International Journal of Fluid Power 13 (2012) No. 2 pp.

51-59

MEASURING AND MODELLING HYDRAULIC FLUID DYNAMICS

AT HIGH PRESSURE – ACCURATE AND SIMPLE APPROACH

Juho-Pekka Karjalainen, Reijo Karjalainen and Kalevi Huhtala

Tampere University of Technology (TUT), Department of Intelligent Hydraulics and Automation (IHA).
P.O.Box 589, FIN-33101, Tampere, Finland.
[email protected], [email protected], [email protected]

Abstract

Dynamic properties of hydraulic fluids have to be taken into account in ever increasing fluid power applications.
The main reasons are increasing accuracy demands in control and modelling, as well as increasing operating pressure
and temperature ranges. Moreover, the already wide spectrum of different hydraulic fluids is also expanding all the
time. However, information on dynamic hydraulic fluid behavior is still very difficult to be obtained. On the other hand,
existing fluid models tend to be either too inaccurate, or at least highly non-generic for most practical applications.
This article introduces simple, yet accurate approaches for measuring and predicting the most important dynamic
fluid parameters: bulk modulus, density and speed of sound in fluid. The methods are basically applicable to any stan-
dard hydraulic fluid, without any extra system-related constraints, at least at the presented conditions. The studied pres-
sure range reaches 1500 bar, and the temperatures cover a normal operating range of industrial applications. Examples
of both measured and predicted results for selected commercial hydraulic fluids are given. The results have also been
found to be in excellent agreement with existing reference data.

Keywords: adiabatic, bulk modulus, density, dynamics, high-pressure, hydraulic fluid, isothermal, measuring, modelling, second order polynomial,
speed of sound

1 Introduction
Most fluid power engineers agree on hydraulic fluid On the other hand, it would be important to have use-
being one of the most important components in every ful tools for predicting different fluid parameters. There
fluid power system. In addition to e.g., lubrication, heat are a number of models also for fluid parameters, but
transfer and contamination control, fluid is first and unfortunately they usually fall into two different catego-
foremost the power transmitting medium. Therefore, it ries. Some models are too simplified and therefore inac-
might be even argued that fluid is the most important curate. Other models might possibly be accurate, but
single component. they are impossible to use in practice, due to the parame-
Possibly, fluid properties have not been as signifi- terization, which assumes information, which is very
cant design parameters in former applications as they difficult or usually even impossible to find. Both of these
are today. Therefore, there is very limited or no reli- kinds of models are, in effect, as useful from a system
able, measured information on fluid behavior available designer’s point of view as no fluid model at all.
for system designers – especially at pressure levels In this article, the effects of pressure, temperature
over 300 bar, or for different types of hydraulic fluids. and fluid type on dynamic fluid parameters are studied.
However, operating pressures have been increasing An accurate, yet simple and cost effective approach for
throughout the fluid power field. Moreover, the selec- measuring the most important dynamic fluid parameters
tion of different hydraulic fluids is increasing all the is presented, and a method for predicting the observed
time – mineral oil based fluids are being replaced e.g., behavior in a very generalized manner is suggested – the
with vegetable oil or synthetic ester based fluids. fluid parameters being speed of sound in a fluid, fluid
Therefore, fluid parameter variation has become an density and fluid bulk modulus. The presented pressure
important design aspect in many cases. range reaches 1500 bar, and the temperature range of 40
- 70°C covers normal industrial solutions.
This manuscript was received on 2 February 2012 and was accepted
after revision for publication on 5 March 2012

© 2012 TuTech 51
 Juho-Pekka Karjalainen, Reijo Karjalainen and Kalevi Huhtala

2 Measuring Method 2.1.3 Density

Also in Gholizadeh (2011), an inclusive survey of
2.1 Reference Methods density measuring methods was given. Most of the
There are many reported methods for measuring density measuring methods are based on some special
different dynamic fluid parameters. A short overview laboratory equipment which is not that commonly
of the most typical reference methods is given in the usually available for an average fluid power engineer.
following. Reference methods have also been surveyed Moreover, this kind of equipment may be quite expen-
in Karjalainen (2011b). sive. One such accurate and commercially available
device, vibrating tube densitometer, is e.g., based on
2.1.1 Bulk Modulus measuring oscillation periods inside a fluid filled tube.
On the other hand, density may be defined e.g., from
In Gholizadeh (2011), a relatively thorough litera- mass of a fluid and using a similar type of compression
ture survey was recently performed concentrating par- device as used in defining isothermal bulk modulus.
ticularly on fluid bulk modulus measurement. Gholi-
zadeh (2011) also creditably explains how there are 2.2 The Studied Measuring Method
four different definitions for fluid bulk modulus – it
can be either secant or tangent, and isothermal or adia- One possibility of defining all the three parameters
batic. Which value of bulk modulus should be used in with the same system is given in Karjalainen (2011b).
which cases is a somewhat controversial subject. This The measuring system consists of typical fluid power
article will not focus on that part, some discussion may components and is relatively easy and cost effective to
be found e.g., in Gholizadeh (2011) and its references, build. It is also easy to maintain which is a benefit,
also in Karjalainen (2011b). particularly when dealing with previously unknown
As it is also stated in Gholizadeh (2011), the most fluids or operating conditions. The method is based on
typical ways to measure fluid bulk modulus are based measuring speed of sound in fluid directly, fluid den-
on using some kind of fluid compressing device, or on sity and bulk modulus may be determined iteratively.
defining the speed of sound in a fluid. Using compres- This article is based on the described method.
sion (e.g. ASTM-D6793-02) leads usually to isother-
mal secant bulk moduli, whereas speed of sound meth- 2.2.1 The Studied Measuring System for Speed of
ods are expected to lead to adiabatic tangent values. Of Sound
course, there are certain relationships between different A schematic picture of the studied measuring system
definitions of bulk moduli, which might allow defining for speed of sound in a fluid line is shown in Fig. 1.
another values based on another. However, for exam-
ple, heat capacity factors are not commonly available
for many hydraulic fluids.
Using compressing devices, the reached pressure
ranges are usually reported higher, evidently even up to
690 MPa. However, compression methods may be
quite sensitive to errors, e.g., due to unexpected struc-
tural compliances of the measuring device. Moreover,
compression methods are more expensive and difficult
to apply in many situations e.g., when studying new
and unknown fluids.

2.1.2 Speed of Sound in Fluid

Fig. 1: The schematic picture of the studied measuring
In Gholizadeh (2011), also the most typical ways to system
measure speed of sound in fluid were surveyed. For
example, an ISO standard (ISO 15086-2:2000) presents A closed, straight, fluid filled measuring pipe is
two possible methods. However, these methods are pressurized with a pneumatically driven single-piston
known to be tested more specifically at pressure levels pump, which acts as a pressure intensifier. The high-
under 500 bar. Therefore, elevated pressure levels pressure measuring pipe is enclosed in an outer, oil-
present many challenges to the equipment depicted filled low-pressure pipe for temperature control circu-
e.g., in the standard. In particular, producing continu- lation. With a PID controller, the temperature of the
ous pressure fluctuation at high pressures may be diffi- actual measuring pipe can be maintained accurately
cult and very expensive. It is also worth mentioning independent of possible changes e.g., in room tempera-
that the possible viscosity range of measured fluids will ture. Oil temperature is measured at the both ends of
be narrowed, if e.g., a hydraulic pump is used for pro- the measuring pipe. The temperature circulation is
ducing the demanded pressure fluctuation. Some other driven with a simple gear pump. The pressure relief
possible aspects needed to be considered with continu- valve in the measuring line is a safety valve, which
ous pumping have been discussed e.g., in Karjalainen remains closed in normal operation.
(2005, 2011b). Once a desired pressure level is reached, the pneu-
matic pump is shut and a single pressure transient is
produced into the statically pressurized measuring pipe

52 International Journal of Fluid Power 13 (2012) No. 2 pp. 51-59

 Measuring and Modelling Hydraulic Fluid Dynamics at High Pressure – Accurate and Simple Approach

by knocking a hydraulic cylinder. Two fast piezo- 2.2.2 The Iterative Procedure for Bulk Modulus and
electric pressure transducers record the pressure wave Density
at two points, at known distance from each other. With
The originally used iteration procedure for fluid
cross-correlation algorithm, the delay of the pressure
bulk modulus and density was published e.g., in Kar-
wave propagation can be found from the phase shift of
jalainen (2005, 2006, 2007, 2009, 2011a) leading to a
the two pressure transients. And since pressure wave is
systematic 1 - 2 % referenced maximum error at pres-
traveling at a speed of sound in a fluid line, this will
sure range of 1500 bar (Karjalainen, 2011b; Kuss,
lead to the effective value of the speed of sound.
1976). However, the procedure was slightly revised to
The removal of the effect of system compliances is
its final form in Karjalainen (2011b) leading to e.g.,
explained in Section 2.2.3. After that, following an
referenced maximum error of density of less than
iterative procedure (explained in Section 2.2.2) it will
0.5 % (Karjalainen, 2011b; Kuss, 1976). The revision
lead to the desired values of speed of sound in fluid,
process has been explained in detail in Karjalainen
adiabatic tangent fluid bulk modulus, and fluid density.
(2011b). The revised final iteration procedure is pre-
The reliability of this non-standard measuring system
sented in the following.
has been justified in detail in Karjalainen (2011b).
Once a speed of sound in the fluid line is measured
The repeatability of measuring pressure wave propa-
according to Section 2.2.1, there are two equations,
gation has been found to be excellent. Figure 2 shows an
Eq. 1 (Merritt, 1967) and Eq. 2 (Karjalainen, 2011b;
example scatter of three consecutive measurements
Garbacik, 2000) for iterative determination of effective
(t1 - t3) with Shell Tellus VG 46 mineral hydraulic oil at
adiabatic tangent bulk modulus and fluid density. As
40°C temperature. The dashed line represents a second
already mentioned in Section 2.1.1, the heat capacity
order p ial fitting of the reported delay. The maximum
factors in Eq. 2 might be difficult to find for every
error in repeatability at the same operating pressure is
fluid. Therefore, also an estimated equation Eq. 3 (Kar-
one sample period – in this case only 20 μs with 50 kHz
jalainen, 2011b) may be used instead of Eq. 2 with
sampling rate. During the testing phase of the measuring
good accuracy in practice. The ratio of the fluid heat
system the repeatability in normal operation has been
capacity factors has been replaced with an estimate that
found to be within ± one sample period even with statis-
isothermal tangent bulk modulus is approximately
tically larger investigation.
200 MPa smaller than the corresponding adiabatic
tangent value. In practice, this is a fair assumption
according to e.g., Hodges (1996) and Borghi (2003).
The results of this study have been defined using the
estimate equation, with the above reported accuracy.
Beff a
c = (1)
ρn
p − patm
1 Cv
( ∑ Beffa )⋅
n Cp
ρn = ρatm,T e n =1
(2)
p − patm
1
( ∑ Beff a )−0.210
⋅
9
n
∧

Fig. 2: Example of the repeatability of measuring pressure ρ n = ρatm,T e n =1

(3)
wave delay in a mineral hydraulic oil.
The iteration procedure should be started at the at-
Unlike e.g., in (ISO 15086-2:2000), Johnston mospheric pressure, or close to it, and the first values
(1991), Kojima (2000) and Yu (1994), this method for effective bulk modulus and density are received.
does not have a hydraulic pump for producing pressure After that, all the preceding effective bulk moduli, as
fluctuation. Therefore, this approach has certain advan- well as the presently iterated effective bulk modulus,
tages. The viscosity range of a measured fluid is not so should be summed in either of the used equations Eq. 2
limited. With the presented system, fluids with kine- or Eq. 3. Presented mathematically the above would
matic viscosities of 1 - 1100 cSt have been measured mean:
successfully. Moreover, temperature can be controlled Beffmean = mean (Beffatm ,Beffp1 ,Beffp2 ,...,Beffp )
more easily when there is no volumetric flow through
the measuring line. where the value Beffp is the currently iterated adiabatic
Basically, the presented method does not make any value of the effective bulk modulus at the pressure in
restrictions for measuring temperatures. Of course, question. All the other values are received from the
some method of cooling the system is needed for lower preceding pressure steps.
temperatures. Moreover, fluid should be in such a state The size of consecutive pressure steps can e.g., fol-
in the measuring pipe that a detectable pressure tran- low normal pressure steps of these kinds of measure-
sient can still be produced with the cylinder. The stud- ments. In Karjalainen (2011b) it has been shown that
ied pressure range so far has reached 1500 bar. How- this will lead to similarly good accuracy than by using
ever, there are components available for raising the very small pressure steps, which would only lead to
pressure level safely at least up to 2500 bar. more tedious calculation routines.

International Journal of Fluid Power 13 (2012) No. 2 pp. 51-59 53

 Juho-Pekka Karjalainen, Reijo Karjalainen and Kalevi Huhtala

2.2.3 The Removal of System Compliances as standard industrial hydraulic fluids. Shell Tellus TX
is basically the same fluid as the previous ones, but it
As in this studied case, the measuring line should
has viscosity index enhancing additives. Comet SAE
be designed to have no dead volumes. Moreover, the
and Shell Rimula are diesel engine motor oils. Shell
measuring line should be de-aerated with thorough
Calibration fluid is a standard calibration fluid used
flushing. In this study, the flushing of the system was
e.g., in injection motors. Shell Naturelle HF-E is a
performed at a pressure level of over 1000 bar. With
synthetic ester fluid, and Comet ECO Pine is pine oil
the above conditions, system compliances can be esti-
based natural ester fluid.
mated very well with equations Eq. 4 and Eq. 5 (Kar-
jalainen, 2011b; Merritt, 1967) for a rigid thick-walled
3.2 Speed of Sound in Fluid
pipe.
1 The measured speeds of sound are presented in
B = (4) Fig. 3 and 4. As it can be seen, only calibration fluid
2 ⎡ ( 1 + µ ) ⋅ D + (1 − µ ) ⋅ d ⎤
⋅⎢ ⎥ stands out, clearly having the lowest value for speed of
E ⎣⎢ (D − d ) ⋅ (D + d ) ⎦⎥ sound. All the other fluids present somewhat similar
absolute values regardless of different base fluids,
B ⋅B
B = (5) viscosity grades or additives. In practice, the trend of
B −B speed of sound behavior will not change between the
fluids or temperatures. In fact, the measurements are
following uniform second order polynomials (Kar-
jalainen, 2009; 2011b) – this discovery has been used
3 Measurement Results in defining the prediction method of Section 4.3.
Eight different commercial hydraulic fluids were
selected for this study. Speed of sound in fluid, adia-
batic tangent fluid bulk modulus and fluid density were
measured according to Chapter 2. The studied pressure
range was 100 - 1500 bar which covers e.g., state-of-art
common rail injection pressures. The studied tempera-
ture range of 40 - 70°C was selected to cover the nor-
mal range of industrial fluid power applications. The
selected fluids and the received results are presented in
the following.

3.1 The Selected Fluids

Fig. 3: The measured speeds of sound in the studied hy-
Different commercial fluids were selected to cover draulic fluids at 40°C
the most typical spectrum of the hydraulic fluids being
used at the moment – water had to be abandoned at this
stage e.g., due to possible rusting problems. Water
viscosity would not have been a problem. Different
base fluids were selected for finding out whether it
would affect the fluid dynamics. The effect of viscosity
grades and additives were researched by selecting dif-
ferent mineral oil based fluids.

Table 1: The fluid characteristics of the studied hy-

draulic fluids (Karjalainen 2011b)

Fig. 4: The measured speeds of sound in the studied hy-

draulic fluids at 70°C

Speed of sound in fluid is known to set the upper

limit for the fastest possible frequency response
(Smith, 1960). However, it does not necessarily mean
that the fastest possible response could be used in prac-
tice – it might even lead to unstable control response.
Moreover, in sections 3.3 and 3.4 it is shown that there
are significant differences in bulk moduli and densities
between fluids with similar measured speeds of sound.
The fluid characteristics of the selected fluids are
listed in Table 1. The six first fluids are mineral oil
based. Shell Tellus VG 32 and VG 46 can be regarded

54 International Journal of Fluid Power 13 (2012) No. 2 pp. 51-59

 Measuring and Modelling Hydraulic Fluid Dynamics at High Pressure – Accurate and Simple Approach

3.3 Adiabatic Tangent Bulk Modulus

The measured adiabatic tangent bulk moduli are
presented in figures 5 and 6. It can be seen that the pine
oil has the highest bulk modulus. Also the HF-E fluid
has a bit higher bulk modulus than the studied mineral
oils. The calibration fluid has the lowest bulk modulus,
whereas the other five mineral oils are presenting
highly similar behavior – despite different viscosity
grades or additives. Therefore, base fluid would seem
to have the biggest effect on fluid’s bulk modulus.

Fig. 7: The measured densities of the studied hydraulic

fluids at 40°C

Fig. 5: The measured adiabatic tangent bulk moduli in the

studied hydraulic fluids at 40°C

Fig. 8: The measured densities of the studied hydraulic

fluids at 70°C

Similar to speed of sound and bulk modulus, the

trend of the density behavior will not change between
the fluids or temperatures. Also these measurements
are following uniform second order polynomials (Kar-
jalainen, 2009; 2011b) – this discovery has been used
in defining the prediction method of Section 4.2.

Fig. 6: The measured adiabatic tangent bulk moduli in the

4 Prediction Method
studied hydraulic fluids at 70°C
As mentioned in the introduction, there are different
There are differences in absolute values between models also for dynamic fluid parameters. However,
the fluids – definitely big enough to affect certain ap- the existing models are usually too simplified and inac-
plications. However, the trend of the bulk modulus curate, or the models require parameters, which cannot
behavior will not change between the fluids or tem- be realized in practice. On the other hand, the high-
peratures. Again, the measurements are following uni- pressure behavior of hydraulic fluid dynamics is still
form second order polynomials (Karjalainen, 2009; quite an unknown research area. Even any measured
2011b) – this discovery has been used in defining the data is very rare for pressure levels over 300 bar –
prediction method of Section 4.1. especially when the fluid is not mineral oil based.
Quite simple prediction method has been developed
3.4 Density at IHA for speed of sound in fluid, fluid bulk modulus
and fluid density. As stated in Chapter 3, measure-
The measured densities of the studied fluids are ments seem to follow a uniform second order polyno-
presented in figures 7 and 8. The absolute values may mial trend. Therefore, with simple second order poly-
vary significantly between the fluids. However, the five nomials it is possible to predict the dynamic fluid be-
fluids (Tellus VG 32, Tellus VG 46, Tellus TX, Comet havior. Furthermore, the most important aspect with
SAE, Rimula x30) having highly similar mineral oil this method is that the parameterization is possible for
base fluid are forming one group. Pine oil has the high- any standard hydraulic fluid – only ISO standardized
est density, followed by the HF-E fluid. Calibration fluid characteristics is needed. The actual development
fluid clearly has the lowest density. Based on the re- of the models has been explained more in detail in
sults it seems quite clear that base fluid has the domi- Karjalainen (2009, 2011b).
nant effect on fluid density, rather than viscosity grade Despite the simple models, they have proven to be
or additives. very accurate at the presented operating range of 40 -

International Journal of Fluid Power 13 (2012) No. 2 pp. 51-59 55

 Juho-Pekka Karjalainen, Reijo Karjalainen and Kalevi Huhtala

70°C and up to 1500 bar, with many different types of density correction coefficients α are listed e.g., in
fluids. In Chapter 5 of this article, the received predic- (ASTM/IP) and Hodges (1996). For the studied fluids
tion accuracy is demonstrated. More results have been the coefficients are also listed in Table 1. In Eq. 10,
reported in Karjalainen (2011b). When the operating temperature is expressed in [°C], α in [kg/(m³·°C)] and
conditions clearly differ from the presented ones some density in [kg/m³].
restrictions may occur with the presented equations. CD (patm ,T) = ρatm,15 C − α ⋅ (T − 15 C) (10)
This is discussed in Section 4.4. Further research is in
progress for narrowing down these restrictions.
4.3 Speed of Sound Model
4.1 Bulk Modulus Model As with bulk modulus and density, the second and
According to the second order polynomials fitted to first order terms of speed of sound model have been
measured data, the second and first order terms may be selected for the best fit. The constant term is a tempera-
estimated quite well for the best fit. The constant term ture and fluid dependent factor. The equation for speed
is a fluid and temperature dependent factor. The equa- of sound in fluid is given in Eq. 11 (Karjalainen, 2009;
tion for fluid tangent bulk modulus is given in Eq. 6 2011a; 2011b). Pressure is expressed in [bar] and speed
(Karjalainen, 2011b). The pressure is expressed in [bar] of sound is expressed in [m/s].
and bulk modulus in [MPa]. c(p,T) = - 0.0001 ⋅ p 2 + 0.48 ⋅ p + CC (patm ,T) (11)
2
B(p,T) = - 0.0001 ⋅ p + 1.2 ⋅ p + C B (patm ,T)(6) The constant term of speed of sound equation can
The constant term of the second order model of be calculated from the previously presented constant
adiabatic tangent bulk modulus can be estimated using terms of fluid tangent bulk modulus (Eq. 7 or Eq. 8)
Eq. 7 (Borghi, 2003; Karjalainen, 2011b). Eq. 8 (Bor- and density (Eq. 10), using also the previously pre-
ghi, 2003; Karjalainen, 2011b) may be used for iso- sented Eq. 1. Combined, this will lead to Eq. 12 (Kar-
thermal tangent bulk modulus. All the parameters of jalainen, 2009; 2011a; 2011b).
the constant term equations can be found for any stan- CB (patm ,T)
dard hydraulic fluid. The kinematic viscosity at atmos- CC (patm ,T) =
CD (patm ,T)
pheric pressure and 20°C may e.g., be calculated with (12)
the well-known Walther equation (Hodges, 1996). In
these equations, kinematic viscosity is expressed in 4.4 Current Restrictions
[cSt] and temperature in [°C].
The presented temperature range covers 40 - 70°C.
CB (patm ,T) = 0.1 ⋅ ⎡⎣1.57 + 0.15 ⋅ log(νatm,20 C )⎤⎦ Further research has been also performed for tempera-
20−T
(7) ture range of 20 - 130°C. Based on first results with the
4+
⋅10 417 expanded temperature range, the density equation (Eq.
9) may be modified for a better accuracy by giving some
CB (patm ,T) = 0.1 ⋅ ⎡⎣1.3 + 0.15 ⋅ log(νatm,20 C )⎤⎦ ⋅ simple temperature relation for the first order term. It
20−T
(8) would seem clear that the first order term slightly in-
4+
10 435 creases with increasing temperature – the effect of which
is insignificant with the presented range of 40 - 70°C.
4.2 Density Model The speed of sound and bulk modulus models would not
seem to be that affected by temperature.
The first version of the density model was already However, some further research has also been made
published in Karjalainen (2009, 2011a). However, the using mineral oil and petrol based fluids with densities
second and first order terms of the model were slightly as high as 1000 kg/m³. With these fluids the constant
revised in Karjalainen (2011b). The first version can be terms of speed of sound and bulk modulus models
expected to lead to an additional systematic maximum might need some kind of simple density related factor.
full scale error of about 1 - 2 %, and the model pre- It would seem that with high-density fluids there is a
sented in this article should be used instead for better systematic uniform shift between the measured and
accuracy. predicted values – the predicted ones being somewhat
Also for fluid density model, the second and first lower than the actually measured ones. Same kind of
order polynomial terms may be estimated for the best behavior could also be discovered with the pine oil
fit, at the presented temperature range. The constant studied for this article – the results and received model-
term is a fluid and temperature dependent factor. The ling accuracies of the pine oil are given more in detail
equation for fluid density is given in Eq. 9 (Kar- in Karjalainen (2011b).
jalainen, 2011b). Pressure is expressed in [bar] and Fluid behavior at temperatures below 0°C will
density in [kg/m³]. clearly need more research, but in theory the basic fluid
ρ(p,T) = - 1 ⋅ 10 −5 ⋅ p 2 + 0.056 ⋅ p + C D (patm ,T)
(9)
behavior should remain unchanged with fluids de-
signed for lower temperatures. Fluid behavior at pres-
For the constant term, there is a commonly used sures over 1500 bar will be researched in near future –
procedure for finding a density value at atmospheric it is possible that elevated pressure will slightly modify
pressure and at operating temperature. Eq. 10 (Hodges, the presented equations. However, its effect is insig-
1996) is based on fluid heat expansion, and gives di- nificant at the pressure range presented in this article.
rectly the needed constant term for the model. The

56 International Journal of Fluid Power 13 (2012) No. 2 pp. 51-59

 Measuring and Modelling Hydraulic Fluid Dynamics at High Pressure – Accurate and Simple Approach

5 Prediction Results
In the following, the measured results of Chapter 3
are compared to the prediction results, which were
received with the equations of Chapter 4. Due to the
length of the material to be considered, only the results
of Shell Tellus VG 46 are shown graphically in this
article. Shell Tellus VG 46 was selected due to its gen-
erally well-known behavior. More results can be found
in Karjalainen (2011b). Nevertheless, the received
modelling errors for all the eight studied fluids are
given in Section 5.4.
Fig. 10: The comparison of measured and predicted adia-
5.1 Speed of Sound in Fluid batic tangent bulk moduli of Shell Tellus VG 46, at
40°C and 70°C
Figure 9 illustrates the comparison of measured
speeds of sound in the selected Shell Tellus VG 46
mineral hydraulic oil (data points) with the predicted
ones (continuous lines), at 40°C and 70°C. With visual
examination, the prediction seems to perform well at
the presented operating range. There are no significant
uniform shift errors, and the predicted polynomials
clearly follow the measurements. The values for mod-
elling errors are given in Table 2, in Section 5.4. Possi-
ble restrictions of the presented model were discussed
in Section 4.4.

5.2 Adiabatic Tangent Bulk Modulus

Figure 10 shows the comparison of measured adia- Fig. 11: The comparison of measured and predicted densi-
batic tangent bulk moduli of the selected Shell Tellus ties of Shell Tellus VG 46, at 40°C and 70°C
VG 46 mineral hydraulic oil (data points), with the
predicted ones (continuous lines), at 40°C and 70°C. 5.4 Modelling Error
As it can be seen, the prediction method performs very The modelling errors between the prediction
well at the presented operating range. The modelling method of Chapter 4 and the measurement results of
errors are given in Table 2, in Section 5.4. Possible Chapter 3 are listed in Table 2, for all the eight studied
restrictions of the presented model were discussed in fluids. The average modelling errors in Table 2 have
Section 4.4. been calculated as percentages from the maximum
value of the fluid parameter at 1500 bar pressure (from
5.3 Density full scale, ‘FS’).
Figure 11 represents a comparison of measured Apart from few exceptions, all the studied full scale
densities of the selected Shell Tellus VG 46 mineral errors are within two percent. The results are excellent,
hydraulic oil (data points) with the predicted ones (con- especially keeping in mind the universally applicable
tinuous lines), at 40°C and 70°C. Again with visual nature of the models.
inspection, the prediction performs well at the pre-
sented operating range. The modelling errors are also Table 2: The average prediction errors of the second
given in Table 2, in Section 5.4. Possible restrictions of order polynomial models
the presented model were discussed in Section 4.4.

In Karjalainen (2011b) the average modelling errors

have also been calculated as percentages from the ab-
Fig. 9: The comparison of measured and predicted speeds
of sound in Shell Tellus VG 46, at 40°C and 70°C solute value variation of the fluid parameters between
pressures 0 - 1500 bar. In most of the studied cases,

International Journal of Fluid Power 13 (2012) No. 2 pp. 51-59 57

 Juho-Pekka Karjalainen, Reijo Karjalainen and Kalevi Huhtala

even this much tighter condition of error evaluation ing system. Also, results with water should be tested –
gives an error under five percent. there is quite good reference data available for water
When the operating conditions clearly change from dynamics.
the presented ones, some modifications to the pre-
sented models may be needed for achieving similar
accuracy – this was discussed in Section 4.4. Nomenclature

Beff Effective bulk modulus [Pa]

6 Conclusions Beff Effective adiabatic bulk [Pa]
a
modulus
In this article, simple yet accurate methods for Bn Fluid bulk modulus [Pa]
measuring and predicting the most important dynamic Bp Pipe bulk modulus [Pa]
hydraulic fluid parameters were introduced. The fluid
parameters; speed of sound in fluid, adiabatic tangent C Speed of sound [m/s]
bulk modulus and density; were studied at the normal CB(PatmT ) Constant term of tangent bulk [Pa]
operating temperatures of industrial fluid power sys- modulus
tems. The studied pressure range was up to 1500 bar. CC(PatmT ) Constant term of speed of [m/s]
These methods were applied to eight different com- sound
mercial hydraulic fluids. It was shown that at the pre- CD(PatmT ) Constant term of fluid density [kg/m³]
sented conditions both the measuring system and the
prediction method performed very well. Cp Fluid heat capacity factor at [-]
Based on the results, the studied fluids were com- constant pressure
pared in terms of dynamic fluid behavior. It is quite Cv Fluid heat capacity factor at [-]
evident that base fluid has the dominant effect on these constant volume
fluid parameters. Viscosity grade or additives did not dp Pipe inner diameter [m]
seem to have any significant impact. Practically, all the Dp Pipe outer diameter [m]
studied fluids were discovered to behave similarly in Ek Modulus of elasticity [Pa]
terms of changing temperatures. Moreover, all the fluids L Measuring pipe length [m]
were presenting similar pressure trends. However, sig- P Pressure [Pa]
nificant differences in absolute values were recorded Patm Atmospheric pressure [Pa]
between the fluids at the same operating conditions. T Temperature [ºC]
The experimental results of e.g., fluid bulk moduli Α Density correction coefficient [kg/m³·°C]
might seem to be counterintuitive. For example, it may μp Poisson’s ratio for hydraulic [-]
be surprising that the slope of bulk modulus has been pipe material
discovered to reduce as pressure is increased, when it ρn Density at measured pressure [kg/m³]
could be expected to even increase due to increasing and temperature
molecular forces of a compressed fluid. The actual  Density estimate at measured [kg/m³]
physics behind this phenomenon would clearly need ρ n
pressure and temperature
more research which is beyond the scope of this article. ρatm,T Density at atm. pressure and at [kg/m³]
However, similar reducing trend in speed of sound temperature T
measurements at elevated pressures has also been dis- ρatm,15°C Density at atm. pressure and at [kg/m³]
covered in e.g., Beyer (1998), where speed of sound in 15°C temperature
diesel fuels was discovered to follow decreasing fourth νatm,20°C Kinematic viscosity at atm. [m²/s]
order polynomials. Moreover as stated in Section 2.2.2, pressure and temperature of
the measured densities of this article have been found 20ºC
to be in excellent agreement with references even in
numerical values. These will give additional confi-
dence also to the experimental discoveries and observa-
tions of this article.
Future work is already in progress for expanding
the operating range of the measuring system. Pressure
level will be raised up to about 2500 bar. The tempera-
ture range has already been raised up to 130°C, results
will be published later. In addition, research on ex-
panding the presented prediction models to cover a
wider operating range with similar accuracy has been
started. At this stage, it would not seem to demand any
major modifications – naturally these future modifica-
tions will not affect the results of this article at the
presented operating conditions.
Fluid behavior at temperatures below 0°C needs to
be researched. It will need some controlled cooling
method which will be added to the presented measur-

58 International Journal of Fluid Power 13 (2012) No. 2 pp. 51-59

 Measuring and Modelling Hydraulic Fluid Dynamics at High Pressure – Accurate and Simple Approach

References Karjalainen, J. - P. 2011b. High-Pressure Properties of

Hydraulic Fluid Dynamics and Second Order Poly-
ASTM-D6793-02. 2002. Standard Test Method for nomial Prediction Method. Dr.(Tech.) thesis. Tam-
Determination of Isothermal Secant and Tangent pere University of Technology. Finland. (164 p.)
Bulk Modulus. USA: ASTM. (5 p.) Kojima, E. and Yu, J. 2000. Methods for Measuring
ASTM/IP. Petroleum Measurement Table 53. the Speed of Sound in the Fluid in Fluid Transmis-
sion Pipes. SAE Technical Paper 2000-01-2618.
Beyer, T. 1998. The Measurement of Diesel Fuel Society of Automotive Engineering, Inc. (10 p.)
Properties at High Pressure. M.Sc.(Tech.) thesis.
Georgia Institute of Technology. USA. (141 p.) Kuss, E. 1976. pVT-Daten bei hohen Drücken.
DGMK- Forschungsbericht, 4510/1975. (69 p. +
Borghi, M., Bussi, C., Milani, M. and Paltrinieri, F. appendixes)
2003. A Numerical Approach to the Hydraulic Flu-
ids Properties Prediction. Proceedings of SICFP‘03, Merritt, H. E. 1967. Hydraulic Control Systems.
pp. 715 - 729. Tampere University of Technology. USA: John Wiley & Sons Inc. (358 p.)
Finland. Smith, jr. L. H., Peeler, R. L. and Bernd, L. H. 1960.
Garbacik, A. and Stecki, J. S. 2000. Developments in Hydraulic Fluid Bulk Modulus – Its Effect on Sys-
Fluid Power Control of Machinery and Manipula- tem Performance and Techniques for Physical
tors. Fluid Power Net publication, pp. 227 - 257. Measurement. NFPA Publication. (19 p.)

Gholizadeh, H., Burton, R. and Schoenau, G. 2011. Yu, J., Chen, Z. and Lu, Y. 1994. The Variation of Oil
Fluid Bulk Modulus: A Literature Survey. Interna- Effective Bulk Modulus with Pressure in Hydraulic
tional Journal of Fluid Power, Vol. 12, No. 3, pp. Systems. Transactions ASME, Journal of Dynamic
5 - 15. Systems, Measurement & Control, Vol. 116, pp.
146 - 150.
Hodges, P. 1996. Hydraulic Fluids. London: Arnold.
(167 p.)
ISO 15086-2:2000. 2000. Hydraulic Fluid Power De-
termination of the Fluid-borne Noise Characteris- Juho-Pekka Karjalainen
tics of Components and Systems -Part 2 (27 p.) Born in August 1978. Received his Dr. Tech.
degree from Tampere University of Tech-
Johnston, D. N. and Edge, K. A. 1991. In-situ Meas- nology (Finland) in 2011. He is currently
working as a research fellow in the Depart-
urement of the Wavespeed and Bulk Modulus in ment of Intelligent Hydraulics and Automa-
Hydraulic Lines. Proc.I.Mech.E., Part 1, Vol. 205, tion (IHA) of the university. His primary
pp 191 - 197. research fields are hydraulic fluid dynamics,
different hydraulic fluid types and fluid
Karjalainen, J. - P., Karjalainen, R., Huhtala, K. effects on fluid power applications.
and Vilenius, M. 2005. The Dynamics of Hydraulic
Fluids – Significance, Differences and Measuring. Reijo Karjalainen
Proceedings of PTMC 2005, pp. 437 - 450. Univer- Born in February 1952. Received his Lic.
Tech. degree from Tampere University of
sity of Bath. UK. Technology (Finland) in 1996. He is cur-
rently working as a laboratory manager in
Karjalainen, J. - P., Karjalainen, R., Huhtala, K. the Department of Intelligent Hydraulics and
and Vilenius, M. 2006. High-pressure Properties of Automation (IHA) of the university. His
Hydraulic Fluids – Measuring and Differences. primary research fields are testing and
analysis of different hydraulic fluid types
Proceedings of PTMC 2006, pp. 67 - 79. University and different test equipment designs for fluid
of Bath. UK. power applications.

Karjalainen, J. - P., Karjalainen, R., Huhtala, K.

and Vilenius, M. 2007. Fluid Dynamics – Com- Kalevi Huhtala
Born in August 1957. Received his Dr. Tech.
parison and Discussion on System-related Differ- degree from Tampere University of Tech-
ences. Proceedings of SICFP’07, pp. 371 - 381. nology (Finland) in 1996. He is currently
Tampere University of Technology. Finland. working as a professor in the Department of
Intelligent Hydraulics and Automation (IHA)
of the university. He is also head of depart-
Karjalainen, J. - P., Karjalainen, R., Huhtala, K. ment. His primary research fields are intelli-
and Vilenius, M. 2009. Second Order Polynomial gent mobile machines and diesel engine
Model for Fluid Dynamics in High Pressure. Pro- hydraulics.
ceedings of ASME DSCC 2009. Hollywood, CA.
USA. (7 p.)
Karjalainen, J. - P., Karjalainen, R., Huhtala, K.
and Vilenius, M. 2011a. Comparison of Measured
and Predicted Dynamic Properties of Different
Commercial Hydraulic Fluids. Proceedings of
SICFP ’11, pp. 281 - 295. Tampere University of
Technology. Finland.

International Journal of Fluid Power 13 (2012) No. 2 pp. 51-59 59