure.
To generate a lookup table, an arbitrary 1 cell geometry is needed. The reaction mechanism is the same
as what will be used for plasma simulation by fluid model. For this 1 cell geometry, the spatial variation of
the energy distribution function is neglected; therefore local approach should be activated in kinetic
module. The detailed GUI setup for LUT generation is as follows.:
1. In PT, active flow + chemistry + kinetic module.
2. IN MO->shared: set the total number of the time steps and the time step
IN MO->kinetic:
1. Check local Boltzmann + lookup table generation+ Coulomb collision. Coulomb collision
is included to take account of the effect of e-e collision, which also gives the dependence
of rate coefficients on ne.
2. Specify the number of energy grid, and set the energy range. Choose the boundary
conditions for minimum and maximum energy (zero total flux boundary condition is
recommended).
3. Specify number of ne, E and the limits for both variables.
4. Specify the tolerance and time step ramping for kinetic module.
4. Next, please choose from the following 2 options:
1. In VC->VC Setting Mode=Properties->Kinetic, for the Fluid, set diffusion coefficients and
convective velocities to be from plasma.
2. In VC->VC Setting Mode=Kinetic, for the Fluid domain, the box for Kinetic Source need to
be checked.
5. BC is symmetric on 4 walls.
6. In IC->kinetics, set an initial function of EDF.
7. In SC-> relaxation, you need to put a large value (eg, 100) for the initial relaxation of species (ion
or neutral) to freeze the solution of species fraction.
In sc->advanced, put kinetic start time step 1 and frequency 1, which means the kinetic module is
solved every step starting from time step 1.
7. In Out, specify monitor point for EDF output.
Note : For LUT generation, the maximum iteration number for each (ne, E) pair is determined by total
time steps divided by the number of (ne, E) pairs. If the tolerance of LUT calculation is set to be a specific
value in step 2d, the calculation of LUT for each pair will run until the tolerance is reached, otherwise, it
will run to the maximum iteration number. Therefore the total time steps consumed when the LUT
calculation is done can be different from the number specified in MO->shared.
After LUT calculation, 5 files are generated with the information of the reaction rate and transport
coefficient, namely model_lut.dat, model_lut.in, model_md_lut.in, model_lut_E.in,and
model_lut_EbyP.in.
1. File Model_lut.dat has all the data for reaction rate R and Dr, and .. The format of the data is
as follows: starting from column 1, is the electron density, electric field strength, electron
211
CFD-ACE_V2010.0_Modules_Manual_Part2
temperature, forward rate for reaction step1, backward rate for reaction step1,..., forward rate for
reaction step N, backward rate for step N, mobility, and diffusion coefficient.
V/m eV
GUI setup for plasma simulation with the use of lookup table (LUT)
To use the LUT together with the fluid model, simply follow the following steps.
1. Determine which .in file has to be used, modify the volume reactions in reaction database for the
simulation file (eg. Model_Fluid_LUT.DTF).
1. If using model_md_lut.in file, add two dummy steps at the end of the volume reactions,
and then specify the names to be "lookup_table_mobe” and "lookup_table_De”;
2. If using the model_lut_E.in file, add three dummy steps at the end of the volume
reactions, and then specify the names to be "lookup_table_E_Te”,
"lookup_table_E_mobe” and "lookup_table_E_De” respectively;
212
Modules
3. If using the model_lut_EbyP.in file, add three dummy steps at the end of the volume
reactions, and then specify the names to be "lookup_table_EbyP_Te”,
"lookup_table_EbyP_mobe” and "lookup_table_EbyP_De” respectively.
2. Upload the selected LUT (eg, modle_lut_E.in) to the simulation file (eg, Model_Fluid_LUT.DTF)
using command:
DTF -ud Modle_Fluid_LUT.DTF < model_Lut_E.in
3. Run the updated simulation file (eg. Modle_Fluid_LUT.DTF ) with CFD-ACE+plasma model.
Theory
Same as ICP, 2-T thermal plasma module solves the CFD equations for heavy particles. The electron
density is obtained from equation (15-26) based on the quasi-neutrality condition. Electron temperature
comes from the electron energy balance equation (15-3). Instead of using equation (15-28) to get
electrostatic field, the 2-T thermal plasma module solves the electric conduction equation to obtain the
electric potential and electric field for the domains of metal conductors and plasma.
(15-52)
where electric conductivity σe is assumed to be known for metals. As for plasma, it can be modeled as
the function of gas temperature and pressure. The data for some gas mixture may be found from
publications. If the data are not available, the conductivity can also be derived from the plasma solution
as:
(15-53)
The electron conduction current used in the electron energy Equation (15-52) can be expressed as
(15-54)
2. To turn off the electric Joule heating for gas, since it is on electron. Default = 1
JouleHeatingElectric int 1 0
213
CFD-ACE_V2010.0_Modules_Manual_Part2
6. To set electric conductivity of plasma (fluid VC) as defined in GUI. Default = 1 (ie, e-cond from
plasma model).
econd_from_plasma int 1 0
Sheath model
1. Sheath voltage drop can be considered in the electric conduction equation for the solution of
electric potential. The sheath model (for the potential drop) can be defined at solid/fluid interface
only (not for outer wall). To activate the sheath model at solid/fluid interface in GUI, =>BC =>BC
Setting Mode =>General, click on the interface, =>Plasma =>Sub Type =>Sheath Model. The
sheath voltage drop can be defined as Constant or User-Subroutine.
2. The Constant sheath voltage drop can be input in the GUI as "DC Bias”, with the same value of
sheath voltage drop, but in negative, ie., DC_base = - Vsh_drop.
3. The User-Subroutine calculated sheath voltage drop can be activated by command (with the
default selection as false, ie., 0).
vsheath_userdef int 1 1
214
Modules
for boundary named "name_string”. Please notice since v2004, the "NAME_STRING” needs to be in
CAPITAL for all letters.
The primary source (i=1) needs to be input as the Sinusoidal (Fixed Potential) BC for Electric in CFD-
ACE+GUI, i.e.:
The additional sine terms can be input via model.in file as follows:
multi_sin_bc_names string 1 name_string
multi_sinN int 1 (N-1)
multi_sinV1 double (N-1) V2 V3 V4 ...VN
multi_sinF double (N-1) f2 f3 ... fN
multi_sinPhi double (N-1) φ2 φ3 ... φΝ
How to assign different capacitance values to capacitors connecting different electrodes and their
RF voltage sources?
In CFD-ACE-GUI, one can only specify the same value of capacitance per electrode area to capacitors
connecting RF voltage sources to their different electrodes. To allow different values of unit capacitance,
one can use the following script commands:
RF_Vsin_Cb_Num int 1 N
RF_Vsin_Src_Freq double N f 1 f 2 ....f N
RF_Vsin_Cb_per_Area double N C 1 C 2 ....C N
The first line specifies that there are N capacitors, which are associated with the RF source frequencies
2
as defined in the Sinusoidal BC. The corresponding unit capacitances (in F/m ) are specified in the last
line.
What solver control parameters can be used to improve the CCP solution convergence?
In the solution for RF CCP discharges, about two-order residual drop on E-potential, Ne and Te is
necessary to ensure time accuracy. In the case that the residual drop (shown in model_name.RSL) is not
satisfactory, ∆t CCP , CCP_Iteration (under SC/Relax/Adv) and Max_Iteration (under SC/Relax/Iter) are
main solver control parameters to be used to improve the convergence of CCP solution. Decrease in
CCP time step size ∆t CCP is the most effective way to improve the convergence. Generally the maximum
3
electron CFL number is required to be smaller than 10 to ensure a good convergence. To
check the CFL number, use the script command "cfl_check_ccp int 1 1” under Tools/Special DTF
Update in GUI. 20 steps/Cycle is the minimum for a 13.56MHz and should be increased (up to > 100) if
divergence persists.
Another way is to increase the numbers of those iterations to make the non-linear solution more accurate.
How can I tell if the RF CCP solution reaches periodic steady-state?
Basically there are two factors that can be used to determine if the periodic steady-state solution is
reached for RF CCP: 1) monitor points show the flattened changes; 2) bulk sources and boundary fluxes
are fully balanced over CCP RF-cycle.
How can I create a metastable species if the energy difference from the available ground state is
known?
215
CFD-ACE_V2010.0_Modules_Manual_Part2
For example, if the energy difference of metastable A* from the ground state A is ∆ε (in Volts). To modify
species A to obtain A* in the DataBase Manager, click on Tools => DataBase => "Gas Species”. Copy
the species A from "ESI-CFD” folder to the folder named after current DTF file. After changing the copied
species name to A*, click the tab "Thermal” to modify the value of "a6” for the JANNAF Coefficients (both
lower and upper limits) according to: a6* = a6 + 11605*∆ε. The factor 11605 is the factor to convert 1 V
to Kelvin.
What is the sheath model is used for?
The sheath model is used to calculate the effects of an electrode bias on the ion energy distribution for
energy dependent reaction rates and yields. If the feature scale coupling is requested at a point where the
sheath model is active, the ion energy and angular distributions are provided as additional output for use
in feature scale simulations. See Chemistry Module-Summary Output for information on the energy and
angular distribution output files.
Two-Fluid Module
216
Modules
• Solves for flow and heat transfer of a two fluid system where the fluid have independent
properties.
• Includes gravity force and rotating reference frame body forces.
• Offers several different correlations for inter-phase drag force based on systems of gas-liquid
(bubbly flow), liquid-liquid, or solid gas.
• Provides user specified second phase (dispersed phase) slip or non-slip on the wall.
• Enables you to activate the virtual mass force.
• Solves phase change problems.
Two-Fluid Theory
Two-Fluid Module
Theory-Introduction
The fundamental theory of the two-fluid model to solve multi-phase flow problems is the assumption that
the two fluids are continuous media. The transport equations that describe the conservation (those of
mass, momentum, and energy) can be written for the individual phases. Since the interphase boundaries
are practically intractable, it is necessary to invoke certain averaging procedures to obtain the
macroscopic forms of the equations for which solutions are sought. The result of this operation is the
introduction of a new variable generally called the Volume Fraction α, sometimes referred to as phase
fraction. In addition to two velocities, there are two volume fractions for each phase. The governing
equations are:
(16-1)
(16-2)
(16-3)
where:
k phase indicator
τk
th
is the k phase stress-strain tensor
F additional forces
A closed set for u k , α k , and p is formed by equation 16-1, equation 16-2, and equation 16-3.
Two-Fluid Module
Theory-Two-Phase Model Improvement
A number of algorithms can be used for solution of the phase-fraction equation. In the One Equation
Model, the continuity equation for the dispersed phase is solved to obtain alpha and the constraint
equation, i.e. the sum of the phase fractions must be unity, is used to obtain the solution for the primary
phase. This method does not guarantee boundedness for both phase fractions. In the Two Equation
217
CFD-ACE_V2010.0_Modules_Manual_Part2
Model, the continuity equations of both phases are solved separately. The solution is then renormalized
to ensure a unity sum,
(1)
(2)
The principle disadvantages of the Two Equation model are that this approach does not guarantee
continuity of the individual phases until overall convergence, due to the renormalization, and the
increased computational effort required for solution of an additional scalar equation.
In the new release, we followed the approach of composite solution of and mentioned in Vaidya et
al. (1). This approach is based on Carver (2, 3) and derives an equation for the phase fraction by taking
the difference of the density-weighted continuity equations. The continuity equation for each phase is
solved with the constraint . The difference of the density-weighted continuity equations
takes the form
(3)
(4)
(5)
The discrete form of this equation, with transient and source terms, is used in CFD-ACE+.
The model formulation presented in the following sections is a general form for the Eulerian two-fluid
model with extensions to address transport in porous media.
The selected channel, Figure 1, is a Ballard Power Systems design that is approximately hemispherical in
cross-section, with a hydraulic diameter of 0.533 mm and length of 600 mm. The convergence of steady
218
Modules
state two-phase flow simulations for inlet gas velocity of 6 m/sec, inlet liquid water velocity 0.1 m/sec, inlet
3
liquid volume fraction 0.5, and constant densities of 3.3 and 1000 kg/m for the gas and liquid,
respectively was evaluated for each of the algorithms. The convergence histories, shown in Figure 2,
clearly demonstrate the advantage of algorithm C for this particular application.
Figure 1. Ballard Power Systems channel geometry utilized for algorithm evaluation, shaded by a
typical pressure distribution.
Figure 2. Convergence histories for tested formulations of the two-phase flow continuity and
phase fraction equations. Labels correspond to the formulations in Table I.
Two-Fluid Module
Theory-Interphase Forces
In equation 16-2, the term F includes various types of interphase forces. Some of these are lift, virtual
mass force, and drag force. The two-fluid module considers only drag (friction) and virtual mass force.
Two-Fluid Module
Theory-Drag Force
In most two-phase systems, it is assumed that the dispersed phase is present in the form of spherical
particles (bubbles, droplets). Thus formulation for the drag coefficient, C D , for a single particle is known
as:
(16-4)
(16-5)
219
CFD-ACE_V2010.0_Modules_Manual_Part2
F d is in per volume unit, is a force acting on a single particle, and D is the particle diameter.
For bubbly flow, C D is derived based on the curve fit of the data from the terminal velocity of a single
bubble rising in water [Wallis, 1969]. The equivalent C D -Re relation is expressed as:
(16-6)
where .
For solid particles in air, Ishii’s correlation is used:
(16-7)
For solid-liquid or liquid-liquid systems, an additional correction factor, , multiplies the above
equation.
Two-Fluid Module
Theory-Virtual Mass Force
The virtual mass force can be written as:
(16-8)
where the coefficient C vm takes a default value of 0.5. An implicit form of the term is implemented in the
code ensuring stability.
Two-Fluid Module
Theory-Enthalpy Method
The enthalpy method is a homogeneous flow model in which the two phases are treated such that there
is no slip between the two phases and the two phases have equal velocity and temperature. Therefore,
one set of governing equations (mass, momentum and energy) are solved for two-phase mixture (such as
liquid-vapor mixture). The major advantage of this method is that the numerical treatment of the phase
change can be achieved through simple extensions of existing single-phase numerical methods and /or
software. A phase fraction is introduced to keep track of the composition and interface of the two-phase
mixture. The local phase fraction is determined from the total enthalpy by consulting with the phase
diagram.
The latent heat evolution associated with the phase change process is accounted for in the energy
equation by defining a total enthalpy H.
(16-9)
where is a reference temperature; is the boiling temperature of the fluid; , denote the
specific heats of liquid and vapor, respectively; L is the latent heat, f is the mass fraction of vapor.
220
Modules
(16-10)
In general, the continuity and momentum equations are solved in conjunction with the enthalpy equation
as in single phase problems whenever the flow is involved. The thermal-physical properties of the mixture
such as density, viscosity, specific heat, thermal conductivity, etc. are computed following the averaging
relation:
(16-11)
Where Ω is the generic symbol for the thermal-physical property, α is the volume fraction of the liquid.
The mass fraction f of vapor is determined from the phase diagram (see Sample Phase Diagram):
(16-12)
Where H 1 and H 2 are the enthalpy of liquid and vapor at boiling temperature T, respectively; the
difference H 2 - H 1 =L is the latent heat.
(16-13)
221
CFD-ACE_V2010.0_Modules_Manual_Part2
may occasionally need dual pressure. This is not supported at this time. Another limitation is that only two
fluids are supported; that is, only two phases are allowed in the system. For the enthalpy method:
• Enthalpy method can handle the heat and mass transfer between two phases of the same fluid.
• In CFD-ACE+, the liquid phase should be specified as fluid2.
Two-Fluid Module
Implementation-Grid Generation
You can apply the Two-Fluid Module to any geometric system (3D, 2D planar, or 2D axisymmetric).
Furthermore all grid cell types are supported (quad, tri, hex, tet, prism, poly).
The general grid generation concerns apply. For example, ensure that the grid density is sufficient to
resolve thermal gradients, minimize skewness in the grid system, and locate computational boundaries in
areas where boundary values are well known.
Two-Fluid Module
Implementation-Model Setup and Solution-Introduction
CFD-ACE+ provides the inputs required for the Two-Fluid Module. Model setup and solution requires data
for the following panels:
• Problem Type
• Model Options
• Volume Conditions
• Boundary Conditions
• Initial Conditions
222
Modules
• Solver Control
• Output
Two-Fluid Module
Implementation-Model Setup and Solution-Problem Type
Click the Problem Type [PT] tab to see the Problem Type Panel. See Control Panel-Problem Type for
details.
Select Two-Fluid to activate the Two-Fluid Module. The Flow Module is also required when the Two-Fluid
module is activated.
Two-Fluid Module
Implementation-Model Setup and Solution-Model Options-Introduction
Click the Model Options [MO] tab to see the Model Options Panel. See Control Panel-Model Options for
details. The Two-Fluid Model Options include:
• Model Options-Shared
• Model Options-Fluid2
Two-Fluid Module
Implementation-Model Setup and Solution-Model Options-Shared
There are no settings under the Shared tab that affect the Two-Fluid Module. See Model Options for
details.
Model Options-2-Fluid
Two-Fluid Module
Implementation-Model Setup and Solution-Model Options-2-Fluid-Introduction
All of the model options for the Two-Fluid Module are located under the Fluid2 tab. There are two
methods available for the two-fluid model:
• General Method (shown here)
• Enthalpy Method
223
CFD-ACE_V2010.0_Modules_Manual_Part2
224
Modules
(16-14)
• Virtual Mass Force - The are two options available: No Virtual Mass and Constant. If you select
the Constant option, you must a value: C vm .
• Lift Force - This option is currently not available.
Two-Fluid Module
Implementation-Model Setup and Solution-Model Options-2-Fluid-Enthalpy Method
When you select the Enthalpy Method, the following inputs are available:
225
CFD-ACE_V2010.0_Modules_Manual_Part2
:
Model Options - Enthalpy Method
• Phase Change - Select Phase Change and enter the saturation temperature and Latent Heat.
226
Modules
Two-Fluid Module
Implementation-Model Setup and Solution-Volume Conditions
Click the Volume Conditions [VC] tab to see the Volume Condition Panel. See Control Panel-Volume
Conditions for details. Before you can assign any property values, pick one or more volume condition
entities from either the Viewer Window or the VC Explorer.
With the volume condition setting mode set to Properties select any volume conditions and ensure that
the volume condition type is set to Fluid2. Only volume conditions that are of type Fluid2 need to have
Two-Fluid properties specified (since there is no flow in solid or blocked regions there are no Two-Fluid
Module properties for those regions.)
There are two volume condition properties required by the Two-Fluid Module; density and viscosity.
227
CFD-ACE_V2010.0_Modules_Manual_Part2
Two-Fluid Module
Implementation-Model Setup and Solution-Boundary Conditions-Introduction
Click the Boundary Conditions [BC] tab to see the Boundary Condition Panel. See Control Panel-
Boundary Conditions for details. To assign boundary conditions and activate additional panel options,
select an entity from the viewer window or the BC Explorer.
The Two-Fluid Module is fully supported by the cyclic, thin wall, and arbitrary interface boundary
conditions. See Cyclic Boundary Conditions, Thin-Wall Boundary Conditions, or Arbitrary Interface
Boundary Conditions for details on these types of boundary conditions and instructions for how to
implement them.
All of the general boundary conditions for the Two-Fluid Module are located under the Fluid2 tab and can
be reached when the boundary condition setting mode is set to General. Each boundary condition is
assigned a type (e.g., Inlet, Outlet, Wall, etc.). See BC Type for details on setting boundary condition
types.
The Two-Fluid Boundary Conditions section includes:
• Boundary Conditions-Inlet
• Boundary Conditions-Outlet
Two-Fluid Module
Implementation-Model Setup and Solution-Boundary Conditions-Inlets
For any inlet boundary condition the Two-Fluid Module ultimately needs to know how to set the velocity,
density and volume fraction of the second fluid for each cell face on the boundary condition patch. There
are various ways to specify this information and there is one method (known as subtypes) available:
Fixed Velocity.
Fixed Velocity Inlet Boundary Condition Subtype Specification
This inlet subtype allows you to set the velocity pressure (used only to calculate inlet density), and
temperature for each boundary face on the inlet to a fixed value (this effectively fixes the mass flow rate).
The velocity vector is specified directly and the code calculates the density using the specified values of
pressure (P) and temperature (T) and the selected density method (specified in the volume condition
settings). For constant density flows, the pressure value is not used.
Inlet Velocity Direction Specification
All of the inlet boundary condition subtypes allow for velocity directions to be specified in various ways.
There are several ways to specify the velocity directions at inlets:
• Cartesian
• Normal
• Cylindrical
The differences in these velocity direction specification modes is given below.
• Cartesian - Allows you to specify the velocity magnitude in xyz components (U, V, W) for fixed
velocity inlets, or the velocity direction components (Nx, Ny, Nz) for fixed mass flow or fixed total
pressure inlets.
• Normal - The code calculates the velocity direction based on the boundary face normal direction.
(The face normal always points into the computational domain).
• Cylindrical - Allows you to specify the velocity direction in axial, radial, and tangential components
(Va, Vr, Vt). The axis of the cylindrical coordinate system is always the x-axis.
228
Modules
The following table summarizes the above information by listing the available inlet boundary condition
subtypes (with different velocity direction specification options). The table also shows the variables
required for each subtype.
3D,2Ds 2Ds
Fixed Velocity (Cartesian) P, T, U, V, [W] , [Omega]
3D,2Ds
Fixed Velocity (Cylindrical) P, T, Va, Vr, [Vt]
3D
Available for 3D simulations.,
2Ds
Available for axisymmetric 2D swirl simulations.
Two-Fluid Module
Implementation-Model Setup and Solution-Boundary Conditions-Outlets
For any outlet boundary condition the Two-Fluid Module needs to know how to set either the static
pressure or the mass flow rate for each cell face on the boundary condition patch. The various ways to
specify outlet boundary conditions are the same as those for the Two-Fluid Module except that, where
necessary, you need to supply the information for the second fluid velocity.
The table summarizes the available outlet boundary condition subtypes. The table also shows the
required variables and optional variables for each subtype.
Outlet Boundary Condition Subtypes and Variables
Available Subtypes Required Variables Optional Variables
Fixed Pressure P T
3D,2Ds 2Ds
Farfield P, T, U2, V2, [W2] , [Omega]
3D,2Ds 2Ds
Fixed Velocity P, T, U2, V2, [W2] , [Omega]
Extrapolated P, T
3D
Available for 3D simulations.
2Ds
Available for axisymmetric 2D swirl simulations.
Two-Fluid Module
Implementation-Model Setup and Solution-Initial Conditions
Click the Initial Conditions [IC] tab to see the Initial Condition Panel. See Control-Panel-Initial Conditions
for details.
The Initial Conditions can be specified as constant values or read from a previously run solution file. If
constant values are specified then you must provide initial values required by the Two-Fluid Module. The
values can be found under the Fluid2 tab and the following variables must be set; U2, V2, T2, and Alpha.
Although the Initial Condition values do not affect the final solution, reasonable values should be specified
so that the solution does not have convergence problems at start-up.
229
CFD-ACE_V2010.0_Modules_Manual_Part2
For problems with fixed pressure outlet conditions it is best to set the initial pressure to the outlet pressure
and the initial velocities to some reasonable value. For problems with total pressure inlet conditions it is
often desirable to set the initial pressure equal to the inlet pressure and the initial velocities to zero.
Two-Fluid Module
Implementation-Model Setup and Solution-Solver Control-Introduction
Click the Solver Control [SC] tab to see the Solver Control Panel. The Solver Control Panel provides
access to the various settings that control the numerical aspects of the CFD-ACE-Solver and the output
options. The Two-Fluid Solver Control section includes:
• Solver Control-Spatial Differencing Scheme
• Solver Control-Solver Selection
• Solver Control-Under Relaxation Parameters
• Solver Control-Variable Limits
Two-Fluid Module
Implementation-Model Setup and Solution-Solver Control-Spatial Differencing Scheme
Under the Spatial Differencing tab, you may select the differencing method to be used for the convective
terms in the equations. Activating the Two-Fluid Module enables you to set velocity, volume fraction, and
density calculations. The default method is first order Upwind. See Spatial Differencing Scheme for more
information on the different differencing schemes available. Also see Discretization for numerical details
of the differencing schemes.
Two-Fluid Module
Implementation-Model Setup and Solution-Solver Control-Solver Selection
Under the Solvers tab you may select the linear equation solver to be used for each set of equations.
Activating the Two-Fluid Module enables you to set the velocity and volume fraction correction equations.
The default linear equation solver is the conjugate gradient squared + preconditioning (CGS+Pre) solver
with 50 sweeps for the velocity equations and the volume fraction equation. The default convergence
criteria is 0.0001. See Solver Selection for more information on the different linear equation solvers
available and Linear Equation Solvers for numerical details of the linear equation solvers.
Two-Fluid Module
Implementation-Model Setup and Solution-Solver Control-Under Relaxation Parameters
Under the Relaxation tab you may select the amount of under-relaxation to be applied for each of the
dependent (solved) and auxiliary variables used for the flow equations. Activation of the Two-Fluid
Module allows settings for the velocity and volume fraction dependent variables, as well as the auxiliary
variables; density, and viscosity. See Control Panel-Under Relaxation Parameters for details on the
mechanics of setting the under relaxation values and see Numerical Methods-Under Relaxation
Parameters for numerical details of how under relaxation is applied.
The velocity and volume fraction equations use an inertial under relaxation scheme and the default values
are 0.2. Increasing this value applies more relaxation and therefore adds stability to the solution at the
cost of slower convergence.
The calculations for density and viscosity use a linear relaxation scheme and the default values are 1.0.
Decreasing this value applies more relaxation and therefore adds stability to the solution at the cost of
slower convergence.
230
Modules
The default values for all of the relaxation settings will often be sufficient. In some cases, these settings
will have to be changed, usually by increasing the amount of relaxation that is applied. There are no
general rules for these settings and only past experience can be a guide.
Two-Fluid Module
Implementation-Model Setup and Solution-Solver Control-Variable Limits
The Limits tab enables you to set minimum and maximum variable values. CFD-ACE+ ensures that the
value of any given variable will always remain within these limits by clamping the value. Activating the
Two-Fluid Module allows limits to be set for U2, V2, Density2, Viscosity2, and Volume Fraction. The
default limit range for volume fraction is 0-1. See Control Panel-Solver Control-Variable Limits for details
on how limits are applied.
Two-Fluid Module
Implementation-Model Setup and Solution-Solver Control-Output Options
There are no settings under the Output tab that effect the Two-Fluid Module. See Control Panel-Output
Options for details.
Two-Fluid Module
Implementation-Model Setup and Solution-Output-Graphical Output
Under the Graphics tab, you can select the variables to output to the graphics file (modelname.DTF).
These variables enable you to view and analyze output in CFD-VIEW. Activating the Two Fluid Module
outputs these variables:
Two-Fluid Module Graphical Output
Variable Units
Velocity Vector2 m/s
Laminar Viscosity2 kg/m-s
3
Density2 kg/m
2
Stream Function2 m /s
Volume Fraction -
Two-Fluid Module
Implementation-Model Setup and Solution-Output-Printed Output
Under the Summary tab, select the summary data to be output by the solver.
There are no settings under the Summaries tab that are exclusive to the Two-Fluid Module, but the Mass
Flow Rate and the Force and Moment Summaries of the Flow Module are clearly applicable to and
available for the Two-Phase Module since these summaries are determined entirely from the flow field
and the fluid properties. See Control Panel-Summary Output for details on the general summary output
options, including boundary integral output, diagnostics, and monitor point output.
Two-Fluid Module
Implementation-Post Processing
CFD-VIEW can post-process the solutions. When the Two-Fluid Module is invoked, the velocity and
pressure fields are usually of interest. Use CFD-VIEW’s vector plot and stream trace features to visualize
the velocity field. You can view the pressure field with surface contours and analyzed through the use of
point and line probes. A complete list of post processing variables available as a result of using the Two-
Fluid Module is shown in the following table.
231
CFD-ACE_V2010.0_Modules_Manual_Part2
Post-Processing Variables
Variable Description Units
U2 X-direction velocity of 2nd m/s
fluid
V2 Y-direction velocity of 2nd m/s
fluid
W2 Z-direction velocity of 2nd m/s
fluid
RHO2 Density of 2nd fluid kg/m3
H2 Enthalpy of 2nd fluid kg-m2/s2
Visc2 Viscosity of 2nd fluid kg/m-s
T2 Temperature of 2nd fluid K
Alpha Volume fraction of 2nd fluid -
2
Stream2 Stream Function of 2nd fluid m /s
The Mass Balance Summary and the Force and Moment Summary that are output by the flow module
can also be used with the Two-Phase Module to assess the global behavior of the solution and the extent
of convergence. This usage can be pursued in the same manner and for the same reasons given in the
Flow Module.
Two-Fluid Module References
[1] Voller, V.R., Swaminathan, C.R., and Thomas, B.G., "Fixed Grid Techniques for Phase Change
Problems: A Review.” Int. J. Numerical Methods Engineering, 30, 875, 1990.
[2] Swaminathan, C.R., and Voller, V.R., "A General Enthalpy Method for Modeling Solidification
Processes.” Metallurgical Transaction B, 23B, 651, 1992.
[3] Swaminathan, C.R., and Voller, V.R., "On the Enthalpy Method.” Int. J. Num. Meth. Heat Fluid Flow, 3,
233, 1993.
Kinetic Module
(17-1)
This equation describes the temporal evolution of the PDF due to diffusion and drift in the physical and u-
spaces, and different sources due to various processes. The dimensionality and physical meaning of the
u-space depend on the nature of the problem.
The Kinetic Module includes:
• Kinetic-Applications
• Kinetic-Features
• Kinetic-Theory
• Kinetic-Numerical Implementation
232
Modules
• Kinetic-Limitations
• Kinetic-Implementation
• Kinetic-References
Kinetic Module Applications
The FP equation appears in many areas of physics and chemistry. Under certain conditions, the
Boltzmann transport equation is reduced to the FP equation in a four-dimensional coordinate-energy
space. The FP approach is widely used for describing electron kinetics in collisional gas discharge
plasmas and semiconductor devices. The Kinetic Module is a multi-purpose module that can be used for
various applications as a stand alone application or coupled to other modules. The Kinetic Module
simulates electron kinetics in plasmas.
Kinetic Module Features
The Kinetic Module solves the FP equation for a four-dimensional probability distribution function f. By
integrating f over u-variable, kinetic module calculates macroscopic quantities such as particle density,
flux, rates of chemical reactions. These quantities can be used by other modules, such as the Plasma,
Chemistry and Electric Modules, to calculate electromagnetic fields, chemistry, etc.
Kinetic-Theory
Kinetic Module
Theory-Introduction
The Kinetic Module solves the FP equation (see equation 17-1) to calculate the time-dependent four-
dimensional probability distribution function (PDF) f. This equation represents a continuity equation and
expresses conservation of f in the course of particle transport due to drift and diffusion.
The first term is the transient term which describes the temporal evolution of the distribution function. The
second term describes transport in configuration (physical) space and the third term describes transport
in u-space. A uniform grid is assumed for the u-space which can represent different physical variables. In
particular, the u-space can be one-dimensional with u being energy (for particle transport problems),
radius of pores (for electroporosity problems), etc. The Kinetic Module Theory section includes:
• Theory-Electron Kinetics in Plasmas and Semiconductors
• Theory-Plasma Chemistry for Kinetic Module
Kinetic Module
Theory-Electron Kinetics in Plasmas and Semiconductors
The FP equation is derived from the Boltzmann transport equation in the case of collision-dominated
transport when the velocity relaxation time due to collisions is much smaller than the energy relaxation
time. Such a situation is rather common and corresponds, for example, to the transport of light particles
(electrons) through a background of heavy particles (neutral atoms in a gas or lattice in a solid). You can
use a two-term spherical harmonic expansion (SHE) to express the distribution function in the form:
(17-2)
The first term in equation 17-2 is a spherically symmetric (isotropic) part f0 responsible for the scalar
characteristics of the particle ensemble such as density, temperature, etc. The second term describes
vector quantities such as current density.
233
CFD-ACE_V2010.0_Modules_Manual_Part2
(17-3)
where:
The isotropic part f 0 is determined by the Fokker-Planck equation in a four-dimensional space (three
spatial coordinates plus energy)
(17-4)
Here u plays the role of a Lame coefficient in velocity space and the different terms describe (from left to
right)
234
Modules
• Transient
Kinetic Module
Theory-Electrostatic Field Effect
The presence of an electrostatic field, E=∇φ, where φ is the electrostatic potential, results in additional
terms in the FP equation:
(17-5)
The cross-derivative terms create difficulties for the numerical solution of the FP equation. There are two
methods to overcome these difficulties. The first one reduces the cross-terms to convection terms by
introducing log derivatives as:
(17-6)
(17-7)
This method is currently used in the Kinetic Module. The non-linear terms and are calculated in an
iterative manner.
If considering a homogeneous system with negligible spatial variations, the second term in equation (17-
5) can be neglected. This is known as the local approach. The local approach greatly reduces the
computational effort for solving kinetic equations. It is often used to generate a lookup table. The local
approach can be activated in MO by selecting the local Boltzmann option.
The second method uses total energy as an independent variable instead of the kinetic energy
u. This eliminates the cross derivative term and reduces the FP equation 17-5 to the following form:
(17-8)
This second method is currently known as the total-energy option, which can be selected though module
options.
Kinetic Module
Theory-Coulomb Collisions
235
CFD-ACE_V2010.0_Modules_Manual_Part2
In the cases where Coulomb interactions between electrons are important, the corresponding collision
terms can be included in the FP equation. Coulomb interactions among electrons are described by non-
linear drift and diffusion terms in energy space:
(17-9)
and:
(17-10)
Kinetic Module
Theory-Plasma Chemistry
For Plasma simulations, the electron induced reaction rates are calculated according to:
(17-11)
where f is the electron energy distribution function (EEDF), σ is the collision cross section, u is electron
energy, and v is electron velocity. If the Kinetic Module is activated, the EEDF is calculated by solving the
FP equation and the electron induced reaction rates are calculated according to equation 17-11 and
passed to the Chemistry Module. In the Reaction Manager, three types of electron induced reactions can
be specified. The first type is:
where A and B are different species (neutral, ion, excited). The properties of these species are specified
in the Property Manager and the electron energy loss specified in the Reaction Manager should be set to
zero. These reactions define the source terms in the FP equation.
The second type of reactions is:
where A denotes the same species. These reactions describe quasi-elastic collisions such as excitation
of molecular vibration and rotations. You have to enter the energy loss for these reactions in the Reaction
Manager. This type of reaction is used when you choose not to solve for the transport of these individual
states (species) in the chemistry module, but need to include the important electron energy loss in these
processes. These reactions produce a contribution to the energy convection term V u in the FP equation.
The third type of reaction is elastic collision. For these reactions, the energy loss is set to zero in the
Reaction Manager. The collision cross sections specified in the Reaction Manager and used to calculate
the diffusion coefficient D r in the FP equation The energy loss is calculated from the mass ratio of
colliding particles and the corresponding term is added to the V u . If the Kinetic Module is activated, then
you must specify the cross section for each electron induced reaction.
Kinetic-Numerical Implementation
Kinetic Module
Numerical Implementation-Introduction
236
Modules
This section describes how the numerical implementation of the FP solver is done in the Kinetic Module,
which is important to know in order to be able to successfully control the solution. The Kinetic Numerical
Implementation section includes:
• Numerical Implementation-Operator Splitting Method
• Numerical Implementation-Finite Volume Approach
• Numerical Implementation-Fluxes on Faces
• Numerical Implementation-Coulomb Collisional Integral
Kinetic Module
Numerical Implementation-Operator Splitting Method
For the numerical solution of equation 17-1 the operator splitting method is used (also called the method
of fractional steps). The splitting methods for the numerical solution of partial differential equations have
been used for two decades and consist in replacing the multi-dimensional problem by a succession of
simpler (for example, one-dimensional) problems. In order to use this method, the FP equation 17-1 is
written in an operator form as
(17-14)
where F r and F u are differential operators in coordinate and u spaces, correspondingly. A residual
function is introduced as:
(17-15)
(n)
and the iterative method to calculate f is used. Let f denote the value of f at an n-th iteration. To find the
n+1 *
value of f for the next iteration, we first solve for f in coordinate space:
where α is a linear relaxation parameter (1<α <0), and β is the inertial relaxation parameter (βδ ij term
**
implies that the relaxation applies only to the diagonal terms). Then we solve for f in u-space
Here θ is an implicitness parameter. For θ =1, the numerical scheme corresponds to a backward Euler
scheme, for θ =0 it is a forward Euler scheme. For θ =1/2, the numerical scheme becomes accurate to
Kinetic Module
Numerical Implementation-Finite Volume Approach
For the numerical solution, the FP equation 17-1 is discredited using the finite volume approach. The
finite volume discretization is obtained by integrating this equation over an arbitrary polyhedral
computational cell with arbitrary polygonal faces. In this way the divergence of a flux is reduced to flux
over the surface, and the resulting equation may be expressed in the form:
237
CFD-ACE_V2010.0_Modules_Manual_Part2
(17-
16)
where N r is the number of faces in physical space, A α is the area vector for face α, N u is the number of
faces in the u-space, A β is the area vector for face β. The superscripts n and n+1 denote the time step
number, subscript c denotes the cell centered values of variables, subscripts a and b denote the face
centered values of variables, and V c is the cell volume. The estimation of the values of the unknown at
faces is obtained by interpolation from the cell-centered values. The accuracy of the interpolation
determines the accuracy of the scheme and its computational cost. Higher order accuracy typically
requires more computing time and is usually less stable.
Kinetic Module
Numerical Implementation-Fluxes on Faces
The fluxes on faces are evaluated as the following. The diffusive flux at a representative face denoted by
a subscript e may be evaluated as:
where the diffusion fluxes and represent main diffusion and cross diffusion fluxes. By discrediting
the gradient of f at the cell face center e with a central difference scheme, we obtain:
where f c1 and f c2 are the values at the corners and can be expressed as the value at the neighboring cell
center f nb of p.
Convective flux at the face e is evaluated as:
where and f e is the value at the face center. To calculate f e , the following schemes can be
used:
1. upwind scheme
where g P = 0.5
238
Modules
*
where f is the current value of f, a nb are the link coefficients with the neighbor cells, and e fac is the
relaxation parameter which provides a link between the new values of f c and the old values of to
improve convergence.
Now we define
Assuming F rα = a rα and F uβ = a rβ , we can follow the procedure described in the previous section.
Currently, the described procedure is implemented for one-dimensional u-space.
Kinetic Module
Numerical Implementation-Coulomb Collisional Integral
For the numerical implementation of the Coulomb collision integral terms in the FP equation, we use the
following relationship (index ee is omitted for brevity):
Using this relation, the Coulomb collision terms calculated on faces in u-space are:
239
CFD-ACE_V2010.0_Modules_Manual_Part2
• Grid Generation
• Problem Type
• Model Options
• Volume Conditions
• Boundary Conditions
• Initial Conditions
• Solver Control
• Control Parameters
• Capacitively Coupled Plasma
• Global Model
• Results
Kinetic Module
Implementation-Grid Generation
The standard procedure of grid generation using CFD-GEOM can be followed to create geometry and
generate structured or unstructured grids in configuration (physical) space. The u-space grid is specified
in the Model Option Settings.
Kinetic Module
Implementation-Problem Type
Click the Problem Type [PT] tab to see the Problem Type Panel. See Control Panel-Problem Type for
details.
Select Kinetic to activate the Kinetic Module. For Plasma simulations, the Fokker-Planck solver requires
as inputs gas composition, electrostatic potential profile, and the RF electric field. The diffusion and
convection coefficients are calculated taken into account elastic collisions, Coulomb interactions, Joule
electron heating by electromagnetic fields. The source terms include inelastic and super-elastic collisions
and generation of new particles (ionization, recombination, attachments, etc.). The solver calculates the
EEDF, the rates of electron induced chemical reactions, electron density and flux, and provides these
data to Chemistry and other modules.
Kinetic Module
Implementation-Model Options
Click the Model Options [MO] tab to see the Model Options Panel. See Control Panel-Model Options for
details.
The kinetic tab enables you to enter data for different kinetic module options.
Local Boltzmann: The local Boltzmann option corresponds to the local approach of kinetic solver. It
assumes the simulation domain is spatially homogenous, and there is no variation of the PDF in physical
space. This option reduces the commutation effort and can be used to generate a lookup table for plasma
simulation.
Total Energy: The total energy option is based on equation (17-8). Since it includes the spatial variation
of the EEDF, it is generally used more than the local Boltzmann approach. The distribution function
obtained is related to the total energy of the electrons, ie, the kinetic energy plus the potential energy.
Coulomb Collisions: If the box for coulomb collisions is checked, the electron-electron coulomb collision
will be included in the simulation.
Energy grid and Boundary Condition Specification: The figure below shows how to specify
computational grid and boundary conditions in energy for calculations of the EEDF using the MO Panel.
We currently use uniform grid in the energy axis. The low energy bound is assumed to be at zero energy,
240
Modules
the maximal energy is specified by the user (in eV). The boundary conditions at min and max energies
include Fixed EEDF (Expression), Fixed Gradient, Total Flux (Convection + Diffusion), and Mixed
(A*df/de+B*f=C). For the Fixed EEDF case, the user can specify the value of f(r) using any combination
of standard analytical functions (note that vx denotes u variable in the GUI when setting Expression). The
Total Flux BC allows the user to specify the values of the particle flux along the energy axis in the FP
equation at the min and max energies. Using Mixed BC, the user can specify the values of constants A,
B, and C to define a combination of f(r) and its energy derivative at the boundaries. For the local
Boltzmann option, the energy refers to the kinetic energy. For the total-energy option, the input energy is
for kinetic energy plus potential energy.
241
CFD-ACE_V2010.0_Modules_Manual_Part2
down Delta t is set to 1, the initial time step ∆t will linearly decrease for increasing E0; if the value is 0,
constant time step is used.
Kinetic Module
Implementation-Volume Conditions
The figure below shows how to specify kinetic transport properties (diffusion and convection coefficients,
and source terms in the FP equation) in different computational domains. The transport properties are
specified as volume conditions in the VC Panel. For fluid domains, choose one of three evaluation
methods: Constant, Plasma, and Semiconductor. For plasma simulations, select the Plasma option for
the fluid domains. For solid domains, select a large value for the Spatial Diffusion Coefficient, and put
zero for Energy Diffusion Coefficient to enable zero value for the EEDF in the solid domains. Also, using
the VC Setting Panel, select Kinetic Source (Plasma) for Fluid domains to activate the source terms in the
FP equation. To account for new electrons generated in the ionization events, select the Secondary
Electrons button.
Kinetic Module
Implementation-Boundary Conditions
To assign boundary conditions and activate additional panel options, select an entity from the viewer
window or the BC Explorer. The figure below shows how to define boundary conditions in physical space.
The boundary conditions in physical space include Fixed EEDF (Expression), Fixed Gradient, Total Flux
(Convection + Diffusion), and Mixed (A*df/de+B*f=C).
For the Fixed EEDF case, you can specify the value of f(u) using any combination of standard analytical
functions (Expression). The fixed gradient BC enables you to set the value of normal derivative at the
boundary. The Total Flux BC enables you to specify the value of the spatial particle flux in the FP
242
Modules
equation at the boundary. Using Mixed BC, you can specify the values of constants A, B, and C to define
a combination of f(r) and its normal derivative at the boundary.
Kinetic Module
Implementation-Initial Conditions
The figure below shows setting up initial conditions for the kinetic module. Initial conditions can be
specified on the Volume by Volume basis. You can specify an arbitrary dependence of f(u) as an initial
condition using a set of standard analytical functions.
Initial Conditions
Kinetic Module
Implementation-Relaxation
Relaxation: The Relaxation Panel enables you to specify Relaxation Parameters affecting the
convergence process. For the Kinetic module, you can specify the EEDF Inertial Relaxation. Since the
Inertial Relaxation procedure does not necessarily conserve the number of particles, it is advisable for
-5
transient simulations to use a small value of the Inertial Relaxation (as low as 10 ).
Relaxation Parameters
Adv: Once the Adv tab is activated, the start time step and the frequency to calculate the EEDF can be
specified.
243
CFD-ACE_V2010.0_Modules_Manual_Part2
Kinetic Module
Implementation-Output
Graphical output: Under the Graphic Tab, you can select if you want EDF to be output to the graphic file
(modelname.DTF). A complete list of output variables using kinetic module is shown in the table. Theses
variables can be visualized and analyzed in CFD-VIEW.
3
N_e_B Electron number density 1/m
2
Jy_Ne_b Current density, y-component A/m
2
Jz_Ne_b Current density, z-component A/m
Monitor: The window shows how to select monitor points and monitor variables. Monitor points play a
double role when using the Kinetic module. As for all other modules, the monitor points are used to
monitor the values of variables during the simulation process. In addition, for the kinetic module, the
monitor points are used to visualize simulation data for f(u) at selected points in physical space (see the
Visualization section below).
244
Modules
Implementation-Control Parameters
Kinetic Module
Implementation-Control Parameters-Introduction
To control the solution in the Kinetic Module, we have introduced several parameters that can be
specified via an input file (command: DTF -ud model.DTF < model.in). To enter these parameters into a
model.DTF file, create an ASCII file (for example, module.in) containing these lines and then update the
DTF file by executing the following command: DTF -ud model.DTF < model.in. The plasma related entries
of the input file are (in this examble, # is comment, 0 = no, 1 = yes):
# Specify energy as independent variable in FP equation
Energy int 1 0
# Include RF Heating
RF_Heating int 1 0
# RF Field frequency (Hz)
RF_frequency_Boltz double 1 0
# RF Field amplitude (V/m)
RF_Field_Boltz double 1 100
Here, the first line specifies (kinetic) energy as an independent variable in the FP equation. If the plasma
module is activated, this option is turned on by default. The sixth line turns on/off RF field heating of
electrons; the seventh line gives the RF field frequency in Hz and finally the last line -- its amplitude in
V/m.
The Kinetic module can stand alone or together with other modules (electrostatic, chemistry, plasma,
electromagnetic). The electrostatic module provides with the electrostatic potential. The chemistry module
provides with the collision cross sections required to calculate the diffusion coefficient and sources. These
data are stored in the Reaction Manager. The electromagnetic module provides with the electromagnetic
fields responsible for electron heating in ICP. For Plasma applications the energy diffusion coefficient is
calculated by four different ways.
245
CFD-ACE_V2010.0_Modules_Manual_Part2
Kinetic Module
Implementation-Control Parameters-Positive Column
When Plasma is activated and a PC option is chosen, the energy diffusion coefficient is calculated as:
Kinetic Module
Implementation-Control Parameters-Inductively Coupled Plasma
Electron heating in an ICP is produced by RF electromagnetic fields. These fields are calculated by the
electromagnetic module.
where is the induced electric field provided by the electromagnetic module, and ω is the field
frequency.
Kinetic Module
Implementation-Capacitively Coupled Plasma
For CCP simulations, the electrostatic potential φ is separated into a steady-state component and an
oscillating component :
,
where ω is the RF frequency. The steady-state component is responsible for electron confinement and
the oscillating component determines the electron heating. The electric field determines the
energy diffusion coefficient:
(17-17)
Kinetic Module
Implementation-Global Model
For the Global model, the energy diffusion coefficient is calculated from the total power adsorbed in the
Plasma.
Kinetic Module
Implementation-Results
Visualization of multi-dimensional kinetic simulation data for f(r,u) is a challenging task. We have
developed two ways to visualize these data.
246
Modules
• Spatial distributions of f(r,u) for different u are saved as different variables into a DTF file (as
variables F_1 through F_N, where N is the number of grid points in energy) and can be viewed
using CFD-VIEW.
• The data for f(r,u) in selected monitor points in physical space are written into separate ASCII
files (named *.K.L.BOL, where * denotes the name of the DTF file, K is the number of time step
(for transient simulations), L is the number of monitor point, and BOL is the file extension). You
can view these using any XY plotting routines (we use Grace and Tecplot).
Kinetic Module References
Bringuier, E., "The current equation in strong electric field." Philosophical Magazine B 79,1659 (1999)
Kolobov, V., and Godyak, V.A., Nonlocal Electron Kinetics in Collisional Gas Discharge Plasmas, IEEE
Trans Plasma Science 23, 503 (1995)
Risken, H., The Fokker-Planck Equation, Springer, Berlin 1984
Van Kampen, N.G., Stochastic Processes in Physics and Chemistry, 1992
Semi-Device Module
247
CFD-ACE_V2010.0_Modules_Manual_Part2
Figure 1: Sample Octree Mesh for a Basic MOSFET Transistor and Predicted Electron Density
Distribution Using Semi-Device Module
Figure 2: ACE+ Semi Device Simulation Results for VCSEL Device: Model Setup, Energy Band
Diagram, and Predicted Electron and Hole Densities
Semi-Device Module Features
The Semi Device module has been incorporated in CFD-ACE+ in a very general form with flexibility in
geometry, mesh, materials, complexity of physics, high fidelity of numerical methods and boundary
conditions and links to other physics modules of CFD-ACE+. The features and capabilities of the Semi
Device module are:
• 1D/2D/3D geometry and all types of meshes - structured, unstructured and octree
• Steady and transient simulations
• Fermi-Dirac and Boltzmann statistics
• Drift Diffusion (DD) and Hydrodynamic (HD) Device model implementation
• Coupling with Lattice Heating and Heat Spreading
• Silicon and Compound Semiconductor Materials
• Heterostructures and Quantum Wells
• Ohmic and Schottky Contacts
• Mobility Models dependent on longitudinal and transverse electric fields, temperature, doping,
mole fractions (for compound materials) and so on.
248
Modules
• Impact Ionization
• Link to circuit models (SPICE) for mixed-mode simulations
• Photon number density equation for optoelectronic devices
• Link to Electromagnetic Optics and Scalar Optics (WIM)
• Link to Electromagnetics for RF devices (under development)
• Scharfetter Gummel Numerics
• Sequential Gummel and Full Newton Solution Algorithms
Semi-Device-Theory
Semi-Device Module
Theory-Introduction
The fundamentals of semiconductor device physics can be found in text books such as Sze and device
modeling is very well discussed by Selberherr. More advanced quantum physics relevant to modern
semiconductor devices, including opto-electronics, has been discussed in Hess and Ferry. The objective
of this chapter is to provide a short summary of the structure and operation of electronic devices, the
basics of semiconductor physics and a detailed description of semiconductor equations and modeling
techniques used in the Semi Device module.
The Semi-Device Theory section includes:
• Theory-Structure and Operation of Semiconductor Devices
• Theory-Physics of Electrons in Semiconductor Materials
• Theory-Semiconductor Equations
• Theory-Drift-Diffusion (DD) and Hydrodynamic (HD) Transport Models
• Theory-Lattice Energy Equation
• Theory-Insulator Equations
• Theory-Optical Models
• Theory-Boundary Conditions
• Theory-Mixed-Mode Simulation: Device+Circuit (SPICE)
Semi-Device Module
Theory-Structure and Operation of Semiconductor Devices
A pure semiconductor material is essentially neutral. It contains no free electrons in its conduction bands.
By adding very small amounts of selected additives (impurities) to them (doping process) one creates
either
• n type semiconductors with excess electrons and donor impurity ions
• p type semiconductors with excess holes and acceptor impurity ions
The purpose of semiconductor doping is to increase the number of free charges that can be transported
by an externally applied voltage.
Although the n-type material has an excess of free electrons, it is still electrically neutral because the
donor atoms are left with positive charges. For the same reason, the p-type material is also electrically
neutral because the excess holes are balanced by the negative acceptor ions. The holes and electrons
are free to move in these materials because they are only loosely bound to their parent atoms, whereas
the impurity ions are fixed in place in the crystal lattice structure and cannot move. The electrons and
holes are also called charge carriers.
249
CFD-ACE_V2010.0_Modules_Manual_Part2
The p-n junction, a fundamental building block in microelectronics, consists of two adjacent
semiconductor regions of opposite doping. It shows a pronounced rectifying behavior and is therefore
often called a p-n diode. It is also an essential part of Metal-Oxide-Semiconductor-Field-Effects-
Transistors (MOSFETs), Bipolar Junction Transistors (BJTs), solar cells, photodiodes, LEDs and even
laser diodes.
In an unbiased p-n junction the electrons from the n material diffuse across the junction into the p material
and fill some of the holes. At the same time, the holes from the p material diffuse across the junction into
the n material and are filled by electrons. Because there is a depletion of free electrons and holes in an
area around the metallurgical junction, it is known as the depletion region. The figure below illustrates the
p-n junction formation.
250
Modules
The current flow across the barrier, however, is not quite zero because of the minority carriers crossing
the junction. There are minority charge carriers in both regions: holes in the n material and electrons in
the p material, that are created by thermal generation. Under reverse bias, electrons in the p-type
material are repelled towards the junction by the negative terminal. As the electrons move across the
junction, they neutralize positive ions in the n-type material. Similarly, holes in the n-type material are
repelled by the positive terminal towards the junction. As the holes crosses the junction, they neutralize
negative ions in the p-type material. This movement of minority carriers generates a minority current.
Key components of modern integrated circuits (ICs) are three terminal devices: Bipolar Junction
Transistors (BJT) and Field Effect Transistors (FET). A small signal applied to one of the terminals (gate)
can modulate large changes of current through the other two terminals (contacts).
The BJT in the following figure is composed of two p-n junctions in series (e.g. p-n-p or n-p-n). The three
regions are referred to as: emitter, base and collector, and are connected by means of ohmic contacts to
external circuits. The operation of a BJT can be described in the following manner:
• The emitter current is determined by the injection of majority carriers (holes in this case) from the
emitter to the base. This is done by heavy doping of the emitter compared to base.
• A small fraction of these holes recombine with electrons in the base but most of them diffuse to
the reverse biased collector-base depletion layer where they are whisked away to the collector by
the high field in the depletion layer.
• The majority carriers in the base (electrons for p-n-p) are trapped in the potential well, with a
barrier on each side. This explains why the minority carriers carry most of the current.
A p-n-p bipolar transistor with forward bias emitter-to-base voltage V EB and reverse bias base-to-
collector voltage V BC .
The device is called bipolar since its operation involves both types of mobile carriers, minority and
majority (electrons and holes).
Another very important device is the Metal-Oxide Semiconductor Field Effect Transistor (MOSFET). The
n-channel MOSFET (see following figure) consists of a source and a drain, two highly conducting n-type
semiconductor regions which are isolated from the p-type substrate by reversed-biased p-n diodes. A
metal (or polysilicon) gate covers the region between the source and the drain, but is separated from the
semiconductor by a gate oxide.
251
CFD-ACE_V2010.0_Modules_Manual_Part2
Cross section, Operation Principle, and Contact Circuits of an n-type Metal Oxide Semiconductor
Field Effect Transistor (MOSFET)
The voltage applied to the gate controls the flow of electrons from the source to the drain. A positive
voltage applied to the gate, attracts electrons to the interface between the gate oxide and the
semiconductor. These electrons form a conducting channel just below the gate oxide between the source
and the drain, called the inversion layer. Since the gate oxide blocks any carrier flow, no gate current is
required to maintain the inversion layer at the interface. The current between the drain and the source is
controlled by the voltage which is applied to the gate.
The MOSFET provides current and voltage gain (amplification). The current gain results from the fact that
no gate current is required to maintain the inversion layer and the resulting current between drain and
source. The device has therefore an infinite current gain in DC. The current gain is inversely proportional
to the signal frequency. The voltage gain of the MOSFET is due to the fact that the current saturates at
higher drain-source voltages, and therefore a small drain current variation can cause a large drain voltage
variation.
Both n-channel and p-channel MOSFETs can be formed on p-type substrates in which localized n-type
wells are created by diffusion (see following figure). When properly connected in pairs these n- and p-
channels form complementary or CMOS circuits.
Semi-Device Module
Theory-Physics of Electrons in Semiconductors
Fundamentally, electron transport in crystals has to be described in terms of quantum physics
(Schrödinger equation). However, for practical devices, the solution of Schrödinger's equation for each
carrier is both mathematically challenging and computationally intensive. In semiconductors, where a
large number of electrons are involved, we use Fermi-Dirac statistics to describe the average occupation
number of a state of energy E as
252
Modules
(18-1)
where E F is the Fermi energy (chemical potential), and k Β is the Boltzmann constant. For energies above
E F , i.e. (E-E F )>>k Β T, the exponential factor is large and +1 can be neglected. Thus, we get the
Boltzmann statistics (distribution):
(18-2)
These statistics are used in the derivation of semiconductor properties and equations.
In a semiconductor, the electrons reside within two energetic values or bands, viz. the valence band (EV),
and the conduction band (EC). Separating these two bands is a band of forbidden states, the energy
bandgap, EG=EC-EV. The energy band diagram (see figure below) determines the electrical and optical
characteristics of a semiconductor. In a metal the valence and the conduction bands overlap and no
forbidden states exist. In an insulator material the band gap is very large (Eg > 3eV)
The left half shows a simplified energy diagram of semiconductors with intrinsic, n-type and p-
type properties. EC - energy of conduction band, EV - energy of valence band, EF - Fermi level
energy, ED, EA - energy of donor or acceptor dopants respectively. The right half of the figure
illustrates light absorption by an intrinsic semiconductor. The impinging photon induces an
electron transition from the valence band into the conduction band, if hν>Eg.
At room temperature some electrons from the valence band are excited into the conduction band leaving
behind positive holes. In the simplest case, the equilibrium concentrations of electrons and holes are both
equal, n 0 =p 0 . The electrons in the conduction band and the holes in the valence band are both
responsible for transmission of electric current. Such an ideal material is called an intrinsic
semiconductor. The thermal excitation of electrons into higher energetic levels is described by the
Boltzmann distribution and allows the equilibrium concentrations of electrons and holes to be calculated
as:
(18-3)
(18-4)
where N C and N V are the effective density of states in the conduction and valence bands. The Fermi
energy of an intrinsic semiconductor lies midway between the conduction and valence band energies at a
temperature of T=0K. Similarly, at higher temperatures, excited electrons must be taken into account
resulting in:
253
CFD-ACE_V2010.0_Modules_Manual_Part2
(18-5)
Additional electrons in the conduction band and holes in the valence band can be generated by adding
(doping) small amounts of acceptor or donor elements (dopants) into the semiconductor to produce
extrinsic semiconductors.
Arsenic (a group V element) behaves as an electron donor when introduced into the intrinsic Si (a group
IV element) lattice. This results in a donor level (E D ) near the conduction band as shown above. In such a
material most of the electrical conductivity is due to the conduction band electrons which are the majority
carriers. The holes make a small contribution to the conductivity and are called the minority carriers. A
material doped with donor atoms is called an n-type (negative-type) semiconductor.
Conversely, introducing an acceptor element (e.g. Ga-group III element) in intrinsic Si creates an
acceptor level (E A ) near the valence band. In this case electrons are thermally excited from the valence
band into these acceptor sites, leaving behind mobile holes which become the majority carriers. Electrons
are now the minority carriers and the material is called a p-type (positive-type) semiconductor.
For doped material the Fermi level (E F ) location depends strongly on the doping level N A or N D . For
17 -3
moderately or heavily doped n-type solids (N D >10 cm ), E F lies just below the conduction band edge
and similarly, for heavily doped p-type materials, E F lies just above the valence band edge.
In optoelectronic devices, lattice atoms and carriers interact with light (leading to generation, emission,
absorption and scattering). When an intrinsic semiconductor is irradiated with light of energy larger than
the band gap energy (hν>Eg), an electron from the valence band can be promoted into the conduction
band (see figure b).
Semi-Device Module
Theory-Semiconductor Equations
The Semi Device Module has two transport models: drift diffusion (DD) and hydrodynamic (HD). The
lattice temperature can be treated either as a constant or can be solved by the ACE+ energy equation
module. The basic equations solved in the Semi-Device simulator are the Poisson equation:
(18-6)
(18-7)
(18-8)
The unknown quantities of this equation system are the electrostatic potential psi and the electron and
+ -
hole concentrations, n and p respectively. N D and N A denote the net concentration of the ionized
dopants, ε is the electrical permittivity of the semiconductor, and G and R are the carrier generation and
recombination rates.
Semi-Device Module
Theory-Drift Diffusion and Hydrodynamic Transport Models
The DD and HD current relations can be derived from the Boltzmann transport equation by the method of
moments and are calculated in ACE+ Semi Device as:
254
Modules
These current relations account for position-dependent band edge energies, E c and E v , and for position-
dependent effective masses, m n and m p , which are included in the effective density of states, N n and N p .
In the DD model carrier temperatures are assumed to be equal to the lattice temperature T L .
In the HD model additional energy balance equations are solved for electrons and holes in addition to the
lattice energy equation. This model is particularly useful for ultra small devices which can potentially
exhibit overshoot:
(18-9)
(18-10)
Here, w n and w p are the carrier kinetic energy densities, τ wn and τ wp denote the energy relaxation times,
and S n and S p are the carrier energy fluxes which are computed as follows:
(18-11)
(18-12)
The carrier thermal conductivities, k n and k p , are calculated from a generalized Wiedemann-Franz law.
(18-13)
(18-14)
Semi-Device Module
Theory-Lattice Energy Equation
255
CFD-ACE_V2010.0_Modules_Manual_Part2
The lattice temperature, T L , is solved by the CFD-ACE+ Heat module and is used in the Semi-Device
module. Strong (implicit) two way coupling is maintained. The model accounts for self-heating effects in
semiconductor devices by solving the lattice heat flow equation self-consistently with the DD or HD
transport equations. The CFD-ACE+ energy equation is:
(18-15)
where h is the lattice enthalpy, k is the lattice thermal conductivity, R is the carrier recombination rate and
Eg is the band gap. Note that the last two terms represent Joule heating and carrier recombination.
The CFD-ACE+ Heat model offers a link to other physical processes such as conjugate heat transfer,
thermal stress, piezoelectric, and thermoelectric effects, electrochemistry and others.
Semi-Device Module
Theory-Insulator Equations
In insulating materials, only the electrostatics equation and the lattice heat flow equation need to be
solved. The presence of doping and mobile carriers can be neglected (the carrier continuity equations can
be omitted). This assumption also simplifies the Poisson equation to the Laplace equation for ψ. In the
Semi-Device module, you can solve the carrier equations in insulators (e.g. SiN type materials or to study
hot electron effects or charge trapping) with appropriate material properties.
Theory-Optical Models
Semi-Device Module
Theory-Optical Models Introduction
The physics of vertical-cavity surface-emitting lasers (VCSELs) and edge-emitting lasers (EELs) includes
strong nonlinear coupling of time-dependent and space-dependent optical fields, electric carriers (electron
and holes) distributions in semiconductors, and temperature. Moreover, multiple optical modes are
coupled and energy transfers to the modes are better supported by carrier concentration distribution at a
given time moment. The Semi-Device Module uses an iterative approach to solve this stiff system of
equations. The following sections describe basic equations and methods applied in the semiconductor
laser modeling software.
Optical models include the Weighted Index Method (scalar) and the Method of Lines (scalar and
vectorial). Both edge-emitting lasers and VCSELs can be efficiently modeled. This enables you to study
the transient behavior of lasers with time-dependent and spatially inhomogeneous gain. Multiple quantum
wells and multiple transverse modes are included. For each time step, a calculation loop over electrical,
thermal, gain, and optical solvers is performed.
Semi-Device Module
Theory-Optical Models-Weighted Index Method (WIM) for VCSELs
Solution for VCSEL transverse and axial field distribution in the modes as well as their lasing frequencies
consists of solution of an eigenvalue problem for Helmholtz equation in cylindrical coordinates, and
boundary value problem set by the boundary conditions at VCSEL layer and radial region interfaces.
To solve the optical eigenvalue problem for VCSEL the Weighted Index Method (WIM) (Robertson 1989),
(Noble 1998) was applied. This method solves the Helmholtz equation for axisymmetric problem using
separable eigenmode fields and harmonic azimuthal field dependence:
256
Modules
where are respectively radial, axial, and azimuthal coordinates, E, G, F are electric field complex
amplitudes, ω is angular frequency, c is speed of light, is the complex relative permittivity, and m is
azimuthal mode number.
The original Helmholtz equation is replaced by two equations for axial and radial fields.
where , are propagation constants in axial and radial direction, respectively, and indices j, i
denote material layer perpendicular to optical axis, and radial region respectively.
The propagation constants , couple these equations. We find VCSEL linearly polarized (LP)
eigenmodes by alternatively solving axial and radial constraint equations, updating weighted propagation
constants, and the weighted permittivity. This iterative improvement of solutions can be understood as an
enhancement of the popular effective index method.
The solutions are of the form:
where and are Bessel functions of the m-th order of the first kind and modified
Bessel function of the m-th order of the second kind, is the oxide aperture radius, and a, b and c are
unknown constants.
The result of the WIM analysis is a set of eigenmodes (bound in radial direction and radiative in the
axial direction), m is azimuthal and n is radial mode number.
The unknowns are, for each lasing eigenmode, the F(ρ), G(z), the resonant frequency ω, and the
imaginary part of permittivity in the active region (related to material gain).
Semi-Device Module
Theory-Optical Models-Photon Rate Equations
For each mode, the photon rate equation is solved including optical losses, and spontaneous and
stimulated emission terms taking into account spatially and time-dependent material gain:
where is number of photons in the mode, t is time, is stimulated emission rate per photon in
the mode, is the quantum wells volume, is the photon lifetime describing losses through the
mirrors and absorption, is spontaneous emission rate, is a factor describing geometrical and
spectral coupling of the spontaneous emission into a mode.
257
CFD-ACE_V2010.0_Modules_Manual_Part2
Corresponding stimulated and spontaneous emission terms are responsible for carrier relaxation in the
electrical model.
Theory-Boundary Conditions
Semi-Device Module
Theory-Boundary Conditions-Introduction
The CFD-ACE+ Semi-Device Module provides a range of practical boundary conditions (BCs) required to
model very complex geometries. At each boundary BCs are needed for all dependent variables: ψ, n, p,
T n , T p and T L . Semi Device provides both external and internal boundaries (e.g., hetero-interfaces) and
treats them fully implicitly. The following boundaries are described next: zero flux, semiconductor-metal,
insulator metal, insulator-insulator and semiconductor-semiconductor.
The Semi-Device Theory Boundary Conditions section includes:
• Zero Flux Boundaries
• Semiconductor Metal Boundaries-Ohmic Contacts
• Semiconductor Metal Boundaries-Schottky Contacts
• Insulator Metal Boundaries
• Semiconductor Insulator (Floating Contact) Boundaries
• Semiconductor Hetero-interface Boundaries
• Lumped Circuit Elements Contact
Semi-Device Module
Theory-Boundary Conditions-Zero Flux Boundaries
Zero flux boundary conditions are used on symmetry planes, far field boundaries and other external
planes. At the insulator-semiconductor interface, in the absence of surface charges, zero gradient electric
field is applied. The Neumann boundary condition (zero normal gradient) ensures that the simulation
domain is self-contained and there are no fluxes across the boundary.
Semi-Device Module
Theory-Boundary Conditions-Semiconductor Metal Boundaries-Ohmic Contacts
An ohmic contact is fabricated to have negligible contact resistance compared to the bulk material. At
Ohmic contacts, Dirichlet boundary conditions for the contact potential and carrier concentrations n and p
are applied. In the HD simulation case, the contact carrier temperatures, T n and T p , are set equal to the
lattice temperature. The electrical boundary conditions are derived from the assumption of quasi neutrality
at the contact and by assuming that the carrier quasi-Fermi potentials are equal to the applied voltage
bias f n = f p =V appl . The contact potential at the semiconductor boundary is the sum of the applied voltage
and built in potential:
(18-16)
258
Modules
(18-17)
Here, C is the net concentration of dopants and other charged defects at the contact boundary. The
additional constants are defined by
(18-18)
and:
(18-19)
The carrier concentrations in the semiconductor are equal to the equilibrium carrier concentrations at the
contact:
(18-20)
and:
(18-21)
In the above equation, the temperature can correspond to either individual carrier or lattice temperatures
depending on whether the HD or DD model is used. If the semiconductor material near the contact has
low doping or if high current injection is needed, the above ideal ohmic contact is not valid. In some
applications and materials the model will not inject carriers through the contact and unrealistic currents
will be predicted. It is recommended to use high doping in the contact vicinity.
Semi-Device Module
Theory-Boundary Conditions-Semiconductor-Metal Boundaries-Schottky Contacts
Schottky contacts have a rectification property i.e. they conduct a large current if the junction is forward
biased and a very small current under reverse bias. The surface potential is calculated as:
(18-22)
259
CFD-ACE_V2010.0_Modules_Manual_Part2
where Φ MS = Φ M -Φ S is the difference between the work functions of the metal and semiconductor (user
specified). Note that Φ S is the electron affinity of the semiconductor. N C is the conduction band density of
states, and n i,eff is the effective intrinsic concentration in the semiconductor. For the carrier concentrations
normal flux conditions are set:
(18-23)
and:
(18-24)
where v n and v p are the thermionic recombination velocities. The carrier concentrations are calculated as:
(18-25)
and:
(18-26)
Semi-Device Module
Theory-Boundary Conditions-Insulator Metal Boundaries
The insulator contact potential is the difference between the metal quasi Fermi level (equal to the applied
voltage) and the metal work function difference potential Φ MS
(18-27)
Semi-Device Module
Theory-Boundary Conditions-Semiconductor-Insulator (Floating Contact) Boundaries
At the semiconductor insulator interface, not connected to a voltage source (floating contact), the normal
component of dielectric displacement and continuity of potential are used as the boundary condition:
(18-28)
260
Modules
and:
(18-29)
where E is the electric field and σ is the surface charge. By default σ=0. Insulating contacts connected to
a voltage source are treated similar to the Schottky contacts. Electron and hole concentrations within the
insulator and at the interface are set to zero.
Semi-Device Module
Theory-Boundary Conditions-Semiconductor Hetero-interface Boundaries
This is the internal boundary at the interface between two different semiconductor materials (e.g. GaAs
and AlGaAs). The electrostatic potential boundary condition is set automatically inside the code using
similar approach as in the Semi-Insulator boundary described above. Calculation of carrier concentration
fluxes is nonlinear and more involved. CFD-ACE+ Semi Device uses the thermionic emission model:
(18-30)
and:
(18-31)
(18-32)
A similar equation is applicable for holes. In the above equations, the emission velocity and the barrier
height lowering are calculated as:
(18-33)
and:
(18-34)
The barrier height lowering depends on the normal electric field and the effective tunneling length. This
model accounts for tunneling effects through the heterojunction barrier by introducing a field dependent
barrier height lowering.
Semi-Device Module
Theory-Boundary Conditions-Lumped Circuit Elements Contact
Lumped R,L,C elements are used to represent parts of semiconductor devices or interconnects. In some
cases the physical contacts are very far away from the active region (e.g. channel or quantum wells) and
261
CFD-ACE_V2010.0_Modules_Manual_Part2
thus, a large number of grid cells would be required to mesh the entire region. For example, for MOSFET
substrate contacts one does not need to mesh all the way to the wafer backside. Those areas also are
ohmic and can be represented as resistors or capacitive elements. Capacitance increases with the device
width while resistance decreases.
Lumped element models (and any other electronic circuits) may be attached to device contacts (boundary
conditions) by means of the "mixed-mode simulation” capability described in the next section.
Semi-Device Module
Theory-Mixed-Mode Simulation: Device+Circuit (SPICE)
ESI CFD has developed a software coupling mechanism and procedures to enable 2D or 3D multiphysics
CFD-ACE+ simulations coupled with an external circuit simulator (the popular SPICE from UC Berkeley).
Such a methodology (simultaneous solution of device and circuit equations) has been called mixed-mode
or mixed-level simulation (Rollins 1988, Mayaram 1993).The coupling mechanism is enabled by an
additional program, CFD-AceSpice, also provided by ESI CFD in the CFD-ACE+ package.
CFD-AceSpice is the controlling interface between the CFD-ACE+ (Semi-Device Module) and the
SPICE3 code. It starts instances of both solvers, coordinates information exchange between the two
solvers, and controls the progress of the simulation based on the convergence criteria, tolerances, etc.,
specified in the input circuit (.cir) netlist file.
Semi-Device-Physical Models
Semi-Device Module
Physical Models-Carrier Mobility Models
Semi-Device uses various levels of electron-hole mobility (µ) models for silicon (Si):
Level 1 (const)
(m = constant)
Level 1 (const)
µ = const. = µ
0
Level 2 (L)
The mobility in this model is referred to as µ because it reflects Lattice Scattering, i.e. the interaction of
L
carriers (electrons and holes) with the thermally generated vibrations of the atoms of the crystal. The
following power law is used for both electron and hole mobility calculations:
Level 3 (LIC)
µ = µ(T, C, n, p) - dependent on temperature, doping and electron-hole density, which reflects Lattice,
Impurity and Carrier-Carrier Scattering.
262
Modules
Level 3 (LIC)
µ = µ(T, C, n, p) = µ
LIC
LIC
This mobility model (µ ) accounts for three basic phenomena:
• µ - reflecting Lattice Scattering (as in level 2) - depends on temperature T
L
• µ - reflecting Ionized Impurity Scattering, i.e. the interaction of carriers (electrons and holes)
I
with ionized atoms of dopants (acceptors and donors) - depends on the total doping density C =
N D + N A;
• µ - reflecting Carrier-Carrier Scattering (CCS), i.e. the interaction between free charge carriers
C
Level 4 (LICE)
This mobility model includes the level 3 model µ , plus dependence on electric field value, E. It reflects
LIC
the saturation of carrier drift velocity under high electric fields. We use the formula proposed by Caughey
and Thomas (1967) and later verified by Alley (1983):
where:
vsat = 1.05 * 109 m/s for electrons at T= 300 K
8
vsat = 9.50 * 10 m/s for holes.
The temperature dependence of the saturation velocity, v sat , is given by:
vsat ~ (T/300)-0.87 for electrons (Canali, 1975)
263
CFD-ACE_V2010.0_Modules_Manual_Part2
Level 5 (LICES)
This mobility model includes the Level 4 model µ , plus surface scattering, S. This effect is of
LICE
fundamental importance in all devices where current flow takes place primarily along a surface of the
device, like for instance, MOS transistors. The influence of surface scattering is accounted for as follows
(Selberherr, 1984, p.101):
ref
In this equation, x denotes the distance normal to the interface, and x represents a characteristic length
describing the range of influence of the surface. As is intuitive, the reduction in mobility is maximum
ref
directly at the surface (x=0) and becomes less important as the normal distance x increases. x is
calculated as:
where:
ref
The above range of x values is derived in terms of the longitudinal electric field between the carrier
source (emitter) and the drain (collector). As the magnitude of this electric field (E n,p ) increases, the range
of influence of surface scattering effects decreases and velocity saturation effects become important. In
other words, carriers traveling with the saturation velocity (hot carriers) do not experience surface
scattering effects to the extent that other low velocity carriers (cold carriers) do.
Surface scattering could be due to two phenomena: (a) electrostatic forces imposed on carriers by the
transverse electric field and (b) surface roughness. Parameter b in the mobility equation above indicates
the strength of influence of surface scattering, and is given as:
where:
represents the transverse electric field, i.e. the projection of the electric field normal to the surface on
the direction normal to carrier flow (Selberherr, 1984)
Other materials: Carrier mobility values for compound semiconductors (e.g. AlGaAs and SiGe) are
calculated according to the mole fractions of the constituent elements. Temperature dependence of
mobility is accounted for in SiGe and GaN. In other cases (e.g. III-V compounds such as GaAs, InP etc.)
constant values available in the literature are utilized.
264
Modules
Semi-Device Module
Physical Models-Carrier Generation-Recombination Models
The Semi Device Module offers the following generation-recombination (G-R) models:
1. Shockley-Read-Hall (SRH) model (involving two species - electrons and holes)
2. Auger model (involving three carriers)
Apart from these G-R models, the following specialized generation models are also available:
1. Photo generation model (which accounts for excitation of semiconductor devices by
electromagnetic waves, e.g. light)
2. Impact Ionization (which accounts for the avalanche generation of excess carriers under the
influence of very high internal electric fields)
The Carrier Generation-Recombination Models section includes:
• SRH Model and Parameters
• Auger Model and Parameters
Semi-Device Module
Physical Models-Carrier Generation-Recombination Models-SRH Model and Parameters
(18-35)
Here n and p are the electron and hole concentrations and n ie is the effective intrinsic concentration,
which may take the energy band gap narrowing phenomena into account.
2
n 1 and p 1 usually obey the law of mass action i.e. n 1 .p 1 = n ie . When the carrier trap level responsible for
generation-recombination is located near the middle of the energy band gap, n 1 = p 1 = n ie and R SRH is
maximum. The lifetime parameters τ n and τ p can be modeled on three levels:
• Level 1 - constant values, typically in the range 1-10 µs
• Level 2 - lifetimes dependent on total doping density (N D + N A ):
(18-36)
with typical values of the parameters being: τ 0 = 1e-5 s, N ref = 3e17 cm , κ = 0.3 -0.6.
-3
However, these values are highly dependent on the specific device structure. In particular,
the basic lifetime value τ 0 can vary significantly depending on technology (Turowski et al,
1991).
• Level 3 - lifetimes dependent additionally on temperature (T): The temperature dependence of the
lifetime parameters, as derived from experimental results in Dorkel (1982), for the temperature
range 300-450 K, can be described by the following formula:
τ(T) = τ(300 K) * [1 + 0.0125*(T-300)] (Turowski, 1992)
265
CFD-ACE_V2010.0_Modules_Manual_Part2
Semi-Device Module
Physical Models-Carrier Generation-Recombination Models-Auger Model and Parameters
Unlike the SRH parameters, Auger coefficients are almost independent of doping, carrier density and
temperature (Beck, 1973), (Dziewior, 1977).
(18-37)
The R SRH parameters proposed by Dziewior and Schmid (1977) are often used in modeling programs and
are summarized in the following table:
Au Au
Temperature Cn Cp
6 -1 6 -1
[K] [cm s ] [cm s ]
300 2.8e-31 9.9e-32
400 2.8e-31 1.2e-31
The intrinsic concentration (n ie ) in the equations for R SRH and R Aug can be calculated from the following
formula:
(18-38)
(Adler,
1981)
10 -3
For T = 300 K, the above formula gives the typically used value for silicon: n i = 1.45⇒10 cm .
The above equation for n ie does not include band-gap narrowing effects. For high doping concentrations
18 -3
(> 10 cm ), the energy gap between the valence and conduction energy bands in semiconductors
becomes narrower, resulting in higher intrinsic concentration of carriers than in semiconductors with low
doping levels. This effect is called band-gap narrowing (BGN) and is modeled by using effective intrinsic
concentration n ie , which can be calculated by the empirical formulae given in Dorkel (1982). Note that
N DA = N D -N A :
for N-type silicon (N DA > 0)
(18-39)
266
Modules
(18-40)
Semi-Device Module
Physical Models-Carrier Mobility Models-Photo Generation
The generation of excess carriers due to excitation of semiconductors by electromagnetic waves such as
light, is modeled in the Semi Device Module as follows :
G = ηW(x,y)/hν
where G is the number of excess carriers generated per second per m , η is the internal quantum
3
3,
efficiency (normally equal to 1), W(x,y) is the power dissipated in Watts/m , h is Planck's constant (equal
to 6.634e-34 Js) and ν is the frequency of incident light.
The power density absorbed at each point (i.e. W(x,y)) in the equation above is calculated using the
Magnetic Module available in CFD-ACE+. The excess carriers due to photo generation are then
incorporated into the carrier continuity equations in the Semi Device Module to observe the effect of
electromagnetic wave (light) impingement on a semiconductor device.
The following figures illustrate representative results of photogeneration simulations done on a Metal-
Semiconductor-Metal (MSM) photodetector:
Electric field distribution inside a standard MSM architecture, calculated by the Magnetic Module
in CFD-ACE+
Semi-Device Module
Physical Models-Carrier Mobility Models-Impact Ionization
267
CFD-ACE_V2010.0_Modules_Manual_Part2
The combination of high bias and small dimensions leads to very high internal electric fields in devices
(e.g. power transistors), especially under reverse bias. These electric fields accelerate mobile charge
carriers (electrons and holes) to very high velocities (and thus, kinetic energies), leading to hot-carrier
effects. Impact ionization is one such effect, and is the process of electron-hole pair (EHP) creation
through the breaking of a lattice bond by a charge carrier whose kinetic energy exceeds the threshold for
bond breaking.
An electric field dependent impact ionization model has been implemented in CFD-SEMI. The carrier
generation rate due to impact ionization is expressed as
(18-41)
or:
(18-42)
where G is the number of carriers generated per cm per second, α i ’s (i = n,p) are the carrier ionization
3
rates and are defined to be the number of EHPs generated by an energetic carrier per unit distance
traveled, n and p are the electron and hole concentrations respectively, v n and v p are the mean velocities
of electrons and holes respectively, J n and J p are the electron and hole current densities, and q is the
electron charge constant. The α i ’s are strong functions of the electric field, and are calculated as:
(18-43)
, i = n,p
where:
(18-44)
In the impact ionization model implemented in CFD-SEMI, α i • is assumed to be a constant and is set
equal to C 1 . The exponents β i lie in the range 1-2. Values of α i and β i for different materials can be found
in Selberherr (1984). T L is the lattice temperature (constant in the drift diffusion model). In the
hydrodynamic model, individual carrier temperatures must be used instead. The dot product E⇒ J
268
Modules
crit
indicates that impact ionization occurs only if the electric field is aligned with the current flow. E i is the
critical electric field and impact ionization occurs when the field due to the applied reverse bias exceeds
this critical value.
Semi-Device Module
Physical Models-Thermal Conductivity
The variation of thermal conductivity with temperature is modeled by means of a simple power law, as
shown below:
(18-45)
where κ represents the thermal conductivity and α is an experimentally determined value (α = -1.65 for Si
at 300 K). Tabulated values for α at other temperatures can be found in literature (Dorkel, 1982).
Semi-Device-Numerical Schemes
Semi-Device Module
Numerical Schemes-Gummel-Scharfetter Scheme
To avoid numerical instability in semiconductor simulations, the Gummel-Scharfetter discretization
method is implemented in the CFD-Semi Device module.
For two given neighboring cell centers p and g, the Gummel-Scharfetter scheme, using the following
function derived from an exact 1D solution, is used to describe the n and p distribution:
(18-46)
where ψ is the electric potential, s represents the arc length from cell center p and U t is the thermal
voltage on the cell interface.
Semi-Device Module
Numerical Schemes-Fully Coupled Newton Method
For Newton’s method it is required to solve a system of linear algebraic equations at each iteration step.
The question arises naturally how accurately these linear systems have to be solved, since the result
obtained thereby is just an incremental correction to the intermediate approximation of the solution. The
accuracy of the solution of the linear systems required to preserve the convergence properties of
Newton’s method has been investigated in, e.g. (Sherman, 1987). For the semiconductor equations, to
simplify the notation we define the following nonlinear system:
269
CFD-ACE_V2010.0_Modules_Manual_Part2
(18-47)
F 1 denotes the Poisson equation; F 2 and F 3 are the continuity equations for electrons and holes
respectively. The correction vector for the k-th iterate is given by:
(18-48)
Newton’s method at the k-th step can then be formally written as:
(18-49)
Various (direct and iterative) linear equation solvers can be used to solve equation 18-49. After δ are
k
(18-50)
Semi-Device Module
Numerical Schemes-Gummel Method
Particularly suited for the semiconductor equations is a block nonlinear iterative algorithm which has been
first suggested by Gummel (Gummel, 1964). This algorithm is therefore most frequently called Gummel’s
method. Instead of solving the fully coupled linear system it solves the following decoupled equations:
(18-51)
270
Modules
(18-52)
(18-53)
Since only a single equation is solved at a time, the Gummel method is much more efficient in memory
and CPU time per iteration.
Semi-Device-Implementation
Semi-Device Module
Implementation-Introduction
The Implementation section gives details about how to set up a model for simulation utilizing the Semi
Device Module of the CFD-ACE-Solver. The Grid Generation section provides tips on what types of grids
are allowed and general gridding guidelines. The Solution Setup section describes all the Semi Device
Module related inputs to the CFD-ACE-Solver. The Post Processing section provides tips on what to look
for in the solution output. The Semi-Device Implementation section includes:
• Procedures
• Grid Generation
• Doping File Format
• Model Setup and Solution
• Problem Type
• Model Options
• Volume Conditions
• Boundary Conditions
• Initial Conditions
• Solver Control
• Output
• Running Simulations
• What to do if a Solution Does Not Converge
• Post Processing
Semi-Device Module
Implementation-Procedures
To set up, solve, and post-process a semiconductor device problem using the Semi Device
Module:
1. Use CFD-GEOM or CFD-Micromesh to create a solid geometry model. Build a valid
structured/unstructured grid based on the solid model. Save the grid file in the DTF file format.
2. Create a text file describing the doping profile of the device, specified by an analytic formulation.
The doping file format and input data are described in Doping File Format. The created doping
profile text file, named <model_name>.doping, should be placed in the same directory where the
DTF model file is stored.
271
CFD-ACE_V2010.0_Modules_Manual_Part2
Semi-Device Module
Implementation-Grid Generation
The Semi Device Module can be applied to any geometry (3D, 2D planar or 2D axisymmetric).
Furthermore all grid cell types are supported (quad, tri, hex, tet, prism, poly). The general grid generation
concerns apply, i. e., and you have to ensure that the grid density is sufficient to resolve solution
gradients, minimize skewness in the grid system and locate computational boundaries in areas where
boundary values are well known. See CFD-GEOM or CFD-Micromesh manuals or relevant ESI CFD
tutorials for detailed information and guidelines.
Semi-Device Module
Implementation-Doping File Format
Doping profiles in the Semi Device Module are described by means of wells. The wells can merge into
each other to describe overlapping doping areas (multiple diffusion etc.). Each single well represents a
doping profile described by a Gaussian distribution using the following analytical expression:
(18-54)
where:
f(x) =
3
C = the doping density (in 1/cm )
272
Modules
The values of x l (X left), x r (X right), y l (Y low), y h (Y high), z t (Z top), and z b (Z bottom) define a bounding
box where C has the constant value C 0 . Outside the bounding box, C has a Gaussian distribution.
NOTE: All dimensions in the doping file should be in centimeters.
(18-55)
In extreme cases a bounding box can be a surface, a line, or a point. Special cases are illustrated below.
273
CFD-ACE_V2010.0_Modules_Manual_Part2
274
Modules
1.E20
Sizes(X_left, X_right, Y_low, Y_high, Z_top, Z_bottom)
0. 2.0e-4 0. 3.e-4 0. 0.
Decay Distances (Dx, Dy, Dz)
0.3e-4 1.e-6 0.3e-4
Well # 4 =================================== (N-Drain)
Method(1-Rectangular Well; 2-Radius Well)
1
Acceptor(-)/Donor(+) Concentration
1.E20
Sizes(X_left, X_right, Y_low, Y_high, Z_top, Z_bottom)
4.0e-4 6.0e-4 0. 3.e-4 0. 0.
Decay Distances (Dx, Dy, Dz)
0.3e-4 1.e-6 0.3e-4
Semi-Device Module
Implementation-Model Setup and Solution
CFD-ACE+ provides the inputs required for the Semi Device Module. Model setup and solution requires
data for the following panels:
• Problem Type
• Model Options
• Volume Conditions
• Boundary Conditions
• Initial Conditions
• Solver Control
Scaling
All dimensions and corresponding constants in Semi Device Module are in centimeters (cm). Therefore,
enter the geometry scaling factor such that all the dimensions of your meshed model will be in
centimeters. The corresponding physical constants and results also use cm as the dimension unit, for
2 -3
example, current density is in A/cm , electron/hole concentration in cm , etc. For example, if the initial
model dimension unit is meter, then you need to apply Grid Scaling equal to100.
In CFD-ACE+, select the Set Grid Scaling tab and enter the appropriate Scale factor for the problem.
Semi-Device Module
Implementation-Problem Type
Click the Problem Type [PT] tab to see the Problem Type Panel. See Control Panel-Problem Type for
details.
Select Semi Device to activate the Semi Device Module. The Heat Transfer (Heat) module will be
activated automatically.
Semi-Device Module
Implementation-Model Options
Click the Model Options [MO] tab to see the Model Options panel. See Control Panel-Model Options for
details.
275
CFD-ACE_V2010.0_Modules_Manual_Part2
Shared
When running a 2D case, choose between the Axisymmetric and Non-Axisymmetric options in the
Shared section. Also available is Steady/Transient time dependence. Select the Transient option and
enter the number of time steps and the size of each time step.
Heat
Semi
In the Semi panel, select appropriate models for the simulation. There are four choices available:
1. Physical Model
• Drift Diffusion (DD)
• Hydrodynamic + Energy Balance Equation (DD+EB)
• Hydrodynamic Full
2. Other physical models - You can select specific physical models to be used in the simulation
• Photon Rate Equation - See Photon Rate Equation
• Impact Ionization - Used in device regions with very high electric fields (e.g. in power devices);
also referred to as avalanche breakdown (see Impact Ionization)
• Ion Strike - Simulates a heavy ion strike on a semiconductor region; used to study radiation
effects on devices
• Photogeneration - Incorporates the effects of excess carriers generated by optical excitation (see
Photo Generation.)
3. Algorithm
• Newton (direct solution of all the equations together), see Fully Coupled Newton Method; this is
the default solver in Semi Device Simulations.
• Gummel (sequential, iterative solution of Poisson’s and continuity equations). For more
information see Gummel Method. The Gummel method is more appropriate for large problems
(large number of mesh cells) and all unstructured grid types. It typically requires quite strong
numerical damping parameters (see Solution Control Settings -> Relax) with Linear Relaxation
values typically below 0.5. Convergence of the Gummel method is relatively slow, requiring
typically a few hundred iterations. The Newton method is more appropriate for smaller problems
(small number of mesh cells) because it generates a large matrix for the solver. It typically
converges very fast (~10 iterations), with Relax parameters set to 1.0.
4. Linearization of Potential Equation - Boltzmann/Mixed. Use the defaults if you do not have
specific requirements.
Semi-Device Module
Implementation-Volume Conditions
Click the Volume Conditions [VC] tab to see the Volume Conditions Panel. See Volume Conditions for
details). Before you can assign volume conditions, you must pick an entity from the viewer window or the
VC Explorer.
Select all the volumes from the VC list in the Explorer table at the bottom (by Shift-clicking on volume
names). Click the Solid option in the panel on the right, and then click Apply.
Properties
276
Modules
Select Properties from the VC Setting Mode pull down menu. The following properties relate to the Semi
Device Module and are set under the Heat and Semi tabs that appear in the VC page.
Enter corresponding data under the Heat and Semi sections. No data are required in the Shared section.
You can use the Group/Ungroup function for specifying VCs. Before specifying VCs, you can group a set
of volumes that have the same VC type. To group a set of VCs, hold down the Ctrl key and left-click on
the VC handles in the grid drawing or click on volume names in the Explorer table. Then click Group in
the lower left corner. When a set of VCs is grouped, you can specify VCs for the whole set as one VC.
For example, select all the silicon volumes in the model (using Ctrl-click), group them together (click
Group), and then in the Semi panel apply the suitable properties for Si. Next, select all the Si-Oxide
volumes, group them together and in the Semi panel select Si_Oxide as the material and click Apply.
Be sure to select an appropriate Mobility model for the semiconductor material (for example,
Function_LICE for Si). See the Physical Models section for a detailed description of mobility models
available in the Semi Device Module.
You can also specify carrier lifetime values for all semiconductor materials. This input feature is included
to provide maximum flexibility for device simulations, since physical properties of semiconductor materials
can vary significantly under different conditions. In case of compound semiconductors, you can also
specify the mole fraction of each component. The available materials are shown in the figure below.
Semi-Device Module
Implementation-Boundary Conditions
Click the Boundary Conditions [BC] tab to see the Boundary Conditions Panel. See Control Panel-
Boundary Conditions for details. To assign boundary conditions and activate additional panel options,
select an entity from the viewer window or the BC Explorer.
The Semi Device Module fully supports the Cyclic and Arbitrary Interface boundary conditions. (See
Cyclic Boundary Conditions or Arbitrary Interface Boundary Conditions for details on these types of
boundary conditions and instructions on how to implement them). The Semi Device Module does not
support the Thin Wall boundary condition.
277
CFD-ACE_V2010.0_Modules_Manual_Part2
Most of the general boundary conditions for the Semi Device Module are located under the Semi tab
(except for a few settings under the Heat tab) and can be reached when the boundary condition setting
mode is set to General. There are three types of boundary conditions available in the Semi Device
Module: Wall, Symmetry and Interface. There are no special subtypes for the Symmetry and Interface
BCs. For the Wall BC, there are three possible BC subtypes : Ohmic Contact, Schottky Contact and
Insulator (No current normal to the wall).
In the lower left panel, sort the BC elements by name (click Boundary Name header in the table). Then
select particular boundaries (or a group of them as described previously in the VC section) by choosing a
boundary name from the Explorer table, or clicking on the boundary handle in the grid image. Specify
boundary condition data in the Heat and Semi sections. In the Heat panel, set at least one boundary to
Isothermal, with a specified temperature. In the Semi panel the most important aspect is setting biases
(Electric Potential) on contacts. Electron and hole densities are not required. They are calculated initially
from the thermal equilibrium condition. Again the Group/Ungroup method can be applied here. If you
select the Hydrodynamic (DD+ EB) model, under Physical Model in the Model Options page, then the BC
options for electron and hole temperatures also appear in the Semi panel. If specifically required, you can
set these to different values.
Semi-Device Module
Implementation-Initial Conditions
Click the Initial Conditions [IC] tab to see the Initial Conditions Panel. See Initial Conditions for details.
The Initial Conditions can either be specified as constant values or read from the solution file of a
previous run. If constant values are specified then the user must provide initial values required by the
Semi Device Module. The values can be specified under the Heat and Semi tabs.
Select initial conditions For All Volumes (or Volume by Volume if you need so). If it is a new model, enter
initial values in the Heat panel. In the Semi panel, initial values of Electric Potential, Electron Density, and
Hole Density are not required. They are initialized by thermal equilibrium condition. If the Hydrodynamic
(DD+EB) model is selected under Physical Model in the Model Options page, then the IC option for
electron and hole temperatures also appears in the Semi panel. Under most circumstances, default
values can be used for these variables. If it is not a new model, the user can choose between Constant
and Previous Solution under the IC Sources tab.
Semi-Device Module
Implementation-Solver Control
Click the Solver Control [SC] tab to see the Solver Control panel. The Solver Control panel allows access
to the various settings that control the numerical aspects of the CFD-ACE-Solver as well as the various
output options. There are ten major panels inside this section:
Iter: specify number of iterations and convergence criteria. You may run 1 iteration to save the doping
profile in DTF file and see it in CFD-VIEW, or otherwise set Max. Iterations to 30 as a first guess and
leave other defaults. If the total contact currents in Current Summary in the output file do not balance, this
typically means your solution is not fully converged, and you should decrease the Convergence Criteria
value (for example, to 1e-6) and increase the Max. Iterations number.
Spatial: specify the numerical discretization scheme for solving variables. (initially, leave defaults here)
Solvers: specify linear algebraic solvers for solving variables. (initially, leave defaults here)
Relax: specify relaxation parameters for solving variables. The definitions of Inertial Relaxation and
Linear Relaxation can be found in the CFD-ACE+ theory manual. (initially, leave defaults here, but see
also the section below What to Do If Solution Does Not Converge)
Limits: specify lower and upper limits on solving variables. The smaller the range, the better the
convergence will be. But if the range is set too small, it will result in a non-physical solution and lead to
convergence problems; The Temperature limits require special treatment. The first simulation of any
semiconductor structure should be performed as an isothermal case, by setting the Temperature
extremes to the same value (for example, Minimum = 300, Maximum = 300). If you are eventually
278
Modules
interested in heat generation and temperature calculation for that structure, then after having calculated a
stable and converged solution (proper balance of currents on all contacts) for the isothermal case, use
the DTF file with this isothermal solution as an initial condition (IC -> IC Sources -> Previous Solution),
and then set the temperature limits to a wider range (e.g. Min = 100, Max = 1000). Such a procedure
prevents numerical instability in current and temperature calculations for initial iterations.
Monitor: you may specify monitor points (at selected X, Y, Z location) to monitor Heat Transfer variables
in those points.
Adv (Advanced Solver Settings): specify CFL (local adaptive) relaxation option.
Semi-Device Module
Implementation-Output
Output
If this is a Steady simulation case, click on Output Results, and choose to save them into DTF file at the
End of Simulation or at Specified Interval; if this is a Transient simulation case, then specify the frequency
with which results should be saved.
Summary Output
Specify global outputs; (you may turn on the Energy Balance Summary in the Heat Transfer section
and/or the Current Summary in the Semi section. ) You may also use the Boundary Integral Output option
to output the applied bias, contact current, and electric potential (with built-in potential correction) at each
electrical contact. This will collect all simulation output in a single, formatted data file, and will facilitate
post-processing of the data to obtain I-V curves and other system characteristics.
Graphical Output
Specify data to be saved into the DTF file for post processing; (we recommend the selection of all
variables in the Semi section, and turning off all other variables except Temperature).
Semi-Device Module
Implementation-Running Simulations
Click the RUN tab. Click Submit to launch the CFD-ACE-Solver with the Semi Device Module to solve the
problem. After job submission, you can view residuals and/or output files, and stop the running job.
To set up a model and simulation parameters using CFD-ACE+, you only need to modify required
variable values and simulation settings. All other parameters are automatically set to correct default
values by CFD-ACE+.
Semi-Device Module
Implementation-What to Do if a Solution Does Not Converge
If a solution does not converge:
1. Check the units. At present, the Semi Device Module accepts only cm, s, A,...
2. Check the scaling factor to make sure the problem is correctly scaled.
3. Adjust Linear Relaxation values (SC -> Relax). Smaller relaxation values make the solution
process more stable. The drawback of reducing relaxation values is that the convergence rate will
be slower. The normal suggested range for linear relaxations is 0.1 5 - 1.0. For example, for a
steady-state simulation of a generic nMOSFET transistor (channel L=1.5mm) we found the
following Linear Relaxation parameters to be appropriate for the Gummel scheme: Temp.: 0.1, V:
0.2, N:0.2, P: 0.2. For the Newton scheme, the best starting option is to set all the Linear
279
CFD-ACE_V2010.0_Modules_Manual_Part2
Relaxation parameters to 1.0, and only if the solution does not converge, slightly decrease those
values.
4. Use smaller voltage step applied at the contact (use several smaller voltage increments to
achieve a higher bias on the contact).
5. Adjust Inertial Relaxation values if adjustments on Linear Relaxations do not work. Increasing
Inertial Relaxations will add more diagonal dominance into linear algebraic equation. Hence the
solution procedure becomes more stable. But it has the same drawback as decreasing the linear
relaxation values. The normal suggested range for Inertial Relaxations is 0-0.001.
6. Restart from a converged lower bias solution. If the solution still does not converge, please call
ESI CFD for support.
Semi-Device Module
Implementation-Post Processing
When the job is terminated normally, it is time to post-process the results using CFD-VIEW. See CFD-
VIEW for detailed information. Here are some hints for a quick start with CFD-VIEW:
1. Click the Launch CFD-VIEW button [ ] in the CFD-ACE+ top tool bar. A new VIEW window
displaying the device structure appears. NOTE: Within the CFD-VIEW window, you may use LB
to rotate the object, MB to zoom in or out, and RB to move (pan) the object. This feature is
consistent among all ESI CFD software tools such as CFD-GEOM, CFD-ACE+, CFD-VIEW, etc.
2. Click the Select All Surfaces button [ ] located below the list of zones at the lower right side.
5. Click the Legend button [ ] located in the upper right section of the CFD-VIEW window to see
the values of the displayed variable.
6. Click RB on the legend map that appears, and then click Properties.
7. Check the Show Flooded box, and click Close.
Color contours representing distribution of the selected variable should appear.
We recommend that you choose Doping as the first variable to view, to see if your doping profile has
been correctly defined in the <filename>.doping file. Next, you may look at Voltage to check the applied
bias.
Press and hold the left mouse button to rotate the 3D model with results, middle mouse button to zoom in
or out the figure, and right mouse button to shift the image.
Please refer to ESI CFD Semi-Device tutorials to learn detailed steps of building a device model, running
simulations, post-processing results, and running parametric simulations to obtained various
characteristics of semiconductor devices.
Semi-Device Module References
Adler, M.S., Possin, G.E., "Achieving accuracy in transistor and thyristor modeling.” IEEE Trans. Electron
Devices. 28.9(1981): 1053-1059.
Alley, G.D., "High-voltage two-dimensional simulations of permeable base transistors.” IEEE Trans.
Electron Devices. 30.1(1983): 52-60.
Beck, D.J., Conrad, T.R., "Auger recombination in Si.” Solid State Commun. 13(1973):93-95.
Canali, C., Majni, G., Minder, R., Ottaviani, G., "Electron and hole drift velocity measurements in silicon
and their empirical relation to electric field and temperature.” IEEE Trans. Electron Devices.
11(1975):1045-47.
280
Modules
Caughey, D.M., Thomas, R.E., "Carrier mobilities in silicon empirically related to doping and field." Proc.
IEEE 55(1967): 2192-2193.
Dorkel, J.M., Leturcq, P., "Carrier mobilities in silicon semi-empirically related to temperature, doping and
injection level.” Solid-St. Electron., 24(1981): 821-825.
Dorkel, J.M., "Physique des redresseurs de puissance aux tres forts niveaux de courant - aspects
electriques et thermiques.” Ph.D. Thesis, Université Paul Sabatier, Toulouse, France, 1982.
Dziewior, J., Schmid, W., "Auger coefficients for highly doped and highly excited silicon.” Appl. Phys.
Lett., 31(1977):346-348.
Ferry, D., Semiconductors. New York: Macmillan, 1991.
Gummel, H.K., "A Self-Consistent Iterative Scheme for One-Dimensional Steady State Transistor
Calculations." IEEE Trans. Electron Devices. 11(1964): 455-465.
Hess, K., Leburton, J. and Ravaioli, E., eds. Computational Electronics, Kluwer, 1991.
Mayaram, K., Chern, J.H., and Yang, P., "Algorithms for Transient Three-Dimensional Mixed-Level Circuit
and Device Simulation,” IEEE Trans. on CAD, No. 11, 1993, pp. 1726-1733.
Noble, M. J., Loehr, J. P., and Lott, J. A., "Analysis of Microcavity VCSEL Lasing Modes Using a Full-
Vector Weighted Index Method”, IEEE J. Quantum Electron., vol. 34, pp.1890-1903, Oct.1998.
Robertson, M. J., Kendall, P. C., Richie, S., McIlroy, P. W., and Adams, M. J., "The Weighted Index
Method: A New Technique For Analyzing Planar Optical Waveguides." J. Lightwave Technol., vol. 7,
pp.2105-2111, 1989.
Rollins, J.G., and Choma, J., "Mixed-Mode PISCES-SPICE Coupled Circuit and Device Solver,” IEEE
Trans. on CAD, No. 8, 1988, pp. 862-867.
Selberherr, S., Analysis and simulation of semiconductor devices. Springer-Verlag, Wien - New York,
1984.
Sherman, A.H., "On Newton-Iterative Methods for the Solution of Systems of Nonlinear Equations." SIAM
J. Numer. Anal. 15(1978): 755-771.
Slotboom, J.W., De Graaf, H.C., "Measurements of bandgap narrowing in Si bipolar transistors.”, Solid-St.
Electron. 19(1976): 857-862.
Sze, S.M., Physics of Semiconductor Devices. J. Willey, 1981.
Turowski, M., "Two-dimensional simulation of power semiconductor devices.” Ph.D. Thesis, Tech.
University of Lodz, 1992.
Turowski, M., Napieralski, A., and Dorkel, J.M., Two-Dimensional Turn-off Process Simulation of a GTO
with Realistic RLC Load. Proc. Int. Conf. Materials and Devices for Power Electronics. EPE-
MADEP'91, 2-6 Sept. 1991, Florence, Italy, pp.363-368.
281
CFD-ACE_V2010.0_Modules_Manual_Part2
Generation Rate of
2
Reaction_Rate Product/consumption Rate of moles/m /sec BioChem
Substrate
Wall_conc_AnalyteName Wall concentration M BioChem
282
Modules
Ω m
-1 -1
sig_i Imaginary part of conductivity Electric
Ω m
-1 -1
sig_r Real part of conductivity Electric
2 2
H0 Total Enthalpy m /s Heat
T Temperature K Heat
283
CFD-ACE_V2010.0_Modules_Manual_Part2
2
Wall_Rad_Flux Wall radiative flux W/m Heat
284
Modules
285
CFD-ACE_V2010.0_Modules_Manual_Part2
286
Modules
287
Glossary
A
amplification: An increase in the number of copies of a specific DNA fragement; can be in-vivo or in-vitro
analyte: The substance which a laboratory test aims to detect. In cholesteral testing, the analyte would
be 'cholesterol'. In genetic testing, the analyte could be a specific allele or genetic mutation.
antibody: A protein produced by certain white blood cells in response to a foreign substances (antigen).
Each antibody can bind only to a specific antigen. The purpose of this binding is to help destroy
the antigen. Antibodies can work in several ways, depending on the nature of the antigen. Some
antibodies disable antigens directly, while others make the antigen more vulnerable to destruction
by white blood cells.
antigen: Any foreign or "non-self" substance that, when introduced into the body, causes the immune
system to create an antibody.
assay: A laboratory test to identify and/or measure the amount of a particular substance in a sample
D
DNA Sequencing: Determining the exact order of the nucleotide base paairs (adenine, cytosine,
guanine, or thymine) in a segment of DNA.
DNA/RNA Hybridization: A technique for selectively binding specific segments of a single-stranded (ss)
DNA or RNA by base pairing to complementary sequences on ssDNA molecues.
E
electrophoresis: The separation of charged particles using their different rates of migration in an electric
field
enzyme: A protein that acts as a catalyst, speeding the rate at which a biochemical reaction proceeds,
but not alterning the direction or nature of the reaction
G
Genomics: The study of genes and their function
I
immunoassay: An assay that measures and identifies a specific biological substance such as an antigen
in-vitro: Outside a living organism
in-vivo: Latin phrase meaning "in life"; in the living organism
L
ligand: A molecule that interacts with a receptor
289
CFD-ACE_V2010.0_Modules_Manual_Part2
polymerase: Any enzyme that catalyzes the formation of DNA or RNA from its component molecules
Proteomics: The study of the full sent of proteins encoded by a genome
R
reaction kinetics: The individual steps that make up a chemically reacting system and the specification
of the rates at which those steps progress.
receptor: A protein or group of associated proteins in a cell or on its surface that selectively binds a
specific substance (call a ligand). Upon binding its ligand, the receptor triggers a specific
response in the cell.
S
Smagorinsky Model: A model where the eddy viscosity is proportional to the square of the grid spacing
and the local strain rate.
SPC: SPC is the acronym for specialized point constraint and is a CFD-ACE+ file type.
sticking probability: The probability that a molecule will adsorb upon collision with the reacting surface.
It is defined as the rate of adsorption divided by the collision frequency with the surface.
stoichiometry: The description of the conservation of mass and elements.
U
under-relaxation: A constraint on the change of a dependent or auxiliary variable from one solution
iteration to the next.
290
Index
2 Boundary Conditions Section ........................ 148
2D axisymmetric ............ 42, 77, 81, 96, 222, 272 Boundary Element Method ........................ 22, 44
A Boundary Trace Method .................................. 46
ab-initio ........................................................... 148 Brownian Motion ............................................ 102
Absorbed Power ............................................. 186 Buffer ............................................................... 80
Absorbed/Generated ........................................ 35 Buffer Conductivity ........................................... 54
AC Conduction ...........................................26, 44 C
AC Single Frequency .......... 74, 78, 81, 188, 201 Capacitance Matrix .................................... 46, 49
AC Single Frequency Magnetic Field Option .. 81, Capacitively Coupled Plasma ....... 165, 166, 184,
201 185, 186
AC Source ................................................81, 201 Capillary Flows ............................................... 136
AC+DC .....................................................81, 202 Capillary Force ............................................... 225
Added Mass ................................................... 102 Capillary-Wave Damping ....................... 156, 157
Adjust Inertial Relaxation ............................... 279 Carreau .......................................................... 163
Adjust Linear Relaxation ................................ 279 Carreau Law Model........................................ 163
Advanced Input .............................................. 157 Carrier Generation-Recombination Models ... 265
Advanced Settings ........................................... 44 Carrier-Carrier Scattering .............................. 262
Advanced Tab ........................................156, 157 Caughey................................................. 262, 280
advection ................................................145, 153 Cavitation ....................................................... 184
Algebraic Multigrid Solver ................................ 84 CCP ......165, 166, 170, 177, 182, 183, 184, 185,
ambipolar................................................187, 209 186, 188, 202, 204, 205, 214, 246
Ambipolar Field .............................................. 205 CCP Cycle Frequency ................................... 205
ambipolarity .................................................... 171 CCP Iterations ................................................ 204
AMG ................................................................. 84 CCP Physical Model ...................................... 166
Ampère’s Law .................................................. 72 CCP Plasma .................................................... 63
Amperes ...................................................81, 201 CCP Self Consistent Simulation .................... 182
Amps ................................................................ 81 CCP timestep ................................................. 205
Arbitrary Interface .............................62, 152, 277 CFL Number .......................... 145, 157, 160, 162
Arbitrary Interface Boundary Conditions ... 228, CFL Relaxation .............................................. 204
277 CGS ................................................................. 84
Artificial Spray Dispersion .............................. 115 CGS+Pre............................................ 66, 84, 230
Axial Direction ..........................................81, 201 Characteristic Cell Length .............................. 162
Axisymmetric ............. 77, 81, 116, 201, 256, 275 Chemical/Biological.......................................... 88
axisymmetric 2D .....................................228, 229 Chemistry Module .......................................... 236
B Chemistry/Biochemistry ................................. 135
Back Projection .............................................. 145 Chemkin ......................................................... 202
Backward-Euler Scheme................................ 153 Circuit Leakage ................................................ 27
corresponds ................................................ 153 Collisional Gas Discharge Plasmas ............... 247
BEM....................... 22, 23, 28, 44, 54, 62, 63, 66 collision-dominated ........................................ 233
BEM Electrostatics Solution Techniques ......... 28 combustors ...................................................... 88
BEM Solver ................................................44, 67 Complementary MOS .................................... 249
Bessel............................................................. 256 Compound Semiconductor Materials............. 248
biochips ............................................................ 23 compressibility ............................................... 136
bioelectronic ................................................... 247 Compressible Flow .......................................... 90
bioMEMS .......................................................... 23 Computational Electronics ............................. 280
Bipolar Junction Transistors ........................... 249 Computational Physics .................................. 164
Bipolar Transistors ......................................... 247 Concentration..................................... 54, 80, 272
Blank’s Law .................................................... 214 Condensation ................................................. 216
specifying .................................................... 214 conducting ...................................................... 249
Boltzmann ............. 179, 187, 233, 248, 252, 254 Conduction ................................................. 79, 82
Boltzmann/Mixed ............................................ 275 Conduction Current Density ............................. 68
Boundary Condition Subtypes ...............228, 229 Conductivity Calculation .................................. 79
291
CFD-ACE_V2010.0_Modules_Manual_Part2
292
fluid_dynamics_back
293
CFD-ACE_V2010.0_Modules_Manual_Part2
294
fluid_dynamics_back
295
CFD-ACE_V2010.0_Modules_Manual_Part2
296
fluid_dynamics_back
297
CFD-ACE_V2010.0_Modules_Manual_Part2
298
fluid_dynamics_back
299
CFD-ACE_V2010.0_Modules_Manual_Part2
300
fluid_dynamics_back
multi-dimensional ........................................ 237 Semi Device .. 247, 248, 249, 254, 258, 261, 271,
Required Inputs .............................................. 227 275
Required Variables .................................228, 229 Launch ........................................................ 271
Resistivity ...................................................54, 80 Semi Device Module ..... 254, 265, 267, 271, 272,
Resistivity Function ....................................54, 80 275, 276, 277, 278, 279
Resistivity Poly ................................................. 54 relate .......................................................... 276
Resistivity Polynomial ...................................... 80 utilizing ....................................................... 271
Reverse Biased p-n Junction ......................... 249 Semi Device Simulations ............................... 275
reversed-biased p-n ....................................... 249 Semi-Conductor ............................................. 267
Review............................................................ 232 Semiconductor Devices ......................... 249, 280
RF... 85, 169, 170, 181, 182, 183, 186, 240, 245, Operation .................................................... 249
246, 247, 248 Physics ....................................................... 280
applying ...................................................... 170 Semiconductor Materials List......................... 276
generates .................................................... 170 Semiconductor Structure ............................... 249
gives ........................................................... 245 semiconductor-metal...................................... 258
produces ..................................................... 182 Semiconductors ..................... 233, 242, 252, 280
RF Field .......................................................... 245 semiconductor-semiconductor ....................... 258
Rosin-Rammler ................................................ 94 Semi-Device ........................... 254, 255, 256, 262
Rosin-Rammler Distribution ............................. 94 Semi-Device Theory ...................................... 249
Rotating .......................................................... 280 Semi-Insulator ................................................ 261
3D ............................................................... 280 Sequential Gummel ....................................... 248
Rotating Walls ................................................ 202 Set Grid Scaling ............................................. 275
Round Well ..................................................... 272 Setting Mode .................................................... 51
defining ....................................................... 272 Setting Mode list .................................... 200, 201
Round Well Defined ....................................... 272 Setting Mode menu .......................................... 81
Top View ..................................................... 272 Settings Mode .................................................. 79
Royal Society ................................................. 164 Setup ........................................................ 51, 247
Philosophical Transactions ......................... 164 Shared ...........43, 44, 74, 78, 185, 223, 275, 276
RUN................................................................ 279 Shared - Buffered Output ............................... 204
S Shared, Flow .................................................. 185
Saffman Lift .................................................... 102 Sharfetter-Gummel ........................................ 182
Sample Doping Input File nMOSFET.doping . 272 SHE ................................................................ 233
Sample Octree Mesh ..................................... 247 Shear Modulus ................................................. 52
Basic MOSFET Transistor .......................... 247 Sherman ................................................ 269, 280
Sample Phase Diagram ................................. 220 sherwood ................................................... 91, 97
Scalar Optics .................................................. 248 calculating .................................................... 97
scales ............................................................. 275 set ................................................................. 97
Scharfetter Gummel Numerics ....................... 248 Sherwood Number ........................................... 97
Schmidt ............................................................ 91 Shift-clicking ................................................... 276
Schottky..................................................259, 260 Shockley-Read-Hall ....................................... 265
Schottky Contact ....................................248, 277 Simple External Circuit .................................. 171
Schrödinger .................................................... 252 simplifications ................................................. 160
Secondary Breakup........................................ 102 making ........................................................ 160
Secondary Electrons button ........................... 242 simplifies ........................................................ 256
Secondary Fluid Flux ..................................... 145 Poisson ....................................................... 256
second-invariant ............................................. 163 simulating ............................................... 247, 275
Seebeck .................................. 33, 35, 36, 40, 49 1D ............................................................... 247
Seebeck Effect ................................................. 33 simulation ....................................................... 261
Selberherr.............................. 249, 262, 267, 280 Simulation ................................................ 84, 279
Select Capillary Force .................................... 225 End ....................................................... 84, 279
Select Phase Change .................................... 225 Single Capacitor............................................... 46
self-capacitance ............................................... 30 Single Line Interface Construction ................. 138
Self-Consistent Iterative Scheme ................... 280 Sinusoid ........................................................... 62
One-Dimensional Steady State Transistor Si-Oxide ......................................................... 276
Calculations ............................................ 280 SLIC ...............136, 138, 145, 153, 157, 160, 162
Semi ...................... 272, 275, 276, 277, 278, 279 compared ................................................... 157
301
CFD-ACE_V2010.0_Modules_Manual_Part2
302
fluid_dynamics_back
303
CFD-ACE_V2010.0_Modules_Manual_Part2
Viscosity Evaluation Methods ........................ 227 Volume Conditions ...... 51, 56, 79, 113, 154, 186,
VOF ...... 23, 25, 32, 64, 135, 136, 137, 138, 145, 200, 227, 228, 242, 275, 276
150, 153, 157, 158, 159, 160, 163, 164, 221 Volume Fraction ..................................... 217, 231
Volt ................................................................... 46 Volume Trace Method ..................................... 46
Voltage ........................................................... 280 W
Voltage Difference............................................ 29 Wallis ............................................................. 219
Electric Field Due ......................................... 29 Weber Number............................................... 161
Volume Charge ................................................ 69 Z
Zero Gradient ................................................. 202
304
305