3D Flow Modeling of the First Trifurcation

Made in Nepal
RK Malik, Paras Paudel
RK Malik

Paras Paudel

Abstract: The foremost objective of the study was to find out the most efficient profile of trifurcation in given
constraints of pressure, velocity and layout of the overall geometry. The study was done for the 3.2 MW Madi Khola
Hydropower Project of Gandaki Hydropower Development Co. Pvt. Ltd. The 3 Dimensional Flow modeling of the
trifurcation was based on the application of Computational Fluid Dynamics (CFD).
The loss in the Trifurcation greatly depends upon its geometrical configuration. The research started with a general
profile and the flow pattern generated inside it was studied with the help of 3 Dimensional Flow modeling .The extent
of vortex zone formation inside the trifurcation indicates the loss inside trifurcation. The profile of the trifurcation
was hence changed to reduce the vortex formation as far as possible, till we get minimum possible loss. The profile
under study should meet maximum flow efficiency under the physical constraints of fabrication. The flow efficient
profile was then analyzed to capture the stress amplification near junction. The reinforcing element in the form
of steel T-section was added of different sectional values till the stress was within allowable limits under severe
conditions.
Key words: Symmetrical Trifuraction, trifurcation, Computational fluid dynamics, hydropower, Nepal

Introduction
Project layout and powerhouse orientation decide what
kind of penstock branching would be most suitable for the
highest safety and minimum head loss. Usually in case of
branching for 3 or more units, a number of unsymmetrical
bifurcations are used one after another along the penstock
alignment. It involves relatively less analytical works
during design and is easy to fabricate. But project layout
can dictate the other way around, as in case of 3.2 MW
Madi Khola Hydropower Project in Kaski district, Nepal,
which is already nearing the completion of construction.
In this project orientation of the powerhouse with respect
to the penstock alignment is such that only a symmetrical
trifurcation can be used to feed water to three equal
capacity Pelton turbines. Madi trifurcation has become the
first trifurcation designed and fabricated within Nepal. It
was manufactured and installed by Radha Structures and
Engineering Works Pvt. Ltd.
Design of an element of water conveyance system
constitute of hydraulic and structural analyses. Hydraulic
analysis for penstock branching (bifurcation and
trifurcation) is considered unimportant and is found
mostly avoided in small hydropower projects of Nepal.
But the fact is that hydraulic analysis here is as important
as structural analysis. Structural analysis optimizes the
initial cost through right selection of steel thicknesses,
whereas hydraulic analysis minimizes the head loss
through selection of best possible geometry. Head loss
in the branching entails a constant loss of money for as
long as the plant runs, and this loss, in long run, is many
times higher than the cost of the structure itself. Large
hydropower projects do not only conduct hydraulic analysis

56

but also go for model tests. But small project of capacities

1-10 MW are recommended at least to conduct hydraulic
analyses for the critical components of water conveyance
system. Bifurcation and trifurcation are among such
components. They are used near the powerhouse under
high pressure head added by pressure surge due to water
hammer. Besides, in case of Pelton turbines, there are free
water jets downstream to the manifold, which convert
almost all of the available head into velocity head. Under
this condition the water exerts force on manifolds due to
change in momentum of water and can be evaluated from
principle of conservation of momentum. This net force
must be resisted by the manifold system and the concrete
block holding manifold.
The structural analysis of the manifolds is necessary
but not sufficient if we consider the long term operational
benefits in terms of power outcome and the performance
of the plant. The vibration problem caused due to
unnecessary eddies developed can cause huge losses of
pressure head downstream of the manifold. The profile of
the manifolds affects the loss in the available water head
significantly. This loss can decrease the potential plant
capacity. The profile selection process can be done either
by experimental analysis on reduced scale manifolds model
test at lab or by numerical modeling of the fluid flow. The
former option is rather expensive and may not be feasible
every time. It is preferable to select best profile by tuning it
with CFD solver and then follow reduced scale model test
for the confirmation of flow parameters.
At the time of valve closure the velocity reduces to zero
within time interval of valve closure. This phenomenon
now converts velocity head to pressure head again. The

HYDRO NEPAL

ISSUE NO. 5

JULY, 2009

pressure magnifies due to abrupt change in the velocity.

The magnified pressure moves to up stream with certain
velocity which depends up on bulk modulus of water and
its density. The trifurcation has to resist this magnified
pressure. The hoop stresses and longitudinal stress and
the combined stress due to increased pressure must be
below the tolerable limit as specified in Steel Structure
Design Codes. At the zone of the junction of bifurcation
the simple analysis approach can not catch the local stress
concentration and there are chances of underestimation of
such valuable stress concentration. Finite element method
can be used to capture Stress Concentration near junction
of trifurcation using finer mess near junction.

If the net head of water and discharge at inlet is h and

Q respectively then from the continuity equation for the
equal discharge among three branching pipe,
Expected Discharge through each nozzle = Q/3

Design & Analysis Criteria

The pressure losses at junction in manifolds are not
analyzed and neglected which is the normal practice
in hydropower of Nepal. These losses are significant in
obtaining high plant capacity. Before starting design and
analysis of the trifurcation, the constraining parameter,
such as space available and the position of turbine from the
existing site condition, are used to select the trifurcation
layout. The normal practiced profile is first used in detailed
numerical analysis to understand how the profile affects
the losses. The outcomes in analyzing such manifolds
provide necessary data for modification. A number of
trial and errors to minimize loss was carried out. Finally
the best profile was adapted and was used for the stress
analysis to minimize the stress below the allowable limit at
the junction by adding reinforcement at junctions.
CFD is used for the analysis of the velocity and pressure
distribution at different section of manifolds. The pressure
and velocity distribution will then be used as a criteria
modifying profile of manifold body.
Flotran CFD features of the Ansys v11 is used for the
mathematical modeling of the flow through the profile.
AutoCad 2006, Ansys Workbench 11 features is used
for the mesh generation of the water volume inside the
manifolds. The volume is imported in Ansys Flotran CFD
for the analysis.
The jet velocity was expected as an output from the
analysis

Where : density of water

Q : Discharge in branching upstream
Q1 , Q2 , Q3 : Discharge in Branching
v1 , v2 , v3 : Velocities in branching
Q , , P are known input parameters, while Q1 , Q2 , Q3
and v , v , v can be determined from the Ansys Flotran
1
2
3
CFD output by surface integral of velocity distribution of
cross section near jet.

ANSYS Flotran CFD

Types of FLOTRAN Analyses
We can perform these types of FLOTRAN analyses:

Laminar or turbulent

Thermal or adiabatic

Free surface

Compressible or incompressible

Newtonian or Non-Newtonian

Multiple species transport

These types of analyses are not mutually exclusive. For
example, a laminar analysis can be thermal or adiabatic. A
turbulent analysis can be compressible or incompressible.
For the first trial, we use either of the Laminar Flow
model or Turbulent Flow model. If differences in pressure
and velocity field at a section appear significant after the
first trial, we have to choose turbulence flow model for the
next trial. Laminar and turbulent flows are considered to
be incompressible since density is constant i.e. the fluid
expends little energy in compressing the flow.

Fig.1. Ansys meshing of the trifurcation fluid control

volume

HYDRO NEPAL

ISSUE NO. 5

JULY, 2009

57

Flotran CFD analysis

Applying Boundary Conditions

Determining the Problem Domain

The analysis of the losses in penstock pipe line up to
trifurcation upstream was done to evaluate pressure and
velocity as an inlet boundary condition for trifurcation.
Boundary condition at out let of trifurcation is rather
difficult process. It is preferable to model up to nozzle
where there is free jet condition. Such modeling requires
excessive computer memory almost difficult to solve with
normal PC. If we focus on the losses due to trifurcation
only then logically we can locate free jet zone some how
near to the junction provided that it will not affect the
junction velocity and pressure distribution.

The free body diagram of the manifolds was considered

for the analysis.
The pipe wall was replaced by the zero velocity
constrained.
The net pressure near inlet surface of the manifold was
directly applied (177m water head).
The trial velocity at inlet is applied directly as 3m/sec,
4m/sec. the velocity to generate input discharge
corresponding to plant capacity.
The nozzle of the different trial dia. was attached with
the trifurcation outlet. This helps us to apply pressure
boundary condition at nozzle outlet as zero equivalents to
atmospheric pressure.

Determining the Flow Regime

We need to estimate the character of the flow. The
character is a function of the fluid properties, geometry,
and the approximate magnitude of the velocity field. The
Reynolds number can be used to decide weather the flow
will be laminar or turbulent. Similarly, Mach no criteria can
be used to evaluate the compressible and incompressible
flow.
Reynolds number = inertial force/viscous force, it
should be greater than 2000 for turbulent flow. Mach no =
Velocity of fluid/velocity of sound in fluid, which should be
greater than 0.7 to activate the compressible flow model.

Creating the Finite Element Mesh

For the most accurate results, we should use mapped
meshing. It more effectively maintains a consistent mesh
pattern along the boundary. In some cases, we can use
hexahedral elements to capture detail in high-gradient
regions and tetrahedral elements in less critical regions.
For flow analysis, especially turbulent, we should not use
pyramid elements in the region near the walls because it
may lead to inaccuracies in the solution.
In our case free meshing with tetrahedral element was
used for simplicity in meshing it leads some degree of
mass imbalance. The mass imbalance can be controlled by
further fine meshing.

Fig.3. Boundary condition for most efficient

trifurcation control volume
Setting FLOTRAN Analysis Parameters
This step includes selection of the flow models as
laminar incompressible, laminar compressible, turbulent
incompressible, turbulent compressible etc. The number
of iterations, convergence option, and fluid properties
should also be assigned in this step.

Solving the Problem

This step includes the solving process. We can monitor
solution convergence and stability of
the analysis by observing the rate of
change of the solution and the behavior
of relevant dependent variables. These
variables include velocity, pressure,
temperature, and (if necessary)
turbulence quantities such as kinetic
energy (degree of freedom ENKE),
kinetic energy dissipation rate (ENDS),
and effective viscosity (EVIS).

Examining the Results

Fig.2, Tetrahedral meshing of fluid

control volume near junction

58

Fig.4.Different section of interest

for recording output data

The results are in the form of velocity

distribution and pressure distribution
and turbulence quantities distributions.
The numerical integration is used to
convert velocity distribution to average

HYDRO NEPAL

ISSUE NO. 5

JULY, 2009

velocity at outlets. The average velocity at inlet and outlets

are then used to evaluate the loss in manifold.
The loss in energy is made as small as possible by
adjusting the geometrical profile of manifold at joint.
This is the trial and error process. The modification is
made till the loss in the manifold is equivalent to 0.42%.

Fig.5.Velocity distribution at section 1-1

Conclusions from most efficient profile

Left jet
Area
Discharge
Velocity
Velocity head
Right jet
Area
Discharge
Velocity
Velocity head
Middle jet
Area
Discharge
Velocity
Velocity head
Main Pipe
Area
Discharge
Velocity
Velocity head
Pressure
Pressure head
total energy u/s to
bifurcation
Total output energy
Total input energy
Total Loss
% Loss
loss in terms of head

Energy
0.022532
1.303
57.846
170.552

m2
m3/sec
m/sec
m

0.022532
1.307
58.006
171.494

m2
m3/sec
m/sec
m

0.022532
1.295
57.494
168.48

m2
m3/sec
m/sec
m

1.125
3.677
3.268
0.544
1770000
177

m2
m3/sec
m/sec
m
N/m2
m

177.544

6520696.58 J
6548275.47 J
27578.89 J
0.4211
0.748

%
m

2180739 J

2198827 J

2141131 J

6548275 J

Fig.6.Pressure Distribution at section 1-1

Uniform pressure of 177m water head

Fig.8.Velocity distribution between section 1 and 2

Fig.7.Velocity distribution between section 1 and 2

The velocity distribution profile for section at junction

indicates the causes of loss of energy near junction. These
velocity profiles are for the most possible efficient geometry
of the trifurcation junction.

HYDRO NEPAL

ISSUE NO. 5

JULY, 2009

Fig.9.Velocity distribution in section 2-2

Stress Analysis at Junction

According to the ASME Code, the non-embedded penstock
pipe may be designed under the following condition.

59

Fig.10.Velocity distribution in sections 5-5

Fig.13. Mesh generation of manifold for stress analysis.

Fig.11.Velocity distribution in sections 5-5

Fig.12.Velocity distribution in sections 6-6

Normal condition:
This condition gives the allowable stress=138.67 Mpa
Intermittent condition:
This condition gives the allowable stress=184.89 Mpa
Emergency condition:
This condition gives the allowable stress=250 Mpa
Exceptional conditions:
It includes malfunctioning of control equipment in most
adverse manner and shall not be used as the basis of
design.

60

If the maximum stress does not exceed the specified

minimum ultimate tensile strength, the structural integrity
of the penstock is reasonably assured.
Precautions must be taken to minimize the probability
of occurrence and effects of the exceptional condition.
In this analysis the stress was supposed not to be
increase above 138 Mpa in normal water pressure + water
hammer pressure. For the combination of normal flow
condition and the earthquake effects the whole penstock
unit should be studied.
The Von-misses stress criterion was used for checking
the yield at the zone of stress concentration

Fig.14. Von-misses stress distribution showing stress

concentration near junction.

Examining the Results

The stress at the junction of the manifold was observed
exceeding at normal running condition. The problem was
resolved by adding extra reinforcing plate at junction in
the form of T section

Conclusions
Around 20 models were checked for the loss of the head
due to geometry of manifold between common profiles

HYDRO NEPAL

ISSUE NO. 5

JULY, 2009

surface developments of trifurcation for Madi

Khola small hydropower.
Corresponding address: paras.paudel@gmail.
com

References
ANSYS Fluid Analysis Guide, ANSYS Release
10.0.002184, ANSYS,Inc. Southpointe 275
Technology Drive Canonsburg, PA 15317,
August 2005.
Chen Y.S., Kim S.W., 1987, Computation
of Turbulent Flows Using an Extended k-,
Turbulence Closure Model, NASA CR-179204

to most efficient profile. Only result for the most efficient

profile is presented in this paper.
The Study indicates that the profile can be better tuned
with the finite element application for fluid flow (CFD).
The flow distribution in case of common profile is rather
disturbed to longer zone of the trifurcation system. In case
of most efficient profile the flow distribution is disturbed
slightly to smaller length of the trifurcation system.
The second profile was taken for the fabrication to limit
loss in head less than one meter head of water
R. K. Mallik completed his M.Sc. in Structural Engineering
from IOE in 2005. He has worked for Radha Structure
Engineering & Works and many consultancy as a Structural
Engineer. These days he is involved as research engineer
at NSET. He has started the finite element modeling for the
analysis of the complex profile of hydropower for the first
time in Nepal such as Bifurcation modeling for Heldung
Hydropower and Lower Piluwa Hydropower.
Besides other different researches in structural field, he was
the developer of the analysis and design methodology for the
3-Dimensional flow modeling in the trifurcation for Madi
Khola small hydropower for the first time in Nepal.

Bunti Ogor I., Dietze S., Ruprecht A.,

Numerical Simulation of Flow in Turbine -99
Draft Tube, Proceedings of the third IAHR/
ERCOFTAC workshop on draft tube flow 8-9 December 2005,
Porjus, Sweden
T.J. Chung Computational Fluid Dynamics, University of
Alabama in Huntsville, Cambridge University Press, First
published at 2002
Hydropower in Norway , Mechanical Equipment , A survey
prepared by Arne Kjlle , Professor Emeritus Norwegian
University of Science and Technology. December 2001
Xiaodong Wang, On Mixed Finite Element Formulations
for Fluid-Structure Interactions, PhD Thesis in Applied
Mechanics, MIT, June 1995
F. Nobile: Numerical Approximations of Fluid-Structure
Interactions with Applications to Haemodynamics, Ph. D.
Thesis, EPFL Lausanne (CH), 2001
Ellen Kock , Lorraine Olson, Fluid-Structure Interaction
Analysis By The Finite Element Method-A Variational
Approach, International Journal for Numerical Methods in
Engineering,Volume-31,Issue-3, 24 June 2005
University of Alberta ANSYS Tutorials - www.mece.ualberta.
ca/tutorials/ansys

Corresponding address: [email protected]

Paras Paudel completed B.E. in Mechanical Engineering
in 2006. He was involved in various research activities in
Robotics club of IOE. He is currently working as a Mechanical
Engineer at Nepal Airlines Corporation.
He was involved in the 3-Dimensional flow modeling and

HYDRO NEPAL

ISSUE NO. 5

JULY, 2009

P. Colella and E.G. Puckett, Modern Numerical Methods for

Fluid flow, UC Berkeley E266A course notes
By Thomas H. Pullia, Solution Methods in Computational
Fluid Dynamics, Based on notes from a VKI lecture series
in 1986.

61