See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/237620158

Experimental Investigation on Swash Plate Axial Piston Pumps With Conical

Cylinder Blocks Using Fuzzy Logic Control

Article · January 2002

DOI: 10.1115/IMECE2002-39325

CITATIONS READS

0 156

4 authors, including:

Medhat Kamel Bahr Khalil Rama B. Bhat

Milwaukee School of Engineering Concordia University Montreal
23 PUBLICATIONS   113 CITATIONS    367 PUBLICATIONS   2,974 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Crashworthiness of helicopters View project

It is just a text book I published View project

All content following this page was uploaded by Rama B. Bhat on 23 July 2014.

The user has requested enhancement of the downloaded file.


FPST TOC
ASME International Mechanical Engineering Congress & Exposition
EXPERIMENTAL INVESTIGATION ON SWASH PLATE AXIAL
PISTON PUMPS WITH CONICAL CYLINDER BLOCKS
USING FUZZY LOGIC CONTROL
M.K. Bahr Khalil*, J. Svoboda** and R. B. Bhat***
*Graduate student ** Professor ***Professor
Mechanical Engineering Department, Concordia University
1455 De Maisonneuve Boulevard West, Montreal, Quebec H3G 1MG, Canada
Abstract
Analytical studies on the dynamic performance of
variable displacement swash plate axial piston
pumps show that the pump performance can
significantly improved by replacing the
conventional PD process controller with fuzzy
logic controller. Electrically controlled, constant
power regulated, swash plate pumps with conical
cylinder blocks have been recently extensively
studied. Comparative study has been carried out
to compare the pump dynamic performance when
the conventional PD process controller is replaced
by a proposed fuzzy logic one. The study reveals
some superior performance characteristics when
fuzzy logic controller is used.
In the present study an experimental setup is built
to measure the dynamic performance of a typical
9-piston pump that has conical shaped cylinder
block. The pump is of 40 cc/rev geometrical size,
type A4VSO, that is manufactured by Rexroth.
The experimental setup consists of a hydraulic
test bed interfaced with real time control and data
acquisition system. The setup is used firstly for
testing the pump static characteristics.
Subsequently, the setup is used to measure the
time response of the pump, which is equipped
with the conventional PD controller, to the
stepwise changes in the load pressure. Pump
model verification is then discussed based on the
comparison between the theoretical and
experimental results. The pump is afterwards
interfaced with real time control software for
prototyping the proposed fuzzy logic controller
that replaces the currently used PD one. With the
fuzzy logic control, measuring the pump time
response under the same loading conditions is
repeated. Experimental results are presented,
compared with the analytical findings and
discussed.
Keywords
Axial, Piston, Pump, Swash Plate, Fuzzy,
Experimental, Variable and Power.
1 Copyright © 2002 by ASME
Proceedings of IMECE2002
November 17–22, 2002, New Orleans,
>, Louisiana
IMECE2002-39325
Introduction
Swash plate pumps with conical cylinder blocks
have generated some interest recently, in view of
their improved static and dynamic characteristics.
Kassem and Bahr [1] carried out theoretical study
led to a recommended valve plate configuration
that suppresses cavitation and has low pump
discharge fluctuations in the piston chamber
pressure. Kassem and Bahr [2] shows
theoretically that the lateral moment that acts on
the swash plate increases nearly linearly with the
increase in the load pressure and/or the decrease
in the swash plate inclination angle. Kassem and
Bahr [3] proposed a frizzy logic controller to
replace the conventional PD controller currently
in use. Their study shows that the fuzzy logic
controller yields better pump dynamic
performance under both the design and off-design
parameters of the control system. Bahr, Svoboda
and Bhat [4] carried out a vibration analysis of the
pump. Results show that in order to keep the
swash plate inclination angle nearly constant at
the desired position, under the effect of the
excitation of the lateral moment, the actuator
spool vibrates periodically with the same
harmonic content as those contained in the lateral
moment. Frequency response of that kind of
pumps was studied by Bahr, Svoboda and Bhat
[5]. Results show significant amplitude
attenuation beyond 15 Hz. Bahr, Svoboda and
Bhat [6] studied the dynamic load on the drive
shaft bearing due to the pressure force. Results
can be used for shaft bearing design. Kaliafetis
and Costopoulos [7] studied both theoretically
and experimentally the static and dynamic
characteristics of a variable displacement swash
plate pump with constant pressure regulator. N.D.
Marring, and R.E. Johnson. [8] carried out
modeling and designing a variable geometric
volume axial piston pump.
In the present paper, pump dynamic performance
with PD controller is measured, pump model is
verified and fuzzy logic controller is prototyped
using real time control software. Pump dynamic
performance with fuzzy controller is compared to
that one with PD controller
Pump Mathematical Model

The pumping mechanism of an axial piston pump

that has conical shaped cylinder block as shown
in Fig. 1. Frames of reference, shown in Fig.2,
were defined to describe the general space motion
of the piston. The initial frame of reference
XoYoZ0 is chosen such that its origin coincides
with the swash plate pivoting point Oo. The axis
Z0 coincides with the pump driving shaft axis and
Y0 coincides with the axis around which the
swash plate is swinging. Five steps of frame
transformation, starting from XoYoZo and ending
by XsYsZs, are then carried out to get the
components of the position vector [rsk]O that Fig. 1 Piston pump with conical cylinder block
denotes to the kth piston spherical head center at
any angular position Okrelative to the initial frame
of reference. Position vector [rsk]0is then given by
] FL3kCOS0kSin~ + R2COS0 k-
[rsk ]0 = [~551 | = | L3kSin0ksinl3 + R2sin0k (1)
LZ,k J L -LakC°Sp+L1

With the constraint that the pistons move guided

by a plane parallel to the swash plate surface and
passing through the swash plate pivoting point,
the kth piston displacement inside the
corresponding piston chamber was first
introduced in [1] to be give by Fig.2 Frames of reference
s k = L 3 - ( E L/cos13) (2)

0.5D 2 cos0ktanct +L~

where L 3 =
cos 13- COS0k sin 13tanco'
Ok =(00t+2n(k-1))/N 13=tan "l 0.5 D1 "D2
' L2
Applying the continuity equation to the piston
chamber control volume (C.V.) shown
schematically in Fig.3, it can be shown that: Fig.3 Piston chamber parameters and variables

Qsk +ApSk =Qdk +Pk /RL -I-Vck[)k/ B (3) force, piston inertia and the component of the
where: Q dk = CdAdk (2lp
k l . Pd Ii)YAsgn (P k-Pd), piston centrifugal force in the direction of piston
P line of action. In the present paper, a new
approach is followed considering the piston
Qsk = cdgsk (21Ps
.I Pk Ii.)g sgn (Ps-Pk), and absolute acceleration during the piston general
P space motion relative to the inertial frame of
Vck =Ap(0.5L c --Sk)+ Vo reference.

It was shown in [1 to 6] that knowing the pump Pressure force due to single piston that acts on the
dimensions, both the piston chamber swash plate along Z5 axis is denoted by [Fpk]5,and
instantaneous pressure Pk and the delivery flow has, relative to the 5th frame of reference, the
rate Qak can be determined by solving the components of 0, 0 and ApPk. This force can be
forgoing equations numerically. Knowing Qdk, the redefined, relative to the inertial frame of
pump instantaneous delivery flow rate due to all reference, to be equal to
pistons can be determined.
I I cos 0k sial31
Forces acting on the swash plate due to single [Fpk]0 = g p P k / - - sin 0k sing3 / (4)
piston, which presented in [2 to 6], were
calculated as the algebraic sum of the pressure L cosp j

2 Copyright © 2002 by ASME

As shown in Fig.4, position vector [rck]0 that
denotes the center of mass of the kth piston can be
calculated as follows
[rck]0 = [rSk]0 + [rcg]6 (5)
Where vector [rck]6 is the fixed magnitude
position vector [rck]5 as redefined relative to 6th
frame of reference that has origin coinciding with
the 5th frame origin, and axes parallel to those of
the inertial frame of reference. It can be calculated
as follows
L F- COS0ksin~l
[rck ]6 = - ~ - [ - - sin 0k sin 13[ (6)
L cosl3 J
Hence, the piston absolute acceleration, due to its
general space motion relative to the inertial frame
of reference, is given by
[aok]0=6ox[rck]0+oox(oox[rok]o)
+ 2~ox [i'ok]o + [i:ck]0 (7) Fig.4.Piston general space motion
The first two terms represent the piston
acceleration due to its rotation while the last two Figures 5 and 6 show the layout and schematic
terms represent the piston acceleration due to its representation of the pump control unit,
translation. The first term is the piston tangential respectively. The control unit is used to achieve
acceleration that equals zero due to pump rotation constant power operation of the pump regardless
with constant angular speed. The second term is of the load pressure variation encountered by the
the piston normal (centripetal) acceleration. The pump. The control system was fully described in
third term is the piston coriolis acceleration. The [3 and 4].
last term is the second derivative of the position
vector [rck]O. When the proportional valve solenoid receives a
control current iv, a force proportional to this
The resultant force vector by which the kth piston current; namely kiiv, acts on the valve spool and
acts on the swash plate has the components Fxk, causes it to move. The equation of motion of the
Fyk and Fzk. It can be calculated as follows valve spool was first introduced in [3] and given
by
[Fk] 0 =[Fpk]0 + m p [a~k]O (8)
m v~v + fv~v + kvs,, = k~iV (10)
This force causes moment that, due to single
piston and relative to the inertial frame of
reference, is given by ~ ' Ele~ctronic
Cort~ol
i0 J0 k o ' Card ' I
[Mk]o =[rsk]0 x[Fk]0 = Xsk Ysk Zsk (9)
Vyk ....... . . . . . . . . . . . . . iv ....................
i

The total moment components acting on the

swash plate can be found by superimposing [Mk]o L~- ~ - ]LI..,......~....].................
~.. ~::JLS:; _ ~.......~..........
components due to the whole piston group. The
total moment has thus the components Mx,
My and Mz. The moment My tends to change the
swash plate inclination angle and should be
balanced by the control system. The moment M,
equals the pump driving torque, while the
resultant of the two moment components Mx and
Mz acts on the swash plate bearing system. Fig.5 Symbolic representation of the control unit

3 Copyright © 2002 by ASME

Due to valve spool displacement, assuming
constant discharge coefficient and negligible
valve leakage, we get
Qa =CdW(Svmax--Sv)J~Pc~ pT[ sgn(p~ "Pr )(11)
I r
Qb --~-CdWSvl 2 IPvpPellsgn( pv - pc, )
! 2lPv-Po2I
Q =Caw(s . . . . -Sv)~ 9 sgn(pv-Pc2)(13)
Qd = CdWS,,I~P¢~ Pr[ sgn(pc2 - pT )
Applying the continuity equation to the side
chambers of the control piston, it can be shown
that the pressures P¢i and Pc2 are given by
B corresponding percentage of its maximum
Pc, = ~ - I(Qb - Q , -Acp :~cp-P~/Rg)dt,
B 100%. At a certain time, after reaching the steady
p ~ 2 = ~ - I ( Q -Qd+AcpRcp+pc/RL)dt
where gcl= Vci "1-AcpXcp andV~2 = Vci- AcpXcp
The pressure difference on the two sides of the
control piston drives it to a new equilibrium
position. The instantaneous angular speed and
swash plate inclination angle are first introduced
in [4] and are given by
1 dynamic performance with the newly developed
6t = i--~- I[(pcl - pe2)Aeprs + My - f~&
- k,~ (or + 0.09)]dt
ot = I&dt
$vntll~t
+$V
P T ~ ~
Fig.6 Schematic representation of the hydro-
mechanical part of the control unit
Simulation of the Pump Dynamic
Performance
Software based on Matlab was developed to
simulate the dynamic performance of the pump in
constant power operation. Figure 7 shows the
layout of the main blocks of which the simulation
(12) program contains. It contains subsystems that
simulate the dynamics of both the pump and the
proportional valve, in addition to the main
subsystems that form the amplifier card (A) used
currently and supplied with the pump.
(14)
The swash plate is assumed to be initially at
minimum inclination angle that satisfies the pump
self-lubrication requirements. Then, the pump is
subjected to a step change in the load pressure
with a value, which according to the constant
power operation drives the swash plate to a
inclination angle; namely 25%, 50%, 75% and
(15) state, the load pressure was assumed to increase
in a stepwise manner to its maximum value that
swings the swash plate again to the minimum
position. The model is validated in this study
based on the agreement between the theoretical
and experimental results as shown later.
Fuzzy logic control was first constructed in [3]
and proposed to replace the currently used PD
process controller. Simulation of the pump
controller was repeated under the same loading
conditions. A comparison of the settling time of
(16)
the swash plate while it raises from zero to 100%,
(17) as presented in Fig.8, shows that the swash plate
with the PD controller is slightly faster. Important
advantage of using such newly developed fuzzy
controller is that the proportional valve spool is
iv
subjected to less vibration, which increases the
service life of the valve and reduces the heat
generated in it.
Experimental Measuring of the Pump
Dynamic Performance
Experimental setup is built to measure the pump
to measure the time response of the pump to the
stepwise changes in the load pressure. It consists
of a hydraulic test bed interfaced with real time
control and data acquisition system, as shown in
Fig.9 in a block diagram form.
For testing purposes, the contents of the amplifier
card (A) are replaced by real time control
software to facilitate changing the design
parameters of the original PD controller or
prototyping the proposed fuzzy controller.
4 Copyright © 2002 by ASME
 .... Parameters ...........................................................
, Simulation Subsystems .....................................................................................................
,
Setting and Load Pressure _1
Monitoring [Dist~baace(~ ] i
, ~ AmplilierCard A . . . . . . . . . . . . . . . . . . . . . . . . i t------------------------~

i a s se i ',ConLrolprossure
IQmax!'~'Jl"~}!]~]8]8]ll:, t-~'~ oL+ " e l-------I ~ i r--~ rI °~'°~
p'~s'~el~ is(Q,

t~--iltaX '[__~___]- " ~/FI.IZ~ SVl"" l l 'l

l L I

Fig.7 Block diagram of the simulation program subsystems

So, the real time control and data acquisition

system is composed of software and hardware
parts. The hardware part is the I/O interfacing 100 V g = 40 c d r e v ," "1
card and newly developed amplifier card (B) that 9~ ................................. ~ ................................................................ ~ll'lllllllllll l ]

has no controller and used just for driving the

proportional valve solenoid.

Figure 10 shows the schematic layout of the

i 90
• 80 ,
..................
i...................
.V..L - _/.,q
................................... .................................. l
experimental setup including the hydraulic test bed
that consists of the tested pump that has the same
design parameters and specifications used in the 65-~ ...................'~
................. -PD
analytical investigation, the control pressure
supplying unit, load disturbance unit and oil 6o i ! !
25 50 75 100
conditioning unit. The setup is used first to
measure the time response of the pump with the Swash plate inclination angle (%)
conventional PD controller. Subsequently, with the Fig.8 Settling time of the swash plate response in
fuzzy logic controller, pump testing under the case of different using different controllers
same loading conditions is repeated.

..............................................................................
Real TimeCoaWolSoftware ........................................-~
Data Setliag, ~{onitoring /JO Card
and AcquisitionUnitt °/
Ipd ~J,

Iq,,u ,fl I ~,~F---,q ~ I r-----, ~,-,--~.

• A

Fig.9 Real time control and data acquisition system

Figures 11 and 12 show a comparison of the mathematical model. The slight deviation
theoretical and experimental swash plate step between the theoretical and experimental results
response, of the pump that is equipped with the PD is referred to the effect of the time response of the
controller, while the inclination angle is raising loading unit, which is not considered in the
and falling, respectively. Similar results are model. Results shows that the swash plate
presented in Figs. 13 and 14 with the proposed response with the PD controller is slightly faster,
fuzzy controller. while with fuzzy controller is smoother. It is to be
noted that, in case of falling conditions with fuzzy
The evident agreement between the theoretical and controller, the swash plate reaches zero position at
experimental results fairly validates the developed nearly zero velocity meaning less impact.

5 Copyright © 2002 by ASME

 R,,,|~.~, C.~t:a! ~ . ~ , A t ¢ ~ g n Sy,_~=.........................................................................................................................
..........................................
, R . ....... I #~

II~II ;~/~;, "-,~l~,L_._ i,~ ~,o~

1777;77 / vo~c~
Load Disturbunce Unit

Hydranl/c Tested Pump

Fig. l 0 Experimental setup

Ioo f" 10o

9O /t 9O
80 80

70 ~,i 7 = < . ~ " - - " -

~ ~°

N 5o
40
30
3O
20
10
• ..~.4-~- - - Theoretical 2O

10
0 ............... ......

10 20 30 ~0 ~0 60 70 ~0 90 100 0
Time (,ins) I0 20 30 40 50 60 G0 80 90 100
Time(m~)
Fig. 11 Swash plate time response with PD Fig.13 Swash platc time rcsponsc with fuzzy
controller, raising conditions controllcr,raisingconditions
~oO. 100

gO~ \ Theoretical ~o Theoretical

80 Expei:Jnelltal • ~ Expedmcum!
80 • k
¢~, 70X \ \ ) , i ~
~ so
° i
"~ 40
3O
20

10
10
10 20 30 40 50 60 70 80 90 100 o • ~ X - ~ - . . - ~ "~=2
.- "--~:~. ~
Time{ms) 1o ?o 30 .~o so 6o ~o so 9o 1oo
Time (ms)
Fig. 12 Swash plate time response with PD Fig. 14 Swash plate time response with fuzzy
controller, falling conditions controller, falling conditions

6 Copyright © 2002 by ASME

Conclusion
A mathematical model has been developed to
describe the dynamic performance of swash
plate axial piston pumps with conical cylinder
blocks. The model has been used to simulate the
dynamic performance of electrically controlled
constant power regulated pumps. The model
was validated on the basis of the good
agreement between the simulation and the
experimental results.
Using this model, fuzzy logic control is
proposed to replace the conventional PD one,
currently used in practical application. A
theoretical study was carried out to compare the
dynamic performance of the pump when it is
equipped with the newly developed fuzzy
controller, with the pump performance in case
of using the PD controller.
In view of the simulation results, it was found
that using such proposed fuzzy controller
resulted in smoother dynamic performance.
When fuzzy controller is used, the swash plate
moves with less speed of response and that the
steady state oscillation of the proportional valve
spool is eliminated.
Hydraulic test bed is interfaced with real time
control and data acquisition system to measure
the actual pump dynamic performance. The
good agreement between the theoretical and
experimental results validates the model.
The developed model, software simulating the
pump performance and real time control
software can be used at the design stage of any
swash plate axial piston pump of various
geometric volume.
References
1. S.A. Kassem and M. K. Bahr "Effect of Port
Plate Silencing Grooves on Performance of
Swash Plate Axial Piston Pumps", 7th
Mechanical Design and Production Congress
MDP7, Pergamon Press139-148, February
NOMENCLATURE
Symbol Description
Ad(As) Delivery (Suction) port area
Acp Area of the control piston
Ap Piston cross-section area.
lackJo K th piston absolute acceleration
B Effective bulk modulus
7 Copyright © 2002 by ASME
2000, Faculty of Engineering, Cairo University,
Cairo, Egypt.
2. S.A. Kassem and M. K. Bahr, "On the Dynamics
of Swash Plate Axial Piston Pumps with Conical
Cylinder Blocks", Sixth Triennial International
Symposium on Fluid Control Measurement and
Visualization Flucome2000, August 13-17, 2000,
Sherbrooke University, Sherbrooke, Canada.
S.A. Kassem and M.K. Bahr, "Fuzzy Logic
Control of Constant Power Regulated Swash
Plate Axial Piston Pumps", International
Mechanical Engineering Congress and
Exposition ASME-ME2001, Nov. 11-16, 2001,
New York, USA.
4. M. K. Bahr, J. Svoboda and R.B. Bhat,
"Vibration Analysis of Constant Power Regulated
Swash Plate Axial Piston Pumps" accepted for
publication in the Journal of Sound and
Vibration.
5. M. K. Bahr, J. Svoboda and R.B. Bhat,
"Response of Constant Power Regulated Swash
Plate Axial Piston Pumps to Harmonic and
Random Inputs", International Conference on
Multidisciplinary Design in Engineering, CSME-
MDE2001, November 21-22, 2001, Concordia
University, Montreal, Canada.
6. M. K. Bahr, J. Svoboda and R.B. Bhat, "Dynamic
Loads on the Drive Shaft Bearings of Swash
Plate Axial Piston Pumps with Conical Cylinder
Blocks" CSME Forum2002, May 21-24, 2002,
Queen's University, Toronto, Canada.
7. P. Kaliafetis, and T. Costopoulos, "Modeling and
Simulation of an Axial Piston Variable
Displacement Pump with Pressure Control"
Mechanical Design and Control Section &
Machine Design Laboratory, , Partition 42, 599-
612, 1994, National University of Athens,
Greece.
8. N.D. Manring, and R.E. Johnson, "Modeling and
Designing a Variable-Displacement Open-Loop
Pump", J. of Dynamic Systems, Measurement,
and Control, Vol. 118, 267-272, 1996.
Value Unit
m2
8.1x10 "4 m2
2.27 xl0 4 m2
lxl09 Pa
 Cd Coefficient of discharge. 0.611
Di/Ri Pitch circle diameter/radius of the cylinders' arrangement 0.07175/0.0359 m
at the base of the cylinder block
D2/R2 Pitch circle diameter/radius of the cylinders' arrangement 0. 0602/0.0301 m
at the top of the cylinder block
Fxk,yk,zk Components of the resultant force acting on the swash N
plate due to the k th piston
[Fpk]0/[Fk]o Pressure / resultant force vector due to single piston
fv Proportional valve viscous friction coefficient 90 N.s/m
f~ Control piston viscous friction coefficient 1.5 Nm/(rad/s)
Ie Equivalent moment of inertia of the swash plate .0039 kg.m 2
iv Proportional valve solenoid current A
k Piston number in the piston group arrangement
ki Proportional solenoid force-current constant 2.5 N/A
kv Proportional valve spring stiffness 20000 N/m
lq Control piston spring stiffness 72 Nm/rad
Ll/L2 Lengths, referred in Fig. 1 .0766/.0661 m
L3 Variable length. m
Le Cylinder length .0573 m
Lp Piston length .0591 m
mp Piston mass 0.118 kg
mv Proportional valve spool mass 0.1 kg
[Mk]0 Moment acting on the swash plate in vector form
Mx,y,z Components of the resultant moment acting on the swash plate N.m
N Number of pistons 9
Pe Pressure difference across the control piston Pa
Pcl,2 Pressure at the two sides of the control piston Pa
Pd Pump delivery pressure Pa
Pk Piston chamber pressure Pa
Ps Pump suction pressure .05x105 Pa
Pr Tank line pressure lxl05 Pa
Pv Control pressure Pa
Qa, b,c,d Flow rates through proportional valve ports m3/s
Qd Delivery flow of one cylinder m3/s
Qs Suction flow rate into one cylinder ma/s
RL Leakage resistance Ixl013 Pa/(m3/s)
[rek]0/[rek]6 Position vector indicates to the kth piston' center of mass
relative to inertial/6 th frame of reference
[rSk]0 Position vector indicates to the kth piston' spherical head
center relative to the inertial frame of reference
rs Radius of swash plate swinging .055 m
Sk Piston displacement m
Sv(max) Proportional valve spool displacement (maximum) (.001) m
t Time S
Vel,2 Control volume on the two sides of the control piston m3
Wci Initial control volume 13x106 m3
Vek Instantaneous cylinder volume of the kth piston m3
Vo Additional piston chamber volume l x l 0 -6 m3
W Proportional valve area factor 4.8x10 3 m
Xcp (min,max) Control piston displacement (minimum, maximum) (0, 0.015) m
Xsk, Ysk, Z5k Cartesian coordinates of piston spherical head center m
relative to the inertial frame of reference
Sk, Xcp, Sv Velocity of k th piston, control piston, valve spool m/s
d Swash plate angular velocity S-1
C~p) Swash plate angle of inclination (set point value)
Cylinder block cone angle 5 deg
Ok Angular position of the kth piston
P Oil density 850 kg/m3
~/~" Pump angular velocity/velocity S-1

8 Copyright © 2002 by ASME

View publication stats