International Journal of Engineering (IJE), Volume (4) : Issue

International Journal of
Engineering (IJE)
Volume 4, Issue 2, 2010
Edited By
Computer Science Journals
www.cscjournals.org
Editor in Chief Dr. Kouroush Jenab
International Journal of Engineering (IJE)

Book: 2010 Volume 4, Issue 2
Publishing Date: 31-03-2010
Proceedings
ISSN (Online): 1985-2312
This work is subjected to copyright. All rights are reserved whether the whole or
part of the material is concerned, specifically the rights of translation, reprinting,
re-use of illusions, recitation, broadcasting, reproduction on microfilms or in any
other way, and storage in data banks. Duplication of this publication of parts
thereof is permitted only under the provision of the copyright law 1965, in its
current version, and permission of use must always be obtained from CSC
Publishers. Violations are liable to prosecution under the copyright law.
IJE Journal is a part of CSC Publishers

http://www.cscjournals.org
©IJE Journal
Published in Malaysia
Typesetting: Camera-ready by author, data conversation by CSC Publishing

Services – CSC Journals, Malaysia
CSC Publishers
Editorial Preface
This is the second issue of volume fourth of International Journal of

Engineering (IJE). The Journal is published bi-monthly, with papers being
peer reviewed to high international standards. The International Journal of
Engineering is not limited to a specific aspect of engineering but it is devoted
to the publication of high quality papers on all division of engineering in
general. IJE intends to disseminate knowledge in the various disciplines of
the engineering field from theoretical, practical and analytical research to
physical implications and theoretical or quantitative discussion intended for
academic and industrial progress. In order to position IJE as one of the good
journal on engineering sciences, a group of highly valuable scholars are
serving on the editorial board. The International Editorial Board ensures that
significant developments in engineering from around the world are reflected
in the Journal. Some important topics covers by journal are nuclear
engineering, mechanical engineering, computer engineering, electrical
engineering, civil & structural engineering etc.
The coverage of the journal includes all new theoretical and experimental
findings in the fields of engineering which enhance the knowledge of
scientist, industrials, researchers and all those persons who are coupled with
engineering field. IJE objective is to publish articles that are not only
technically proficient but also contains information and ideas of fresh interest
for International readership. IJE aims to handle submissions courteously and
promptly. IJE objectives are to promote and extend the use of all methods in
the principal disciplines of Engineering.
IJE editors understand that how much it is important for authors and
researchers to have their work published with a minimum delay after
submission of their papers. They also strongly believe that the direct
communication between the editors and authors are important for the
welfare, quality and wellbeing of the Journal and its readers. Therefore, all
activities from paper submission to paper publication are controlled through
electronic systems that include electronic submission, editorial panel and
review system that ensures rapid decision with least delays in the publication
processes.
To build its international reputation, we are disseminating the publication

information through Google Books, Google Scholar, Directory of Open Access
Journals (DOAJ), Open J Gate, ScientificCommons, Docstoc and many more.
Our International Editors are working on establishing ISI listing and a good
impact factor for IJE. We would like to remind you that the success of our
journal depends directly on the number of quality articles submitted for
review. Accordingly, we would like to request your participation by
submitting quality manuscripts for review and encouraging your colleagues to
submit quality manuscripts for review. One of the great benefits we can
provide to our prospective authors is the mentoring nature of our review
process. IJE provides authors with high quality, helpful reviews that are
shaped to assist authors in improving their manuscripts.
Editorial Board Members

International Journal of Engineering (IJE)
Editorial Board
Editor-in-Chief (EiC)
Dr. Kouroush Jenab
Ryerson University, Canada
Associate Editors (AEiCs)

Professor. Ernest Baafi
University of Wollongong, (Australia)
Dr. Tarek M. Sobh
University of Bridgeport, (United States of America)
Professor. Ziad Saghir
Ryerson University, (Canada)
Professor. Ridha Gharbi
Kuwait University, (Kuwait)
Professor. Mojtaba Azhari
Isfahan University of Technology, (Iran)
Dr. Cheng-Xian (Charlie) Lin
University of Tennessee, (United States of America)
Editorial Board Members (EBMs)

Dr. Dhanapal Durai Dominic P
Universiti Teknologi PETRONAS, (Malaysia)
Professor. Jing Zhang
University of Alaska Fairbanks, (United States of America)
Dr. Tao Chen
Nanyang Technological University, (Singapore)
Dr. Oscar Hui
University of Hong Kong, (Hong Kong)
Professor. Sasikumaran Sreedharan
King Khalid University, (Saudi Arabia)
Assistant Professor. Javad Nematian
University of Tabriz, (Iran)
Dr. Bonny Banerjee
(United States of America)
AssociateProfessor. Khalifa Saif Al-Jabri
Sultan Qaboos University, (Oman)
Table of Contents
Volume 4, Issue 2, May 2010.
105 - 118
The modeling of UPFC based on circuit elements in an exact
transmission line model
Ali Akbar Motie Bjandi, Kauomars Sabzawari
119- 138
Vibration Attenuation of a Thin Cantilevered Beam Using LQG-
Based Controller and Inertial Actuator
Ameen El-Sinawi
139- 148 A New Strategy to Achieve Radical Combustion Through Exhaust

Port Throttling for Two Stroke Engine
Saqaff Ahmed Alkaff , Mohamed Khan
149 - 159 Bi-level Control and Chopper Control Methods for Improving the
Dynamic Performance of Stepper Motor
Walid M. Emar, Ziad Sobih, Musbah Aqel, Mahmoud S. Awad
160 -178 Three Dimensional Bearing Capacity of Shallow Foundations

Adjacent to Slopes Using Discrete Element Method
Mona Arabshahi, Ali Asghar Mirghasemi, Ali Reza Majidi
A modified free decay test apparatus for the characterization of soft

179 - 190
magnetic gels in the presence of magnetic fields
Venkateswara Rao. P, Maniprakash. S, Srinivasan. S.M
191 - 200 Modeling of moisture diffusion and heat transfer during softening in
wood densification
Donghua Jia, Muhammad T. Afzal, Meng Gongc, Alemayehu H.
Bedane
International Journal of Engineering, (IJE) Volume (4) : Issue (2)

Ali akbar Motie birjandi , Kauomars Sabzawari
The modeling of UPFC based on circuit elements in an exact

transmission line model
Ali akbar Motiebirjandi [email protected]

Faculty of Electrical Engineering / assistant professor
Shahid Rajaee University/ Tehran, 67616, Iran
Kauomars Sabzawari [email protected]

Faculty of Electrical Engineering / M.Sc
Shahid Rajaee University/ Tehran, 67616, Iran
Abstract
UPFC is considered and known as one of the best FACTS devices. It is a

combination of series and parallel quick compensation, and can therefore
provide active and reactive control to achieve maximum power transfer, system
stability and improve power quality and reliability. Therefore, presenting a
suitable model for UPFC which enables studying the network and load flow in
energy transmission lines, has gotten the experts’ attention. This paper presents
a new model of UPFC in steady state based on circuit elements in an exact
transmission line model. Considering the fact that the elements like UJT and
tunnel diode in negative resistance region cause power increase, the model
determines the value of the active and reactive power injected by the series
converter into the network, and then the series converter of UPFC is simulated
by means of a negative resistor and an induction or capacitor reactance.
Resistance and reactance are expressed as functions of voltage of buses, load
angle, voltage injection and fire angle of series converter. The relations of this
model have been achieved in a two-bus system and have been simulated in 14-
and 30- bus standard.
Keywords: FACTS, UPFC, Load Flow, Reactive Power Control, Active Power Control.
1. INTRODUCTION
Unified power flow controller is a multi-functional inter-mediator in FACTs devices family. UPFC
can act as a parallel or series compensator, voltage regulator, phase replacer or power controller
in different conditions by simultaneous control of line parameters [1]. Different models of UPFC
are presented in steady state, including:
UPFC model based on voltage sources and series impedance: in this model the series and
parallel converters are replaced by a voltage source and series impedance [2],[3].
UPFC model based on ideal voltage and current sources: in this model the series converter is
replaced by a voltage source and the parallel converter by an ideal current source [4].
UPFC injection mode: in this model the series and parallel converters are replaced by a voltage
source and series impedance and are then combined with the π-line model [5].
UPFC hybrid model: in this model the series converter is replaced by a series voltage source and
leakage impedance related to the series transformer, and the parallel converter is replaced by
two ideal current sources; one expresses the shunt converter’s active power and its loss, and the
other indicates the reactive power of the shunt converter and its loss. [6]
International Journal of Engineering, (IJE) Volume (4): Issue (2) 105

This paper presents a new model of UPFC in steady state based on the circuit elements in the
exact transmission line model in a way that the series converter injection active power can be
simulated with a negative resistance and the series converter injection reactive power can be
simulated with a capacitor or inductive reactance, and the parallel converter reactive power will
be shown as a connected power to the bus. Equations of the resistance and reactance are
computed in two–bus single machine power system in the approximate transmission line model.
The model is then generalized to exact transmission line model and simulated in 14- and 30–bus
power system.
2. UPFC model based on circuit elements

UPFC series converter injects active and reactive power to the network by injecting series
voltage into the transmission line with controllable phase amplitude and angle. The series
converter provides the required active power through the capacitor link and the parallel converter
through the network. Also, each converter can inject reactive power to the network or receive
from it independently (see figure 1).
In the existing UPFC models, converters are replaced by voltage or current sources and the
controller final model is shown as the injection power to the bus which causes changes in the
Jacobean matrix and the load flow program structure. The goal of this paper is to model the
UPFC series converter with a resistance -which expresses the injected active power of series
converter to the line- in steady state, and also to model the injected reactive power with a
reactance so that the reactive power of parallel converter can be shown as connected power to
the bus .
FIGURE 1: UPFC is placed in two- bus power system
In this model, we assume that the active and reactive power exchanged between the series
converter and the network is known. Therefore, knowing the amplitude of buses’ voltages and
series converter injection voltage, we can compute the resistance and reactance values to
replace the series converter.
Modelling of the Series Converter

For modelling of the UPFC series converter in steady state, the following assumptions are
considered:
1- First, we consider the UPFC voltage source model in a two-bus one-machine system as
depicted in figure 2 and by calculating the line current, we will obtain the active and reactive
power injected by the series converter.
2- Then, we model the active power injected by a series converter with a negative resistance.
The series converter reactive power is also modelled with a capacitive or inductive
reactance, considering the type of the series converter.
Calculation of injected active and reactive power of the series converter

First, we obtain the electrical equations for figure 2 and calculate the series converter injection
power.

VS VSE Xl VR  0
I S
Qsh
FIGURE 2: UPFC in a two–bus network
Flowing current in the line is calculated as follows:

VS   VSE   VR 0
I (1)
jX L
The active and reactive power injected by the series converter is expressed by the following
equation:
*
S SE  VSE * I (2)
Replacing the first equation in the second one and expanding it, the active and reactive power
injected by the series converter is achieved:
VS * VSE V * VSE
PSE  sin(   )  R sin( ) (3)
XL XL
VS * VSE V * VSE V2
Q SE  cos(  )  R cos()  SE (4)
XL XL XL
The active and reactive power of the series converter can be modelled with a resistance and a
reactance, respectively.
VS  R XL VR 0
V SE
Qsh X I
FIGURE 3: Modeling of the series converter with parallel
2
VSE
R (5)
PSE
2
VSE
X (6)
Q SE
To prove our assumptions, we calculate the R bus absorbed power for the both systems shown
in figures 2 and 3 and show that the value of absorbed power by the R bus in the series
converter voltage model and our own model are the same.
Calculation of the active and reactive power absorbed by the R bus in the UPFC
voltage source model

The power of the R bus in the two–bus system of figure 2 is calculated using the following
equation:
*
S R  VR * I (7)
Replacing equation 1 in the above equation we have:
VS  VSE   VR 0 *
SR  VR * ( ) (8)
jX L
Expanding this equation yields the following equation:
S R  PR  jQ R (9)
So, the active and reactive power absorbed by the R bus can be expressed by these equations:
VS * VR V * VSE
PR  sin( )  R sin( ) (10)
XL XL
VS * VR V * VSE V2
QR  cos()  R cos()  R (11)
XL XL XL
In the next section, the active and reactive power absorbed by the R bus will be calculated with
the new UPFC model and we will show that the results are the same.
Calculating the active and

reactive power absorbed by the R bus in the UPFC circuitry elements model
The flowing current in the two-bus system of figure 3 are as follows:
. .
I  IR  IL (12)
VSE 
IR  (13)
R
V SE 
IL 
jX
(14)
Replacing the equations 13 and 14 in equation 12, we conclude:
VSE  VSE   90
I  (15)
R X
By expanding the above equation:
VSE V V V
I cos   SE sin   j SE sin   j SE cos  (16)
R X R X
*
Using this equation and S R  V R * I , we will have:
VSE VR V V V V V V
SR  cos   SE R sin   j SE R sin   j SE R cos  (17)
R X R X
The active and reactive power absorbed by the R bus:

VSE * VR V * VR
PR  cos   SE sin  (18)
R X
VSE * VR V * VR
QR   sin   SE cos  (19)
R X
Inserting equations 5 and 6 in 18 and 19, we have:
VS * VR V * VSE
PR  sin( )  R sin( ) (20)
XL XL
VS * VR V * VSE V2
QR  cos()  R cos()  R (21)
XL XL XL
Comparing the equations 10, 11, 20 and 21, we can say that the active and reactive power
absorbed by the R bus in the systems shown in figures 2 and 3 are the same. Consequently
UPFC series converter can be modeled with resistance and reactance.
Direct calculation of resistance and reactance equivalent to series converter

Inserting relations (10) and (11) in (5) and (6) we have :
VSE * X L
R (22)
VS sin(   )  VR sin 
VSE * X L
X (23)
VS cos(  )  VR cos   VSE
In this relationship , VS and V R are the amplitude of buses voltage ,  is the angle between
buses voltage , V SE ,  are the amplitude and angle of injected voltage to series converter and
X L is the line reactance .
Assuming the series converter and buses voltage and also load angle are fixed , the resistance
and reactance are only a subordinate of series converter fire angle and can replace the series
converter.
In order that this model can be simulated in load flow program , parallel elements convert to
series elements.
VS  VR 0
XL
Qsh
R UPFC X UPFC I
FIGURE 4: Modelling the UPFC in two –bus power system
The impedance equivalent to parallel elements equals to:
R * X2
R UPFC  (24)
R 2  X2

X*R2
X UPFC  2 (25)
R  X2
Therefore , by having buses voltage and the injected voltage of series converter values and the
active and reactive power value the series converter exchanges with network , the values of
resistance and reactance can be calculated , and put instead of series converter and also show
the reactive power of parallel converter as connected power to bus.
Modelling the UPFC in the exact transmission line model

In the previous section, we obtained the UPFC model based on the circuitry in the approximate
transmission line model. In this section, we model it in the exact transmission line model. To
perform this modelling, we consider the UPFC in the two–bus single-machine system in the exact
transmission line model in which the effects of the line capacitors are also considered. Figure 5
depicts the described model.
VS  VSE 
VR  0
VS' X L
IS Y Y IR
Q sh 2 2
FIGURE 5: UPFC placed in exact line model
To calculate the active and reactive power exchanged between the series converter and the
network, we first obtain the current I S considering the equations between the current and
voltage of the line model which is a bipolar network:
 V ' S  A B   VR 
    (26)
 S   C D  I R 
I
in which:
V ' S  VS   VSE  (27)
YZ
A  D  (1  ) (28)
2
BZ (29)
YZ
C  Y(1  ) (30)
2
Considering figure 5, we can write:
Z  jX L (31)
j
Y  j c  (32)
XC
Using equation 26, the transmitter line current equation can be written:
D ' CB  AD
IS  VS  VR (33)
B B

Inserting equations 27 to 32 in equation 33, the current equation is achieved as follows:
VS   VSE   VR 0 VS   VSE 
I  (34)
jX L  j2X C
The active and reactive power that the series converter exchanges with the network is obtained
as follows:
*
PSE  jQ SE  VSE *I (35)
in which:
*
V SE  VSE    (36)
Inserting equations 34 and 36 in equation 35 yields the following results:

VS * VSE V * VSE
PSE  jQ SE  (2    90)  R (    90) 
XL XL
2
(37)
VSE V2 V * VSE
  90  SE 90  S (    90)
XL 2X C 2X C
By expanding the above equation, the active and reactive power injected by the series converter
is achieved as follows:
the active power that the series converter injects into the network:
VS * VSE V * VSE V * VSE

PSE  sin(   )  R sin( )  S sin( ) (38)
XL XL 2X C
the reactive power that the series converter exchanges with the network:
VS * VSE V * VSE V2 V2 V * VSE

Q SE  cos(  )  R cos()  SE  SE  S cos(  ) (39)
XL XL X L 2X C 2X C
According to the figure 3, we can model the series converter active and reactive power with a
negative resistance and a capacitor or inductive reactance, respectively. It means:
2
VSE
R (40)
PSE
2
VSE
X (41)
Q SE
Inserting equations 38 and 39 in equation 40 and 41, we can write:
VSE
R (42)
VS V V
sin(   )  R sin   S sin 
XL XL 2X C

VSE
X (43)
VS V V V V
cos(  )  R cos   SE  SE  S cos(  )
XL XL X L 2X C 2X C
Using equations 24 and 25, the parallel reactance and resistance elements are converted to
series elements and according to figure 6, the UPFC series converter in the exact line model, is
modeled.
VS
VR 0
XL
R UPFC X UPFC I S
Y Y
Q sh 2 2
FIGURE 6: The UPFC series converter model in two-bus power system exact transmission model line
3. Simulation
In this section, we study the UPFC model based on circuitry elements in the 14- and 30-bus
power system. Performing load flow using the Newton-Raphson method [7]in the 14- and 30-
buses power system, the results of flowed power in the lines and voltage angle and amplitude of
buses is calculated. The results are summarized in table 1-9.
Load flow results in 14-bus system without UPFC series converter

The results show that the bus number 5 has the lowest voltage and based on voltage, is a weak
bus. Therefore we put UPFC converter as an example in line 1-5. By putting UPFC and changing
the amplitude and angle of voltage injected by the series converter, the active and reactive power
passing from line 1-5 will be controlled. The reactive power range is shown in figure 7 with
respect to the active power. From this curve, it is obvious that using UPFC, we can change and
control the active power in the range of 40 to 78 MW and reactive power between -8 to 42
MVAR.
bus bus
bus
voltage voltage
number
amplitude angle
1 1.06 0
2 1 -5.043
3 0.999 -13.362
4 .989 -11.428
5 .969 -9.151
6 1.02 -15.233
7 1.016 -14.628
8 1.05 -14.628
9 .998 -16.333
10 .998 -16.563
11 1.007 -16.095
12 1.012 -16.337
13 1.009 -16.526
14 .995 -17.72
TABLE 1: buses voltage amplitude and angle.

First End
MW MVAR
bus bus
1 2 161 73.391
1 5 57.7 40.61
2 3 74.7 17.859
2 4 63.8 25.351
2 5 40.8 36.253
4 3 19.5 -7.468
4 7 27.4 -2.434
4 9 15.6 4.373
5 4 46.3 -24.413
5 6 44.5 8.218
6 11 7.7 7.208
6 12 7.77 3.345
6 13 17.8 9.045
7 9 27.4 16.563
8 7 0 21.284
9 10 4.72 0.712
9 14 8.89 1.102
11 10 4.27 5.193
12 13 1.68 1.569
13 14 6 4.303
TABLE 2: Active and reactive power lines without UPFC.
FIGURE 7: Changes in reactive power in terms of active power in line 1-5
For instance, we want the active power passing from line 1-5, to increase from 57.7 to 75.2 MW
which is about 30 percent, and the reactive power of line 1-5 to decrease from 40.6 to 39.8
MVAR, which is around 1.7 percent. Therefore, assuming that the voltage amplitude injected by
the series converter is fixed and equals to 0.1, using equations 38 and 39, we obtain the voltage
phase angle of the series converter as follows:
and
Knowing the voltage phase and amplitude injected by the series converter, the resistance and
reactance values that are equivalent to the series converter, is calculated using equations 24, 25,
42 and 43:
And

Inserting the equivalent UPFC series converter resistance and reactance values in the line 1-5
and performing load flow calculations using Newton-Raphson method, the voltage and angle of
buses and also the active and reactive power of lines can be calculated. The results are shown in
tables 3 and 4.
Results of load flow with UPFC series converter in 14-bus power system
The results of tables 3 and 4 show that this model, like the other models of UPFC, is also able to
control the active and reactive power of lines, stabilize the buses voltage and use the
transmission lines in the optimized way. The flowed active and reactive power in line 1-5 is
depicted in figures 8 and 9, with and without UPFC.
bus bus
bus
voltage voltage
number
amplitude angle
1 1.06 0
2 1 -4.476
3 0.999 -12.4
4 1.006 -10.155
5 1.007 -7.604
6 1.02 -13.583
7 1.024 -13.203
8 1.05 -13.203
9 1.006 -14.843
10 1.004 -15.041
11 1.01 -14.51
12 1.013 -14.698
13 1.01 -14.896
14 1 -16.156
TABLE 3: buses voltage amplitude and angle
First End
MW MVAR
bus bus
1 2 143.00 65.939
1 5 75.615 30.761
2 3 71.173 12.748
2 4 57.794 8.887
2 5 32.342 6.467
4 3 23.027 3.563
4 7 26.786 3.373
4 9 15.349 6.799
5 4 54.968 2.282
5 6 45.565 28.477
6 11 8.384 5.31
6 12 7.861 3.066
6 13 18.12 8.068
7 9 26.786 17.204
8 7 0 15.662
9 10 4.116 2.564
9 14 8.519 2.322
11 10 4.884 3.322
12 13 1.761 1.291
13 14 6.381 3.057
TABLE 4: Active and reactive power lines with UPFC

FIGURE 8: the curve of line 1-5 in active power with and without UPFC
FIGURE 9: the curve of line 1-5 in active power with and without UPFC
In this system the load flow calculations are performed in the approximate transmission line
model and its results are compared with the exact transmission line model. (See table 5)
approximate model exact model percentage

line line change
Voltage bus 5 (pu) 1.003 1.007 +0.39%
Active power line 1 - 5 (MW) 75.19 75.615 +0.56%
Reactive power line 1 - 5

(MVAR) 39.88 30.761 -22.86%
TABLE 5: changes of voltage and active and reactive power of line 1-5 with UPFC in approximation and
exact transmission line model
Comparing the results, we conclude that the exact transmission line model yields more exact
results. So, use of the exact transmission line model based on circuitry elements is preferred in
comparison with the approximate model.
In the next section, the load flow calculations in 30-bus power system in presence of UPFC
model in exact transmission line model will be investigated.

Results of load flow on 30–bus system without UPFC series converter

Performing the calculations of the load distribution, buses’ voltage angles and amplitudes and
also the active and reactive power passing from the lines are computed. The results for voltage
and angle of buses number 2 and 5 and also the flowed power in line 2-5, are shown in tables 6
and 7.
bus number bus voltage bus voltage
amplitude angle
2 1.043 -4.97
5 1.005 -13.54
TABLE 6: buses voltage amplitude and angle
First
End bus MW MVAR
bus
2 5 78.7 25.2
TABLE 7: The active and reactive power line2-5 without UPFC
Results of load flow on 30–bus system in the presence of UPFC series converter
Using UPFC in line 2-5, we can control the passing active and reactive power from this line.
Calculating the load flow by MATLAB software and use of Newton-Raphson method and for
changes of voltage phase and amplitude injected by the series converter, the curve of reactive
power changes has been plotted with respect to the active power in line 2-5 in figure 10
.
FIGURE 10: The reactive power changes in terms of active power in line 2-5
As this figure determines, by using the UPFC new model, the active power of line 2-5 of 30-bus
power system can be controlled and changed between 63 to 97 MW and reactive power between
-23 to 20 MVAR.
Therefore, for each value of voltage amplitude and phase of the series converter and using
equations 24, 25, 42 and 43 the series converter equivalent resistance and reactance value will
be calculated and used instead of the UPFC series converter . For example in work point:
and
The UPFC series converter equivalent series resistance and reactance values can be achieved
as follows:

And
Putting the resistance and reactance values equivalent to the series converter and performing
load flow, the results will be obtained according to tables 8 and 9.
bus number bus voltage bus voltage

amplitude angle
2 1.043 -5
5 1.044 -11.1
TABLE 8: buses voltage amplitude and angle with UPFC
First End bus MW MVAR

bus
2 5 94.7 5.2
TABLE 9: The active and reactive power line2-5 with UPFC
Figures 11 and 12 show the curve for the active and reactive power passing from lines 2-5 with
and without UPFC.
FIGURE 11: Active power curve passing line 2-5 in the presence and without UPFC
FIGURE 12: Reactive power curve passing line 2-5 in the presence and without UPFC

4. Conclusion
Considering the results of the simulation in figures 6 to 13 and tables 1 to 9, we can conclude:
 Like the other UPFC models, this model is also able to control the flowed power in
transmission lines and as previously proved the results of this model and voltage injection
model is the same. Therefore we can replace UPFC model on the basis of circuitry
elements with other models.
 In this model a resistance is used instead of an injected active power, that the resistance
value and polarity show the injected active power’s value and direction; negative
resistance shows the power injection and positive resistance shows the active power
absorption.
 In this model instead of series converter reactive power, a capacitor or inductive

reactance is used and its value indicates the rate of the reactive power. If the series
converter injects reactive power into the network, the reactance will become capacitive
(negative), and if the series converter absorbs reactive power from the network, the
reactance will be inductive (positive).
 Unlike the previous models in which the active and reactive power of series converter
could be modelled as connected power to bus and this leads to change of program
structure and modification of Jacobian matrix, in this model the injected active and
reactive power can be modelled with a resistance and reactance and in load flow
programs only the network admittance matrix will change.
 Using the UPFC model based on circuitry elements in the exact transmission line model
and adding the capacitors effects, yields more accurate load flow calculation results, so
using the UPFC model in the exact transmission line model is preferred in comparison
with the approximate model.
5. REFERENCES
1. N.G. Hingorani, Gyugyi. “Understanding Facts”. NewYork. IEEE.PRESS, pp. 1-300(2000)
2. M.H. Haque, and C. M. Yam. “A simple method of solving the controlled load flow problem of
a power system in the presence of UPFC”. Elsevier Science, 65(1):55-62,2002
3. M. Noroozian, and M. Ghandhari. “Use of UPFC for optimal power flow control”, IEEE
Transaction on Power Delivery,12 (4): 1629-1634 (1997)
4. M.H. Haque. “Power flow control and voltage stabilitylimit:regulating transformer versus
UPFC”, IEEProc-Gener.Transm,151(3): 291-304 (2004)
5. M. Alomoush. “Exact Pi-Model of UPFC-Inserted Transmission Lines in Power Flow Studies”,

IEEE Trans,22(1): 54-56 (2002)
6. H.L. Sheng, and C. ChiChu. “Comprehensive UPFC Models for Power Flow Calculations in
Practical Power Systems”, IEEE Trans,1 (1): 27-32 (2000)
7. C.R. Fuerte-Esquivel. and H.Ambriz." A Comprehensive Newton- Raphson UPFC Model for
the Quadratic Power Flow Sloution of Pratical Power Networks", IEEE Transaction on Power
Systems, 15 (1): 102-109 (2000)

A.H. El-Sinawi
Vibration Attenuation of a Thin Cantilevered Beam Using LQG-

Based Controller and Inertial Actuator
A. H. El-Sinawi [email protected]
Associate Professor
Department of Mechanical Engineering
American University of Sharjah,
Sharjah, UAE, P.O. Box 26666
Abstract
This study considers the problem of attenuating the vibration at a certain location
on a flexible cantilevered beam mounted on a vibrating base which is the tip in
this case. Attenuation is achieved without the need for sensor placement at that
location. A modal state-space model of the flexible beam is constructed from the
beam’s first ten modes of vibration. A reduced-order optimal observer is utilized
to estimate the deflection of the beam’s tip from measurements of vertical
deflections at mid-span and actuator locations. An inertial actuator is mounted on
the beam itself, provides the control effort necessary for attenuating the tip
vibration, resulting from shaker excitation. Experimental and simulation results
have demonstrated the effectiveness of the proposed control technique.
Keywords: Inertial actuator, optimal control, LQG, estimation, cantilevered beam, modal analysis.
NOTATION
A dynamics matrix of full order modal model

Ac dynamics matrix of the controller
Ar dynamics matrix of reduced-order modal model
Aa dynamics matrix of augmented system
b actuator damping coefficient
Bf input vector of transmitted force
Bm input matrix of full order modal model
Br input matrix of reduced-order modal model
B rc control distribution modal matrix
Bc controller input vector
Cc controller output matrix
C md modal displacement output matrix
C mv modal rate output matrix
Cr output matrix of reduced-order system

A.H. El-Sinawi
D nodal damping matrix

D ur reduced-order feed through matrix
e2 model reduction error
Fi Excitation force from the shaker (support)
Fc actuator control force
Gc Actuator transfer function
Gr transfer function of reduced model
I identity matrix
J quadratic cost performance index
k actuator stiffness
K nodal stiffness matrix
Kc controller modal Matrix of gains
L Estimator gain matrix
M nodal mass matrix
m number of actuators
mo mass of inertial actuator
n dimension of nodal model
N dimension of modal model
Nr dimension of reduced order modal model
p number of sensors
Pc solution of the controller algebraic Riccati equation
Pe solution of the estimator algebraic Riccati equation
Q Controller weighting matrix
q vector of nodal displacements
R controller weighting matrix
s Laplace variable
t time
u input to transmitted force model
Uc control law
V estimator weighting matrix
v measurements noise
w vector of external inputs
W estimator weighting matrix
Wc controllability grammian
Wo observablility grammian
y(t) displacement at actuator location
Yr nodal output vector of reduced model
Ys nodal output vector of full model
z modal states
Δ modal damping matrix
η, η modal displacement and velocity vectors
Φ matrix of eigenvectors
International Journal of Engineering (IJE), Volume (4), Issue (2) 120

A.H. El-Sinawi
Ω diagonal matrix of eigenvalues
1. INTRODUCTION
Accuracy of the model of a flexible structure plays a major role in the design of an
appropriate control system that will force the structure to perform certain tasks or
attenuate structural vibration. Flexible structures are inherently distributed parameter
structures with infinite degrees of freedom. A model of such structure will have to be a
reduced order model but adequate enough to yield a controller that will carry out the
necessary manipulation and/or vibration attenuation.
Research efforts in this field have provided a wealth of techniques for modeling and
control of flexible structures. Once the model is constructed, the accuracy and precision
of control systems, needed to manipulate the structure, or to perform certain tasks,
depends solely on the accuracy of the structure’s model. Since, models of flexible
structures are reduced order, the control scheme has to either, focus on certain dynamic
characteristics and try to control the structure based on the available model, or it has the
ability to somewhat estimate or predict the un accounted for dynamics. In either case,
the model and the controller will have limited ability in predicting the actual structural
behavior [1]. As it is known, any active control scheme must keep track of the behavior
of the physical system in order to adjust its effort to suite the current status of the
structure and perform the predefined requirements on that structure. This also poses
another problem in large flexible structures, since sensor placement on large flexible
structures increases the complexity of both the model and the controller. Therefore, to
actively control a flexible structure, one has to keep in mind all the aforementioned
challenges.
From modeling standpoint, many research efforts have utilized FEM to obtain models of
flexible structures that can be used for control purposes. However, modal models
obtained by FEM are not sufficient for providing a reliable model for control purposes for
many reasons, one of which is that, FEM provides no information about damping in the
structure [1]. This aspect is crucial to the model since damping dictates the transient
behavior of the structure as well as the phase of the control effort in relation to structural
vibration.
In modeling of flexible structures, FEM technique is widely used for constructing

analytical models of flexible structures. This method is well established but may not be
accurate for controller design due to drawbacks mentioned above [2, 3]. Another
approach for modeling flexible systems is to determine the model directly from
experimental data, such as experimental modal analysis [2]. Analytical methods have
also been used to obtain working models for flexible structures [4-6].
As for the control of flexible structures, considerable research has been done in this field
with various approaches and techniques. A common form of vibration control of flexible
structure is done using passive means such as using viscoelastic materials, and passive
damping [7, 8]. Other control approaches utilize various forms of feedback control
techniques to achieve certain manipulation, attenuation, suppression or isolation
requirements [9-13]. Other techniques involve robust control, optimal and fuzzy control
[14-19]. Control techniques of flexible structures focus mainly on improving the agility,
efficiency and bandwidth of the controller in the presence of noise as well as modeling
and excitation uncertainties. A comprehensive review of shape control of flexible

A.H. El-Sinawi
structures with emphasis on smart structures using piezoelectric actuators is presented

by [20, 21].
In this work, vibration attenuation of the tip of a cantilevered beam subjected to

excitation at the support is presented. A novel control technique uses a non-collocated
sensors and actuators to attenuate the vibration at the beam’s tip. The control effort is
determine by an LQG-based controller that uses feedback measurements at locations
other than the tip to determine the required control force needed to attenuate the
vibration at the tip. A non-reactive actuator (i.e. inertial actuator) is mounted close to the
support and used to apply an appropriate control force to attenuate the tip vibration. The
sensor-actuator non-collocation is utilized in this work to accommodate situations where
sensor placement at the control location is prohibitive. Inertial actuators react directly off
a mass; therefore, they can be placed directly on the vibrating structure without the need
for a reactive base [11, 12, 22]. LQG is utilized to estimate the control effort needed
without the need for sensor placement at the control location. LQG-based controller
reduces computational cost and provides robustness in the presence of model
uncertainties as well as measurements and excitation noise. Dynamic modal model of
the flexible structure is constructed from the first ten modes of vibration obtained via
finite element analysis of the structure. To improve the agility of the controller and
reduce its sensitivity, the proposed control technique uses a reduced order modal model
of the structure. The latter eliminates the least effective modes of vibration at the control
location, i.e., modes with the least H 2 norms are truncated from the modal model [23].
Both simulation and experiments are carried out to verify the integrity of the proposed
control technique.
2. CONTROL STRATEGY DESCRIPTION
In the following sections, the linear quadratic Gaussian (LQG) based controller is
constructed. The LQG has two parts, namely, an optimal observer, and an optimal
controller. The observer generates estimates of the vertical nodal displacement of three
locations on the beam, which are the actuator location, mid-span, and the beam tip. The
controller output is a control force lateral to the beam axis. As for the inputs, the
observer is subject to all external inputs that the actual beam is subject to. Those inputs
are the force applied to the beam at the support Fi , and the vertical control force Fc .
Moreover, optimal observers in general require feedback of errors between state
estimates and state measurements.
The demand on LQG is to reduce the vibration of the beam tip. To fulfill this demand,
the LQG has to perform the two nearly-simultaneous tasks of, (a) producing optimal
estimates of the vibration at three locations along the beam’s span and (b) generating a
control vector that will drive the inertial actuator to minimize the force transmitted to the
tip and subsequently reduce its vibration.
3. INERTIAL ACTUATOR
To better explain the control scheme proposed by this study. The layout of the dynamic
system is shown in Figure (1-a). In Figure (1-b) a free body diagram of the beam is
shown where the inertial actuator is applying force Fc to cancel out the effect of force Fi
on the section of the beam to the right of the inertial actuator. The actuator is particularly

A.H. El-Sinawi
targeting the attenuation of the vibration at the tip through transferring a force to the
beam with required magnitude and phase such that the tip’s vibration is minimized.
Modeling the actuator as a spring mass damper system, the transmitted force from
actuator to the beam can be expressed in the form
 mo 
z Fc (1)
where mo is the mass of the actuator and z is the actuator’s mass displacement.
Therefore,
Fc  c( z y)  k ( z  y ) (2)
where y is the displacement of the beam at the actuator’s location. Assuming that the
ultimate objective of the controller is to nullify the displacement of the actuator mass
relative to the beam (i.e., z  0 ), will yield an a required actuator force of the form,
FC  c ( y)  k ( y ) (3)
With an actuator stiffness that is relatively large compared to the stiffness of the beam,
and assuming only natural damping in the beam and actuator, the force acting on the
actuator mass will vanish only if the vibration of the beam at the actuator location
vanishes. However, since the purpose of the control is to eliminate the vibration at the
beam tip not the actuator location, the controller will assume the actuator’s mass
vibration is indirectly caused by the vibration of the tip, and therefore, must keep track of
the tip’s vibration despite the fact that physical measurements at that tip are
inaccessible. Therefore, if estimates of the tip vibration is fed to the actuator, the result
will be a control force that will minimize the tip vibration with Fc in Equation (3) replaced
by
FC  c ( y
ˆ t )  k ( yˆ t ) (4)
Where ŷt is the estimate of the tip’s vibration and the transfer function mapping the tip’s
displacement estimates to the control force is,
Fs ( s )
G( s)   bs  k  . (5)
Yˆt ( s )
this resembles a PD controller with proportional and derivative gains equal to the
stiffness and damping of the actuator, respectively.
It is clear, that actual implementation of the inertial actuator will cause the dynamics of
the latter to be part of the overall dynamics of the system. Therefore, actuator dynamics
will be later on augmented with those of the structure (i.e., beam) during the design of
the LQG-based controller.
4. STATE-SPACE REPRESENTATION OF THE BEAM
To implement the proposed concept of controlling the tip vibration using inertial actuator,
a state-space model is formulated, in this case, for a distributed parameter system such
as the beam shown in Figure (1). The beam is considered as an n-dimensional model of
a modally damped flexible structure having (m) actuators and (p) sensors, not
necessarily collocated. The beam system retrofitted with an actuator is shown in Figure

A.H. El-Sinawi
(1). The beam’s structure is represented in nodal coordinates by the following second-
order matrix differential equation:
Mq
 Dq Kq  Bw (6.a)
Y s  C d q  C v q (6.b)
In the above equation q is the n  1 displacement vector, w is the m  1 external input

vector, Y s is the p  1 nodal output vector, M , D , and K are the n  n mass, damping,
and stiffness matrices, respectively. C d , and C v are respectively, the p  n output
displacement and output velocity matrices. The mass matrix is positive definite and the
stiffness and damping matrices are positive semidefinite. The damping matrix D is
assumed to be proportional to the stiffness matrix K without any significant effect on the
integrity of the model [23, 24]. Dynamic model such as the one shown in Equation (6) is
usually obtained from finite-element codes and has the dimension n which is
unacceptably high to use in producing a state-space model suitable for structural control.
Therefore an alternative approach is to use an N-dimensional second-order modal
model of the system where N  n .
A second order modal-model of the system can be expressed as
 2ΔΩη Ω 2 η  B m w( t )

η (7.a)
Y s  C md η  C mv η (7.b)
where η  Φq , and Φ is the n  N modal matrix, Ω is the N  N diagonal matrix of

modal natural frequencies, Δ is the N  N modal damping matrix, B m is the N  m
modal input matrix, C md and C mv are the p  N modal displacement and rate matrices,
respectively. Defining the state vector z  z 1 z 2   η η , a flexible structures
T T
having point force(s) as the input(s) and point displacement(s) as outputs will have the
state space representation,
 0 I   0 
z  2  z w (8.a)
 Ω  2ΔΩ  B m 
Y s  C md 0 z  D u u (8.b)
where, Ys (t ) and w(t ) are the nodal output and input vectors, respectively. Matrix D u is
the p  m feed through matrix. Equation (8) can also be represented by the following
compact form,
z A(  ) z  B(  ) w (9.a)
Y s  C(  ) z  D u (  ) w (9.b)
where A , B , C , and D u matrices are functions of the system (natural frequency,

damping ratio, and mode shapes (i.e.,   f  i ,  i , and  i i 1 N ) . The dimension of
this state-space representation is 2 N and it is much more manageable than the 2n
state-space model obtained from the corresponding nodal model. Detailed information

A.H. El-Sinawi
on constructing state-space model from natural frequencies and mode shapes is given in
[15, 23].
5. MODEL REDUCTION
To show the effectiveness of the proposed control strategy, the 2 N -dimension state
space model of the beam-hub system constructed in Equation (9) will be reduced even
further to a 2 N r -dimension state-space representation. This will reduce the
computational cost and further validate the effectiveness of the proposed technique. In
this study the reduction is based on retaining only half the number of modes
(i.e., N r  (N / 2) ) and the modes retained will be those with the largest H 2 norm [25]. In
such case, the H 2 reduction error is expressed as:
e2  G  G r 2
(10)
where G is assumed to be the transfer function of the full model corresponding to

equation (9), and G r is the transfer function of the reduced model. The square of the
mode norms are additive [25], therefore the norm of the reduced system with N r modes
is the root-mean-square sum of the mode norms
Nr
2 2
Gr 2
  Gj (11)
2
j
and the reduction error is,

N
2 2 2
e22  G 2  G r 2
 G i 2 (12)
j  N r 1
The term Gi 2
is the modal cost of Skelton [26], which for the jth mode has the form
Bj Cj
2 2
Gj  (13)
2
2  j j
It can be seen from Equation (12) that near optimal reduction is attained if the truncated
mode norms for j  N r  1,  N are the smallest. It should be noted that after
truncation, modes should be rearranged such that the reduced model has a modal
natural frequency  f  diag[1 ,   k ] where, 1 has the highest H 2 norm. This also
implies rearranging the corresponding modal states as well. The new state space
representation of the truncated system is:
z A r (  ) z  B r (  ) w (14.a)
Yr  C r (  ) z  D ur (  ) w (14.b)
where the subscript r denotes a reduced –order model. This state-space model has the
dimension 2 N r .
6. AUGMENTATION OF THE CONTROLLR WITH THE STRUCTURE

A.H. El-Sinawi
As it was mentioned in the Section (3) above, the dynamics of the actuator must be
augmented with the dynamics of the beam because the actuator will e mounted on
the beam itself. The augmented system will have the following state-space matrices,
 A 0  Bc 
Aa   c , B  , C  Dur  Cc C r , and Da  [ Dur  Dc ]
 Br C c Ar  a  Br  Dc  a
(15)
Where Ac , Bc , Cc , Dc , are the state space matrices of the controller given in

Equation (5). Notice that the transfer function in Equation (5) is unrealizable at its
current state, but it can be with a second order filter in the denominator assuming that
measurements of acceleration rather than displacement are to be fed to the actuator;
see [15] for details.
7. OPTIMAL OBSERVER-CONTROLLER SYNTHESIS
In this section, the optimal observer-controller is formulated. The formulation is based on

the augmented model of Equation (15). Measurements of the beam’s acceleration in the
vertical direction are assumed to be available at actuator location and at beam’s mid-
span. The beam support is assumed to be mounted on a shaker the cause the excitation
at the support. The optimal observer-controller will be an LQG type for which the plant is
described as follows:
z A a z  B rc U c  B a w (16.a)
Ya  C a z  v (16.b)
The plant model of equation (19) is perturbed by the external input w , and the output Ya
is corrupted with noise v . The term Dur is dropped because the denominator of
Equation (5) has been replaced by the second order filter. The optimal control problem
is formulated by finding the linear quadratic regulator (LQR) gains and the Kalman
estimator gains that will minimize a quadratic performance index,

T
J   ( z T Qz  U c RU c ) dt (17)
0
The estimator dynamics are given by the following state-space matrix differential
equation
z
ˆ  Aa zˆ  BrcUc  L(Yr  Ca zˆ)  Ba w (18.a)
Yâ  C a zˆ (18.b)
Here ẑ is the vector of estimated modal states, Brc is the modal feedback control
distribution matrix, and U c  K c ˆz , where K c , and L are respectively, the LQR and
Kalman gains. The feedback gain vector K c is give by
K c  R 1B Trc Pc (19)

A.H. El-Sinawi
where Pc is the solution of the controller algebraic Riccati equation

T
A a Pc  Pc A a  Pc B rc R 1B Trc Pc  Q  0 , (20)
and the estimator gain matrix L is given by

L  Pe C Ta V 1 (21)
where Pe is the solution of the estimator algebraic Riccati equation

T T T
Aa Pe  Pe Aa  PeCa V1Ca Pe  Ba WBa  0 (22)
In this study, Equation (18) produces the vector


Yâ  qˆ c qˆtip qˆ mid  span 
T
qˆ act _ location where, qc are the controller states. It should be
noted that the matrix K c is the LQR modal gains and the portion of Yâ namely;
 qˆtip qˆ mid  span qˆ act _ location 
T
is the vector of beam’s nodal displacements, and
thus, before utilizing the optimal feedback, this portion has to be transformed back to the
modal domain using the modal matrix Φ . Matrices R and V are positive definite while
matrices Q and W are positive semidefinite. The latter four matrices can be treated both
as weighting matrices and as tuning parameters for the LQG [26, 27].
In this study it is assumed beforehand that qtip is not available for measurements,
therefore, no estimate error for this state is available for feedback. However, estimate
errors for the actuator location and the mid-span states are available and must be
fedback to the estimator through the Kalman matrix of gains L . This implies that poles of
the closed loop system depicted by Equation (18) are placed in such manners that only
optimal LQG estimates of actuator location and mid-span displacements are obtained.
This is a major deficiency and it will cause an arbitrary change in the tip state due to the
lack of information available to the estimator about measured tip displacements. Such
estimator will not be of much use if the tip deflection is to be controlled. Therefore, the
last term of equation (18) (i.e., B a w ) is utilized to both eliminate this deficiency and
achieve the demand on the LQG. In other words, the term Ba w will have to enable the
LQG to (a) force subsequent estimates of tip modal states (i.e., ẑ ) to be dependent on
previous estimates of tip deflection and (b) generate a control force that will minimize the
difference between the actuator mass and tip deflections through manipulation of the
actuator location only. If the inertial actuator utilized to generates a force that is a
function of the difference between the actuator mass displacement (i.e., zero in this case
based on the reasoning in Equations (3-5)) and tip deflections, then equation (18) will
have w  Fi Fc  where,
T
Fi and Fc are the excitation and control forces,
respectively. we obtain,
Fi  Cc (sI  A c ) 1 Bc (0  q̂ act _ location ) , (26)
where s is the Laplace variable, I is a 2  2 identity matrix.

For effective implementation of the proposed control strategy it is important to determine
early on whether the system is controllable (i.e., actuators excite all modes) and
observable (i.e., the sensors detect the motion of all modes). It is clear, however, that

A.H. El-Sinawi
the proposed control strategy may very well have a value of N large enough to prohibit
accurate calculation of the controllability and observability matrices of the system. To
avoid such problem, the controllability and observability grammians are used instead of
the original method proposed by Kalman [22, 23]. The grammians are determined from
the following Lyapunov equations:
AWc  Wc A T  BB T  0,
(30)
A T Wo  Wo A  CCT  0
where A , B , and C , are the dynamic , input, output matrices of the overall model of
the system, respectively. Matrices Wc and Wo are, respectively, the controllability and
T
observability grammians. The system is controllable and observable if Wc  Wc , and
Wo  WoT . The synthesized controller is implemented and the results are presented in
the next section.
8. EXPERIMENTAL AND NUMERICAL SIMULATION RESULTS

Vibration attenuation of the beam tip of Figure (1) is investigated. The beam-actuator
system is subject to excitation shown in Figure (2) exerted on the beam’s support.
Dimensions and properties of the system under study are listed in Table 1. Actual
behavior of the beam-actuator is constructed based on the first ten modes of vibration of
the beam-actuator system for which the modal frequencies and modal norms are listed
in Table (2). The reduced-order dynamics model used in the design of LQG is
constructed from four modes only. The four modes are those with the largest H 2 norms
as shown in Table 2, namely, modes 1, 2, 3, and 5. The reduced order model of the
system is constructed from those four modes and the states are arranged according to
the same sequence.
Using four modes of vibration to synthesize the LQG-based controller indicates that
exact knowledge of the dynamics of the system is not necessary for successful
implementation of the controller; rather a carefully reduced-order model is sufficient. The
controller and estimator gain matrices, K c , and L are generated in Matlab using
weighting matrices Q  α I 1010 , R  I 22 , V  10  I 22 , and W  I 33 ,   100 . Modal
damping of 0.01 is used for all modes of vibration. Notice that the two extra states in K c ,
and L belong to the controller as shown by Equation (15).
The control effort is tuned to reduce the relative vibration between the beam’s tip and the
actuator mass via (1) feedback of two error signals; which are the difference between
measured and estimated displacements at actuator location and mid-span, which are
used to produce better estimates of deflections including that of the tip. And (2)
estimates of the tip deflection, are fed through the inertial actuator with a demand to
produce the control effort necessary to reduce the tip vibration. Block diagram of the
control scheme used in this simulation is shown in Figure (3).
The ideal outcome of the control strategy is to reduce the estimates of the beam tip
vibration to zero. Figures 4 and 5 show significant reduction in the vibration both at the
tip, and at the actuator location. Figure 6 shows the control force needed to achieve the
attenuation and it is within the capabilities of the actuator.

A.H. El-Sinawi
To verify simulation outcomes, experiments are carried out under conditions identical to
those used in the simulation. The experimental setup shown in Figure 7 is utilized to
verify the integrity of the proposed technique. Significant reduction in the magnitude of
the transfer function (approximately 30% at resonant frequencies) is attained as shown
in Figure 8. No abnormal vibration is created at the actuator location when the controller
is targeting the tip vibration as shown in Figure 9. Experimental actuator force is shown
in Figure 10 which is reasonable and within the capabilities of the actuator used.
It should be pointed out that this control technique has draw backs mostly in its
sensitivity to excitation changes, mass of the inertial actuator, and overestimates of the
force which appear to be related to the choice of values for the weighting matrices of the
controller and observer as well as the derivative term in the transfer function of Equation
(5).
9. CONCLUSIONS
A novel approach for attenuating the vibration of a cantilevered beam mounted on a

vibrating base has been presented. The proposed technique is particularly useful
considering the fact that, real-life systems exhibits considerable variations in their
properties. Thus, the characteristics of a structure corresponding to these properties
show some stochastic variations. This makes it necessary to take into account the
uncertainties of the system if a reliable control system is to be implemented. To this
purpose, a robust and effective control system using optimal estimation and control
techniques would be a suitable choice. The control strategy uses available
measurements at various locations on a flexible structure to produce estimates of the
vibration at inaccessible locations. Latter estimates are then used to control the vibration
at those inaccessible locations using inertial actuator. The inertial actuator is non-
collocated with the sensors, and can be mounted directly on the structure, providing
flexibility in locating sensors and actuators according to needs and constraints.
Experimental and simulation results of the proposed method show significant reduction
in the tip vibration without the need for any sensor(s) placement at the control location
(the tip in this case). This is an important feature which could prove useful in applications
where use of sensory devices at any location on the flexible structure is difficult to attain.
The proposed strategy managed to reduce the vibration of the beam’s tip by
approximately 30% at resonant frequencies using only estimates of the tip vibration
rather than actual measurements.
Fine tuning of the weighting matrices should be carried to obtain the optimal results for
the control force and the reduction in the tip vibration. Further work is needed in the
implementation of this technique and two and three dimensions where the processes
can be duplicated for the other two dimensions that were not discussed by this work.
ACKNOWLEDGMENT
The author acknowledges the support of American University of Sharjah.
REFERENCES
[1]. J. Dosch, D. Leo, and D. Inman. “Modeling and control for vibration suppression of
a flexible active structure”. Journal of Guidance Control and Dynamics, 18(2): 340-
346, 1995
[2]. J. N. Juang and R. S. Pappa. “An eigensystem realization algorithm for modal

A.H. El-Sinawi
parameter identification and model reduction (control systems design for large
space structures)”
. Journal of Guidance, Control, and Dynamics, 8: 620-627, 1985
[3]. Gawronski, W. and Williams, T. “Modal reduction for flexible space structures”.
Journal of Guidance, 4: 68–76, 1991
[4]. Y-R Hu and Alfred Ng. “Active robust vibration control of flexible structures”. Journal
of Sound and Vibration, 288: 43-56, 2005
[5]. J-H Han, K-H Rew and I. Lee. “An experimental study of active vibration control of
composite structures with a piezo-ceramic actuator and a piezo-film sensor”. Smart
Materials and Structures, 6: 549-558, 1997
[6]. M. J. Balas. “Active control of flexible systems”. Journal of Optimization Theory and
Applications, 25 (3): 415-436, 1978
[7]. Yen, H-Y. and Shen, M. H. “Passive vibration suppression of a beam and blades
using magneto-mechanical coating”. Journal of Sound and Vibration, 245(4): 701-
714, 2001
[8]. Griffin, J. H. “Friction damping resonant stresses in gas turbine engine airfoil”.
International Journal of Turbo and Jet Engines, 7, pp. 297-307, 1999
[9]. H.A. Malki, M. D. Feigenspan, and D. G. Chen. “Fuzzy PID control of a flexible-
joint robot arm with uncertainties from time-varying loads”. IEEE Transactions on
Control Systems Technology, 5(3): 371-378. 1997
[10]. Krodkiewski, J. M. and Faragher, J. S. “Stabilization of motion of Helicopter rotor
blades using delayed feedback-modeling, computer simulation and experimental
verification”. Journal of Sound and Vibration, 234(4): 591-610, 2000
[11]. L. Benassi and S.J. Elliott. “Active vibration isolation using an inertial actuator with
local displacement feedback control”. Journal of Sound and Vibration, 278: 705–
724, 2004
[12]. L. Benassi and S.J. Elliott. “Global control of a vibrating plate using a feedback-
controlled inertial actuator”. Journal of Sound and Vibration, 283: 69-90, 2005
[13]. N. Hogan. “Impedance control: An approach to manipulation”. Journal of Dynamic
Systems Measurement and Control, 107: 8-16, 1985
[14]. A. H. El-Sinawi. “Vibration attenuation of a flexible beam mounted on a rotating
compliant hub”. I-Mech, Part I, Journal of Systems and Control Engineering,
218:121-135, 2004
[15]. El-Sinawi, A. H. and Hamdan, M. N. “Optimal vibration estimation of a non-linear
flexible beam mounted on a rotating compliant hub”. Journal of Sound and
Vibration, 259(4), pp. 857-872, 2003
[16]. El-Sinawi, A. H. and Kashani, A. R. “Active isolation using a Kalman estimator-
based controller”. Journal of Vibration and Control, 7: 1163–1173, 2001
[17]. M. R. Serbyn. “Active control of cantilever-beam vibration”. Journal of Acoustical
Society of America, 112 (5)5: 2246-2247, 2002
[18]. J. J. Dosch, G. A. Lesieutre; G. H. Koopmann; and C. L. Davis. “Inertial piezo-
ceramic actuators for smart structures”. Smart Structures and Materials, 2447: 14-
25, 1995
[19]. R. Babuska and H. B. Verbruggen. “An overview of fuzzy modeling for control”.
Control Engineering Practice, 4 (11): 1593-1606, 1996
[20]. H. Irschik. A review on static and dynamic shape control of structures by piezo-
electric actuation. Engineering Structures, 24(1): pp. 5-11, 2002
[21]. L.-C. Yao, J.-S. Chen, C,-Y, Hsu, “A Mode Switching Sliding-mode Controller with Observer-
based State-Dependent Boundary Layer and Its Application”. International Journal of
Engineering, 1(1): 39-53, 2007

A.H. El-Sinawi
[22]. C. Paulitsch, P. Gardonio and S. J. Elliot. “Active vibration control using an inertial
actuator with internal damping”. Journal of Acoustical Society of America, 119 (4),
pp. 2131-2140, 2006
[23]. Gawronski, W. and Williams, T. “Modal reduction for flexible space structures”.
Journal of Guidance, 4: 68–76, 1991
[24]. Gawronski, W. K. “Dynamics and Control of Structures”, (Springer-Verlag; New
York: First edition.), 1998
[25]. Bellos, J. and Inman, D. J. “Frequency response of non-proportionally damped,
lumped parameter, linear dynamic system”. ASME Journal of Vibration and
Acoustics, 112, pp. 194-201, 1990
[26]. Skelton, R. E. and Hughes, P. C. “Modal cost analysis of linear matrix second-order
system”. Journal of Dynamic Systems, Measurements and Control, 102: 151-158,
1980
[27]. Maciejowski, J. M. “Multi Variable Feedback Design”. (Addison-Wesley;
Workingham: England.) 1989
Table 1: Simulation and Experimental Data
Component Properties Dimensions

Thin beam Aluminum 6061 Length =0.53 m
E  6.89  1010 Pa Thickness =2.1 mm
Depth=0.035 m
  2710 kg / m 3
Poisson ' s Ratio  0.35
Actuator ETREMA Terfenol-D Stiffness 300 kN/mm
Actuator Mass=0.203 kg
Damping =11560 kg/s
Excursion=10 micro-meter pk-pk
Excitation B&K shaker 1-100 Hz chirp signal + Random
input
Table 2: Beam-Actuator system modal frequencies and modal H 2 Norms

A.H. El-Sinawi
Mode Number Frequency Hz H 2 Norm

1 10.106 0.05043736496038
2 54.732 0.00179075345056
3 139.72 0.00021895088343
4 195.05 0.00021895088343
5 271.71 0.0000000000000012
6 457.76 0.00004501906852
7 461.14 0.00000000000000
8 590.08 0.00002132364349
9 720.38 0.00000521173674
10 1000.4 0

A.H. El-Sinawi
Figure 1: (a) Cantilevered beam mounted on a shaker (b) Free body diagram of the beam

A.H. El-Sinawi
Figure 2: Excitation at the beam’s support
Figure 3: Block Diagram of the control scheme

A.H. El-Sinawi
Figure 4: Tip vibration with and without control
Figure 5: Vibration at the actuator location

A.H. El-Sinawi
Figure 6: Control force, simulation
Figure 7: Experimental Setup

A.H. El-Sinawi
Figure 8: Frequency response function of the beam: tip displacement/shaker displacement
Figure 9: FRF of the beam with and without control

A.H. El-Sinawi
Figure 10: Control force, experimental

Saqaff Ahmed Alkaff & Mohamed Khan
A New Strategy to Achieve Radical Combustion

Through Exhaust Port Throttling for Two Stroke Engine
Saqaff Ahmed Alkaff [email protected]

Faculty of Engineering,
Mechanical Engineering Department
University of Aden, Yemen,
Aden, Yemen
Mohamed A. Khan [email protected]

Faculty of Engineering,
Mechanical Engineering Department
University of Aden, Yemen,
Aden, Yemen
Abstract
Radical combustion is a critical condition behind control. In two stroke engine, a

condition of radical combustion could be achieved through control of the trapped
exhaust gases at a certain condition, might include, the engine load, speed,
concentration of the unburned hydrocarbons, their temperature and other
operational parameters.
An earlier work was conducted towards the achievement of the radical
combustion. The mechanism was made to throttle the opening of the exhaust
port at a range of 1% to 8 % [1]. However, several difficulties were faced, mainly
with the control mechanism of throttling. In addition there is a complexity in
identifying the critical conditions at which the radical combustion could be
achieved.
In this paper, a new strategy was used to control the exhaust port throttling in a
way to manage the amount of exhaust gases trapped and avoiding the
drawbacks of the throttling mechanism in the earlier work. Three trial plates
were used, one-sixth closed, one-third closed and half closed to throttle the
exhaust gases leaving the combustion chamber.
Results reveal the possibility of achieving radical combustion, when using the
one-sixth closed plate under relatively higher load. Therefore, partial trapping of
the exhaust gases should be carried, through the restriction of the opening of the
exhaust port not to exceed 15% of the port exit area. However, it is more
interesting, that throttling of the exhaust gases of the two stroke engine have
clear influence on the quality and stability of the combustion and hence its direct
effect on the fuel consumption and the rate of pollutants expelled to the
environment.
Keywords: Spark Ignition Combustion, Radical Combustion, Exhaust Port Throttling, Two Stroke Engine,
Trial Plates for Throttling Control,
International Journal of Engineering (IJE), Volume (4) : Issue(2) 139

1. INTRODUCTION
The combustion is a chemical reaction in which certain elements of the fuel (i.e. hydrogen and
carbon) combined with oxygen liberating heat energy and causing an increase in temperature of
the gases. The conditions necessary for combustion are the presence of a combustible mixture
and some means of initiating the process (electric spark). Generally, the chemical reaction
occurs more rapidly as the temperature increase. Chain reactions are the most common type of
chemical reactions occurs in internal combustion engines. They consist of a series of
consecutive, competitive and opposing reaction steps with different reaction rate constants [2, 3,
4]. These complex chemical reactions occur in all combustion processes. For many combustion
processes, the specific rate constant for separate reaction steps either are not known or have
been approximated roughly [5, 6, 7]. In this reaction process, the most active species are called
free radicals. In chemical terminology, free radicals are known as unpaired electrons or “bond-
paired” electrons. The hydrogen atom is a free radical, as is illustrated below, where the dots
symbolize electrons.
H:H H+H
If one hydrogen atom H is taken away from CH4, two radicals are formed;
H H
.. ..
H: C: H  H : C
. .
+ H
.. ..
H H
Elementary reactions are called chain initiating or chain-terminating reactions, according as they
produce or destroy free radicals. Also, with regard to the ratio of the number of free radicals in the
products to that in the reactant, elementary reactions are called chain-propagating reactions if the
ratio is equal to 1, and bring chain-branching reactions if the ratio is greater than 1. Some
elementary reactions and their denominations are given below.
A 2  2 A chain initiating (A2 has low dissociation energy)
A + B2  AB + B
B + A2  AB + A chain propagating (very fast rate )
A + AB  A2 + B
B + AB  B2 + A
M + 2A  A2 + M
M + 2B  B2 + M chain termination
A and B are called chain carriers or free radicals and seldom build up in high concentrations. The
elementary reaction
H + O2  OH + O
Is a chain branching reaction, since the number of chain carriers formed is more than a number of
chain carriers used up in the reaction [8, 9, 10].
In this combustion reaction, the element that involved must have energy supply from outside of
the system. The molecules will not react unless heat is produced to some extent from outside.
From the analysis, the combustion not only needs the high temperature but also by fuel
concentration and pressure. Below this temperature, the chain reaction does not take place
because of the few molecules acquire the activation of the combustion reaction have not enough
energy[11, 12, 13].
The most important elements of the fuel are carbon and hydrogen with a small amount of sulphur,
oxygen, water vapour, nitrogen and ash. The fuel elements are bonded to one and another. The
bonding systems have their own energy and are difficult to break. To make sure this hydrocarbon

reacts, the energy must be added to break these bonds. Though there are many kinds of fuel as
well as very complicated combustion mechanisms, fuels usually decompose before the
combustion starts and the oxidation occurs through some basic form. The main substances are
hydrogen, carbon monoxide, gaseous hydrocarbons and solid carbon [14, 15].
2. ACTIVATED RADICAL COMBUSTION THEORY

When the spark plug fires and ignites the fuel mixture, some of the fuel is isolated from the
resulting flame by the exhaust constituents remained in the cylinder from the previous cycle and
does not burn. Activated radical combustion process will burn all the fuel inside the combustion
chamber by using the active radical molecules that are formed in the exhaust gas. When hot
exhaust gases remain in the cylinder, it contains a small percentage of active radical molecules;
when these are combined with the incoming fuel charge, the resulting mixture begins to auto-
ignite at a lower temperature in comparison to that of a pure gasoline/air mixture [14, 17, 18,
19].
3. ACHIEVEMENT OF RADICAL COMBUSTION

Radical combustion is a critical condition behind control. The condition of the radical combustion
is affected by the amount of exhaust gases trapped in the combustion chamber at certain critical
conditions, including the engine load, engine speed, concentration of unburned hydrocarbons
presence in the trapped part of exhaust and its temperature [1, 20, 21].
The earlier work was carried for the achievement of the radical combustion. A control of the
amount of exhaust gasses flow out from combustion space, will lead to trap small amount of
exhaust gasses and kept to the next cycle inside the combustion chamber. Trapping of such part
of exhaust gases may assure the existence of radical agents as OH and H. The exhaust port
throttling mechanism to control the amount of the exhaust gases flow out from the combustion
chamber has been designed and fabricated. A sliding plate has been fabricated to slide within the
machined slot in the cylinder block. Figure 1. illustrates the exhaust throttling mechanism and the
slot machined through the cylinder head and the cylinder block.
Slot
FIGURE 1: Illustration of the Throttling Valve Mechanism.

A very close tolerance slot was machined through the cylinder head and the cylinder block, in
which the control blade move in and out to control the opening of the exhaust port. The machining
process was carried out by Electrical Discharge Machining (EDM). The control blade was made
of stainless steel and machined with very close tolerance with the slot using CNC machine. The
control blade and its slot must have close tolerance to avoid any leakage. A spring is used to
make sure that the control blade can move up or down, to increase or decrease the exhaust port
opening. In this operation, the spring was in turn act to reduce vibration effect of the blade.
In this throttling process, the amounts of exhaust gases flow out can be reduced, hence a part of
it will be kept inside the combustion chamber. Applying this type of throttling process, the occur of
radical combustion was investigated. The engine is tested under different operating condition at
which the exhaust port throttling was set at a range of 2 to 16 mm (exhaust port area of 1 to 8%).
It was proved that radical combustion took place at an engine speed of 3000 to 4000 rpm while
the engine load was in the range of 1 to 2.5 N.m. Specification of the classical two-stroke engine
used in the test, condition of testing and the detailed results of the work were introduced [22, 23].
It was found that when the engine running under radical combustion, assures almost complete
stable combustion. In addition the engine could run under lower temperature and with minimum
emission as it is proved that contents of exhaust gases have much less CO and minimum
Nitrogen Oxides NOx. A typical results of the previous study are illustrated in figures 2 and 3. It is
clear that, lower Specific Fuel Consumption (SFC)) of the engine under radical combustion could
be achieved. This indicate better fuel economy. Figure 3 shows a good indication of relative
increase of the carbon dioxide CO2 under radical combustion which lead to tendency of complete
combustion.
FIGURE 2: Specific Fuel Consumption (SFC) For SI Combustion and Radical Combustion
(Engine Speed 4000)
FGURE 3: CO2 Emission for SI Combustion and Radical Combustion (Speed N=4000 rpm)

Although the above-mentioned method achieve radical combustion, but there are several
difficulties faced with such mechanism of the exhaust port throttling, and summarized as follows:
1. Accumulation of carbon deposits on the moving blade as it is exposed to the combustion
products. In the other hand, the high temperature reached, lead to affect the sliding
mechanism and hence influence the accuracy of the throttling control.
2. As the slot is machined within a narrow space just at a side of the liner on the cylinder
block, which make sealing is difficult as the gasket at that part is thin and might lead to
a serious leakage, which in turn affect the engine performance.
3. This throttling mechanism was relatively costly, as highly sophisticated machines was
used for the machining and producing its components.
4. Manual control of the throttling mechanism will be difficult and not reliable.
The above mentioned difficulties, made such type of throttling control was not successful.
4. A NEW STRATEGY FOR EXHAUST PORT THROTTLING:

In this paper, a new strategy is followed to overcome the above difficulties. The same classical
two stroke engine was used. No modification this time is carried on the cylinder head or the
cylinder block. The throttling control is placed, just at the point where, the exhaust manifold is
connected to the cylinder block. Figure 4: shows clearly the position where to place the
throttling control of a partially closed plate. The throttling process, is carried by means of a
partially closed plate, at the exit of the exhaust port, to restrict and control of the amount of
exhaust gases come out from the combustion chamber.
FIGURE 4: A Partially Closed Plate, To Be Fitted To Control The Flow

Of The Exhaust Gases Come Out Of The Engine.
In this method of the exhaust port throttling, assuring the simplicity of control of the flow of
exhaust gases. Thus displacing the gate of the throttling at the outer side of the cylinder block
overcoming the drawback of the earlier work mechanism.
In this case the throttling process was carried out using three different trial plates of different
opening size, one-sixth closed, one-third closed and half open, to throttle the exhaust gases
leaving the combustion chamber. The configuration of those plates is illustrated in figure 5. The

plates was designed and fabricated to fit easily at the exit of the exhaust port where the neck of
the exhaust pipe (exhaust manifold) is to be fitted.
FIGURE 5. Different Trial Plates of Different Opening.
4.1 Engine Testing

A single cylinder, 100 cc direct scavenging, air cooled two-stroke SI engine was employed in the
present investigation. This engine was coupled to water dynamometer for torque measurements.
A computer exhaust gas analyzer was used to measure the unburned hydrocarbon (UHC),
Nitrogen oxides (NOx) as a part per million (ppm), carbon monoxide (CO), carbon dioxide (CO2)
and Oxygen (O2) as a percentage emission levels in the engine exhaust. Calibrated standard
instrumentation was used for air and fuel flow rates, and exhaust temperature measurement.
Experiments were conducted with ignition system using a normal spark plug, without any
restriction on the exhaust gases, hence the engine was run as a normal spark ignition
combustion. The engine was tested under different engine speed, ranges from 2000 rpm to
4000 rpm.
4.2 Engine Run to Achieve Radical Combustion:

In this stage the engine was run to achieve radical combustion, using the new strategy of control
illustrated above. The partially closed plates are fitted alternatively to investigate the
achievement of the radical combustion. The plate will act to restrict the flow of exhaust gases
come out of the engine, thus lead to trap apart of the exhaust in the combustion space to the next
cycle. The exhaust port opening area is 560 mm2, if the plate is fully open. So with fitting of the
different opening plates, the exhaust gases flow could be controlled. Therefore, the trap of a part
of exhaust gases rich with free radicals, might lead to the achievement of radical combustion.
5. RESULTS AND DISCUSSION

The main goal of this paper is to investigate the criticality of radical combustion and its
achievement using a simple method to throttle the engine exhaust gases. Different partially
closed plates were used in the experiment, to trap a part of the exhaust gases likely rich with free
radicals to the next cycle. Once, the new charge introduced to the engine combustion chamber
will be mixed with the trapped exhaust gases carry radical, lead to achieve radical combustion.
The engine test results was recorded at speed range from 2000 to 4000 rpm, with spark ignition
combustion (SI combustion, normal run). For the same range of speed, the engine results, for the
condition with the exhaust throttle also recorded. The engine different parameters as UHC, SFC,
NOx, CO2 and others are traced at different operating conditions under varying engine load.
Results reveal the difficulty to achieve radical combustion, except with presence of plate one-
sixth closed and at a higher speed. The other trial plates (one-third closed and half open), made
the engine to run with less stable running condition, cause more interrupted running of the
engine. More restriction of the exhaust gases will act to dilute the charge and lead to retard the
combustion.
However, with the one-sixth closed trial plate, after several runs it was found that radical
combustion could be achieved, under specific operating conditions. A sample of some of the

results are introduced to indicate the difference of the condition of normal run (SI combustion)
and radical combustion.
Figure 6; presents the measured unburned hydrocarbon UHC emission versus engine load at
4000 rpm. Restriction and control of the exhaust port opening acts to achieve the radical
combustion. Therefore, more stable combustion consumes most of the hydrocarbons of the fuel
in an efficient way. This clearly indicate the reduction of the UHC emission levels. A noticeable
decrease in the level of UHC ensure better combustion compared to the case of normal run (SI
combustion) of a classical two stroke engine where UHC emission levels is normally higher. It is
found that, there is a good agreement of this result with that obtain in an exhaust gas recycle
(EGR), in which, slow down the speed of chemical reactions. This is favorable at higher speed,
for two stroke engine as it give more chance for complete combustion, thus reduction in fuel
consumption and exhaust emissions [24, 25].
FGURE 6: UHC Emission for SI Combustion and Radical Combustion (Speed N=4000 rpm)
Figure 7; indicate the clear difference of the temperatures of the exhaust gases at normal running
condition (spark ignition SI combustion) and radical combustion running condition. It is observed
that when engine is running on radical combustion the exhaust gases temperature is lower over
the entire load range. The reduction in the exhaust temperature is due to the trap of exhaust
gases in the combustion space by restriction of their flow and mixed with new charge, acting to
dilute the mixture. This might deplete the reaction and lead to noticeable decrease of the exhaust
gas temperature under radical combustion. In addition, the radical combustion is sensitive to the
condition of the exhaust trapped, in particular its temperature, and pressure [26].
FGURE 7: Exhaust Temperature for SI Combustion and Radical Combustion (Speed N=4000 rpm)

The O2 emission under SI ignition and radical combustion is investigated. Figure 8; shows the
clear reduction of Oxygen percentage compared to that under SI combustion at 4000 rpm, in
particular under moderate load. At higher engine speed the oxygen reacted with the hydrocarbon
to form CO. With the restriction of the exhaust flow, the residence time is increases, which will
further convert the CO to CO2 and to the oxides of nitrogen, leading consume more oxygen. It is
an evident that under radical combustion where auto-ignition is predominant more stable
combustion is recorded with less exhaust emissions [27, 28]
FGURE 8: O2 Emission for SI Combustion and Radical Combustion (Speed N=4000 rpm)
Investigating of the engine running, incorporated with this new strategy of exhaust port throttling
(throttling control at the exit of the exhaust port) leads to several interesting results about the
criticality and conditions at which radial combustion is occurring.
1. The throttling controller should be placed adjacent to the exit from the combustion
chamber. Displacement of the location of the controller as conducted in this run of
experiment, made it hardly to achieve the radical combustion. It is believed that
displacement of the throttling control will sweep some of the radical agents to the extra
volume formed by the passage (between the port exit and manifold neck). Therefore
considerable amount of those active radical will leave the combustion chamber hence
reduce the activity of the mixture to achieve radical combustion.
2. Partial trapping of the exhaust gases should be carried through throttling of the exhaust
port with a limit not to exceed 15 % of the exit area. Increase of the amount exhaust
gases trapped lead to dilute the mixture inside the combustion chamber and hence affect
the rate of combustion.
3. As the piston descends and the exhaust port starts to open, most of the unburned
hydrocarbon (UBH) leave with first lot hence, presence of this extra volume due
displacing of the controller will contribute in trapping less active exhaust gases.
4. As the amount of the exhaust gases trapped increase, it was found that radical
combustion is taking place at relatively higher speed and load. Therefore, the volume
and state of the trapped exhaust gases are very critical to achieve radical conditions.
5. Results reveal the importance of the shape and direction of opening of the exhaust port.
Occurring of the radical combustion is strongly affected by the condition inside the
combustion chamber, and the amount of exhaust gases trapped and their content of
active radical.

However, it is more interesting, that throttling of the exhaust for two stroke engines have clear
effect on the quality and stability of the combustion and hence its direct effect on the fuel
consumption and the rate of pollutants expelled to the environment.
6. REFERENCES
1. Alkaff S. A, Ahmed Suhimi and, Hakim A Abbass.” Radical combustion. New concept for two
stroke (SI) engines” . ASEAN Journal On Scientific and Technology for Development
Singapore. 17 (2) :91-99, 2000.
2. V.Ya. Basevich. Chemical kinetics in the combustion processes: A detailed kinetics

mechanism and its implementation . Progress in Energy and Combustion Science, Volume
13,Issue 3, Pages 199-248, 1987.
3. G.P. Huffman, F.E. Huggins, Naresh Shah and Anup Shah. Behavior of basic elements
during coal combustion . Progress in Energy and Combustion Science, Volume 16, Issue 4,
Pages 243-251, 1990.
4. John M. Simmie . Detailed chemical kinetic models for the combustion of hydrocarbon fuels .
Progress in Energy and Combustion Science, Volume 29, Issue 6,Pages 599-634; 2003.
5. B. Lewis and G. Von Elbe , Academic Press,

New York, (1951).
6. John B. Heywood. . Mc-Graw hill Publishing

Company. (1988).
7. Noorfidza Yub Harun; M.T. Afzal and Noorliza Shamsudin.

. International Journal of Engineering (IJE), Volume (3) : Issue(4). P
413-426. November 2009
8. Stone Richard. Mac Millan Publisher Ltd.; London

(1985).
9. Kito, S., Wakai, K., Takahashi, S. and Komori, K. "Ignition Limit of Lean Mixture by Hydrogen
Flame Jet Ignition." Proceedings of 15th Internal Combustion Engine Symposium; in Seoul,
pp.233-238(1999).
10.Takeharu Hasegawa .
. Energies Journal , Volume 3, Issue 3. 335-449; 2010.
11.Semin ; Abdul Rahim Ismail and Rosli Abu Bakar. Combustion Temperature Effect of Diesel
American Journal of Engineering and
Applied Sciences . Vol: 2 Issue: 1; P 212-216;.2009.
12. f combustion parameters and

their relation to operating variables and exhaust emissions in an upgraded multivalve bi-fuel
-01-0983;2004.
13.E. I. Zabryanskii1 and V. P. s and

Chemistry and Technology of Fuels and Oils . Springer New York. P. Volume
8, Number 2;. P. 138-140;1972.

14. -octane oxidation for HCCI

18) 2593 2604;2006.
15.Wakai, K., Shimizu, S., Fuma, H., Hirata T. and Taniguchi, H. "Study on the combustion in a
Divided Chamber (Ignition Mechanism of Main Chamber Mixture)." Transaction of JSME
(Section B) , 51-465, pp.1600-1609;1985.
16.Kuo, Kenneth K. . John Wiley & Sons, Inc., New York, (1986).
17.Spalding, D.B. Pegamon Press, New York, (1979).
18.Y. Ishibashi. -stroke engine

application an . SAE paper 2000-01-1836;2000.
19.H. Zhao, J. Li, T. Ma and N. Ladommatos, -stroke multi-

. SAE paper 2002-01-0420; March 2002
20.Q J. . Allen, D. Law. tuated controlled auto-ignition: speed load maps and
. SAE paper 2002-01-0422 ; March 2002.
21.Steven Ashley. . Mechanical Engineering Journal. P.64-67; 1996.
22.Alkaff S. A, Ahmed Suhimi and, Hakim A Abbass.

. Bureau of Research and Consultancy (BRC) Seminar. MARA University
Technology (UITM), Shah Alam, Malaysia, 1998.
23.Alkaff S. A, Ahmed Suhimi and, Hakim A Abbass. ke (SI)

. Bureau Of Research And Consultancy
Journal, UiTM Malaysia.1, 11-24,1999.
24.y. ohyama. . International Journal of Automotive

Technology, Vol. 2, No. 2, pp. 53
25.Soek Woo Kim and Yasuo Moriyoshi

Schnurle- . KSME International Journal Vol: 18; No. 8, P. 1451-1406;
2004.
26.P. Amneus, D. Nilsson, F. Mauss, M. Christensen and B. Johansson.

. The fourth International Symposium COMODIA 98; P.
567-572;1998.
27. Wijesinghe and G Hong. -ignition combustion operating region of a small

two- . Proc. IMechE Vol. 223 Part D: J. Automobile Engineering; P 651.659
2009.
28.Shunsuk Hashimoto, Yasunori Amino, Koji Yoshida, Hideo Shoji and Atsushi Saima.
Analysis of OH Radical Emission Intensity during Auto ignition in a 2- . The
fourth International Symposium COMODIA 98. P 405-410;1998.

Dr. Walid Emar, Eng. Ziad Sobih, Dr. Musbah Aqel & Dr. Mahmoud Awad
Bi-level Control and Chopper Control Methods for Improving the

Dynamic Performance of Stepper Motor
Walid Emar [email protected]

Faculty of Engineering/Electrical Dept./Power
electronics and Control
Isra University
Amman, 11622, Jordan
Ziad Sobih [email protected]

Faculty of Engineering/Electrical Dept
Northeastern University
Boston, MA02115, USA
Musbah Aqel [email protected]

Applied Science University
Amman, Jordan
Mahmoud S. Awad [email protected]

Al-Balqa Applied University
Amman, Jordan
Abstract
This paper compares between chopper control method and bi-level control
method. Both methods are used for improving the dynamic performance of
variable reluctance stepper motor (VRSM) by modifying its time constant and
thus, increasing its stepping rate. Therefore, the initial torque developed by the
motor is high; the switching from one coil to the next is faster than normal and
consequently, the rotor moves as quickly as it should be. The circuitry discussed
in this paper is connected directly to the motor windings and the motor power
supply, and this circuitry is controlled by a digital system that determines when
the switches are turned on or off. Each class of drive circuit is illustrated with
practical examples, but these examples are not intended as an exhaustive
catalog of the commercially available control circuits, nor is the information given
here intended to substitute for the information found on the manufacturer's
component data sheets for the parts mentioned.
Keywords: Stepper motor, variable reluctance motor, time constant, new topology of chopper converter,
chopper control.

1. INTRODUCTION
The dynamic response of stepper motor may be improved by using special drive and control
circuits yielding better time constant, faster stepping rate and therefore higher torque and well
working rotor [5].
For most industrial applications stepping motors are controlled by using microprocessor
techniques. With the microprocessor any of the types of control can be used such as linear
constant current control, bi-level control or chopper PWM control. It can make the motor to run
through the desired number of steps and can control the acceleration and deceleration of the
motor when required[2-5].
In this paper a bi-level and chopper PWM control will be demonstrated. At the end, a brief
comparison is done to show the priority of each control method and drive circuitry.
2. FUNDAMENTAL PRINCIPLE OF CHOPPER CONTROL OF STEPPER

MOTOR
The typical stepper motor with four windings with the drive waveforms are shown in the schematic
diagram in Figure 1a, with one terminal common to all windings; it is most likely a variable
reluctance stepping motor.
In use, the common wire typically goes to the positive supply and the windings are energized in
sequence [12].
(a)
R L 25 Signa l for T R1 Pulse1. VAL
20
10 30m
15
D1
A ia 10
E1
5
0
0.12 0.13 0. 13 0.14 0. 14 0.15 0. 15 0.16 0.16 t [ s]
Pu l se1
TR1
2 ia. I [A]
1.8
1.5
1.3
1 Coil
0.75 current
0.5 ia
0.25
0
0.12 0.13 0.13 0. 14 0. 14 0.15 0.15 0. 16 0. 16 t [ s]
(b)
Figure 1: (a) Four phase (4-Φ) variable reluctance stepper motor. (b) Fundamental
connection of a chopper down converter and phase current waveforms:
( E1  5  V, L  30mH, R  10Ω ).
Figure 1b shows the equivalent circuit of one phase of the motor and a fundamental chopper
down (step down) converter and the waveforms of the motor phase currents.

The basic principle of operation for a four winding variable reluctance stepper motor is illustrated
in Figure 1.
The stator has 8 poles and the rotor has 6 teeth. When one of the stator coils is energized, the
rotor teeth will align with the energized stator poles. This means, that the rotor will move to a
position of minimum reluctance. Sequentially switching the stator phases produces a rotating
magnetic field which the rotor follows. However, due to the lesser number of rotor poles, the rotor
moves less than the stator angle for each step [1, 3, 6].
For a variable reluctance stepper motor, the step angle is given by:
θ s  360  / N S (1)
Where: θs is stator angle; NS is number of stator poles.
Torq ue
0. 2 A M 1. I [ A]
- 0. 2
1 1 1. 1 1. 1 1. 1 1. 1 1. 2 1. 2 1. 2 t [ s]
Speed
0. 5 vm_um. V [V ]
-1
1 1 1. 1 1. 1 1. 1 1. 1 1. 2 1. 2 1. 2 t [ s]
Ph a s e cu rre n ts
2 io . I [ A]
i1 . I [ A] + 0. 5
i2 . I [ A] + 1
i3 . I [ A] + 1. 5
0
1 1 1 1. 1 1. 1 1. 1 t [ s ]
C o n tro l s ig n a l s
10 S0 + 2
S1
S2 + 4
S3 + 6
0
1 1 1 1. 1 1. 1 1. 1 t [s]
Figure 2: Schematic diagram showing torque, speed and phase current waveforms
of a four phase variable reluctance stepper motor.
θ R  360  / N R (2)
Where: θR is rotor angle; NR is number of rotor poles.

θ ST  θ R  θ S
NS  NR (3)
  360 
NS .NR
Where: θST is step angle. The number of stator poles is the product of number of phases and
number of poles per phase. Thus
N S  m N P (4)
Where: m is number of poles; Np is number of stator poles per phase.
Figure 1 shows that moving from Φ1 to Φ2, etc., the stator magnetic field rotates clockwise. By
reversing the sequence of pulses, the direction of rotation is reversed above right. The direction,
step rate, and number of steps are controlled by a stepper motor controller feeding a driver or
amplifier. This could be combined into a single circuit board [1, 2, 10].
Figure 2 shows the resultant torque, speed and current waveforms of a 4-Φ variable reluctance
stepper motor.
3. IMPROVEMENT METHODS OF DYNAMIC RESPONSE OF STEPPER

MOTOR
It has been mentioned earlier that the basic problem for the drive of a VRSM lies in the
inductance of the stator winding. The time constant (   L R ) of the motor winding prevents the
current to follow the winding voltage pulse. The current rises slowly and does not reach the full
rated value, particularly at high speed. As a result the torque decreases with increase of pulse
rate. Hence, the torque speed performance can be improved by using one of the following drive
methods:
 Bi-level control method.
 Chopper control method
3.1 Bi-level control method
A good approach to improve the dynamic response of stepper motor is to use two power supplies
instead of one. This is called bi-level control as shown in Figure 3. Bi-level drive enables us to
obtain fast rise and fall times of current without using external resistors[5].

30
Control signal for TR1 Pulse1.VAL
20
-10 Pulse1
0.12 0.13 0.13 0.14 0.14 0.15 0.15 0.16 t [s] TR1
Phase Current of VRSM

0.5 ia.I [A]
0.4 L 30m
D2
0.3 D1
0.2 R 10
E1 30
0.1
0
0.12 0.13 0.13 0.14 0.14 0.15 0.15 0.16 t [s] A ia
E2 5
30
Control signal for TR2 Pulse2.VAL
Pulse2
20
TR2
0 Rs 0.00001
-10
0.12 0.13 0.13 0.14 0.14 0.15 0.15 0.16 t [s]
Figure 3: Bi-level control scheme of VRSM drive with current waveforms.
The principle of a bi-level control can be explained as follows: In this system there are two power
supplies E1, E2. The upper supply E1 is about 5 to 10 times higher than E2 which is the rated value
of the motor supply. The current sensing resistor Rs is not considered in this case ( R s  0 ).
At the starting of each step E1 is connected to the motor coil. The coil current rises much faster
than the rated supply and as soon as the coil current reaches the rated value; the supply is
changed from E1 to E2. This continues with TR1 off and TR2 on upto the end of one step. When it
is required to end the pulse TR2 is turned off and the current decays through the load, both
freewheeling diodes D1, D2 and both supply voltages E1, E2. In this way losses in the circuit are
reduced and the overall efficiency is increased.
3.2 Chopper control method
Chopper control may be achieved by using different types of chopper converters. Not all of these
converters help to improve the performance of stepper motors. This paper introduces a new
topology of chopper converters which significantly improves the dynamic performance of stepper
motors.

50 Signal1.VAL
Pu l se1 20
0
Signal for TR1
Pu l se11
Pul se12 -20

TR 1
C1 A ia
-50
D1
0.12 0.12 0.12 0.13 0.13 0.13 t [s]
+ R
E1
va V
50 Signal2.VAL
D2
20
C2 L
TR 2
Signal for TR2
0
Pul se2 -20
Pul se21
-50
0.12 0.12 0.12 0.13 0.13 0.13 t [s]
Pul se22
11 C1.V [V] 0.6 ia.I [A]

C2.V [V]
0.5
10
0.4
Capacitor
Coil current ia
10 voltages 0.3
vc1, vc2 0.2

9.8
0.1
9.5 0
0.12 0.12 0.12 0.12 0.12 0.12 0.13 0.13 0.13 0.13 0.13 0.13 t [s] 0.12 0.120.130.130.130.130.130.140.140.140.140.140.150.150.150.150.150.16 0.16 t [s]
Figure 4: schematic diagram of a series connected double phase chopper down converter
with Current and voltage waveforms.
The system is described with the help of Figure 4. The circuit in this figure consists of a chopper-
down converter with two channels connected in series. In this scheme higher supply voltage E1 is
used. It may be as high as 2 to 4 times of the rated value of the motor voltage. This connection
is also used to increase the voltage-handling capability of the power devices. It also gives with a
variable output voltage the possibility of varying the motor speed by varying its terminal voltage.
Principle of operation of such connection may be explained as follows: The motor is switched on
by the sequencer signal with this high voltage supply and the two capacitors (C1, C2). At the
beginning both switches are on for a time required for the circuit current ia to reach its rated value.
Thus, if the supply voltage had a value of 30-V and if the switches were on all the time, the
resulting current in the circuit would be 30V 10 Ω  3 A . This is much greater than the required
rated current (0.5A). The rate of rise of coil current increases and reaches its rated current much
faster. The time constant of the circuit is again L R  30 / 10  3ms . Thus, the current in the coil
rises at a rate of 30 / 3ms  1000 A/s . The time to reach 0.5A is then 0.5(A)/ 1000(A/s)  0.5ms .
As soon as the current reaches slightly above the rated value, both switches are periodically
turned off and on using pulse width modulation (PWM) technique. The switches connected in
series are usually operating out of phase by the time T/2. During this period each switch is on for
certain time ton to yield the required mean value of the coil voltage and off for the rest of this
operating period, toff. Both transistors have the same operating period Tp which is many times
smaller than the energizing pulse (period) of the coil. When it is required to end the energizing
pulse of the coil, both transistors are simultaneously turned off and the circuit current ia begins
to decrease through diodes D1, D2 as shown in Figure 4.

4. STEPPER MOTOR FOR BOTH DIRECTIONS

The following mechanical model in Figure 5 obtained from Simplorer 6 shows a 4-phase step
motor. Whose electrical part is modeled by 4 electrical circuits in parallel connection.
Troque
Pos ition
A
RT1 CT1
I_CTR_1 I_CTR_2 I_CTR_3 I_CTR_4 +
V IU1
Cw 1
vm_um
(-(E_CTR_0.I) * sin( (p * Cw.V) ) ) EQU

(-(E_CTR_1.I) * sin( (p * (Cw.V - alpha)) ) )
EQU
(-(E_CTR_2.I) * sin( (p * (Cw.V - (2 * alpha))) ) )
(-(E_CTR_3.I) * sin( (p * (Cw.V - (3 * alpha))) ) )
Figure 5: Mechanical part of four phase stepper motor.
There are different ways to model the mechanical part of the motor in Simplorer. Here it is used
according to Simplorer 6 an electrical circuit to describe its mechanical behavior. It can be
modeled as block diagram as well.
The control signal of the step motor is produced by 3 state machines. One for direction control,
e.g. the required step length and moving direction. The others for generating pulse signals, which
depends on the moving direction as shown in Figure 6.
By using static transistor model in Simperlor, the logic control signal can be connected directly to
the transistor as a switch signal. A static transistor model is good enough to simulate the
controlled behavior of the step motor in this case as shown in figure 7. Figure 7 also shows the
waveform of the speed, torque and position of both-directional stepper motor.

SET: := S1:=1 SET: := S2:=1 SET: := S3:=1

SET: := z_left:=z_left + 1
S0 S3
S1 S2 S4
control signal generator 'forward'

SET: := an_left:=0
S5
S_reset_l
SET: := S3:=1
SET: := z_right:=z_right + 1 SET: := S0:=1
SET: := zeropoint:=t SET: := S2:=1 SET: := S1:=1 SET: := zeropoint:=t
S10 S11 S12 S13 S14
control signal generator 'backward'

SET: := an_right:=0
S15
S_reset_r
control signal generator 'direction'

SET: := an_right:=1
ok ((t - wait) >= 0.8)SET: := n_right:=4 ok

SET: := wait:=t SET: := ok:=0
SET: := an_right:=0
SA0 SET: := an_left:=0

SET: := an_left:=1 SA1 SA2
SA3
SET: := n_left:=4
Figure 6: State graph.

Control s ignal
10 S0
15 33m S1 + 2
D1 S2 + 4
E_CTR_0 S3 + 6
5
(-(vm_um.V) * sin( (p * Cw.V) ) )
15
R2
EQU TR1
0
0 0. 5 1 1. 5 2 2.5 3 t [s]
R3 L2
Torque
15 33m
0. 2 AM1. I [ A]
D2 E_CTR_1
0
R4 15
EQU
TR2
-0. 2
R5 L3 0 0. 5 1 1. 5 2 2. 5 3 t [s]
15 33m Speed
D3 0. 5 v m _um . V [V]
E_CTR_2
R6 15
EQU
TR3
-1
L4 0 0.5 1 1. 5 2 2. 5 3 t [s]
R7
15 33m Pos ition

0. 2 C w. V [ V]
D4
E1
E_CTR_3
R8 15
-0. 25
EQU
TR4
-0. 4
0 0. 5 1 1. 5 2 2. 5 3 t [s]
Figure 7: Electrical model and output waveforms.

5. CONCLUSION
The above described application examples of modern control and drive circuits show that
performance and efficiency of variable reluctance stepper motors may be remarkably increased
without any excessive expense increase like before.
A natural limit against any current increase by using very high power supply as in the bi-level
control method is the danger of saturating the iron core and increasing the maximum temperature
rise of the motor, due to the power loss in the stator windings. This shows one advantage of the
modified chopper control method defined in this paper, which, compared to others, has the
motor’s power loss within a reasonable limit since the current in the windings is controlled.
The winding current is chopped and limited within a certain limit and this produces a direct
proportional and positive effect on the torque. At their power loss limit stepper motors with anti-
parallel series four phase chopper control may deliver more torque than stepper motor with other
drive circuits.
Furthermore, if a higher torque is not required, one may either reduce the motor size or the power
loss by utilizing the chopper control method. It gives only as much voltage as needed and makes
the motor to run at as high speed as required and independently from the winding resistance.
On the other hand, the bi-level control method requires higher power supply than other methods
which increases the economical costs of the system and the power supply is more expensive
because it has to deliver 5 times as much power (5E1 instead of E1). Furthermore, the current is
not limited or controlled anyhow and therefore, the magnetic field is not controlled. Consequently,
the torque of the motor is out of control since it is proportional to the current and the magnetic
field of the motor’s windings.
Dedicated integrated circuits have dramatically simplified stepper motor driving. To apply these
ICs designers need little specific knowledge of motor driving techniques, but an under-standing of
the basics will help in finding the best solution.
6. REFERENCES
1. E. Walid, TT. Issam, A. Rateb. “A novel topology of delta modulation technique for
improving the power factor of AC-DC converters”. International journal of engineering
(IJE), CSC journals, Volume 4, Issue 1, January/February 2010.
2. M. F. Rahman, A. N. Poo and C. S. Chang. “Approaches to Design of Ministepping Step

Motor Controllers and Their Accuracy Considerations”. IEEE Trans. Industrial Electronics,
vol. IE-32, No. 3, pp. 229-233, August 1985.
3. M. F. Rahman and A. N. Poo. “An Application Oriented Test Procedure for Designing
Microstepping Step Motor Controllers”. IEEE Trans. Industrial Electronics, vol. 35, No. 4,
pp. 542-546, November 1988.
4. P. C. Sen. “Modern Power Electronics”. Published by S. Chand & Company LTD. New
Delhi, 2004.
5. W. Theodore. “Electrical machines, drives, and power systems”, sixth edition. Published
by Pearson education international, New Jersey 07458, 2006.
6. W. D. Harries and J. H. Lang. “A simple motion estimation for variable reluctance motors,
IEEE Trans. Ind. Appl., pp 237-243, March 1990.

7. P. P. Acarnley. “Stepping Motors: A Guide to Modern Theory and Practice“. IEE Control
Engineering Series 19, Peter Peregrin Ltd., 1982.
8. M. F. Rahman and A. N. Poo. “An Application Oriented Test Procedure for Designing
Microstepping Step Motor Controllers”. IEEE Trans. Industrial Electronics, vol. 35, No. 4,
pp. 542-546, November 1988.
9. D. Carrica, M. A. Funes and S. A. Gonzalea. “Novel Stepper Motor Controller Based on

FPGA Hardware Implementation”. IEEE/ASME Trans. Mechatronics, vol. 8, No. 1,
pp.120-124, March 2003.
10. P. D. Ziogas, L. Morán, G. Joos, and D. Vincenti. "A refined PWM scheme for voltage
and current source converters". IEEE-IAS Annual Meeting, 1990, PP. 997-983.
11. R. Wu, S. B. Dewan, and G. R. Slemon. "Analysis of an AC-to-DC voltage source

converter using PWM with phase and amplitude control". IEEE Transactions on industry
Applications, vol. 27, No. 2, March/April 1991, pp. 355-364.
12. X. Ruan, L. Zhou, and Y. Yan. "Soft-switching PWM three-level converters". IEEE
Transactions on Power Electronics, Vol. 6, No. 5, September 2001, pp. 612-622.
13. M. H. Rashid. “Power Electronics, Circuits, Devices and applications”. Third edition,
2004, Pearson Education, Prentice Hall, Upper Saddle River, NJ 07458.
14. Miftakhutdinov, R. “An Analytical Comparison of Alternative Control Techniques for

Powering Next Generation Microprocessors”. TI Seminar, 2002.
15. G. Moschopoulos, P. Jain. "Single-phase single-stage power-factor-corrected converter

topologies". IEEE Trans. on Industrial Electronics, vol. 52, no. 1, pp. 23- 35, Feb 2005.
1

Mona Arabshahi, Ali Asghar Mirghasemi & Ali Reza Majidi
Three Dimensional Bearing Capacity of Shallow Foundations

Adjacent to Slopes Using Discrete Element Method
Mona Arabshahi [email protected]

Graduate Student of Soil Mechanics and Foundation Engineering
School of Civil Eng., College of Engineering, University of Tehran
Ali Asghar Mirghasemi [email protected]

Associate Professor
School of Civil Eng., College of Engineering
University of Tehran
Ali Reza Majidi [email protected]

Ph.D. of Soil Mechanics and Foundation Engineering
Mahab Ghodss Consulting Engineers
ABSTRACT
In the current study, an effort is made to determine three dimensional bearing capacity
of a shallow foundation adjacent to a slope using discrete element method. The soil
mass under the footing is modeled as discrete blocks connected with a set of normal
and shear springs called Winkler springs. In order to define the geometry of the failure
surface of the soil, six independent angles are considered, which will be obtained by
trial and error. So, the failure surface is not fixed and can be changed by different
characters.
The purpose of the paper is to determine the three dimensional bearing capacity
coefficients of rectangular foundations which are placed adjacent to a slope. In order to
get to such a scope a discrete element program called BCAP3D (Bearing capacity
analysis program in 3D) is employed. The result is presented in relative graphs, versus
the ratio ( ), for each foundation aspect ratio ( ). Besides, a comparison is made
between the current method with other methods used in the same field.
Keywords: 3 Dimensional, Bearing Capacity, Shallow Foundations/Footings, Discrete Element Method.
1. INTRODUCTION
1-1 Field of The Research
Bearing capacity of foundations has always been one of the most interesting research
subjects in geotechnical engineering. In such a field of study, extensive efforts have
been made for bearing capacity in two dimensions. It seems that 2D theoretical
approaches have reached to a relatively satisfactory level for ordinary loading and soil
conditions. Considering that real foundations are not infinitely long and their failure
mechanism is certainly three dimensional, it is necessary to study the real conditions
and try to develop analytical 3D estimations of the bearing capacity.

Among the researches in 3D bearing capacity, a few of them consider the effect of the
ground inclination or a nearby slope in reducing the ultimate load of the footing, which is
the main subject of the current paper.
1-2 Previous Studies
Evaluation of the 3D bearing capacity of shallow foundations is usually assessed by
introducing experimental and empirical shape factors into the ordinary 2D equations for
the strip footings developed by researchers such as Meyerhof [1], Terzaghi and Peck
[2], Hansen [3], de Beer [4], Vesic [5], and the others.
The general bearing capacity relationship suggested by these researches can be
expressed as:
(1)
Where Nc, Nq and N are the bearing capacity coefficients in two dimensional state. For
considering the effect of the lateral surfaces in three dimensional conditions, the
experimental related shape factors will be used.
In order to employ analytical methods in such a research, Shield and Drucker [6]
presented a theoretical evaluation of 3D bearing capacity of rectangular foundations on
homogeneous clay (φ = 0) by means of upper and lower bound solutions. Nakase [7]
used a limit equilibrium method and assumed cylindrical sliding surfaces for rectangular
footings on normally consolidated clays. Narita and Yamaguchi [8], presented a three
dimensional analysis of bearing capacity of rectangular foundations by means of the
method of slices, assuming that sliding surfaces are composed of a set of log-spirals
with different initial radii. There are plenty of other researches in different conditions
such as works of Ugai [9], Michalowski [10], Michalowski and Dawson [11], Zhu and
Michalowski [12] and Salgado et al. [13].
Askari and farzaneh [14] used an upper bound method to determine the bearing
capacity of strip foundations near slopes. Sarma and Chen [15] also studied the same
problem during earthquake using a limit equilibrium technique.
1-3 History of The Discrete Element Method
Cundall and Strack [16] established a Discrete Element Method to study the
micromechanical behaviour of granular materials by modelling assemblies of two
dimensional circular particles. In this method each particle is considered as a distinct
(discrete) element.
The new concept of DEM presented here, falls within the framework of the limit
equilibrium methodology. In two dimensional state, this method was presented by
Chang for analysis of bearing capacity of foundations [17], slope stability [18] and
retaining walls [19]. Using this method, soil mass is modelled as a system of blocks
connected together by elasto-plastic Winkler springs. So in the current method, every
block is considered as a discrete element.
The former state of DEM is an explicit method and the latter one is an implicit method.
1-4 Using DEM for 3D bearing capacity
Majidi and Mirghasemi [20] employed the DEM to study the 3D bearing capacity of
rectangular foundations. The same method is used here to obtain the 3D bearing
capacity of footings adjacent to slopes.

2. DISCRETE ELEMENT MODEL
The method used in order to achieve the 3D bearing capacity coefficients of a

foundation is briefly introduced in here.
2-1 General Information and main equations
To determine the three dimensional bearing capacity of rectangular shallow foundations
by DEM, it is assumed that the soil mass beneath the footing consists of several
discrete blocks connected with an infinite number of Winkler springs, as shown in Fig. 1.
As the load applied on the footing increases, these blocks are going to slip in order to
define the failure mechanism.
Each group of Winkler springs consists of three sets of springs in different orthogonal
directions (Fig.2). One set of springs is located in the direction normal to the contact
surface to simulate the normal stiffness of the soil (E) and the two other sets are placed
within the contact surface, perpendicular to each other, to behave as shear stiffness (G)
of the soil.
The behaviour of the normal and shear springs is assumed to be Elasto-Plastic. As
shown in Fig. 3, the normal springs do not yield in compression. However, in tension
they would yield at the tensile capacity.
Also, based on Mohr-Coulomb failure criteria, the shear springs yield when the shear
strength ( ) is reached, as:
(2)
The initial values of stiffness in the normal and shear directions between blocks can be
estimated using Young’s modulus (E) and shear modulus (G), respectively [18].
According to the soil behaviour, no tension will appear in normal spring and just shear
springs will result in failure of the soil when they reach their shear resistance. Whenever
the force in a shear spring reaches its ultimate value, the initial shear stiffness (kshear)
will be substituted by the reduced secant shear stiffness (k'shear). (Fig. 3-b)
The present method can be used to model progressive failure of a sliding soil mass. As
the applied load increases, the induced stresses in springs may exceed the allowable
stresses. When the shear stress is beyond the admissible stress at an interface, the
local factor of safety is set to be 1 for the interface and the iteration process
redistributes the excessive amount of stress to the neighbouring blocks. The iterative
procedure is carried out until the stresses at all interfaces of blocks are compatible with
their deformations and completely satisfy the stress-displacement relationships.
In discrete element method, the constraint information is given on the centers of blocks.
So, it is necessary to determine the relative displacement of two adjacent blocks
according to the center of their interface. Due to the relative displacement between two
neighbouring blocks, the springs are deformed and their stresses can be calculated. By
integrating on each surface of the blocks, the relative force will be obtained. The forces
acting on all sides of a block should satisfy the force and displacement equilibrium
requirement. The relationship between the forces and the displacements for all blocks
can be written as:
(3)
Where [K] is the global stiffness of the system and the vectors {f} and {U} consist of
body forces and displacements for all blocks, respectively.

Solving the equation (3), normal and shear forces between blocks will be obtained and
so, the overall safety factor of the system can be defined through equation (4).
(4)
Where is the shear strength of the soil and and are the existing shear stresses
on the failure surface.
2-2 The Geometry of The Failure Surface
The failure mass of the soil below the footing consists of three zones similar to the
classical 2D bearing capacity failure surface (Mirghasemi and Majidi, [21]). Each of
them can be divided into several wedges as shown in Fig. 4.
The failure mechanism contains of an active zone below the footing (zone I), which is
pushed downward into the soil mass and a passive wedge (zone III) moves laterally.
The transition between downward movement of the active zone and lateral movement
of the passive zone takes place through the radial shear zone (II). The shape of the
failure surface of zone (II) is assumed to be a logarithmic spiral. The shape of the failure
surface is a function of the footing width (B) and length (L), the internal friction angle of
the underlying soil (φ) and the six independent angles of α1, α2, α3, α4, θ1 and θ2.
The angles θ1 and θ2 as shown in the Fig. 4-b, determine the inclination of lateral failure
surfaces in the three dimensional space. The absolute values of θ1 and θ2 are assumed
to be identical ( ) due to the symmetry in the foundation geometry and loading.
These six angles are not predefined and will be obtained by iteration in order to get the
most critical failure surface. So, a large number of failure surfaces would be examined
to determine the ultimate bearing capacity of the footing.
2-3 General Assumptions
In order to determine the ultimate bearing capacity of a shallow foundation adjacent to a
slope by DEM, it is necessary to clarify the assumptions used in the analyses. They can
be mentioned as:
1. The blocks are rigid and just relative displacement of adjacent blocks will be taken into
account.
2. Every two adjacent blocks will remain in contact and no separation would be occurred.
3. The load is applied to a rectangular rigid foundation with the dimensions .
4. Loading is vertical and centric.
5. The general shear failure will occur in the soil mass.
6. The dry soil density is used.
7. There is a distance ( ) between the footing and the edge of the slope which its
inclination angle is β.
8. In order to obtain the 3D bearing capacity coefficients, the superposition method has
been taken into account.
9. The surcharge applying to the ground is defined just on one side of the foundation
where the failure mechanism would be constructed. No surcharge will be considered on
the inclined ground surface.
10. The radial shear zone (II) is assumed to be divided into 10 blocks. Also, it’s better for
zones (I) and (III) to model as a single block due to their non-shear behaviour [20].
Fig.5 shows the position of a failure mechanism in the relation to the slope inclination.

3. COMPARISON TO OTHER METHODS

3-1 Compare to Classical methods
To compare the 3D bearing capacity of shallow foundation near the slopes resulted
from DEM with other methods, the results of two classic semi-empirical methods which
presented by Hansen [3] and Vesic [5] are used. Fig. 6 to Fig. 8 shows the bearing
capacity of a footing on top of a slope of inclination (β) where: .
3-2 Compare to AASHTO LRFD Bridge Design Specification

AASHTO LRFD bridge design specifications suggests some empirical factors and
graphs to determine the bearing capacity of foundations near slopes, for cohesionless
soils and for saturated clays, separately. [22]
A comparison between the ultimate bearing capacities resulted from DEM and LRFD
specifications is made and shown in Fig. 9 to Fig. 12.
3-3 Compare to experimental results
A comparison is made between the bearing capacity resulted from DEM and an
experimental research of a square footing near a slope [23]. A footing of
is places near a slope of 25° inclination. The soil properties are:
Table 1 shows the comparison between these two methods.
4. THREE DIMENSIONAL BEARING CAPACITY COEFFICIENTS
A nearby slope will definitely decrease the bearing capacity of a rectangular foundation.
In the current study, the bearing capacity coefficients are achieved using DEM with
mentioned assumptions. The results are presented for each of the coefficients versus
the dimensionless quantity ( ), for different foundation aspect ratios.
The bearing capacity can be obtained through the equation:
(5)
In which , and are the modified 3D bearing capacity coefficients of the
rectangular foundation near a slope.
4-1 The coefficient of soil cohesion ( )

In order to have Ncm, we can obtain the bearing capacity of a weightless soil and
disregarding the surcharge on the ground. Figs. 13 to 24 indicate the factor of Ncm.
4-2 The coefficient of soil weight ( )

Nɤm can be defined as the bearing capacity of a cohesionless soil of density equals
to and disregarding the surcharge on the ground. Figs. 25 to 36 indicate the
factor of Nɤm. These Figures are provided for the situation: .
4-3 The coefficient of surcharge ( )
To obtain Nqm, the weight and cohesion of the soil should be disregarded. Figs. 37 to 40
indicate the factor of Nqm, for all the slope inclination.
The analyses indicate that Nqm is hardly a function of the slope inclination and for
various slope inclinations Nqm varies in a small range. This is because of the definition of
the surcharge in the current method.
As the surcharge on the ground is defined to the edge of the slope, for a certain ( ) the
amount of surcharge would be constant for every quantity of slope inclination. So, the
only difference is a small decrease in the failure length of the third region in the failure
mechanism, as the angle of the slope inclination increases.
4-4 Influence of The Foundation Aspect Ratio
An analysis is made to study the effect of the foundation aspect ratio on the bearing
capacity of the footing. As it is shown in Fig. 41 and 42, increasing in footing aspect
ratio ( ) results in the decrease of the bearing capacity. This is because of the fact that
as the foundation aspect ratio increases, the influence of the lateral failure surfaces on
the bearing capacity decreases. And finally, it reaches its 2D ultimate value.
5. SUMMARY AND CONCLUSIONS

In this research, analyses based on Discrete Element Method (DEM) are carried out for
determining three dimensional bearing capacity coefficients of shallow foundations near
slopes. By determining the parameters through relative sensitivity analyses, the final
analyses are made to get 3D bearing capacity coefficients. Besides, a comparison to
other methods that can determine the 3D bearing capacity of a shallow foundation near
a slope is made.
The results obtained from the present study can be summarized as follows:
1. The results of the present method are to some extent in accordance with classical semi-
empirical results proposed by earlier researchers such as Hansen and Vesic. The
present DEM will results in less bearing capacities compare to the classical methods.
2. DEM will results in higher value of bearing capacities comparing to AASHTO LRFD
specifications. However, the difference isn’t a great deal.
3. For higher values of foundation aspect ratios, the comparison between different
methods results in less differences for bearing capacities.

4. Just like other methods, the bearing capacity coefficients obtained by DEM are highly
dependent on the internal friction of the soil, especially for φ values greater than 30°.
5. For each value of the slope inclination, there is a critical value which for
the slope has no effect on the failure mechanism and bearing capacity of the
foundation. This critical distance can be found in the graphs provided for bearing
capacity coefficients.
6. The critical described above, will decrease by decreasing in the soil friction angle. It
also will increase by increasing in foundation aspect ratio.
7. As can be seen in the graphs, for slope inclination more than 30°, Nc will decrease more
slowly in comparison to inclinations less than 30°.
6. Tables and Figures
Difference Q Q Soil
(%) (DEM) (experimental) density
x/B
13.8 9.05 10.5 0.3 1.61

11.2 7.1 8.0 0.3 1.51
Table. 1: Comparison between DEM and experimental bearing capacity
Block B
Block A
Failure surface
Fig.1: Connection of the blocks with the Winkler springs
Shear spring(t)
Shear spring(s)

W
Normal spring(n)
Fig.2: A Winkler spring is consists of three perpendicular springs
Compression
K
1 K shear
1 1 K’ shear
Ft
Tension (a) (b)
Fig.3: The behaviour of normal (a) and shear (b) springs
(a)
(b)
Fig.4: The failure surface below the footing (a) 2D view (b) 3D view

B
L
β
I
II
III
Fig.5: The position of the footing and the slope
Fig. 6: Comparison of the bearing capacity resulted from DEM and Classic methods ( )

Fig. 9: Comparison of the bearing capacity of cohesionless soil resulted from DEM and AASHTO ( )
Fig. 10: Comparison of the bearing capacity of cohesionless soil resulted from DEM and AASHTO ( )

Fig. 11: Comparison of the bearing capacity of saturated clay resulted from DEM and AASHTO ( )
Fig. 12: Comparison of the bearing capacity of saturated clay resulted from DEM and AASHTO ( )

Fig. 13: Ncm for square footing Fig. 16: Ncm for ( )
Fig. 14: Ncm for square footing

Fig. 17: Ncm for ( )
Fig. 15: Ncm for square footing

Fig. 18: Ncm for ( )

Fig. 19: Ncm for ( ) Fig. 22: Ncm for ( )

Fig. 25: Nɤm for square footing Fig. 28: Nɤm for ( )
Fig. 26: Nɤm for square footing

Fig. 29: Nɤm for ( )
Fig. 27: Nɤm for square footing

Fig. 30: Nɤm for ( )

Fig. 31: N for ( ) Fig. 34: N for ( )
Fig. 32: Nɤm for ( ) Fig. 35: Nɤm for ( )
Fig. 33: Nɤm for ( ) Fig. 36: Nɤm for ( )

Fig. 37: Nqm for square footing
Fig. 38: Nqm for ( )

Fig. 41: Influence of the foundation aspect ratio on the bearing capacity ( )
Fig. 42: Influence of the foundation aspect ratio on the bearing capacity ( )

7. REFERENCES
[1] G. G. Meyerhof, “Some Recent Research on the Bearing Capacity of Foundations.” Canadian
Geotech. J., 1( 1):16-26, 1963.
[2] K. Terzaghi and R. B. Peck, “Soils Mechanics in Engineering Practice,” J.Wiley, New York. (1967)
[3] J.B. Hansen, “A Revised and Extended Formula for Bearing Capacity.” Danish Geotech. Inst. Bulletin,
No. 28, Denmark. 1970.
[4] E.E. De Beer, “Experimental determination of shape factors and the bearing capacity factors of
sands,” Géotechnique, 20( 4) :387-411, 1970.
[5] A.S. Vesic, “Analysis of Ultimate Loads of Shallow Foundations.” J. of Soil Mech. and Fndn Div.
ASCE, 99( SM1): 45-73, 1973.
[6] R.T. Shield and D.C., Drucker, “The application of limit analysis to punch-indentation problems.” J. of
Appl. Mech., ASCE: 453-460, 1953.
[7] A. Nakase, “Bearing capacity of rectangular footings on clays of strength increasing linearly with
depth.” Soils and Foundations, 21( 4): 101-108, 1981.
[8] K. Narita and H.Yamaguchi, “Three-dimensional bearing capacity analysis of foundations by use of a
method of slices.” Soils and Foundations, 32( 4): 143-155, 1992.
[9] K. Ugai, “Bearing capacity of square and rectangular footings on nonhomogeneous clays.” J. of
JSSMFE, 25( 4): 179-185, 1985.
[10] R.L. Michalowski, “Upper-bound load estimates on square and rectangular footings.” Géotechnique,
51( 9): 787-798, 2001.
[11] R.L. Michalowski, and E.M. Dawson, “Three-dimensional analysis of limit loads on Mohr-Coulomb
soil.” Fndn of Civ. and Inv. Eng., No. 1: 137-147, 2002.
[12] M. Zhu and R.L. Michalowski, “Shape Factors for Limit Loads on Square and Rectangular footings.”
J. of Geotech. and Geoenv. Eng. ASCE, 131, No.( 2): 223-231, 2005.
[13] R. Salgado, A.V Lyamin, S.W. Sloan, and H.S. Yu, “Two- and three-dimensional bearing capacity of
foundations in clay.” Géotechniqu, 54( 5): 297-306, 2004.
[14] F. Askari and O. Farzaneh, “Upper-bound solution for seismic bearing capacity of shallow
foundations near slopes”. Géotechnique, 53(8): 697-70, 2003.
[15] S. K. Sarma and Y. C. Chen, “ Bearing capacity of strip footings near sloping ground during
earthquakes”. Proc. 11th world conf. on earthquake engineering, 1989.
[16] P.A. Cundall and O.D.L. Strack, “A discrete numerical model for granular assemblies.”
Géotechnique, 29(1): 47-56, 1979.
[17] C.S. Chang, “Discrete element method for bearing capacity analysis.” Comput. and Geotech., 12:
273-288, 1991.
[18] C.S. Chang, “Discrete element method for slope stability analysis.” J. of Geotech. Engng., 118(12):
1889-1905, 1992.

[19] C.S. Chang, “Discrete element analysis for active and passive pressure distribution on retaining
wall.” Compu. and Geotech., 16: 291-310, 1994.
[20] A.R. Majidi, “Development of the Implicit Three Dimensional Discrete Element Method and Using in
Three Dimensional Bearing Capacity Analysis of Shallow Foundations” PHD thesis, University of Tehran,
May 2007.
[21] A.R. Majidi and A.A. Mirghasemi, “Iranian Journal of Science and Technology Transaction B-
Engineering,32 : 107-124, 2008.
[22] “Aashto LRFD bridge design specifications”, American association of state highway and
transportation officials.
[23] A.H. Bozorg Haddad, “3-D study behavior of foundations placed on the slopes by model tests”
Master of Science thesis, University of Tehran.

Venkateswara Rao. P, Maniprakash.S, & Srinivasan.S.M
A modified free decay test apparatus for the characterization of

soft magnetic gels in the presence of magnetic fields
Venkateswara Rao. P, [email protected]

Research scholar / Dept. of Applied Mechanics /
Solid Mechanics division
Indian Institute of Technology Madras
Chennai, 600 036, India
Maniprakash. S, [email protected]
Postgraduate student / Dept. of Applied Mechanics /
Srinivasan. S.M [email protected]

Professor / Dept. of Applied Mechanics /
Abstract
This paper presents the development of a simple free decay test apparatus that
can be a cost effective alternative to the popular expensive dynamical
mechanical analyzers useful for characterization of the dynamic characteristics of
soft magnetic composite gels in the presence of variable magnetic field. This
apparatus also addresses the common difficulty faced in dynamical mechanical
analyzers to conduct the characteristics of deformation dependent mechanical
characteristics especially for large deformations, sometimes to the order of 100%
that may be necessary for highly compliant polymeric materials. In addition, this
apparatus can easily be fitted or modified to facilitate the application of magnetic
field. The apparatus is designed to test thin sheet specimens of the magnetic
gels in the shear mode at room temperature. As an example, magnetic
composite gels prepared with micron sized polarizable particles (carbonyl iron
particles) interspersed in a polymer matrix gel are used to show the effectiveness
of the apparatus. The compliance of this magnetic gel can be varied under the
influence of an external magnetic field. Deviations from the linear material
behavior can be captured using the appropriate equations that relate the linear
assumptions made. Such deviations can then be used in determining the large
deformation dependent characteristics of the gel specimen. Thus, it is
demonstrated that the apparatus is a cost effective and useful tool for purposes
of testing soft and compliant magnetic composite gels used for damping
applications.
Key words: Magneic composite gel, shear mode free decay test, storage modulus and loss factor.
International Journal of Engineering (IJE), Volume (4): Issue (2) 179

1. INTRODUCTION
Magnetorheological elastomers (MREs)/ magnetic composite gels constitute a new generation of
materials used in vibration control and damping devices. Apart from the passive damping
properties, it is possible to control the properties of these materials under the influence of a
magnetic field [1]. Primarily the storage modulus ( stiffness coefficient) and the damping are the
properties that are controlled in the applications. The performance of an MRE is dependent on
the tunability of these properties. Based on unique characteristics of MREs, an adaptive tuned
vibration absorber (ATVA) has been developed by Deng et al. [2]. Elastomeric isolator acts as a
filler, amplifying or reducing the initial frequency content of the seismic motion [3]. There are
also other applications of these MREs such as controllable membranes in micro-pumps [4]. In
structural vibration control system applications, the transmitting force of a vibrating structure to
the base through the support depends on the dynamic characteristics of the support [5]. If the
dynamic characteristics of the supports are tunable then it will be possible to isolate large range
of vibrations at the base of the structure. This tunability can be brought about by using
magnetorheological gels in which the characteristics (typically storage moduli) is varied by
applying an external magnetic field. To estimate these dynamic characteristics of a magnetic gel
specimen that is deformation dependent and nonlinear, a suitable test is to be performed with the
prepared magnetic gels in the presence of a magnetic field.
In the literature, various ways of performing the tests are identified to estimate the dynamic
characteristics of MREs in the presence of magnetic field. Gong et al. [5], Hu et al. [6] and Wang
et al. [7] have measured the mechanical properties of MREs using a custom-made system
developed by them. In this system, the MRE sample sandwiched between copper slabs is
subjected to shear mode forced vibration in the presence of an electromagnet with the help of an
exciter, a power amplifier, a dynamic signal analyzer and piezoelectric acceleration transducers.
Fuchs et al. [8] have obtained the storage and loss moduli of the MREs by the dynamic frequency
scan program of the DMA instrument. Chen et al. [9] have characterized the dynamic
performances of MREs by using a modified dynamic mechanical analyzer (DMA) system. The
DMA system is modified by introducing a self made electromagnet which can generate a variable
magnetic flux density upto 1 T. In this experiment the dynamic strain amplitude is set at 0.3%
only. Zhou [10], conducted the experiment testing the damped free vibration of a system
composed of a MRE and a mass in order to determine the shear storage modulus and damping
factor. In this experiment the MRE is placed between the vertically arranged magnetic pole pieces
in such a way that one side of the MRE is attached to the surface of lower pole and the other side
is attached to the brass cover board. Here, this brass cover board acts as a mass and provides
an initial displacement to the MRE when it is excited by an impulse hammer. This type of
arrangement may not facilitate large deformation measurements in MRE.
Lokander and Stenberg [11] have measured the dynamic shear modulus for a double lap shear
specimen using an Instron 8032 dynamic testing machine equipped with an electromagnet. The
resulting electromagnet induction through the samples due to this electromagnet was 0.24T only.
It was not clear regarding the possibility of more variation of magnetic induction in this test
rrangement. This test was done with the small deformations (small amplitude strain range of 0.6%
to 2.5%) only. Shen et al. [12] carried out the shear test on polyurethane MREs using tensile
testing machine with a negligible shear deformation rate of 1 mm/min. Sun et al. [13] used DMA
and MRE samples were analyzed in tensile mode at a strain of 0.5%. Abramchuk et al. [14]
measured the shear modulus of the MREs using the uniaxial compression method. Popp et al.
[15] investigated the MRE performances under both shear and squeeze modes. But in these tests
the shear strains used is 0.1% to 10%. This set up provides the magnetic field strength of 0.15 T
only when the coil current is 3A. Stepanov et al. [16 ] and Bose and Roder [17] studied the
rheological measurements of MREs in the oscillation mode in a rheometer equipped with a
magnetorheological measuring unit. In the later one, the experiments were carried out at an
amplitude of 1% only.

Mostly, in the literature [2,67,9,13], it was observed that the experiments were conducted using
Dynamic Mechanical Analyzer (DMA) testing machines for identifying the dynamic characteristics
of MR elastomers. However, one important difficulty apart from having to introduce the magnetic
field while using DMA is to be able to characterize the gel that undergoes large nonlinear
deformations. The setup envisaged for use in this work would accommodate for both strong
magnetic field up to 1 T when the coil current is 3A and the large nonlinear deformations.
Therefore, for the characterization of dynamic characteristics of magnetic composite gels, a
dynamic shear mode experiment (free decay test) is to be conducted in the presence of magnetic
field.
Thus, the objective of this work, as a first step, is to prepare isotropic magnetorheological gels
with different concentrations of well-dispersed micron sized magnetically polarizable particles and
to develop a simple experimental setup to study and obtain the dynamic characteristics of the
magnetic composite gel under varying magnetic fields in the shear mode. The uniqueness of this
experimental set up is that, the experiment is performed on the gels in direct shear mode instead
of indirect shear mode (either by tensile or by compressive loading) in order to obtain the shear
characteristics of the gels. The dynamic characteristics are obtained in terms of storage moduli
and loss factor.
The paper is arranged in three sections: methodology, results and discussion and conclusions.
To show the effectiveness of the apparatus, in methodology as an example; procedures for the
preparation of magnetic composite gel, description of the development of experimental set up
that is designed and fabricated in conducting free decay test and data analysis performed are
discussed. The static shear test is also conducted to observe the strain range above which the
magnetic gel behaves as a nonlinear material. The results obtained for various compositions of
the magnetic gels prepared by conducting free decay test at various magnetic field influences are
discussed before making the concluding remarks.
2. METHODOLOGY
To show the effectiveness of the apparatus, a free decay test is conducted in shear mode for the
determination of dynamic characteristics of soft magnetic composite gels in the presence of
electromagnetic field.
There are three stages involved in finding the dynamic characteristics of soft magnetic composite
gel under dynamic shear conditions. They are
1. Magnetic gel sample preparation,
2. Free decay test under varying magnetic field, and,
3. Data analysis.
And then conducted the static shear test in order to identify the shear strain above which the
response is non linear.
2.1 Magnetic gel sample preparation

The magnetic gel samples are prepared using the constant ratio of polystyrene-hydrogenated
polybutadiene-polystyrene triblock copolymer [18] to mineral oil gel matrix with various
proportions 10%, 30%, 50%, & 70% by weight of carbonyl iron particles of size 2 to 10 µm. The
detailed procedures for the preparation of this magnetorheological gel can be found in ref. [19].
The cured samples at room temperature are cut into small rectangular magnetic gel strips of
20x20x3mm to be suitable for use in mechanical characterization tests. This magnetic gel strip is
then sandwiched between two non-magnetizable thin plates and adhesively bonded to them to
form testable samples.
2.2 Free decay test under varying magnetic field

2.2.1 Experimental setup and test procedure

In order to facilitate large deformations with magnetic composite gel and to accommodate the
magnetic field, a simple free decay test apparatus has been designed and fabricated in-house.
This apparatus is used for obtaining the magnetic field dependent dynamic characteristics such
as storage modulus and loss factor at room temperature and ambient conditions. The
photographs of experimental set up are shown in Figure 1. The magnetic gel specimen part of the
test apparatus is placed between two magnetic poles for the application of the magnetic field. The
direction of the magnetic field is, thus, in the direction perpendicular to the plane of shear action
when the rod oscillates.
FIGURE 1: The magneto-mechanical coupled free decay testing instrument on magnetic gel
under shear mode oscillation.
This instrument consists of a rigid rod with negligible mass hinged at one of its ends and hung by
connecting a tensile spring at the other end. A mass is attached to the rigid rod at a known
distance from the hinge. A load cell is connected to the rigid rod through a tensile spring to
measure the instantaneous force exerted in the load cell. This instantaneous force is proportional
to the displacement of the spring. The force exerted in the load cell is recorded through a data
acquisition card as a computer data file using LabVIEW (Laboratory Virtual Instrumentation
Engineering Workbench) software [20].
One plate of the magnetic gel strip specimen is fastened to the rigid rod while the other plate is
fixed to the frame. The specimen gel plate which is fastened to the rigid rod can be removable
from the rigid rod in order to facilitate in measuring the dynamic properties of the system in the
absence of the gel. Here, the rigid rod, gel plates and the fasteners are chosen as the non
magnetic materials.
The idea of this test is to determine the natural frequency and damping ratio of gels using the
changes in dynamic properties of an oscillating rod in the system. In order to do this, first the
dynamic characteristics of the free oscillating rod (without the magnetic gel specimen attached to
the rod) should be measured so that appropriate changes can be registered and used to obtain
the gel characteristics in the actual tests on the gel specimens. Therefore, as a first step in the
absence of magnetic gel, initial excitation is given to the rod. Due to the presence of the spring

and the mass, the rod starts to oscillate and the response of the system is measured using a load
cell. The transient response due to the initial excitation dies down due to air friction and other
damping conditions.. Once this initial characteristic of the rod-mass-spring assembly is registered,
then, the magnetic gel is placed appropriately in the system that allows a shear mode vibration of
the gel sample. Once again, the system is excited and a fresh response measurement is made
on the system. The above test procedure is repeated under various magnetic fields in the range
of 0 to 0.7 T for different gels prepared.
2.2.2 Data analysis

The dynamic response of the system for the given initial excitation depends primarily on the three
dynamic attributes of the system – the stiffness of spring elements, the dissipation due to internal
friction of the system and the inertia due to mass. If the gel sample is also attached, the stiffness
and the damping associated with the gel specimen also affect the dynamic characteristics. The
testing system is assumed to be linear. Tests on the system dynamic characteristics were
conducted to validate this assumption.
With the mass of the system known, a simple straightforward procedure is adopted to estimate
the other two attributes of the system, namely, the storage part (stiffness of the spring) and the
dissipation part (damping coefficient). Using a free decay test (by introducing a small initial
disturbance / excitation to the system), these attributes are then calculated from the measured
response data obtained from the load cell. The procedure and the principle used to calculate the
above mentioned system attributes are discussed below.
The equation of motion for the system in the absence of the gel specimen can be written using
the moment equilibrium about the hinge as,
mlm 2  cs ls 2  ks ls 2  0 (1)
where m is the mass and, lm and ls are the distances from hinge support to the mass and the
spring respectively. cs and ks are chosen such that the response of mathematical model matches
closely with the obtained response. Typically, the damping frequency of the system, ωd in Hz
and the damping ratio, εd, are used in the response calculations to match with the experimentally
obtained response and then the stiffness, ks, and the damping coefficient, cs, of the system are
calculated by using the following expressions and these are assumed to be constants
irrespective of whether the magnetic gel is mounted or not.
lm 2 d 2
ks  4 2 m (2)
ls 2 1   d 2
lm 2 d  d
cs  4 m 2 (3)
ls 1   d 2
Figure 3 shows the schematic of the vibrating test system with the gel. The gel is represented
using a spring and dashpot in parallel attached to the horizontal bar of the system. With the gel
mounted to the system, the equation of motion of the system can be written using moment about
the hinge in the Fig. 3 as,
mlm 2  (csls 2  cg lg 2 )  (ks ls 2  k g lg 2 )  0 (4)
where lg is the distance of the center of the mounted gel from the hinge support as shown in
Fig.3. Thus, the stiffness and the damping coefficient (kg, cg) of the gel can be calculated with the
known values of cs and ks obtained earlier by using the following expressions [21]..

4 2 mlm 2d 2 K sls 2

kg   2 (5)
(1   d 2 ) lg 2 lg
FIGURE 3: A schematic diagram of the testing system with the

magnetic gel mounted. The magnetic gel is assumed to be of linear
viscoelastic nature.
4 mlm 2d  d csls 2

cg   (6)
 
1   d 2 lg 2 lg 2
The solution of the equation of motion in equations (1) and (4) is in the following form:
  e  dnt sin( 1   d 2 nt ) (7)
Where  is initial amplitude and  is amplitude with time t.

These dynamic properties are used to calculate the storage modulus, G’, the loss modulus, G”
and the loss factor, , of the gel using following equations.
kgt
G'  ; (8)
lb
cg td
G"  ; and (9)
lb
G"
 (10)
G'
where, t is the thickness of the gel strip, l is the length of the gel strip, and b is the breath of the
gel strip.
2.3 Static shear test

The static shear test is performed in order to identify the shear strain above which the response is
non linear. The same basic experimental set up used for free decay test is used even for static
shear test with some modifications. This setup consists of a rigid rod (the same rod, which is
considered for dynamic shear test) with negligible mass hinged at one of its ends and the rod is
hung by connecting a tensile spring at the other end. Then, an LVDT probe for displacement
measurement is placed at a known distance at which bar will balance in straight horizontal
position. A load cell is connected to the rigid rod through a spring to measure the balancing force
caused in the load cell. At this moment, the initial reading (at zero displacement) measured using
LVDT.
Now, the gel plate is connected to the rigid rod while the spring attachment is removed
from the rod. The displacement at this configuration is noted down. The difference in the above
two readings gives the displacement at zero load. Knowing the displacement at the LVDT probe,
the displacement at the gel and hence the shear strain can be calculated. At this zero load
position, the force in the gel can be calculated by using force equilibrium conditions. The force vs.
deflection of the gel specimen is used to find the shear stress to shear strain variation in the gel
specimen.

In this section, on the basis of system dynamic response and the gel specimen dynamic
response, results obtained for one of the various compositions of the magnetic gels prepared by
the earlier mentioned procedures using free decay test under varying magnetic field are
discussed.
3.1 Free decay test-Determination of test system dynamic characteristics with out gel.
(a)

(b)
FIGURE 4: The response of the system measured using load cell in the absence of the gel. (a)
without magnetic field, (b) With a magnetic field of 0.7T.
Figures 4(a) & (b) show the response of the system measured using load cell in the absence of
the gel with and without magnetic field respectively. The voltage output is measured which is
proportional to the displacement of the system. In these figures, dashed line shows the fitted
experimental data obtained using load cell and the curve with solid line shows the matching
response from the solution (eq.(7)) of the mathematical model obtained from the equation of
motion upon choosing appropriate amplitude, frequency and the damping ratio.
The frequency and the damping ratio of the response are obtained to be the same for the system
with and without magnetic field (ωd = 3.722 Hz, damping ratio, εd = 0.024). Using these identified
damped oscillation frequency and the damping ratio, the stiffness and damping coefficients of the
system are found to be . cs= 0.3446 Nm/s and ks=167.9181 N/m. It is observed that there is no
change in the response of the system due to the applied magnetic field. Here a linear spring
response can be proved as a linear material from the exact curve fitting of the solution obtained
from the linear equation to the experimental response
3.2 Free decay test response –with magnetic gel
(a)

(b)
FIGURE 5: The response of the system measured using load cell in the presence of the gel
with 50% by weight of carbonyl iron particles (a)without magnetic field, (b) with 0.7 T
magnetic field. It can be seen that the oscillations die down fast in the presence of magnetic
gel.
With the test system characteristics determined, the gel is mounted on the test system and the
free decay vibration test is conducted with varying magnetic field. Depending on the initial
excitation given, though small, shear strains up to 50% could be realized in the gels. From the
response of the test system with the gel mounted, the damping frequency and the damping ratio
are first determined under 0 to 0.7 T of magnetic field. The stiffness and damping coefficient (kg,
cg) of the gel are then calculated using equations (5) and (6) respectively.
The response of the test system with the gel mounted (dashed lines) under 0 and 0.7T magnetic
field and the matching responses (solid line) from the solution (eq.(7)) of the mathematical model
obtained from the equation of motion are presented in Figures 5 (a) & (b) respectively. Here the
shear strain experiences about 33%. On keen observation, one can identify that the curve
obtained through mathematical model on the basis of equation of motion is exactly fitting with the
experimental response at both extreme ends of the decayed curves. But, in the middle of the
curves, a shift is observed. This could be due to the non linear nature of the gel. This nonlinearity
due to large deformations can be observed from the following static shear test results also.
3.3 Static shear test: stress-strain behavior of gel due to large deformations
As shown in Figures 6, whenever the deformation is reaches about 30% of it’s thickness, the
stress-strain behavior is started showing non linear in nature.

FIGURE 6: Static shear force vs shear deformation of magnetic composite gel with 50% by wt. of
carbonyl iron particles at 0.7 T magnetic field.
3.4 Determination of dynamic characteristics of the magnetic gel
(a) (b)
FIGURE 7: The response of the system measured in terms of (a) natural frequency and (b) damping
ratio with a gel (consists of 50% by weight of carbonyl iorn particles) mounted under varying magnetic
field.
The total system characteristics in terms of the natural frequency and the damping ratio are first
determined from the response curves. The natural frequency with respect to the applied magnetic
field is presented in Fig. 7(a). One can also observe that there is a significant change in the
natural frequency in the range of 0.1 T to 0.4 T in the gels so that the properties of the gel can be
significantly controlled in this range of the magnetic field. There are no observable changes seen
in the damping ratio for varying magnetic fields (see Fig.7(b)). This is an important observation in
the context of application of the magnetic gel to damping applications.
The shear storage modulus and loss modulus are calculated appropriately using equations (8)
and (9) for various magnetic field values. The shear storage modulus, from Fig. 8(a), about 52%
change is observed over 0.6 T of applied magnetic field in the gel prepared with 50% by weight of
carbonyl iron particles. In other words, a change in storage modulus of 41.387 kPa to 65.835 kPa
can be achieved with 0.7 T of magnetic field. This capability is required for vibration control
applications. Here, three tests were run at every magnetic field. From those three tests, the
deviation in the results obtained in storage modulus and loss factor are less than 4%. Due to this
very less deviation, the average of the three tests results were presented in the Figs.8(a) and
(b). As can be seen in Fig.8(b), loss factor obtained from equation (10) does not change
significantly, indicating that the change in damping is marginal.
(a) (b)

FIGURE 8: (a) Storage modulus, and (b) loss factor of the gel with 50& by weight of carbonyl iron particles
under different magnitudes of magnetic field.
It is shown in this work that the improvement in storage modulus due to magnetic field of as high
as 59% is achievable with isotropic soft MR gels. This is significantly higher than reported in
literature [5, 6, 13] (nearly 30% or less improvement in the storage modulus). The authors believe
that this increase could be due to the soft elastomeric matrix used in this work.
4. CONCLUSIONS
Since, dynamical mechanical analysis tests are difficult to conduct in the presence of large
deformations ( of the order of 50%) and strong magnetic fields, a simple, cost effective free decay
test apparatus is designed, fabricated and used to conduct the magnetic field dependent shear
response tests on soft magnetic gels under dynamic conditions at room temperature.
It is shown in this paper, that the dynamic response of the gel can be obtained. The dynamic
characteristics of the gel can be calculated from this response using a linear model for the gel.
The apparatus also provides for the possibility of obtaining deviations from the linear behavior
that, if modeled appropriately, can fetch the nonlinear characteristics of the gel due to large
deformations. The deviation from linearity is also confirmed from the static tests conducted.
For a future work, the authors are planning to incorporate the above deviation in behavior as a
measurement for nonlinear response. In the experimental procedures, while giving a free
excitation to the rigid rod, sometimes the rigid rod may be subjected to minor lateral oscillations.
This may leads to small error in the response of the system. In future, the authors are also trying
to improve the experimental setup in order to avoid this minor lateral oscillations of the rigid rod .
5. REFERENCES
1. Carlson, J. D. and M. R. Jolly (2000). MR fluid, foam and elastomer devices. Mechatronics,
10, 555-569.
2. Deng, H. X. and X. L. Gong (2008). Application of magnetorheological elastomer to

vibration absorber. Communication in nonlinear science and numerical simulation, 13,1938-
1947.
3. Ibrahim,R.A. (2008). Recent advances in nonlinear passive vibration isolators. Journal of

sound and vibration, 314, 371-452.
4. Beskok, A., and Srinivasa, A. R.,“Simulation and analysis of a magnetoelastically driven

micro-pump”, J. Fluid Eng-T, ASME, 123 (2): 435-438, 2001.
5. Gong, X. L., X. Z. Zhang, and P. Q. Zhang (2005). Fabrication and characterization of

isotropic magnetorheological elasomers. Polymer testing 24, 669-676.
6. Hu, Y., Y. L. Wang, X. L. Gong, X. Q. Gong, X. Z. Zhang, W. Q. Jiang, P. Q. Zhang and Z. Y.

Chen (2005). New magnetorheological elastomers based on polyurethane/Si- rubber hybrid.
Polymer testing, 24, 324-329.
7. Yinling wang, Yuan Hu, Xinglong Gong, Wanquan Jiang, Peiqiang Zhang, Zuyao Chen
(2007). Preparation and properties of magnetorheological elastomers based on silicon
rubber/polystyrene blend matrix. Journal of applied polymer science, Vol.103, 3143-3149.

8. Fuchs, A, Q. zhang, J. Elkins, F. Gordaninejad and C.Evrensel (2007). Development and

characterization of magnetorheological elastomers. J. Applied polymer science,
vol.105,2497-2508.
9. Chen, L, X. L Gong, W. Q. Jiang, J. J. Yao, H. X. Deng and W. H. Li (2007). Investigation on

magnetorheological elastomers based on natural rubber. J. mater.sci.,.42, 483-5489.
10. G. Y. Zhou . (2003). Shear properties of a magnetorheological elastomer. Smart materials

and structures, 12, 139-146.
11. Lokander. M and B. Stenberg (2003). Performance of isotropic magnetorheological rubber

materials. Polymer testing, 22, 245-251.
12. Shen.Y, M.F. Golnaraghi and G.R. Heppler (2004). Experimental research and modeling of
magnetorheological elastomers. Journal of intelligent material systems and structures, 15,
27-35.
13. Sun.T. L., Gong, X. L., Jiang. W. Q., Li.J.F., Xu. Z. B., and Li. W. H. (2008). Study on the
damping properties of magnetotheological elastomers based on cis-polybutadiebe rubber.
Polymer testing, 27, 520-526.
14. Abramchuk S S, D.A. Grishin, E.Yu. Kramarenko, G.V. Stepanov and A.R. Khokhlov (2006).
Effect of Homogeneous magnetic field on the mechanical behavior of soft magnetic
elastomers under compression. Polymer science, 48, 138-145.
15. Popp. K.M, X.Z. Zhang, W.H Li and P.B. Kosasih (2009). MRE properties under shear and
squeeze modes and applications. Journal of physics: Conference series 149, 012095, 1-4.
16. Stepanov, G. V., S. S. Abramchuk, D. A. Grishin, L. V. Nikitin, E. Yu. Kramarenko and

A.R.Khokhlov (2007). Effect of a homogeneous Magnetic field on the viscoelastic behavior
of magnetic elastomers. Polymer, 48 (2007) 488-495.
17. H. Bose and R. Roder (2009). Magnetorheological elastomers with high variability of their
mechanical properties. Journal of physics: Conference series 149, 012090, 1-6.
18. Terry V.Pearce. (2006). “Jelly Blocks and Jelly Letters”, U.S. Patent. 7,101,247 B2.
19. J. Venkataraghavan, Compliant mechanisms, Master’s thesis, Mechanical Engineering,

Texas A&M University, College Station, Texas, 2001.
20. J.Travis and J. Kring.(2006). LabVIEW for Everyone: Graphical programming made easy and
fun ed.3. Prentice Hall.
21. Thomson W.T and Dahleh M.D. (2003). Theory of vibration with applications, fifth edition,
Pearson Education Pvt. Ltd. ,Singapore.

Donghua Jia, Muhammad T. Afzal, Meng Gong, & Alemayehu H. Bedane
Modeling of moisture diffusion and heat transfer during

softening in wood densification
Donghua Jia [email protected]

Former Graduate Student
Faculty of Forestry and Environmental Management,
University of New Brunswick, P.O.Box 4400,
Fredericton, NB, E3B 5A3 Canada
Muhammad T. Afzal [email protected]

Associate Professor
Department of Mechanical Engineering
Meng Gong [email protected]

Research Scientist
Alemayehu H. Bedane [email protected]

Graduate Student
Abstract
Mechanical densification of wood involves compressing the wood in radial

direction using heat, water and steam to produce a higher density surface
exhibiting better mechanical properties. The densified wood is an
environmentally friendly product that presents new opportunities for the wood
products industry. Wood surface densification involves both soaking and heating.
The objective of this study is to present a two-dimensional model of moisture
diffusion and heat transfer during softening process in order to understand and
control the degree of surface densification. The governing equations for diffusion
process and heat transfer are solved numerically at non-steady state conditions.
Experimental data was also collected on Aspen and Balsam fir specimens to
determine the moisture profile. The model predicts suitably the moisture content
and temperature in the soaking process. The results showed that the surface to
be softened could be heated to a temperature of 80~90 oC in 3~5 minutes with
an average moisture content of 25 percent in the surface layer.
Keywords: Softening, soaking, densification, moisture diffusion, heat transfer, modeling
International Journal of Engineering (IJE) Volume (4): Issue (2) 191

1. INTRODUCTION
With the importance of wood in man’s environment, the demand for hardwood used for furniture
and interior materials keeps increasing. In Canada, coniferous woods account for about 80% of
the total volume of merchantable timber. Compared with hardwoods, coniferous softwoods show
low density and are generally soft. Therefore, to utilize softwoods for interior materials, it is
necessary to improve their surface properties such as abrasion resistance and hardness as well
as their dimension stability. Inoue et al. (1990) studied the improvement of surface properties of
lumber by surface compression [1]. In comparison with untreated wood, abrasion resistance and
hardness increased by 40 to 50% and 120 to 150%, respectively. Softening is an important step
in the manufacture of surface-densified wood. Softening process involves soaking and heating of
wood. It is vital to study and determine the distribution of moisture content and temperature
across the thickness of a wood board in order to understand and control the degree of surface
densification. This study was aimed at understanding the moisture diffusion and heat transfer in
the coniferous wood boards during softening process, and eventually proper softening
parameters can be determined and optimized.
2. MATHEMATICAL MODEL
2.1 Governing equations

At low temperature, moisture in wood mainly moves by a diffusion process that is driven by a
gradient in the moisture content. Fick’s second law offers one generalized method to describe the
diffusion process [2, 3]. This approach allows the estimation of moisture gradients at any time
during heating period. In the literature, the thermally induced mass transfer, which contributed
considerably to the total moisture flux, was also taken into account in the model construction [4].
Moisture transfer should be considered a coupled process with heat transfer, or described by a
non-isothermal diffusion model.
Non-isothermal diffusion was analyzed by different methods. Siau (1995) indicated that the
thermodynamic model provided the best fit to the experimental data [5]. The model derived from
non-equilibrium thermodynamics was presented by Nelson (1991) [6], and the steady state mass
flux was shown as,
 dM dT dM 
J m  k m    (1)
 dT dx dx 
The temperature gradient coefficient was given as,
dM H M Eb
 (2)
dT RT H T
Where
k m Gm D , (3)
Eb  38,500  290 M (4)

Taking Eq. 1 as the starting point, the differentiated unsteady-state equation was derived as,
M
   Deff ,T T  Deff ,M M  (5)
t
As a complete model for moisture movement, the energy balance equation is also required in
addition to Eq. 5. Heat transport in wood occurs via heat conduction in all the three phases (free
water, bound water, and vapor), and by convection through mass transportation. In the slow
heating condition the vapor and free liquid bulk flows can be ignored, and the term of heat
transfer by convection can also be omitted. With the simplification of conventional heat transfer,
the heat balance equation was derived as,
  wood c pwoodT 
 eff T  (6)
t
In this study, the soaking process makes a two dimensional model necessary. In the experiments,
the water was sucked into wood mainly in the thickness direction. However, the wood specimen
was heated in two scenarios: heat transfer along specimen thickness from the boiling water, and
heat transfer along specimen width from surrounding hot evaporated vapor. Thus, the cross
section (i.e. width and thickness) of a specimen was taken into consideration in the model study.
Boundary conditions
Y
3
4
X
0
2
1
Soaking boundary
T
L Computation domain
FIGURE.1. Configuration of 2-D model computation domain
In Figure 1, there were three types of boundary conditions. Boundary 1 was emerged into the
water; boundaries 2 and 4 were heated by hot evaporated vapor; boundary 3 was exposed to
open air. Using the relation between surface emission coefficient (S) and moisture transfer on the
surface, the external surface, boundaries 2, 3, and 4 can be expressed as,
M
D   S M  M   (7)
x

The boundary condition for heat exchange could be written as,
T
 eff  h heat T  T  , at the external surface (8)
x
At the boundary 1 (in water), the condition is imposed as:
M  M max (9)
o
T=100 C, (10)
2.2 Numerical procedures

The non-isothermal model equations are solved numerically. The finite element method is used to
discretize the mass diffusion and heat transfer equations. The weak form of Eq. 5 and Eq. 6 is
expressed as,
 M T   M T 
D  DTy 
Ni 0    M  Ni 0    DMx x  DTx x  Ni 0    My y y  (11)
V  0 Ni  t CpwoodT  V  0 Ni  x  T
 
V 0
  N  y 
i
T 
 kx  ky 
 x   y 
Following the procedure of integration by parts, Eq. 12 can be obtained:
 M T 
Ni 0    M  0   D Mx  x  DTx  x 
Ni
 0  S 0  dS 
V


N i   t   Cpwood T  N i   k
T 
 x 
 x 
 M T   M T 
 D My  DTy 
 Ni 0   D Mx  x  DTx 
 x dx    N i 0   y y 
   V  y  0 N i    dy
V x  0 N i   kx
T  ky
T 
   
 x   y 
(12)
where Ni is the weighting function.
The finite element approximation can be derived by considering
M  x, y   N M  (13)
T  x, y   N T  (14)
where M, T represents the nodal point values with respect to time. Substitution of these finite
element approximations into the weak form gives,
 M e  M 
  C dV
Ve
 
 t  Te 
  S dV  T
Ve
e
   B dS
Se
(15)
 e 
where C   N  c p N  (16)

S   N  D eff N   (17)
Eq. 18 can be obtained, in terms of central time difference,
n 1 n
1 1 M e  1
1 1  M e 
 S   Ct    B   C t   S  
n
2 (18)
2 t  Te   t 2  Te 
A computer code was developed to solve above algebraic equations.
3. EXPERIMENTAL
Aspen (Populus tremuloides Michx.) and Balsam fir (Abies balsamea (L.) Mill.) specimens were
used for moisture distribution tests. The dimension of each specimen was 280 mm long, 25 mm
thick and 37 mm wide. All specimens were stored in a conditioning chamber set at 20oC and 65%
relative humidity prior to the tests. In the soaking process, the water was heated to the boiling
point, and then a specimen was put on the top of a shelf in the water with 1mm of the specimen in
the radial direction (thickness) being emerged into the water. The whole soaking process took
approximately 5 minutes. After soaking, the specimen was cut into several small slides along the
thickness to determine the moisture profiles.
4.1 Model validation

During softening in wood densification, it is important to know how the heat and moisture transfer
distributed within the surface of the wood. This information can be used to adjust the treatment for
densification parameters, and to control the quality of final product more effectively. The proposed
model predicted the moisture and temperature profile very well. The test results are shown in
Figures 2 and 3 for aspen and balsam fir, respectively. The soaking depth of a specimen shown
in the figures is approximately 8 mm for both species. The average moisture content distribution
along the thickness for aspen after soaking process was increased non-uniformly from unsoaked
surface. In case of balsam fir the average moisture content distribution along the thickness was
increased uniformly from unsoaked surface.

25
20
Moisture Content (%) 15
10
0
0 5 10 15 20 25
Distance from unsoaked surface (mm)
FIGURE.2. Distribution of average moisture content along thickness for three aspen specimens
after soaking process
30
25
Moisture Content (%)
20
15
10
0
0 5 10 15 20 25
Distance from unsoaked surface (m m)
FIGURE. 3. Distribution of average moisture content along thickness for three balsam fir
specimens after soaking process

Experimental results of the moisture profile of wood in the soaking process compared to verify the
model calculation. To simulate the soaking process by model calculation, the following diffusion
coefficient was used in the model for wood [4],
Deff , M  4.78  10 5 exp  3574 / T  (19)
Thermal conductivity of wood with moisture content was approximated by the following empirical
equation [7]:
k  G (0.2  0.004 M )  0.024 (20)
Where G is the specific gravity based on volume at moisture content M.

The heat capacity of wet wood is a function of both moisture content and temperature.
Koumoutsakos (2001) showed that the heat capacity was not related to wood species and
density, and was assumed to be constant over the temperature range of interest [7],
Cpwood  1176  5859M  / 1  M  (21)
The results of model simulation and corresponding experimental data are shown in Fig. 4 and 5.
Both figures show a very good model prediction of temperature and moisture distribution. Fig. 4
illustrates that the temperature gradient along the thickness is minimal; this can be explained by
the effect of hot evaporated vapor surrounding the wood. In Fig. 5, the thickness wetted by
soaking in model calculation is also approximately 8 mm.
120
100
Temperature( oC)
80
60
40 Model prediction
Experimental data for aspen
20 Experimental for fir
0
0 5 10 15 20 25 30
FIGURE. 4. Predicted temperature distribution by model calculation

80
70
60 Model prediction
Moisture Content (%)
50 Experimental data for aspen
40 Experimental data for fir
30
20
10
0
0 5 10 15 20 25
FIGURE. 5. Predicted moisture distribution by model calculation
5. CONCLUSIONS
From the above analysis and comparison between model prediction and experimental data, it can
be concluded that:
(1) The principal mechanism of moisture movement in the soaking process is the
diffusion of moisture caused by moisture and temperature gradients, also the
numerical model for moisture and heat transfer can be suitably used to predict the
variables of moisture content (M) and temperature (T) in the soaking process.
(2) The boundary conditions selected can accurately describe the moisture and heat
transfer processes through the material surfaces.
(3) The average moisture content in the surface layers was approximately 25%.
(4) The surface of a specimen to be softened could be heated to a temperature of
80C to 90C in 3-5 minutes for these two species. The temperature of the
opposite face layers was approximately 60C.
6. NOMENCLATURES

Cp heat capacity, J / Kg .K
2
Deff effective diffusion coefficient, m s
Eb activation energy, J mol
G specific gravity
2
hheat heat transfer coefficient, W / m .K
H latent heat, J / kg
H relative humidity
J fluxes of moisture, kg m .s
2
2
Je fluxes of enthalpy, W m
k thermal conductivity
km heat transfer coefficient, m / s
M moisture content
R gas constant, J mol.K
S surface emission coefficient, m s
T temperature
 density of liquid, kg m
3
eff thermal conductivity, W/m.K
7. ACKNOWLEDGMENT

The authors would like to acknowledge Natural Resources Canada for its support and funding
under its Value to Wood Program.
8. REFERENCES
[1] M. Inoue, M. Norimoto, Y. Otsuka, and T. Yamada. 1990. Surface compression of coniferous
wood lumber. 1: A new technique to compress the surface layer. Mokuzai Gakkaishi 36(11): 969-
975.
[2] W.T. Simpson and J. Y. Liu. 1991. Dependence of the Water Vapor Diffusion Coefficient of
Aspen on Moisture Content, Wood Sci. Tech. 26: 9-21.
[3] W.T. Simpson, 1993. Determination and Use of Moisture Diffusion Coefficient to Characterize
Drying of Northern Red Oak, Wood Sci. Tech. 27: 409-420.
[4] S. Avramidis, P. Englezos, and T. Papsthansiou. 1992. Dynamic nonisothermal transport in

hygroscopic porous media, Moisture diffusion in wood, AIChE J. 38(8): 1279-1287.
[5] J.F. Siau, 1995. Wood: Influence of Moisture on Physical Properties, Virginia Tech., USA.
[6] R.M. Nelson, Jr. 1991. Heats of Transfer and Activation Energy for Bound-water Diffusion in
Wood, Wood Sci. Tech. 25: 193-202.
[7] A. Koumoutsakos, S. Avramidis, and S. G. Hatzikiriakos. 2001. Radio Frequency Vacuum

Drying of Wood. I. Mathematical Model, Drying Technology 19(1): 65-84.

CALL FOR PAPERS
International Journal of Engineering (IJE) is devoted in assimilating

publications that document development and research results within
the broad spectrum of subfields in the engineering sciences. The
journal intends to disseminate knowledge in the various disciplines of
the engineering field from theoretical, practical and analytical research
to physical implications and theoretical or quantitative discussion
intended for both academic and industrial progress.
Our intended audiences comprises of scientists, researchers,

mathematicians, practicing engineers, among others working in
Engineering and welcome them to exchange and share their expertise
in their particular disciplines. We also encourage articles,
interdisciplinary in nature. The realm of International Journal of
Engineering (IJE) extends, but not limited, to the following:
Aerospace Engineering
Agricultural Engineering
Biomedical Engineering
Chemical Engineering
Civil & Structural Engineering
Computer Engineering
Control Systems Engineering
Education Engineering
Electrical Engineering
Electronic Engineering
Engineering Mathematics
Engineering Science
Environmental Engineering
Fluid Engineering
Geotechnical Engineering
Industrial Engineering
Manufacturing Engineering
Materials & Technology Engineering
Mechanical Engineering
Mineral & Mining Engineering
Nuclear Engineering
Optical Engineering
Petroleum Engineering
Robotics & Automation Engineering
Telecommunications Engineering
CFP SCHEDULE
Volume: 4
Issue: 3
Paper Submission: May 2010
Author Notification: June 30 2010
Issue Publication: July 31 2010
CALL FOR EDITORS/REVIEWERS
CSC Journals is in process of appointing Editorial Board Members for

International Journal of Engineering (IJE). CSC Journals would
like to invite interested candidates to join IJE network of
professionals/researchers for the positions of Editor-in-Chief, Associate
Editor-in-Chief, Editorial Board Members and Reviewers.
The invitation encourages interested professionals to contribute into

CSC research network by joining as a part of editorial board members
and reviewers for scientific peer-reviewed journals. All journals use an
online, electronic submission process. The Editor is responsible for the
timely and substantive output of the journal, including the solicitation
of manuscripts, supervision of the peer review process and the final
selection of articles for publication. Responsibilities also include
implementing the journal’s editorial policies, maintaining high
professional standards for published content, ensuring the integrity of
the journal, guiding manuscripts through the review process,
overseeing revisions, and planning special issues along with the
editorial team.
A complete list of journals can be found at

http://www.cscjournals.org/csc/byjournal.php. Interested candidates
may apply for the following positions through
http://www.cscjournals.org/csc/login.php.
Please remember that it is through the effort of volunteers such as

yourself that CSC Journals continues to grow and flourish. Your help
with reviewing the issues written by prospective authors would be very
much appreciated.
Feel free to contact us at [email protected] if you have any

queries.
Contact Information
Computer Science Journals Sdn BhD

M-3-19, Plaza Damas Sri Hartamas
50480, Kuala Lumpur MALAYSIA
Phone: +603 6207 1607

+603 2782 6991
Fax: +603 6207 1697
BRANCH OFFICE 1
Suite 5.04 Level 5, 365 Little Collins Street,
MELBOURNE 3000, Victoria, AUSTRALIA
Fax: +613 8677 1132
BRANCH OFFICE 2
Office no. 8, Saad Arcad, DHA Main Bulevard
Lahore, PAKISTAN
EMAIL SUPPORT
Head CSC Press: [email protected]
CSC Press: [email protected]
Info: [email protected]

International Journal of Engineering (IJE), Volume (4) : Issue

Uploaded by

Copyright:

Available Formats

International Journal of Engineering (IJE), Volume (4) : Issue

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

International Journal of Engineering (IJE), Volume (4) : Issue

Uploaded by

Copyright:

Available Formats

International Journal of

Volume 4, Issue 2, 2010

International Journal of Engineering (IJE)

IJE Journal is a part of CSC Publishers

Typesetting: Camera-ready by author, data conversation by CSC Publishing

This is the second issue of volume fourth of International Journal of

To build its international reputation, we are disseminating the publication

Editorial Board Members

Associate Editors (AEiCs)

Editorial Board Members (EBMs)

Volume 4, Issue 2, May 2010.

139- 148 A New Strategy to Achieve Radical Combustion Through Exhaust

160 -178 Three Dimensional Bearing Capacity of Shallow Foundations

A modified free decay test apparatus for the characterization of soft

International Journal of Engineering, (IJE) Volume (4) : Issue (2)

The modeling of UPFC based on circuit elements in an exact

Ali akbar Motiebirjandi [email protected]

Kauomars Sabzawari [email protected]

UPFC is considered and known as one of the best FACTS devices. It is a

International Journal of Engineering, (IJE) Volume (4): Issue (2) 105

2. UPFC model based on circuit elements

FIGURE 1: UPFC is placed in two- bus power system

Modelling of the Series Converter

Calculation of injected active and reactive power of the series converter

International Journal of Engineering, (IJE) Volume (4): Issue (2) 106

Flowing current in the line is calculated as follows:

International Journal of Engineering, (IJE) Volume (4): Issue (2) 107

Calculating the active and

International Journal of Engineering, (IJE) Volume (4): Issue (2) 108

Direct calculation of resistance and reactance equivalent to series converter

The impedance equivalent to parallel elements equals to:

International Journal of Engineering, (IJE) Volume (4): Issue (2) 109

Modelling the UPFC in the exact transmission line model

FIGURE 5: UPFC placed in exact line model

V ' S  VS   VSE  (27)

International Journal of Engineering, (IJE) Volume (4): Issue (2) 110

Inserting equations 27 to 32 in equation 33, the current equation is achieved as follows:

Inserting equations 34 and 36 in equation 35 yields the following results:

VS * VSE V * VSE V * VSE

VS * VSE V * VSE V2 V2 V * VSE

International Journal of Engineering, (IJE) Volume (4): Issue (2) 111

Load flow results in 14-bus system without UPFC series converter

TABLE 1: buses voltage amplitude and angle.

International Journal of Engineering, (IJE) Volume (4): Issue (2) 112

TABLE 2: Active and reactive power lines without UPFC.

FIGURE 7: Changes in reactive power in terms of active power in line 1-5

International Journal of Engineering, (IJE) Volume (4): Issue (2) 113

TABLE 3: buses voltage amplitude and angle

TABLE 4: Active and reactive power lines with UPFC

International Journal of Engineering, (IJE) Volume (4): Issue (2) 114

approximate model exact model percentage

Voltage bus 5 (pu) 1.003 1.007 +0.39%

Active power line 1 - 5 (MW) 75.19 75.615 +0.56%

Reactive power line 1 - 5

International Journal of Engineering, (IJE) Volume (4): Issue (2) 115

Results of load flow on 30–bus system without UPFC series converter

TABLE 6: buses voltage amplitude and angle

TABLE 7: The active and reactive power line2-5 without UPFC

International Journal of Engineering, (IJE) Volume (4): Issue (2) 116