Proceedings of the ASME/BATH 2014 Symposium on Fluid Power & Motion Control

FPMC2014
September 10-12, Bath, United Kingdom

FPMC2014-7856

EFFICIENCY TESTING OF AN ADJUSTABLE LINKAGE TRIPLEX PUMP

Shawn R. Wilhelm James D. Van De Ven
University of Minnesota University of Minnesota
Minneapolis, MN, USA Minneapolis, MN, USA

ABSTRACT swashplate determines the displacement of the pistons and is

Current state of the art variable displacement pumps suffer from varied to adjust the pump output.
poor efficiency at low volumetric displacement. Additionally, These pumps require a leakage path from the pumping
their performance is strongly dependent on operating conditions. cylinder to the swashplate in order to lubricate the piston-slipper
A new variable displacement linkage pump architecture has been joint and create the slipper-swashplate hydrostatic bearing. As a
developed which can achieve high efficiency across a wide range result, there are 3 leakage paths per piston with 9 or 11 pistons
of operating conditions. Previous work has described the used in a typical pump. The leakage energy loss is relatively
kinematics, energy loss modeling, and experimental validation constant with respect to displacement.
of a low power single cylinder prototype as well as the design of Furthermore, the hydrostatic bearing between the swash
a second generation prototype. The second generation pump plate and slippers results in viscous friction. The viscous friction
employs roller element bearings in its joints to minimize friction energy loss increases with the square of the relative velocity,
losses and the three cylinder design reduces the pressure ripple. which in a typical application is constant and independent of
In this paper, experimental characterization is presented of the displacement. Due to these constant energy loss terms, the
new 21 MPa, 6.75 cc/rev pump. High mechanical efficiency is efficiency of these pumps is poor at low output power [1].
achieved at low volumetric displacements at partial loads as low Additionally, because the working fluid is also the lubrication
as 0.6% of maximum power. Poorly performing cartridge check fluid, applications of the axial piston architecture are primarily
valves result in low volumetric efficiencies at low displacements. limited to non-corrosive fluids, unless exotic materials and
Close agreement was obtained between the model and predicted coatings are used [2].
work input using measured cylinder data as an input into the In an effort to design a pump which does not have the
model. This work shows that the presented pump with properly constant energy loss terms related to the piston-slipper interface,
functioning valves can achieve high efficiency across a wide an adjustable linkage based pump has been developed which
range of operating conditions. Having such versatile pump replaces the sliding joints with pin joints. Prior research has
performance can greatly improve the performance of hydraulic shown that a variable displacement linkage pump has the
systems as well as expand their potential applications. potential to have high efficiency across a wide range of
displacements [3].
INTRODUCTION The linkage, depicted in Figure 1, consists of a base fourbar
Variable displacement hydraulic pumps are used for crank-rocker mechanism with an additional connecting rod-
applications that require a more efficient method of flow control slider dyad, creating a sixbar mechanism. As the input crank
than metering valves. A variable displacement pump can deliver rotates, it causes the rocker to oscillate. A connecting rod joins
the power required by the system, rather than throttling the rocker link to the piston at a common pin, resulting in
unnecessary power across a valve, thus consuming less energy reciprocation. The piston reciprocates in a cylinder, which uses
to complete the same task. They are also used in pressure check valves to cause a pumping action. The adjustable ground
compensated systems in which the displacement control is used pivot the rocker is movable about a point, , associated with the
to maintain a constant system pressure. top dead center position of the piston. By moving the adjustable
The axial piston pump is the most widely used variable ground pivot about this point, the displacement of the piston can
displacement machine due to its compact size and robust design. be varied, while maintaining a constant top dead center.
This pump uses a swash plate set at an angle to the rotational axis Additionally, placing the adjustable ground pivot along the
of a cylinder block. As the cylinder block rotates, the piston sliding axis of the piston, results in zero piston displacement. A
slippers slide along the swashplate on a hydrostatic bearing and detailed description of the mechanism and associated kinematics
the relative angle causes them to reciprocate. The angle of the can be found in [4, 5].

1 Copyright © 2014 by ASME

 The constant top dead center position of the piston results in oscillating and rotary motion depending on the installed joint. A
a constant, small dead volume, minimizing compressibility crosshead bearing was not used in this design because the piston
losses at all displacements. This in contrast to other designs side loading was relatively small reducing the overall size. A
where the dead volume increases with decreasing displacement, previous work describes the design and optimization of the
resulting in higher compressibility losses at lower power outputs. machine [3].
The zero displacement capability makes this pump infinitely In this paper, the three-cylinder pump model will be
variable, allowing for continuous input shaft rotation with no experimentally validated and experimental efficiency data is
pumping output. The pin joints of the linkage allow for low presented. Testing equipment and valve limitations prevented
friction rolling element bearings, the losses of which scale operating the pump at full capacity, so low power data is
linearly with load and rotational speed. provided here. It will be shown that high mechanical efficiencies
can be achieved at low displacement. The paper is split into five
sections. First the energy loss model is presented. Second the
experimental methods are presented including the prototype and
experimental setup. Third, the results of the experiments are
given. The fourth section provides discussion and the fifth
section gives conclusions.

NOMENCLATURE

FIGURE 1 SCHEMATIC OF ADJUSTABLE LINKAGE SHOWING

THE ADJUSTABLE GROUND PIVOT

As a result of these features, the majority of the energy
losses of the adjustable linkage mechanism scale with
displacement and the mechanical efficiency remains relatively
constant for a majority of the displacement range. Additionally,
because hydrostatic bearings are not required, the lubricating
fluid can be separated from the pumping fluid using positive
seals, as is commonly practice for fixed displacement linkage
type pumps. The design can be used for a wide range of
applications and multiple leakage paths are eliminated. A model

validation for a low power, single cylinder pump using bronze
sleeve bearings was previously conducted [6]. This work showed
close agreement between model and experiments and concluded
that the efficiency could be greatly improved if roller element
bearings are used in the pump.
A new, three-cylinder prototype, which uses roller element
bearings, has been constructed for model validation. The pump
is designed to operate at 21 MPa and 3600 rpm, with a maximum
displacement of 6.75 cc/rev. Check valves were used for flow
control for design simplicity. The displacement mechanism is the β
focus of this paper as the volumetric performance depends

heavily on the performance of the valve mechanism which is
under separate development and requires special attention.
A multi-objective genetic algorithm was used to optimize
the dimensions of the linkage and pumping chamber to maximize
efficiency and minimize pin forces. The bearings were selected
for a 10 khr operating life at peak load conditions. This life was
calculated from bearing catalog design equations for both

2 Copyright © 2014 by ASME

 and the pin. The friction energy loss is calculated for each
bearing in the linkage and summed to determine the total roller
bearing energy loss of the mechanism. In addition to the
coulomb friction in the pin joints, there is additional friction
ENERGY LOSS MODELING
loss at the crosshead bearing, described by:
There are four energy losses modeled, including viscous
friction in the piston cylinder interface, coulomb friction in the
pin joints, leakage between the piston cylinder, and 5
compressibility losses due to the dead volume. For this model, it
is assumed that the losses related to eccentricity relative to the 6
cylinder are negligible as shown by Kumar [7], and as a result,
the piston-cylinder clearance is considered constant. Ridges are
added to the piston to aid in maintaining concentricity. The Piston-Cylinder Leakage Energy Losses
working fluid is considered Newtonian with fully developed, The gap between the piston and cylinder allows leakage flow as
laminar flow in the piston-cylinder gap. Isothermal operation is a result of a pressure drop across the piston. As described by
assumed. Additionally, input shaft seal friction and splash Cundiff, the leakage flow rate for a piston clearance seal is [9]:
lubrication energy are not considered. The check valves are not
modeled, as the work presented here pertains to the pumping 7
mechanism and the valves are considered outside the scope of 12
the current work.
The energy loss associated with this leakage is calculated by
Viscous Friction Energy Losses integrating the leakage flow with respect to time:
Viscous friction is cause by the relative motion between the
piston and cylinder, resulting in shearing of the fluid in the gap P 8
between the two. The viscous friction force is described by
Newton’s law of viscosity: Fluid Compressibility
The working fluid of the pump is compressible, as described by
1 the bulk modulus, . As the cylinder is presurized there is a
resulting compression of the working fluid, causing energy to be
stored and then released when cylinder pressure is reduced. The
where is the piston diameter, is the piston length, is the energy loss associated with fluid compressibility over a cycle is
dynamic viscosity of the fluid, h is the radial piston clearance, given as:
and is the relative velocity of the piston and cylinder. The
energy loss due to viscous friction is expressed as: Δ
9
β
2
where ∆ is the change in pressure, is the unswept
Coulomb Friction Energy Losses chamber volume, and is the bulk modulus of the fluid. If active
Rolling friction occurs in the bearings of the mechanism as the valves are used, the compression energy of the fluid in the
inner race rotates relative to the outer race. A Coulomb friction unswept volume results in an energy loss. Because passive
model is used to calculate the friction energy using an equivalent valves are used in the presented prototype, this energy is
coefficient of friction, . The friction energy loss over a cycle recoverable and not considered here. Additionally, because the
is found by integrating the friction torque with respect to the pump has a small, constant, unswept volume, this loss is
angle of rotation between the pin and the link: insignificant with respect to the other energy loss mechanisms
and can be considered negligible.
3 Figure 2 shows the modeled energy loss contibutions for the
optimized pump as a function of displacement for a single
4 cylinder at 21 MPa and 60 Hz operating frequency. The roller
bearing friction dominates. Due to the optimized piston-cylinder
gap height of 20 μm, the leakage and viscosity terms are fairly
where is the resultant friction torque which opposes the similar. The compressibility losses are negligible in comparison
direction of motion at the outer surface of the pin, is the to the other contributions.
magnitude of the force applied to the joint, is the equivalent
radius of the pin as described by Beardmore [8], is the
relative change in angle between the link joint and the pin,
and is the relative angular velocity between the link joint

3 Copyright © 2014 by ASME

 Loss Contributions at 21 MPa and 60 Hz
2
Energy Loss Per Rev per Cylinder (J)

Leakage
Viscous Friction
1.5 Bearing Friction
Compressibility

0.5

0
0.4 0.6 0.8 0 1 0.2
Displacement
FIGURE 2 MODEL PREDICTED ENERGY LOSS FIGURE 3 SCHEMATIC OF THE EXPERIMENTAL TEST SETUP
CONTRIBUTIONS OF A SINGLE CYLINDER AT 60 HZ AND 21
MPA OPERATION PLOTTED VS DISPLACEMENT In a given set of experiments, the shaft speed is set, followed
by pressure. Then the pump is set to maximum displacement. For
METHODS each test, the system is given 30 seconds to reach cyclic steady
The pump efficiency is determined using the experimental state, at which point the accumulator is not storing or supplying
setup depicted in Figure 3. The pump is driven by a 3-phase energy over the course of a cycle. Data is then collected for five
electric motor with a variable frequency drive to control shaft seconds and the displacement is then reduced to the next set
speed. A pilot operated relief valve provides both a pressure point. Ten displacements are measured per pressure and speed
control and a load. The displacement is measured with an setting. The working fluid is ISO grade 46 hydraulic fluid.is used
absolute encoder placed on the control link. A hydraulic actuator at a working fluid with the temperature maintained at 22-25° C.
is used to vary the pump displacement, and is powered by the As of the time of publication the pump has been tested to a
pumps own output through a directional control valve. The input maximum of 7 MPa. The pumping frequency was limited to 10
power is calculated from the shaft speed, which is measured Hz by the response time of the commercially available poppet
using an optical encoder, and the torque, which is measured style cartridge check valves used in the manifold block.
using an inline rotary torque transducer. Output power is Figure 4 is a plot of the pressure volume trace of a single
calculated using a pressure transducer at the pump outlet and a cylinder at 10 Hz, 6 MPa, and 60% displacement, demonstrating
gear flow meter. the effective portion of the stroke. The volume is calculated from
Three additional pressure transducers are used to monitor the measured piston position and the cylinder pressure is
each cylinder for evaluating pumping performance and measured with the cylinder pressure transducer. The lag in
determining piston loads for model comparison. The optical opening of the check valve is demonstrated by the slope of the
encoder was also used to estimate piston position by measuring rise and fall of pressure and is an indication of the valve
the encoder count relative to the piston position. For calibration, performance. Ideally, the rise and fall for an incompressible fluid
the pump was partially disassembled to directly measure the would be vertical lines. Due to the slow performance of the
piston position using an LVDT relative to the position of the valves, the ideal valve model could not be used for model
optical encoder on the input shaft. predictions.
A 1 liter accumulator pre-charged to 1.7 MPa is connected
to the outlet of the pump, before the flow meter to reduce flow
ripple. Eaton CV3-10-003 poppet check valves were used on the
pumping chamber outlets and disk style check valves from a
commercial pump were used on the inlets. The disk check valves
were selected for the inlet due to their fast response time and low
cracking pressure.

4 Copyright © 2014 by ASME

 Input Torque
7
Experimemtal Data
Model Prediction
6

Torque Nm
5

2
0.5 0.55 0.6 0.65 0.7
Time(s)
FIGURE 5 MODEL PREDICTED VS EXPERIMENTALLY
MEASURED INPUT TORQUE USING CYLINDER PRESSURE AS
AN INPUT TO THE MODEL AT 10 HZ 6 MPA AND 60%
DISPLACEMENT
FIGURE 4 EXPERIMENTALLY MEASURED PRESSURE VS.
VOLUME DIAGRAM OF PUMPING CYLINDER WITH ARROWS
INDICATING CYCLE DIRECTION RESULTS
The mechanical efficiency of the pump was greater than
The measured system pressure, flow rate, input torque, and 90% at all measured displacements at pressures above 5 MPa and
shaft speed are used to directly calculate mechanical, volumetric, shaft speeds above 600 RPM. The highest mechanical efficiency
and total efficiencies. Mechanical efficiency is defined as: measured was 98% at 7 MPa and 40% displacement. These data
are shown in Figure 6, which is a plot of mechanical efficiency
as a function of displacement, with multiple traces to represent
10 different operating pressures. The mechanical efficiency trends
2
upwards with increased pressure while the volumetric efficiency
where is the average system pressure, is the total volume trends down with increased pressure as shown in Figure 7.
swept by the three pistons per revolution, is the average input Mechanical Efficiency at 10 Hz
1
torque, and ω is the average shaft speed.
The volumetric efficiency is defined as:
0.9
2
11
0.8
20 Bar
30 Bar
where is the averaged flow rate. All measured values are
m

0.7 40 Bar
averaged over the five second experiment. Total efficiency is 50 Bar
defined as: 60 Bar
0.6
70 Bar
12 0.5

0.4
0.2 0.4 0.6 0.8 1
For model comparison, the experimental cylinder pressure Displacement (D/Dmax)
and piston position are used as inputs to the kinematic model
presented in [6], including the losses mentioned in the modeling FIGURE 6 EXPERIMENTALLY MEASURED MECHANICAL
section. Figure 5 shows a plot of the experimentally measured EFFICIENCY PLOTTED AGAINST DISPLACEMENT AT 10 HZ
torque compared to the model predicted torque at 10 Hz, 6 MPa, OPERATING SPEED AT VARIOUS OPERATING PRESSURES
and 60% displacement. The torque predicted by the model agrees
well with the experimental data.

5 Copyright © 2014 by ASME

 Volumetric Efficiency at 10Hz valve timing into account across a broad range of operating
1 conditions. This close agreement shows that the model can
predict well how the mechanism will perform as a function of
0.8 the pumping head behavior.
Having a pump with such high efficiency at low
displacements is an exciting prospect for current hydraulic
20 Bar
0.6 applications. Additionally, because the pump allows for the
30 Bar
separation of the working fluid from the lubrication fluid, the
v

40 Bar
50 Bar
potential for new applications is great.
0.4
60 Bar
According to a report by Oak Ridge National Labs, the
70 Bar average efficiency of industrial hydraulics systems is ~50%,
0.2 consuming 3% of the total energy use in the United States. They
concluded that the inefficiencies were largely a result of the use
of metering valves [10]. These losses can be greatly reduced by
0
0.2 0.4 0.6 0.8 1
using variable displacement flow control.
Displacement (D/Dmax) For many industrial applications, traditional variable
displacement pumps cannot be used to control flow rates because
FIGURE 7 EXPERIMENTALLY MEASURED VOLUMETRIC the working fluid is not lubricating. Variable frequency drives
EFFICIENCY PLOTTED AGAINST DISPLACEMENT AT 10 HZ
OPERATING SPEED AT VARIOUS OPERATING PRESSURES
are used in some cases, but they add extra components to the
system and can cause premature wear in electric motors. [11].
As seen in Figure 8, when the experimentally measured piston However, this new pump can be used in such cases to eliminate
position, cylinder pressure, and shaft speed are passed as inputs metering valves, and vastly improve the efficiency of these
to the model, the input shaft work well is well predicted. systems.
Work Input Comparison at 10 Hz Additionally, this pump could be used to reduce energy
2000 consumption and allow for non-lubricating working fluid in
40 Bar Experiment applications which currently use variable displacement pumps.
1800
40 Bar Model For example, an injection molding machine could use water
1600 50 Bar Experiment hydraulics which would aid in heat transfer while consuming less
50 Bar Model energy.
1400 60 Bar Experiment Further work is required to increase the power density of this
1200
60 Bar Model new pump design to make it more viable for mobile applications
Energy (J)

70 Bar Experiment which have strict size restrictions.

1000 70 Bar Model In the future, a new power supply will be used to evaluate
800 the pump at its full operating range to further validate the
capabilities. Additional work is required to create high speed,
600 low inertia check valves which will allow higher frequency
400 operations and prevent reverse flow at low displacements.

200 CONCLUSIONS
0
A prototype adjustable linkage pump, which uses roller
0 20 40 60 80 100 element bearings in its joints, demonstrated mechanical
Displacement (%) efficiencies greater that 90% as partial loads as low as 0.6%. The
FIGURE 8 COMPARISON OF MODEL PREDICTED AND
volumetric efficiency of the presented prototype was limited by
EXPERIMNETALLY MEASURED INPUT WORK USING
EXPERIMENTAL CYLINDER DATA AS AN INPUT TO THE poorly performing check valves, but the model was still able to
MODEL predict the mechanisms behavior.
The variable displacement linkage pump is able to eliminate
DISCUSSION the leakage and friction losses associated with the hydrostatic
Slow check valves prevented the pump from producing the bearings by using all rolling element bearings. This paper
expected total efficiencies and reaching higher operating speeds. validates the energy loss models of the bearing friction and
This is shown in the volumetric efficiency measurement. leakage. Energy losses associated with bearings were reduced by
However, the mechanical efficiency was high even at low approximately ten times when compared to the previously
displacements. These measurements are taken at partial loads of presented bronze bushing design, proving their added value.
0.3-6% of the design power, which is encouraging for the Work is already underway to incorporate high speed check
performance at higher power levels. Additionally, the valves into the design allowing for this pump to achieve its full
experimental data matches closely with the model when taking potential. In the future, active valves will also be incorporated to
allow for motoring operation.

6 Copyright © 2014 by ASME

ACKNOWLEDGMENTS
This work is supported by the National Science Foundation
under grant number EFRI-1038294.

REFERENCES
[1] Wilson, W., 1946, "Rotary-pump theory," Trans. ASME,
68(4), 371-384.
[2] Haynes, J. M., 2007, "Axial piston pump leakage modelling
and measurement," PhD, Cardiff University.
[3] Wilhelm, S. R., and Van de Ven, J. D., 2014, "Design of a
Variable Displacement Triplex Pump," International Fluid Power
ExpositionLas Vegas, NV.
[4] Wilhelm, S. R., and Van de Ven, J. D., 2013, "Design and
Testing of an Adjustable Linkage for a Variable Displacement
Pump," Journal of Mechanisms and Robotics, 5(4), 041008.
[5] Wilhelm, S., and Van de Ven, J. D., 2011, "Synthesis of a
Variable Displacement Linkage for a Hydraulic Transformer,"
International Design Engineering Technical Conferences &
Computers and Information in Engineering Conference, ASME,
Washington, DC, 8.
[6] Wilhelm, S. R., and Van de Ven, J. D., 2013, "Efficiency
Modeling and Experimental Validation of a Variable
Displacement Linkage Pump," ASME/BATH Symposium on
Fluid Power and Motion Control, ASME, Sarasota, FL.
[7] Kumar, S., 2010, CFD Analysis of an Axial Piston Pump,
PhD Thesis from UPC, Spain.
[8] Beardmore, R., 2010, "Roller Bearing Friction,"
http://www.roymech.co.uk/Useful_Tables/Tribology/Bearing%
20Friction.html.
[9] Cundiff, J. S., 2002, Fluid Power Circuits and Controls:
Fundamentals and Applications, CRC Press.
[10] Love, L., Lanke, E., and Alles, P., 2012, "Estimating the
Impact (Energy, Emissions and Economics) of the US Fluid
Power Industry," Oak Ridge National Laboratory (ORNL).
[11] 2000, "Variable Frequency Drives," No. 00-054, Easton
Consultants, Inc. .

7 Copyright © 2014 by ASME