You are on page 1of 10

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

net/publication/242415677

Estimated versus Calculated Viscous Friction Coefficient in Spool Valve


Modeling

Article

CITATIONS READS

4 2,733

1 author:

Medhat Kamel Bahr Khalil


Milwaukee School of Engineering
23 PUBLICATIONS   106 CITATIONS   

SEE PROFILE

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

No Projects View project

It is just a text book I published View project

All content following this page was uploaded by Medhat Kamel Bahr Khalil on 14 December 2015.

The user has requested enhancement of the downloaded file.


IFPE 2008 Technical Conference
Paper number: 14.2

Estimated versus Calculated Viscous Friction Coefficient in Spool Valve Modeling

Medhat K. Bahr Khalil, Ph.D.


Professional Education Instructor, Milwaukee School of Engineering
1025 North Broadway, Milwaukee, WI, 53202-3109
Tel: 414-940-2232 Email: khalil@msoe.edu

Abstract:

In most hydraulic component mathematical models developed by design engineers, the term
“viscous friction” is involved. Because of the lack of knowledge on how to calculate the viscous
friction or directly measure it, the value of the viscous friction is usually estimated and adjusted
to force the model to provide the desired performance. The model validity based on this
estimation may be an issue. In addition, viscous friction is function of certain dimensional
parameters and operating conditions. Assuming that it a constant is a harsh assumption that
affects the model validity.

In this paper a study has been made to investigate a proportional valve response where the
viscous friction is calculated based on variable working conditions. Simulation runs were made
to show the effect of variable viscous friction on the valve stability. Results are presented and
discussed.

While the study is applicable for many other hydraulic components, it is developed in this paper
for a proportional directional valve dynamic response as a typical example.

Introduction:

Viscous friction between two surfaces that have relative motion between them depends on
dimensional parameters such as contact area and clearance between the two surfaces, and also
depends on fluid properties, e.g. fluid specific gravity and viscosity. Viscous friction, as will be
shown later, was found linearly proportional to fluid viscosity, fluid specific gravity and the
contact area between the two meeting surfaces. Viscous friction was found inversely
proportional to the clearance between the two meeting surfaces.

Fluid properties are significantly affected by the working temperature, which varies in the best
case scenario within the recommended range of 25 to 65 oC. Clearances between spool and
sleeve in the hydraulic valves are in the order of microns so that the wear due to abrasive
contaminants in the hydraulic fluid has significant effect on this clearance.

1
In modeling and simulation of hydraulic components, in general, product developers often
assume constant viscous friction in order to simplify the model and keep their focus on the main
criteria they are studying. In some cases, considering viscous friction to be a constant value may
result in misleading analysis of the component characteristics and put the model validity in
doubt. Since the damping ratio is directly related to the viscous friction, viscous friction variation
with the operating conditions may affect the stability status of closed loop controlled hydraulic
systems.

Sample references who estimated viscous friction to be constant value are as follows. Khalil,
Svoboda and Bhat [1] modeled and simulated closed loop dynamic characteristics of a
proportional directional spool valve that is integrated with the variable displacement swash plate
axial piston pump. The authors considered constant viscous friction of 90 N.s/m which resulted
in a realistic damping ratio. Grabbel and Ivantysynova [2] have also considered viscous friction
coefficient as constant within a Stribeck friction model in their study of control concepts for
displacement pump actuators. Schoenau, Burton and Ansarian [3] discussed parameter
estimation techniques to predict the spring constant and spring pre-compression in the main
spool of a solenoid proportional hydraulic valve. They also assumed viscous friction coefficient
to be constant, equal to 125 Ns/m.

Some exhaustive researches were conducted to measure the friction between mating surfaces.
Scharf and Murrenhoff [4] have developed a test stand to measure the friction between piston
and bushings of an axial piston displacement unit. The authors concluded that the friction is
influenced by the piston-bushing geometry.

Our focus in this paper is to show the impact of calculating the viscous friction on a hydraulic
proportional valve open loop and closed loop dynamic characteristics versus assuming viscous
friction to be constant.

Mathematical Model:

Figure 1 shows schematically a proportional solenoid spool valve that is used typically as an
electro-hydraulic pressure compensator. The prime objective of such a valve is to control its
upstream pressure PU based on the change in its downstream pressure PD by controlling the spool
displacement. The valve controller works to change the spool position in order to stabilize PU to
be constant value if either the flow through the valve QS and/or the valve downstream pressure
PD changes. So the simple goal of the controller here is to change the valve spool position based
on certain inputs.

As shown in Fig.1, the valve is normally closed and it opens gradually based on the strength of
the input signal. The valve spool is assumed overlapped and the valve should move a
displacement XO before it starts to open. The valve travels to the maximum position Xmax, equal
4 mm, when it receives the maximum input signal.

2
Fig. 1- Schematic of the pressure compensator

The valve can be represented by a single degree of freedom spring-mass system in which the
spool mass mx is driven by a linear motor “proportional solenoid”. When the valve solenoid
receives a control signal Ixi above zero, an electromagnetic force equals Ixiki proportional to the
input signal acts on the valve spool and causes it to move until it is balanced at a displacement X
against the return spring force. A simple second order equation of motion is used to calculate the
spool acceleration as follows.

1
a= [I xi k i - f v v - k x X] (1)
mx

The spool velocity and displacement, respectively, can be found as follows

v = ∫ a dt (2)
X = ∫ v dt (3)

3
Viscous Friction Calculation

Figure 2 shows a spool moving in a sleeve. The resistive force (F) due to the viscous friction is
directly proportional to the velocity (v) and the contact area (A), and is inversely proportional to
the clearance (y). This can be written mathematically as follows

vA
Fα (4)
y

Fig. 2 - Viscous friction calculation

According to the basic definition of a Newtonian fluid, the rate of fluid distortion (dv/dy) in the
clearance (y) between two surfaces due to their relative motion is proportional to the shear stress
acting on the fluid layers. Due to the very small clearance the rate of fluid distortion will be
assumed linear. So, equation 4 can be rewritten to meet this definition as follows

F v
=µ (5)
A y

Where (µ) is the dynamic viscosity. Equation 5 can be rewritten as follows

Av
F = ν ⋅ρ ⋅ (6)
y

Where (ν) is the kinematic viscosity and (ρ) is the fluid density.

As shown in equation 1, the resistive force due to viscous friction is equal the product of the
viscous friction coefficient (fv) times the speed (v). Equation 6 can be used to deduce the viscous
friction coefficient as follows

A × ν × ρ w ×SG
fv = [ ] ×10-6 (7)
y

4
Equation 7 shows, as it was indicated in the introduction, that the viscous friction is linearly
proportional to fluid viscosity, fluid specific gravity and the contact area between the two
meeting surfaces, and inversely proportional to the clearance between the two meeting surfaces.
This demonstrates that it worthwhile to include the effect of these parameters on the viscous
friction, rather than estimating a constant value.

Simulation Results and Performance Analysis:

A Matlab-Simulink model has been developed to simulate the valve spool position in an open
loop. As shown in Fig.3, a potentiometer receives an input signal Xr and generates accordingly a
control signal in voltage form Ixv. The valve amplifier card receives the control signal and
generates the corresponding driving signal Ixi in current form with power enough to push the
valve spool against the return spring. The proportional solenoid receives the driving signal and
the valve responds dynamically based on the previously developed mathematical model. A
normalized approach has been followed to compare the reference signal Xr(%) and the actual
spool position X(%) as referred to a maximum spool displacement Xmax = 4 mm.

Fig. 3 - Valve open loop Matlab-Simulink model

An input ramp function was assumed to feed the program in order to simulate the open loop
steady state characteristic of the valve. Simulation results, shown in Fig.4, confirmed that the
valve design parameters were selected properly to obtain a linear input-output relationship.

An input signal Xr of 50% value was used to investigate the step response of the valve under
open loop conditions. Five combinations of parameters were used to simulate the valve step
response. Simulation results are presented in Figs 5A through 5E. Case 1, Fig.5, considered the
ideal values of fluid viscosity and spool-sleeve clearance that resulted in a viscous friction
coefficient of 1.92 N.s/m and valve settling time of 0.7 s. In the following four cases the fluid
viscosity was assumed to be decreased and the clearance increased. Results are concluded in
Table2. Simulation results confirm that it is unrealistic to consider the viscous friction to be
constant as these parameters change. If the viscous friction is considered constant, the simulation
will show the valve to be a stable system while sometimes, in fact, it is not stable such as case 5
where the valve experiences unstable conditions.

5
Fig. 4 - Valve open loop steady state characteristics Fig.5A - Valve open loop step response, case 1

Fig.5B - Valve open loop step response, case 2 Fig.5C - Valve open loop step response, case 3

Fig.5D - Valve open loop step response, case 4 Fig.5E - Valve open loop step response, case 5

6
Open Loop
Case1 Case2 Case3 Case4 Case5
ν (cSt) 64 48 32 16 16
y (µm) 8 8 8 8 16
Fig. # 5A 5B 5C 5D 5E
fv 1.92 1.44 0.96 0.48 0.24
(N.s/m)
Settling 0.7 1 >1 >3 -
Time (s)
Stability Stable Stable Stable Stable Unstable

Table 1 - Open loop characteristics

Figure 6 shows a simple negative feedback control loop that is intended to accurately control the
position of the valve spool. In the feedback path an LVDT displacement transducer is used to
sense the instantaneous actual spool position and convert it to a calibrated voltage signal of
voltage range (0 to 10) V. The actual spool position is subtracted at the comparator from the
reference signal resulting in an error signal Ixve. Under steady state conditions, the error signal
value is zero and the valve spool is held in its current position against the spring force. If the
error signal is greater than zero, a PID controller drives a control signal in voltage form Ixv,
which in sequence will be received by the amplifier, valve solenoid and then the position
transducer to close the loop.

Fig. 6 - Feedback control loop for valve spool position

An input signal Xr of 50% value is used to investigate the step response of the valve in closed
loop conditions. The same combinations of parameters have been used to simulate the valve step
response. Simulation results are presented in Figs 7A through 7E. As shown in Fig.7A, the fixed
parameter PID controller is tuned to stabilize the valve in 0.5 second. Simulation results of the
following four cases reveal the same trend of results that show the viscous friction value and
consequently the valve response is significantly affected by the working conditions. Results are
concluded in Table 2. Therefore considering the viscous friction to be a constant value may be a
misleading assumption.

7
Closed Loop
Case6 Case7 Case8 Case9 Case10
ν (cSt) 64 48 32 16 16
y (µm) 8 8 8 8 16
Fig. # 7A 7B 7C 7D 7E
fv 1.92 1.44 0.96 0.48 0.24
(N.s/m)
Settling 0.5 0.7 >1 - -
Time (s)
Stability Stable Stable Stable Marginal Unstable

Table 2 - Closed loop characteristics Fig.7A -Valve closed loop step response, case 6

Fig.7B - Valve closed loop step response, case 7 Fig.7C - Valve closed loop step response, case 8

Fig.7D - Valve closed loop step response, case 9 Fig.7E - Valve closed loop step response, case 10

8
Conclusion:

In order to increase the accuracy of a hydraulic component mathematical model, viscous friction
coefficients should be calculated interactively based on actual operating conditions, fluid
properties and the dimensional parameters.

Recommendations for future investigations are as follows. Investigate the effect of viscous
friction on the valve frequency response, include pressure and temperature effects on hydraulic
fluid specific gravity and confirm the results experimentally.

Nomenclature:
Symbol Description Units Value
A Spool-sleeve contact area m2 300 x 10-6
a Spool acceleration m/s2
F Force N
fv Viscous friction constant N.s/m 1.92
Ixi Spool driving signal in current Amp
Ixv Control signal in voltage Volt
Ixve Error signal Volt
ki Electromagnetic force constant N/Amp 4
kx Spring stiffness N/m 3000
mx Mass of the spool kg 0.1
SG Hydraulic fluid Specific gravity - 0.8
v Spool velocity m/s
X Actual Spool displacement m
Xmax Maximum spool displacement m 0.004
y Spool-sleeve clearance m 8 x 10-6
ν Hydraulic fluid kinematic viscosity cSt 64
µ Hydraulic fluid dynamic viscosity N.s/m2
ρ Hydraulic fluid density kg/m3
ρw Water density kg/m3 1000

References:
1. M. K. Bahr Khalil, J. Svoboda and R.B. Bhat, “Modeling of Swash Plate Axial Piston Pumps with
Conical Cylinder Blocks”, Journal of Mechanical Design, ASME Transaction, Vol.126, pp 196-200,
January 2004, USA.
2. Jeorg Grabbel and Monika Ivantysynova, “An Investigation of Swash Plate Control Concept for
Displacement Controlled Actuators”, International Journal of Fluid Power, Vol. 6 Number 2, pp 19-
36, March 2005. Germany.
3. Greg Schoenau, Rich Burton, Alireza Ansarian, “Parameter Estimation In A Solenoid Proportional Valve
Using Ols And Mlh Techniques”, Mechanical Engineering Department of University of Saskatchewan,
Canada.
4. Stephan Scharf and Hubertus Murrenhoff, “Measurement of Friction Forces between Piston and
Bushings of an Axial Piston Displacement Unit”, International Journal of Fluid Power, Vol. 6
Number 1, pp 7-17, March 2005. Germany.
5. Franklin, Powell and Abbas, “Feedback Control of Dynamic Systems”, Fourth Edition, ISBN 0-13-032393-4,
Prentice Hall, New Jersey, USA, 2002

View publication stats

You might also like