International Journal of Fluid Power, 2015

Vol. 16, No. 1, 35–51, http://dx.doi.org/10.1080/14399776.2015.1016839

Port and case flow temperature prediction for axial piston machines
L. Shang* and M. Ivantysynova
Department of Agricultural and Biological Engineering, Purdue University, 225 South University Street, West Lafayette,
IN 47907-2938, USA
(Received 22 August 2014; accepted 5 February 2015)

Researchers at Purdue’s Maha Fluid Power Research Center have developed models that will enable computational
design of piston machines. The core of the in-house developed program forms multi-domain models capturing the fluid–
structure interaction phenomena taking place in the main lubricating interfaces (piston/cylinder, cylinder block/valve
plate, and slipper/swash plate) of axial piston machines. The model allows studying the influence of a given pump or
motor design on machine performance, power loss, and energy dissipation in those main lubricating interfaces. The
behavior of the fluid film in these lubricating interfaces as well as the shape of the solid parts is temperature and pres-
sure dependent. In order to solve for non-isothermal flow and to consider elasto-hydrodynamic effects, port and case
temperatures are needed as boundary condition for the model. In case of analysis and optimization of existing pumps
and motors, those boundary conditions can be taken from steady-state measurements; however, when using the model to
design a new unit, this information is not available. The temperature prediction model proposed in this paper fills this
gap. The model can predict the outlet and case temperature for a chosen inlet temperature based on known fluid proper-
ties and calculated energy dissipation in the rotating group of an axial piston pump. The model also considers the tem-
perature change due to fluid compression/expansion and estimated churning losses for a given axial piston machine.
Keywords: axial piston machines; lubrication; computational pump design; thermodynamic model; heat-transfer model;
outlet temperature; case temperature

1. Introduction However, the FSTI model can also be used to sup-

After decades of research on studying the fluid film port the design of new pumps and motors. At this point,
behavior in the main tribological interfaces of axial pis- measurement data are not available. Therefore, the
ton machines, the authors’ research team proposed a authors’ research team started the development of a port
fluid–structure and thermal interaction model (FSTI) for and case flow temperature prediction model few years
the three main lubricating interfaces of swash plate-type ago. A flow temperature prediction model was proposed
axial piston machine (piston/cylinder, slipper/swash by Grönberg (2011). This model was further developed
plate, and cylinder block/valve plate interfaces), details by Zecchi et al. (2013) A drawback of this model is
can be found in Pelosi and Ivantysynova (2012, 2013), its sensitivity to the user’s inputs, especially to the
Zecchi and Ivantysynova (2013) and Schenk and heat-transfer area and the heat-transfer coefficients. A
Ivantysynova (2014). The FSTI model allows predicting slightly inaccurate user’s inputs will lead to a large error
fluid film behavior in three lubricating interfaces based in the case flow and port flow temperature prediction. As
on non-isothermal flow conditions, and macro- and a consequence, the FSTI model will predict very
micro-motion of parts as well as surface deformation due different fluid film behaviors for different given combi-
to pressure and thermal loading. nations of inlet, outlet, and case flow temperature. In real
In order to solve for non-isothermal flow in lubricat- world, there is only one true combination of these tem-
ing interfaces and to consider surface deformation due to peratures for a given pump design, given fluid, and given
thermal loading, a detailed heat-transfer model of the operating conditions. The authors conducted a simulation
entire axial piston machine is included in the FSTI study with the FSTI model to investigate the sensitivity
model (Pelosi 2012; Zecchi 2013). The three-dimen- of these boundary temperatures on the fluid film
sional (3D) heat-transfer model included in FSTI requires behaviors in the three lubricating interfaces. Figure 1
boundary temperatures to be solved numerically. These shows a comparison of power loss and leakage obtained
boundary conditions are inlet port temperature, outlet for the same pump design and same fluid when varying
port temperature, and case flow temperature. the outlet and case temperature. The power loss and
When the FSTI model is used to analyze or optimize leakage values are outputs of the FSTI simulation
an existing pump or motor, the required boundary tem- model. The FSTI simulations have been conducted for
peratures can be taken from steady-state measurement the same pump at same inlet temperature of the fluid.
conducted on the pump or motor of interest. But outlet port temperature and case temperature were

*Corresponding author. Email: [email protected]

© 2015 Taylor & Francis

36 L. Shang and M. Ivantysynova

For this reason, in this paper, the authors presented

generic and in some way rather simplified thermody-
namic pump model. The goal was to develop a model
that calculates outlet port and case flow temperatures
based on known fluid properties, inlet temperature, and
estimated energy dissipation/power loss for a given
pump design. In order to allow the use of the model for
different pump and motors sizes, linear scaling laws,
which are used for pump design by most of the pump
and motor manufactures, are incorporated in the pro-
posed port and case temperature prediction model. In
order to estimate the energy dissipation in lubricating
gaps, a co-simulation between the FSTI model and the
case and temperature prediction model is required. The
required procedure will be explained in the next chapter.

2. Pump port and case temperatures prediction

model
It is known, since the work of Witt (1974), that the over-
all efficiency of hydrostatic pumps and motors can be
determined with temperature and pressure measurement
combination and an accurate description of the fluid
properties. The determination of overall efficiency based
on thermodynamics requires the determination of
enthalpy difference between inlet and outlet. While Witt
calculates pump loss from measured flow temperature
and known fluid properties, the authors of this paper
Figure 1. FSTI simulation results.
wish to determine port and case flow temperature from
given loss of the pump or motor and known fluid
modified. Table 1 shows the inlet flow temperature, the
properties.
outlet flow temperature, and the case flow temperature
For the presented model, the authors propose to
for these three simulations, labeled HT (higher tempera-
divide the fluid volume of a pump or motor into four
ture), MT (measured temperature), and LT (lower
volumes as shown in Figure 2. The volumes are divided
temperature).
in case volume, displacement chamber volume, inlet port
The simulation results shown in Figure 1 demonstrate
volume, and outlet port volume. The heat and mass
the significance of pump port and case flow temperature
transfer in an axial piston machine under steady-state
for a reliable prediction of fluid film behaviors in these
condition is illustrated schematically in Figure 2. The
important pump interfaces. The upper figure shows the
inlet mass flow rate m_ in enters the unit through the inlet
power losses generated in the three lubricating interfaces
port volume. Part of the inlet mass flow rate transfers
and the sum of the three marked as ‘total power loss.’
through the displacement chamber volume to the outlet
An obvious reduction on power loss can be found with
decreasing temperature. As can be seen from Figure 1,
the outlet port and case temperatures have also a large
impact on the predicted leakage. From this simulation
study, it is obvious that the prediction of lubricating
interfaces’ behavior including energy dissipation due to
viscous flow requires very good estimate of port and
case flow temperatures.

Table 1. Input temperatures for three sets of simulations.

Figure 2. Definition of the control volumes.

 International Journal of Fluid Power 37

port volume, and leaves the unit as the outlet mass flow
rate m_ out , the remaining part of the inlet mass flow rate
transfers through the displacement chamber volume to
the case volume, and leaves the unit as external leakage
mass flow rate m_ l;e . The internal leakage mass flow rate
m_ l;i transfers from the outlet port volume through the
displacement volume into the inlet port volume. Due to
the conservation of mass flow rate, the internal leakage
mass flow rate is added to the mass flow rate from the
inlet port to the displacement chamber, and the mass
flow from the displacement chamber to the outlet port.
The heat-transfer in an axial piston machine running
under steady-state condition is illustrated schematically
in Figure 3. Three red arrows in the housing block repre-
Figure 4. Simplified schematic of the heat and mass transfer
sent the convective heat-transfer between the displace-
in the axial piston machine.
ment volume and the solid parts, the case volume and
the solid parts, and the case volume and the inner hous-
ing surface. The red arrow on the outside of the housing heat-transfer between the solid parts and the displace-
block represents the natural heat-transfer between the ment chamber volume contributes to the temperature var-
housing outer surface and the ambient air. The white iation as well. The red dashed line in Figure 5 connects
arrow represents the radiation. the inlet and outlet state in the enthalpy–entropy diagram
As it is shown in Figure 3, the internal leakage mass of the hydraulic fluid based on the measured pressure
flow rate forms two closed loops between the inlet port and temperature. Due to the heat-transfer, the real com-
volume, the displacement chamber volume, and the out- pression process does not follow a vertical line. Since in
let port volume. The internal mass flow rate has limited a real axial piston machine, the fluid pressure changes
impact on the pump flow temperature because of these only over a very short time period and the heat-transfer
two closed loops. Thus, in order to simplify the pump occurs under constant pressure mostly, the real compres-
flow temperatures prediction model, the internal leakage sion process has been divided into two vectors as shown
mass flow rate is neglected. The inlet port volume and in Figure 5. The vertical black vector represents an adia-
the outlet port volume are merged into the displacement batic compression, and the green vector represents the
chamber volume as shown in the Figure 4. temperature variation due to the heat-transfer at constant
The convective heat-transfer between the case vol- pressure. This allows separate study on the temperature
ume and the inner surface of the housing is neglected variation due to the pressure change and due to the heat-
due to the small temperature difference. Also, radiation transfer.
is neglected due to its limited impact on the port and The case temperature variation can also be divided
case temperatures. All the simplifications mentioned into the adiabatic compression or expansion and the
above are shown in Figure 4. heat-transfer.
Figure 4 shows that the pressure in and out of the Therefore, the proposed pump outlet port and
displacement chamber volume changes from pin to pout, case flow temperatures prediction model includes two
which contributes to the temperature variation due to separated modules, the thermodynamic module, and the
compression or expansion, respectively. The convective

Figure 3. Schematic of the heat and mass transfer in the axial

piston machine. Figure 5. Compression process in the displacement chamber.
38 L. Shang and M. Ivantysynova

heat-transfer module, considering the temperature
!

@v
variation due to the pressure change. dh ¼ cp  dT þ vT   dp (2.8)
@T p

Substituting Equation (2.7) into Equation (2.4) gives:

2.1. Thermodynamic module

dT @v
The thermodynamic module predicts the temperature var- ds ¼ cp   dp (2.9)
iation due to the pressure changes assuming adiabatic T @T p
compression and expansion. The enthalpy and entropy of Integrating Equations (2.8) and (2.9) at To and po
the compressible fluid need to be solved as a necessary yields:
step.

The specific enthalpy and entropy of compressible RT Rp Rp @v
h ¼ ho ðTo Þ þ cp  dT þ v  dp  T  dp
fluid can be described as functions of temperature and To p0 p0 @T p
pressure:

RT c Rp @v
s ¼ so ðTo Þ þ Tp  dT  dp
h ¼ hðT ; pÞ To p0 @T p
(2.1)
s ¼ sðT ; pÞ (2.10)
The total differential of the enthalpy and the entropy Based on the adiabatic compression assumption, the
can be written as: entropy of the fluid before and after the compression or



@h @h the expansion is the same:
dh ¼ dT þ dp
@T p @p


T (2.2)
 s1 ¼ sT2

@s @s RT1 cp Rp1 @v R2 cp Rp2 @v
ds ¼ dT þ dp  dT  dp ¼  dT  dp
@T p @p T To T p0 @T p To T p0 @T p
The derivations of the caloric constitutive equations (2.11)
yield: The temperature of the fluid after the adiabatic com-

 pression or expansion in both displacement and case vol-
@h
cp ¼ (2.3) ume, Tadia_DC and Tadia_Case, can be solved using
@T p
Equation (2.11).
Using the Maxwell relation: Figure 6 shows the enthalpy–entropy diagram of the


 HLP-32 which was calculated from the thermodynamic
@s @v
ds ¼ dT  dp (2.4) module. The enthalpy–entropy diagram in Figure 6 for
@T p @T p HLP-32 shows very comparable values to the published
where v is the specific volume of the fluid and substitut- data by Oppermann (2006).
ing dh and ds into ds = T ⋅ ds + v ⋅ dp
we obtain:
2.2. Heat-transfer module
"

 # "

#
@h @v @s @h The heat-transfer module corrects the result of thermody-
þT v dp ¼ T  dT
@p T @T p @T p @T p namic module using heat-transfer calculation.

(2.5)
At constant temperature:
"

 #
@h @v
þT v dp ¼ 0
@p T @T p


 (2.6)
@h @v
¼vT
@p T @T p
At constant pressure:
"

 #
@s @h
0¼ T  dT
@T p @T p


 (2.7)
@s @h
¼1 ¼ 1 cp
@T p T @T p T

Combining Equations (2.3) and (2.6) and substituting Figure 6. Enthalpy–entropy diagram of HLP-32 calculated
into Equation (2.2), we obtain: from the thermodynamic module.
 International Journal of Fluid Power 39

The heat-transfer module, which forms a part of the the equilibrium of rate of heat flow in the volume of the
model presented in this paper, is based on a large experi- solid parts, we can write:
mental study involving different axial piston designs and
q_ loss ¼ q_ s DC þ q_ s Case (2.12)
pump/motor sizes, and it represents a rather empirical
model. The main source of power loss in axial piston According to the equilibrium of rate of heat flow in
machines is the three lubricating interfaces. When study- the displacement chamber volume, we obtain:
ing the behavior of these lubricating interfaces, it can be
q_ DC ¼ q_ s DC (2.13)
shown that the energy dissipation in the lubricating film
is not evenly distributed. Eccentric loading of the sur- Finally, the equilibrium of rate of heat flow in the
faces together with surface deformation due to pressure case volume yields::
and thermal loads leads to usually rather small areas of
q_ Case ¼ q_ s Case (2.14)
very small film thickness with very high shearing and
locally very high energy dissipation. The heat generated The rate of heat flow leaving the displacement cham-
in those small areas of the lubricating gaps transfers into ber with the outlet flow rate is calculated from the outlet
case volume and outlet port through a rather complex mass flow rate, the temperature differential, and the fluid
heat-transfer combining convection and conduction. specific heat capacity:
Because of the very small film thickness in those small ZTout
areas, the convection effect is limited through the limited q_ DC ¼ m_ out  cp dT (2.15)
leakage flow. Tadia DC
Also, a study of temperature distribution in the solid
parts forming the lubricating interfaces was conducted where Tadia_DC is the fluid temperature after adiabatic
by Zecchi (2013). This study shows that the temperatures compression or expansion in the displacement chamber
in the solid parts are locally higher than both the case volume calculated in the thermodynamic module.
flow temperature (sometimes in order of 20 °C) and out- The rate of heat flow leaving the case volume with
let flow temperature (sometimes in order of 50 °C), the case flow rate and due to the natural heat-transfer
which indicates that a good amount of heat is transferred flow into the air is calculated from the case mass flow
from the solid parts into both the case and displacement rate, temperature differential, fluid specific heat capacity,
chamber. Temperature distribution measurements con- the air temperature, and the overall natural heat-transfer
ducted on the rotating cylinder block of a stock swash coefficient of the pump outer housing surface:
plate axial piston machine by Olems (2001) showed also Z
TCase
areas of locally higher temperatures than the oil in the q_ Case ¼ m_ Case  cp dT þ knatural  ðTCase  Tair Þ
displacement chamber and pump case.
Tadia Case
Based on these observations from conducted mea-
surements on stock axial piston machines, the following (2.16)
simplified assumption was introduced in the heat-transfer where Tadia_case is the fluid temperature after adiabatic
model shown in Figure 7. The rate of heat flow q_ loss compression in case volume calculated in the thermody-
generated by the pump power loss is assumed to be namic module and knatural is the overall natural heat-
transferred through conduction into the solid parts and transfer coefficient of the pump housing surface.
from the solid parts into both the displacement chamber The value of knatural is proportional to the housing
volume and the case volume at the same time. surface, which is proportional to the second order of the
The simplified three volume model is shown in linear scaling factor k as usually used for pump design:
Figure 7, where q_ loss represents the total power loss of knatural ¼ Cnatural ref  k2 ½W=K (2.17)
the pump, q_ s DC and q_ s Case represent the rate of heat
flow from the solid parts into the displacement chamber where Cnatural_ref is a constant value determined based on
volume and the case volume, respectively. According to steady-state measurements of a baseline unit. The term
k2 scales the overall natural heat-transfer coefficient with
the heat-transfer surface area.
The linear scaling factor k has been introduced to
allow for general use of the model, i.e. without the need
of steady-state measurement data for each unit size.
The linear scaling factor can be obtained as:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Vi
k¼3 (2.18)
Vi baseline
where Vi is the displacement volume of the pump or
motor under investigation and Vi baseline represents the
displacement volume of the baseline unit the steady-state
Figure 7. Three volumes heat-transfer model. measurements were taken from.
40 L. Shang and M. Ivantysynova

The rate of heat flow from solid parts to both the dis- square root are used here to sum this two velocity vec-
placement chamber volume and the case volume can be tors together assuming they are perpendicular to each
calculated as: other.
There are four heat-transfer constants, Cnatural_ref,
q_ s DC ¼ kDC  ðTsolid  TDC Þ
(2.19) CDC_ref, CCase_ref, and Clcf, needed to complete this
q_ s Case ¼ kCase  ðTsolid  TCase Þ
model.
where kDC and kCase are the overall heat-transfer coeffi- Clcf is determined from the geometry of the pump:
cients of the displacement chamber surface and the case
p  Ablock  t
volume surface, TDC is the temperature in the displace- Clcf ¼ (2.23)
ment chamber which approximately equals to the aver- 2
age of the inlet flow temperature and the outlet flow where Ablock represents cylinder block outer surface area
temperature. and t is the distance between the cylinder block outer
The value of kDC is a function of Reynolds number surface and the housing inner surface.
on the heat-transfer surface in the displacement chamber Cnatural_ref is determined from the geometry of the
which is proportional to the outlet flow rate and the lin- pump and the air-metal natural convection heat-transfer
ear scaling factor: coefficient of the outer housing surface:

 Cnatural ¼ hc  Ahousing (2.24)
Qout 0:4 2 ref air
kDC ¼ k  CDC ref (2.20)
k where hc_air is a the air-metal natural convection heat-
where CDC_ref is a constant value determined based on transfer coefficient and Ahousing is the outer housing
the steady-state measurement, the scaling reference, and surface area.
fluid properties. The term Qkout scales the overall heat- CDC_ref and CCase_ref are found by conducting optimi-
transfer coefficient of displacement chamber surface with zation on the discrepancies between the simulated pump
the Reynolds number; the order value 0.4 is found to flow temperatures and the steady-state measured pump
match the measurements the best, and the term k2 scales flow temperatures.
the overall heat-transfer coefficient with the heat-transfer Those resulting constants can be used for any given
surface area. size of swash plate-type axial piston pump or motor by
The value of kCase is a function Reynolds number on using the linear scaling law.
the heat-transfer surface in the case which is mainly pro- Concluding the rate of heat flow balance in the pump
portional to the pump speed and the linear scaling factor. heat-transfer model:
The leakage flow rate correction factor (lcf) is used here q_ s DC ¼ q_ s DC ðTsolid ; Tout Þ
to add the influence of the leakage flow rate on the Rey- q_ s Case ¼ q_ s Case ðTsolid ; TCase Þ
(2.25)
nolds number: q_ DC ¼ q_ DC ðTsolid ; Tout Þ
 0:4 2 q_ Case ¼ q_ Case ðTsolid ; TCase Þ
kCase ¼ n  k2  lcf k  CCase ref (2.21)
There are three heat flow rate balance equations:
where CCase_ref is a constant value determined based on 8
the steady-state measurement, the scaling reference and < q_ loss ¼ q_ s DC þ q_ s Case
fluid property. The term n  k2  lcf scales the overall q_ DC ¼ q_ s DC (2.26)
:
heat-transfer coefficient of case volume surface with q_ Case ¼ q_ s Case
Reynolds number; the order value 0.4 is found that with three unknowns Tout, TCase, and Tsolid.
match the measurements the best, and the term k2 scales
the overall heat-transfer coefficient with the heat-transfer
surface area. lcf is a leakage flow rate correction factor 2.3. Pump port and case flow temperatures prediction
which calculates the leakage flow rate influence on the model inputs
Reynolds number in the case volume. lcf is a function of The thermal model requires two types of inputs, the fluid
the pump speed, the case flow rate, and the linear scaling properties, and the operating conditions.
factor: The required fluid properties include the density, the
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

ffi
QCase  Clcf 2 specific heat capacity, and the heat-transfer constants
lcf ¼ 1 þ (2.22) which were discussed in the previous section. Note the
k3  n heat-transfer constants have been found based on steady-
where Clcf is a constant value based on the scaling refer- state measurements of a baseline unit. Linear scaling fac-
Clcf
ence. The term QCase
k3 n
is the ratio of the fluid velocity tors are needed to apply them for different unit sizes.
contributed by the leakage flow rate over the fluid veloc- Table 2 summarizes the required fluid properties:
ity contributed by the rotation of the rotating group. By The required operating conditions include the pump
assuming the fluid velocity contributed by the rotational size, the pump speed, the inlet pressure, the outlet pres-
motion of the rotating group is perpendicular to the fluid sure, the case pressure, the inlet temperature, the ambient
velocity contributed by the leakage flow, a square and a temperature, the outlet flow rate, the case flow rate, and
 International Journal of Fluid Power 41

Table 2. Required fluid properties. temperature obtained from the pump port and case flow
temperatures prediction model at each iteration step, until
Fluid Properties Comments
both the outlet flow temperature and the case flow
Density ρ Function of pressure and temperature converge.
temperature The pump power loss can be estimated using an
Specific heat cp Function of temperature
extended version of Maha’s FSTI simulation program.
capacity
The FSTI model calculates the energy dissipation in the
three main lubricating gaps of a swash plate-type axial
piston machine. As already explained in the introduc-
the total power loss. Note that the pump size is used to tion of this paper, the precise calculation of energy dis-
calculate the linear scaling factor, and the displacement sipation in the three lubricating interfaces of swash
is reflected in the outlet flow rate. The outlet flow rate, plate-type axial piston machines requires port and case
the case flow rate, and the total power loss cannot be temperatures as inputs in order to use them as thermal
obtained directly from the pump design; hence, the FSTI boundaries to solve for non-isothermal flow in the fluid
simulation model has to be used to estimate these values. film and also to consider the deformation of solid parts
An iterative process is required between solving the pro- due to thermal loading and its impact on fluid film
posed outlet and case temperature model and the FSTI thickness and energy dissipation in these three lubricat-
simulation model to determine the remaining parameters, ing interfaces. More details about this fluid–structure
as shown in Figure 9. Table 3 summarizes the required interaction model and the in-house developed program
operating conditions: can be found in Pelosi and Ivantysynova (2012), Zecchi
and Ivantysynova (2013) and Schenk and Ivantysynova
(2014). The FSTI model has been extended with a
2.4. Pump outlet port and case flow temperatures ‘churning loss model’ and a ‘remaining loss estimator
prediction model solution scheme model.’ The remaining loss estimator model is used to
estimate the losses created in the shaft bearing and the
The pump flow temperatures prediction model calculates shaft sealing.
the outlet port temperature and the case flow temperature In order to estimate the total power loss of a given
for the given pump size at given operating conditions, swash plate axial piston machine, a co-simulation
for a given fluid, and the estimated power loss of the between the extended FSTI program and the port and
pump or motor. The thermodynamic module calculates case temperature prediction program is required. The
the outlet flow temperature variation and the case iterative simulation scheme of the coupled models is
temperature variation by assuming adiabatic compression shown in Figure 9. The extended FSTI model can be
or expansion as the first step. In a second step, the used to calculate the outlet flow rate, the case flow rate,
heat-transfer module is used to correct the outlet flow and the total power losses for a given pump design,
temperature and the case flow temperature considering given inlet port temperature, given fluid properties, and
the heat-transfer. estimated outlet port and case flow temperatures. In a
However, the outlet flow temperature and the case second step, the pump outlet port and case flow tempera-
flow temperature are needed to specify the fluid proper- ture prediction model is used to calculate the outlet port
ties used in the pump flow temperatures prediction temperature and the case flow temperature based on the
model. The pump flow temperatures prediction model power loss and the outlet and case flow rate from the
solution scheme is shown in Figure 8. The fluid proper- FSTI model. In a third step, the obtained outlet port and
ties which are functions of temperature are initialized by case flow temperature will be used to rerun the extended
applying the inlet flow temperature, and updated by FSTI model to update the power loss and the outlet and
applying the outlet flow temperature and the case flow case flow rate. The described iteration cycle between the
two models will be repeated until the outlet and case
flow temperature converge.
Table 3. Required operating conditions.

Operating conditions Comments 3. Comparison study to measurements

Pump size Vi Known In order to verify the accuracy of the proposed pump
Pump speed n Known outlet port and case flow temperatures prediction model,
Inlet pressure pin Known four sets of measurement data have been used for com-
Outlet pressure pout Known
Case pressure pCase Known parison, which include four different pump designs with
Inlet temperature Tin Known different maximum displacement volumes and working
Ambient temperature Tair Known with different fluids and at different pressures, speeds,
Outlet flow rate Qout FSTI and at different swash plate angles.
Case flow rate QCase FSTI All the measurement data which were used for this
Power loss q_ loss FSTI
study are summarized in Table 4:
42 L. Shang and M. Ivantysynova

Figure 8. Pump flow temperatures prediction model solution scheme.

Figure 9. Pump flow temperatures prediction model coupled with FSTI.

Table 4. Measurement sets for comparison.

Pump size Displacement Pump speed Pressure differential

Set # Fluid type [cc] [%] [rpm] [bar]
1 Type I 42 100, 80, 60, 40, 20 200, 500, 1000, 2000, 3000, 4000, 4600 20, 50, 100, 200, 300, 400
2 Type I 130 100, 50, 20 1000, 2000, 2800 50, 100, 200, 300, 400
3 Type I 75 100, 20 500, 1000, 1500, 2000, 2800 100, 200, 300, 400
4 Type II 4 100, 10 8750, 10500 200
 International Journal of Fluid Power 43

Fluid Type I
Table 5. Linear scaling factor k.
Fluid Type II
880
Pump size [cc] Linear scaling factor λ
42 0.69
Density [kg/m3]

860
130 1.00
840 75 0.83
4 0.31
820

800
400 Table 6. Comparison result for 130 cc pump at 100% dis-
80 placement.
200 70
60
Pressure [bar] 50
0 40 Temperature [oC]

Figure 10. Densities of both types of fluid.

2250
Specific isobar heat capacity [W/kg oC]

Fluid Type I
2200 Fluid Type II

2150

2100

2050

2000

1950
40 50 60 70 80
Temperature [oC]

Figure 11. Specific isobar heat capacities for both types of Table 7. Comparison result for 130 cc pump at 50% displace-
fluid. ment.

3.1. Fluid properties

Two types of hydraulic fluids have been used within this
study.
Figure 10 shows the density change with pressure
and temperature for both types of fluid. Two meshes in
the figure are almost parallel with each other and have
about 35 kg/m3 offset.
Figure 11 shows the change of the specific isobaric
heat capacity with temperature.

3.2. Overall heat-transfer coefficients

The value of the overall heat-transfer coefficients for the
displacement chamber surface, the case volume surface,
and the overall air natural heat-transfer coefficient of the
housing surface are calculated from Equations (2.20),
(2.21), and (2.17). The constant values CDC_ref, CCase_ref,
and Cnatural_ref are determined based on the steady-state
measurement of a 130 cc pump which has been used as
a baseline/scaling reference for the comparison study. 3.3. Comparison results
The linear scaling factors k, which have been used Following figures and tables show the comparison results
for different sizes of pumps in the comparison study are of the model to four sets of measurements. Since the
summarized in Table 5. comparison results to the set #1 measurement cover a
44 L. Shang and M. Ivantysynova

Table 8. Comparison result for 130 cc pump at 20% displace- Thermal Model
Measurement
ment.
120

Case Temperature [oC]

100

80

60

40

20
100
100
50 80
60
40
Normalized Speed [%] 20
0 0 Normalized Pressure [%]

Figure 13. Case flow temperature comparison at 100%

displacement.

large amount of different operating conditions, 3D Thermal Model

Measurement
graphs have been chosen to present the results. In those 64
graphs, the x axis and the y axis represent the pressure Outlet Temperature [oC] 62
differential and the rotating speed, and the z axis shows 60
the temperature. The remaining three model comparison 58
results are shown in tables. 56
The pressure differential and the pump speed have 54
been normalized to maximum pressure and maximum 52
100
speed for each set of measurements. 100
The power losses which are used in the pump flow 50 60
80
40
temperatures prediction model have been calculated from Normalized Speed [%] 20
0 0 Normalized Pressure [%]
the measurements:
q_ loss ¼ m_ out  Cp  ðTout  Tout adia Þ þ m_ l;e  Cp Figure 14. Outlet flow temperature comparison at 80%
 ðTCase  TCase adia Þ (3.1) displacement.
The calculated power loss reflects the required power
loss to heat up the outlet flow and case flow to the mea- Thermal Model
Measurement
sured outlet port and case temperatures.
100
Case Temperature [oC]

90
3.3.1. Set#1: 42 cc pump with fluid type I 80

Figures 12–21 show the comparison results of both the 70

outlet flow temperature and the case flow temperature. 60
50
40
100
Thermal Model
100
Measurement 80
62 50 60
Outlet Temperature [oC]

40
Normalized Speed [%] 20
60 0 0 Normalized Pressure [%]
58
56
Figure 15. Case flow temperature comparison at 80%
54 displacement.
52
50
100 The red meshes show the simulated temperature, and the
100
80 blue meshes show the measured temperature.
50 60
40
Normalized Speed [%] 20
0 0 Normalized Pressure [%]
3.3.2. Set#2: 130 cc pump with fluid type I
Figure 12. Outlet flow temperature comparison at 100% Tables 6–8 show the comparison results to the measure-
displacement. ments on a 130 cc pump with the fluid type I. The first
 International Journal of Fluid Power 45

Thermal Model
Measurement
Thermal Model
64 90
Outlet Temperature [oC]

Case Temperature [oC]

Measurement
62 80
60
70
58
60
56
54 50

52 40
100 100
100 100
80 50 80
50 60 60
40 Normalized Speed [%] 40
Normalized Speed [%] 20 20
0 0 Normalized Pressure [%] 0 0 Normalized Pressure [%]

Figure 16. Outlet flow temperature comparison at 60% Figure 19. Case flow temperature comparison at 40%
displacement. displacement.

Thermal Model
Measurement

90 90

Outlet Temperature [oC]

Case Temperature [oC]

Thermal Model
Measurement
80 80

70 70

60 60

50 50
100 100
100 100
50 80 80
60 50 60
40 40
Normalized Speed [%] 20 Normalized Speed [%] 20
0 0 Normalized Pressure [%] 0 0 Normalized Pressure [%]

Figure 17. Case flow temperature comparison at 60% Figure 20. Outlet flow temperature comparison at 20%
displacement. displacement.

Thermal Model
80 Measurement
Outlet Temperature [oC]

Thermal Model 100

Case Temperature [oC]

Measurement
75 90
70 80
65 70
60 60
55 50
50 40
100 100
100 100
50 80 80
60 50 60
Normalized Speed [%] 40 40
20 Normalized Speed [%] 20
0 0 Normalized Pressure [%] 0 0 Normalized Pressure [%]

Figure 18. Outlet flow temperature comparison at 40% Figure 21. Case flow temperature comparison at 20%
displacement. displacement.

three include the swash plate angle, the pump speed, and 3.3.3. Set#3: 75 cc pump with fluid type I
the pressure differential in percentage to the maximum Tables 9 and 10 show the comparison results to the mea-
value. The last six columns show the comparison results surements of a 75 cc pump with fluid type I. The first
which include the measured port and case temperatures, three columns include the swash plate angle, the pump
the simulated port and case temperatures, and the differ- speed, and the pressure differential in percentage to the
ence between the measurement and the simulation. maximum value. The last six columns are the measured
46 L. Shang and M. Ivantysynova

Table 9. Comparison result for 75 cc pump at 100% displace- temperatures, the simulated temperatures, and the differ-
ment. ence between the measurement and the simulation.

3.4. Conclusion of the comparison study

The comparison study verified that the proposed pump
outlet port and case flow temperatures prediction model
is able to predict the outlet port flow temperature and the
case flow temperature with a reasonable accuracy of
±5 °C for outlet flow temperature and ±10 °C for case
flow temperature. This accuracy is sufficient for the use
of the model as input parameter estimator for the FSTI
simulation model.

4. Sensitivity study
In order to save simulation time and effort within a com-
putational pump design utilizing the proposed iterative
method of coupling the pump outlet port and case flow
prediction model with the extended FSTI model as
briefly discussed in chapter 2, a study has been con-
Table 10. Comparison result for 75 cc pump at 20% displace- ducted to investigate the sensitivity of pump outlet flow
ment. and case flow temperature on torque loss, volumetric
loss, and operating conditions of a given pump. The out-
come of this sensitivity study will be used as guidance
for estimation of temperature boundary conditions (outlet
and case flow temperature) within a computational
design process. With other words, the sensitivity study
was conducted to determine in which cases the impact of
design change on port and case flow temperatures is sig-
nificant and/or negligible.
In this sensitivity study, the internal leakage was
ignored due to its minor impact on the pump port and
case flow temperatures. The outlet mass flow rate fol-
temperatures, the simulated temperatures, and the differ- lows the simplified equation:
ence between the measurement and the simulation. m_ out ¼ m_ th  m_ leakage (4.1)
The theoretical flow rate is calculated from the pump
3.3.4. Set#4: 4 cc pump with fluid type II size and the operating condition.
Table 11 shows the comparison results to the measure- The total power loss is simplified into two parts, the
ments conducted on a 4 cc pump with fluid type II. The volumetric power loss and the mechanical power loss.
first three columns are the swash plate angle, the pump The volumetric power loss is generated by the leakage
speed, and the pressure differential in percentage of the flow rate and the pressure drop from the displacement
maximum value. The last six columns are the measured chamber pressure to the case pressure, and the mechani-
cal power loss captures all the rest of the energy dissipa-
tion rate.
Table 11. Comparison result for 4 cc pump with another type
of fluid. Ploss ¼ Pvolumetric þ Pmechanical
(4.2)
Ploss ¼ Qleakage  Dp þ Pmechanical

4.1. Sensitivity study on leakage and mechanical loss

Fifty sets of outlet flow temperatures and case flow tem-
peratures were obtained at different leakage flow rates and
mechanical losses using the pump outlet port and case
flow temperatures prediction model. Figures 22 and 23
 International Journal of Fluid Power 47

Figure 22. Outlet flow temperatures for different leakages and mechanical losses.

Figure 23. Case flow temperatures for different leakages and mechanical losses.

show two 3D meshes for the simulated outlet flow temper- mechanical loss. The design marked with red dot has
atures and case flow temperatures. high mechanical loss, and the design marked with yellow
In Figure 22, the 3D mesh shows the outlet flow dot has high leakage.
temperatures over different leakages and mechanical The two pictures on the right-hand side show two
losses. The leakages displayed in the picture are normal- side views of the 3D picture. From these pictures, two
ized to the theoretical delivered flow rate, and the trends can be observed clearly: the outlet flow tempera-
mechanical losses displayed in the picture are normalized ture increases with both the leakage and the mechanical
to the theoretical delivered power. loss.
The green dot in the 3D picture shows the ideal Figure 23 shows the case flow temperatures over dif-
design which has minimum leakage and minimum ferent leakages and mechanical losses. Two trends can
48 L. Shang and M. Ivantysynova

be clearly observed for the case flow temperature: the

case flow temperature decreases with the leakage and
increases with the mechanical loss.
The higher mechanical power loss transferred into
the displacement chamber volume and the case volume
increases both the outlet flow temperature and the case
temperature.
The case flow temperature decreases with the leakage
because the case flow temperature variation is inversely
proportional to the case flow rate which is leakage flow
rate. The outlet flow temperature increases with the leak-
age because higher leakage produces higher total power
loss.
Figure 23 also shows that the case flow temperature
mesh is flat in the high leakage region and steeper in the
low leakage region. This is because the sensitivity of the Figure 25. Outlet flow temperature meshes for different
displacements @ 2000 rpm and 200 bar.
case flow temperature decreases with increasing case
flow rate, i.e. increasing leakage.
Figure 24 shows an example of using the sensitivity
study to guide the initial temperature guess. For exam-
ple, if the design goal is to reduce the leakage flow rate
of a pump from point 1 to point 2 by optimizing the
pump design. The sensitivity study results in Figure 24
show that the case temperatures at point 1 and point 2
are almost the same. Thus, the case temperature at point
1 can be used as the initial case temperature guess at
point 2. In case of an original pump design with higher
mechanical loss which places the original design at point
3, and the pump design optimization should reduce the
leakage from point 3 to point 4, the impact on case tem-
perature is not negligible. Thus, higher case temperature
should be used as the initial case temperature guess.
Figure 26. Case flow temperature meshes for different
displacements @ 2000 rpm and 200 bar.
4.2. Sensitivity study on displacement
In the sensitivity study on displacement, the outlet flow
temperature and the case flow temperature were pre- and 20% displacement. The crosses and circles on the
dicted at 100, 50, and 20% displacement over a larger meshes in Figure 25 and Figure 26 show the measured
range of leakage and mechanical loss. temperatures and the simulated temperatures at corre-
Figures 25 and 26 show the outlet flow temperature sponding operating conditions. The simulated tempera-
meshes and case flow temperature meshes at 100, 50, tures are very close to the measured temperatures which
indicate that the temperature meshes obtained from the
simulation are accurate enough to represent the sensitiv-
ity of the real pump.
Both Figures 25 and 26 show that the temperature
meshes for different displacements are very close to each
other. However, the measured temperature points indicate
that the pump has very different normalized leakages
and normalized mechanical losses for different displace-
ments. This concludes that the different thermal behav-
iors of the pump running at different displacements are
mainly caused by the different normalized leakages and
normalized mechanical losses.

4.3. Sensitivity study on pressure differential

Figure 24. Case flow temperatures for different leakages and Figures 27 and 28 show the outlet flow temperature
mechanical losses. meshes and case flow temperature meshes for 400 bar
 International Journal of Fluid Power 49

Figure 27. Outlet flow temperature meshes for different pres- Figure 29. Outlet flow temperature meshes for different pump
sure differentials @ 2000 rpm and 50% Vi. speeds @ 200 bar and 100% Vi.

Figure 28. Case flow temperature meshes for different pres-

sure differentials @ 2000 rpm and 50% Vi. Figure 30. Case flow temperature meshes for different pump
speeds @ 200 bar and 100% Vi.

Figure 29 shows that the outlet flow temperature

pressure differential and 200 bar pressure differential meshes for different pump speeds are very close to each
between the outlet and inlet. Again, the measured tem- other. The measurement points show that the normalized
perature and the corresponding simulated temperature leakages and the normalized mechanical losses are very
can be found in these figures. similar at different pump speeds as well. The pump run-
Figure 27 shows that the outlet flow temperature ning at different pump speeds will have a similar outlet
meshes are very different for different pressure differen- port flow temperature.
tials. That is because the temperature variation is propor-
tional to the pressure change. The fluid temperature will
increase more when running at higher pressure differen-
tial.
Figure 28 shows that the case flow temperature mesh
at higher pressure differential is higher than the mesh at
lower pressure differential. This is because at the same
leakage and mechanical loss point, the pump running at
higher pressure differential will have higher power loss.

4.4. Sensitivity study on pump speed

Figures 29 and 30 show the outlet flow temperature
meshes and case flow temperature meshes at 1000, 2000,
and 2800 rpm. Again, the measured temperatures and the Figure 31. Zoomed in Figure 30 of case flow temperature
simulated temperatures are marked. meshes for different pump speeds.
50 L. Shang and M. Ivantysynova

Figure 30 shows that the case flow temperature 6. Conclusion

meshes for different pump speeds are very close to each The proposed pump outlet and case flow temperatures
other in the high leakage region and differ more in the prediction model, which considers both the temperature
low leakage region. The three measurement points are variation due to the compression or expansion and heat-
located in the low leakage region. Figure 31 shows the transfer due to energy dissipation/power loss, is able to
zoomed area of measured and simulated points of the predict the outlet port flow temperature and the case flow
case flow temperature for different pump speeds. Even temperature for any given pump or motor size at any
though the pump running at different speeds has similar given operating condition based on the given total power
normalized leakages and normalized mechanical losses, loss and the leakage flow rate with reasonable accuracy.
the different thermal behaviors due to the different pump The results (outlet port and case temperature) can be
speeds separate the case temperature points vertically, used as input parameter for the in-house developed FSTI
which can be observed in Figure 31. simulation program, which solves for non-isothermal gap
For all different speeds, the power losses are pro- flow considering elasto-hydrodynamic effects due to ther-
portional not only to the pump speeds, but also to the mal and pressure loading of the main pump surfaces
outlet flow rates and the case flow rates. That explains forming the lubricating interfaces.
why the pump flow temperature meshes of the pump The pump outlet and case flow temperatures predic-
running at different speeds are nearly identical in the tion model coupled with the FSTI model can be used for
high leakage region. The case flow temperature meshes computational design of new pumps or optimization of
show differences in the low leakage region because the given designs without the need of measured port and
low case flow rate causes an increase in the case case temperature data.
temperature. The results of the conducted sensitivity study can be
The temperature of solid parts of the pump running used to reduce the required computation time for port
at higher speed is higher due to the higher power loss. and case temperature estimation as a boundary condition
Since normally the temperature in the case volume is for FSTI model, thus can help to shorten the design pro-
higher than the temperature in the displacement cham- cess while keeping required prediction accuracy.
ber volume, the ratio of solid-case temperature differen-
TCase
tial over solid-DC temperature differential TTsolid solid TDC
increases with the solid parts temperature. Furthermore,
the ratio of the solid-case heat-transfer rate over the Nomenclature
q_
solid-DC heat-transfer rate q_ss Case
DC
also increases with the
solid parts temperature. That explains why the
CCase_ref Case heat-transfer [–]
pump running at higher speed has higher case flow
constant
temperature. CDC_ref DC heat-transfer [–]
Additional to that the pump running at lower speed constant
has lower power loss but has almost the same natural Clcf Leakage flow rate [–]
convection heat flow rate into the air. Therefore, the correction factor constant
Cnatural_ref Natural heat-transfer [–]
pump running at lower speed has relatively greater heat-
constant h i
transfer ratio into the air but not into the fluid. cp Specific heat J
kgK
capacity h i
J
h Enthalpy
4.5. Conclusion of the sensitivity study hkgi
W
kCase Overall case K
The present study captures the sensitivity of the pump heat-transfer
thermal behavior over the normalized leakage flow rate, coefficient h i
W
the normalized mechanical loss, the displacement, the kDC Overall DC heat-transfer K
pressure differential, and the pump speed. The outlet coefficient h i
W
port flow temperature increases with both the leakage knatural Overall natural heat-transfer K
coefficient
and the mechanical loss, and the case flow temperature k Linear scaling factor [–]
decreases with the leakage and increases with the lcf Leakage flow rate correction [–]
mechanical loss. The sensitivity of the case flow tem- factor  
kg
perature decreases with the leakage. The impact of m_ Case Case mass
flow rate s
pump displacement on port and case temperature is  
mainly caused by the different normalized leakages and kg
m_ in Inlet mass
normalized mechanical losses. The impact of pump flow rate s
 
pressure differential on port and case temperature kg
m_ out Outlet mass
differential is mainly caused by the temperature s
flow rate
variations due to the pressure changes. The impact of n Pump speed ½rpm
pump speed on port and case temperature is mainly pCase Case pressure ½bar
caused by the heat-transfer characteristic.
 International Journal of Fluid Power 51

pin Inlet pressure ½bar optimization of displacement machines as well as modeling,

pout Outlet pressure ½bar simulation, and testing of fluid power systems. Besides the book
pDC DC pressure ½bar ‘Hydrostatic Pumps and Motors’ published in German and
q_ Case Heat flow rate out of case ½W English, she has published more than 80 papers in technical
volume with case flow rate journals and at international conferences.
and natural convection
q_ DC Heat flow rate out of DC ½W
volume with outlet flow rate
q_ loss Total power loss ½W References
q_ s Case Heat-transfer rate from solid ½W
parts to case volume Grönberg, D., 2011. Prediction of case temperature of axial
q_ s Heat-transfer rate from solid ½W piston pumps. Master thesis. Chalmers University of Tech-
DC
parts to DC volume nology, Sweden, ISSN 1652-8557.
QCase Case flow rate l Olems, L., 2001. Ein Beitrag zur Bestimmung des Temperatur-
min
verhaltens der Kolben-Zylinder-Baugruppe von Axialkol-
Qin Inlet flow rate l
min benmaschinen in Schrägscheibenbauweise [The temperature
behavior of the piston cylinder interface of swash plate
Qout Outlet flow rate l
hmin i type axial piston machines]. VDI Fortschritt-Berichte.
ρ Fluid density kg
m3 Reihe 1 No. 348. Düsseldorf: VDI. ISBN: 3-18-334801-2.
h i Oppermann, M., 2006. A new approach for failure prediction
s Entropy J in mobile hydraulic systems. Diss. TUHH Hamburg. VDI
kgK Fortschritt-Berichte. Reihe 12 No. 665. Hamburg: VDI.
Tadia_Case Case temperature after ½ C ISBN: 3-18-366512-9
adiabatic compression/expansion Pelosi, M., 2012. An investigation on the fluid-structure inter-
Tadia_DC DC temperature after adiabatic ½ C action of piston/cylinder interface. Thesis (PhD). Purdue
compression/expansion University.
Tair Air temperature ½ C Pelosi, M. and Ivantysynova, M., 2012. Heat-transfer and ther-
TCase Case temperature ½ C mal elastic deformation analysis on the piston/cylinder
TDC DC temperature ½ C interface of axial piston machines. Transaction of the
Tin Inlet temperature ½ C ASME, journal of tribology, 134, 1–15.
Tout Outlet temperature ½ C Pelosi, M. and Ivantysynova, M., 2013. The impact of axial
h 3i piston machines mechanical parts constraint conditions on
m
v Specific volume kg the thermo-elastohydrodynamic lubrication analysis of the
fluid film interfaces. International journal of fluid power,
Vi Pump size ½cc
14 (3), 35–51.
Schenk, A. and Ivantysynova, M., 2014. A transient fluid struc-
Notes on contributors ture interaction model for lubrication between the slipper
and swashplate in axial piston machines. Proceedings of
L. Shang was born on 25 March 1989 in the 9th international fluid power conference (9IFK), 24–26
Tianjin (China). He received his BS degree March, Aachen, Germany, Vol. 1, 398–409.
in Thermal Energy and Power Engineering Witt, K., 1974. Die Berechnung physikalischer und thermody-
from Huazhong University of Science and namischer Kennwerte von Druckflüssigkeiten, sowie die
Technology in 2011 and his MS degree in Bestimmung des Gesamtwirkungsgrades an Pumpen unter
Mechanical Engineering in New Jersey Berücksichtigung der Thermodynamik für die Dru-
Institute of Technology in 2013. He is cur- ckflüssigkeit [The determination of physical and thermody-
rently a PhD student at Maha Fluid Power namic parameters of the hydraulic fluid, and the
Research Center in Purdue University. His determination of pump overall efficiency considering the
main research interests are modeling and thermodynamic behavior of the hydraulic fluid]. Thesis
optimizing of hydraulic pumps/motors. (PhD). TH Eindhoven.
Zecchi, M., 2013. A novel fluid structure interaction and ther-
mal model to predict the cylinder block/valve plate inter-
M. Ivantysynova was born on 11 December face performance in swash plate type axial piston
1955 in Polenz (Germany). She received her machines. Thesis (PhD). Purdue University.
MSc degree in Mechanical Engineering and Zecchi, M. and Ivantysynova, M., 2013. Spherical valve plate
her PhD degree in Fluid Power from the design in axial piston machines – A novel thermo-
Slovak Technical University of Bratislava, elasto-hydrodynamic model to predict the lubricating
Czechoslovakia. After 7 years in fluid power interface performance. The 8th international conference on
industry, she returned to university. In April fluid power transmission and control (ICFP 2013), 9–11
1996, she received a professorship in fluid April, Hangzhou, China, 325–629.
power & control at the University of Zecchi, M., Mehdizadeh, A. and Ivantysynova, M., 2013. A
Duisburg (Germany). From 1999 until novel approach to predict the steady state temperature in
August 2004, she was a professor of Mechatronic Systems at the ports and case of swash plate type axial piston machines.
Technical University of Hamburg-Harburg. Since August 2004, Proceedings of the 13th Scandinavian international confer-
she is a professor at Purdue University, USA. Her main research ence on fluid power (SICFP2013), 3–5 June, Linkoping,
areas are energy saving actuator technology and model-based Sweden.