Sie sind auf Seite 1von 135

M ODELLING AND

CONTROL OF A S AUER
DANFOSS PVG-32 VALVE

5 semester project
Group MCE5-522
Department of Energy Technology
Aalborg university
Autumn/winter 2009
Title: Modelling and control of a Sauer-Danfoss PVG 32 valve

Semester: 5th

Semester theme: Mechatronic System Analysis

Project period: 02.09.09 to 17.12.09

ECTS: 23

Supervisor: Henrik Clemmensen Pedersen

Project group: MCE5-522

SYNOPSIS:
The purpose of this project has been to thor-
oughly understand the principles behind the Sauer-
Thomas Schmidt
Danfoss PVG 32 proportional valve. To achieve this
result, a mathematical dynamic model of the valve
has been developed, and implemented in Matlab
Simulink ™. Also, a linear model have been devel-
Mikael Højen oped, for the purpose to gain understanding of the
interaction between the different parameters and
also to get an estimation of how fast the system
behaves, which is a usefull information when de-
signing a controller for a system applying this valve.
Cemre Yigen
Furthermore, several tests have been carried out
for the purpose of validating the dynamic Simulink
™model. After the data was analysed, and the un-
certain parameters of the model was alteret, the val-
Morten Hyldgaard Sørensen idation was achieved. Also the results from the lin-
ear model have proven to be a good estimation of
how fast the system behaves. The main purpose of
the project is therefore achieved.

Copies: 6

Pages, total: 135

Appendix: 30

Supplements: 1 CD

By signing this document, each member of the group confirms that all participated in the project
work and thereby all members are collectively liable for the content of the report.
I

Preface
The report is written by project group MCE5-522, attending the 5. semester of Mechatronic Energy
Technology at Aalborg University. It is addressed to students on a further scientific education. The
theme for this semester is ”Mechatronic System Analysis”.

Study Guidance
Formulas in the report are numbered by chapter followed be a dot and the equation number. Sec-
tions are numbered to sub sections on the form (Chapter.Section.Subsection). Figures and tabels are
numbered the same way as equations. All references to formulas, sections and figures are referenced
by number and page.
£ ¤
Sources are cited by the Harvard-style, Last name, year , as a reference to the bibliography in the
back of the report. Appendix are stated with A in front of the pagenumber, Enclosures with a B, and
the contents of the enclosed CD with a C. Symbols and their units are described in the section Nomen-
clature in appendix. Appendix D consist of a foldout page. This is meant as a help when keeping track
of the different parts and symbols of the PVG 32. Along with this and the Nomenclature it should
allways be possiple to find the description of any symbol without having to leave the page concerned.

Contents of the CD
• The Matlab Simulink ™model.

• The report.
Table of contents

1 Introduction 1
1.1 Initial problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

2 Problem analysis 3
2.1 The directional control valve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 The internal valves and circuit diagram of the PVG 32 . . . . . . . . . . . . . . . . . . . . 3
2.3 The total structure and the modules of the PVG 32 . . . . . . . . . . . . . . . . . . . . . . 7
2.4 Exploring the PVG 32 under action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.5 Conclusion on the Problem Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

3 Problem statement 11
3.1 Problem statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

4 Problem solving 13
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4.2 Modeling the pumpside module . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4.3 Modelling the compensator spool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
4.4 Modeling the main spool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4.5 Modelling the PVE module . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.6 Modeling the load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.7 Validation of unlinear advanced model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.8 Linear model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
4.9 Controlling the valve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86

5 Conclusion 91

6 Future work 93

Bibliography B1

I Appendix B3

A Govering Equations B5
A.1 Orifice equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B5
A.2 The orifice equation in models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B7
A.3 Translatoric friction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B8
A.4 Flow force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B9
A.5 Continuity equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B10

B Test Reports B15


B.1 Compensator spool in PVP module spring . . . . . . . . . . . . . . . . . . . . . . . . . . . B15

III
IV TABLE OF CONTENTS

B.2 Compensator spool in PVB Module spring . . . . . . . . . . . . . . . . . . . . . . . . . . B17


B.3 LVDT Spring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B19
B.4 Main Spool Spring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B20
B.5 Cavities of Main Spool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B22
B.6 Volumes of the Hoses and Pipes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B23
B.7 Application in Hydraulic System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B25

C Transfer functions B29


C.1 Notes when deriving transfer functions, used in tuning the model . . . . . . . . . . . . B29

D Foldout page with overview of the PVG 32 B33


Chapter 1

Introduction

Today, hydraulics is used in various applications, from the simple hydraulic cylinder known from:
truck mounted cranes, personal lifts, wheel diggers and many other systems where a powerfull linear
force is needed, to more advanced systems, like: Power steering or hydrostatic transmissions.
Some alternatives to hydraulics are pneumatic or electrical systems. The main-problem with pneu-
matics, compared to hydraulics, is that air is much more compressible, which makes it difficult to
control and in some applications very dangerous to use.
Electrical motors and linear actuators is often a cheap alternative to hydraulics, but it is difficult to
achieve the same high linear force and torque that a hydraulic system can deliver.

(a) Mobilecrane.[Danfoss, 2006] (b) Rendegraver.[Danfoss, 2006]

Figure 1.1: Examples on mobile machinery where the PVG-valve is being used.

Some disadvantages exists when comparing hydraulics to electric components. The efficiency for
instance. If the hydraulic pump delivers flow to different hydraulic components, the total efficiency
of the system will drop radically if the load pressure at the various components varies alot.
Hydraulics is nevertheless very popular because of the extreme forces it can transmit. Whether it is a
linear force or a torque that is needed, hydraulics can deliver it with a very compact unit at the point
of action. Usually a large pump, cooler and filter system is needed.
Sauer-Danfoss, is a Danish company, who produces a wide range of hydraulic components. Among
these are the PVG Load independent valves.
The PVG valves are very flexible, because they can be controlled both manually, or by an electrical

1
2 CHAPTER 1. INTRODUCTION

input.

Figure 1.2: [Danfoss, 2006]

In this project, the group is provided with a PVG 32 valve from Sauer-Danfoss to be thoroughly anal-
ysed.
The PVG 32 valve is the smallest in the PVG-series, which also contains a PVG 100 and a PVG 120. The
PVG 32 can operate at the same pressures as it has bigger counterparts, but it can’t handle the same
amount of flow. This particular valve is used on a wide range of mobile-systems like; wheel-diggers,
forestry machines, garbage trucks, tractors and truck-mounted cranes, because of its load-sensing
principle which reduces wear and tear on the other components on the machine[Danfoss, 2006].
Since the PVG 32 valve is widely used, it intensify the need for understanding the valve, which leads
us to the Initial problem.

1.1 Initial problem

In the light of the introduction the following initiating problem can be outlined:

How does the PVG32 proportional valve work?


Chapter 2

Problem analysis

In the following chapter, the initial problem will be investigated. And therefore the general function-
ing of the PVG valve will be explained.
This section of the report will only explain the functions in Steady-state. The valve will be explained
dynamically later when the problem statement has been outlined.
First a general description of how the valve works will be given. Then each individual module of the
valve will be specified. The startup situation of the valve is described, in order to give an example of
how the different parts interact.

2.1 The directional control valve

The PVG 32 is a directional control valve which is used to control the direction, and the amount, of
fluid flow from a pump to its load.
The key part in a directional control valve is the main spool, which is moving inside the housing. The
movement of the spool will open and close for the oil flow to the different ports. The ports can be
connected to different loads, sources and reservoirs, depending on the purpose of the valve in the
hydraulic system. The PVG 32 is a 4/3 valve, which means that it has 4 ports connected to the valve,
and 3 different positions the spool can switch between. The direction of the fluid between the ports is
determined by the position of the spool. Figure 2.1 contains the hydraulic symbol of the mentioned
valve.

B A

T P
Figure 2.1: The figure displays a 4/3-way directional control valve in the middle position centered
by the springs. At this position the flow between the ports are blocked. The valve in the figure is
electro-hydraulic controlled.

2.2 The internal valves and circuit diagram of the PVG 32

A look inside the PVG 32 directional control valve reveals that three internal valves are used to create
the directional valve and to fulfill other features of the PVG 32. The internal valves are respectively a

3
4 CHAPTER 2. PROBLEM ANALYSIS

direction valve called main valve, and two pressure compensator valves.
The modules of the PVG 32 each has a certain task to insure proper function of the valve. The hy-
draulic circuit diagram of the PVG 32 with the division into the three modules is seen in Figure 2.2. A
similar figure can also be seen in Appendix D in the back of the report.

QL QV

Side A Side B
PLS
Pbp Pp
LVDT MM
P5

QB1 AH
P4 AH PVE module
To tank
A5 To tank XM
PK
A2

Pext A4 AK
MK QB3
V1
P1
A6 QB2
Qv
Xk
V2 P2 P0 A 1
PVB module

PP
Mp
AP

A3
P3 V 3 XP

QT QP
VP

RESERVOIR

PVP module

Figure 2.2: The figure illustrates the different parts of the PVG 32 with all the interior and definitions
of areas, fluid flows and pressures.

The main valve


2

The main valve is a vital part of the PVG 32, since this is the part that control the direction of flow in
and out of the PVG valve. The whole structure is build upon this main spool, which can be controlled
remotely, either using a mechanical lever or an electrical control signal. The electrical control system
is managing the position of the spool, inside the main valve. The valve itself, is a housing where the
spool can move freely inside, even though the space between the housing and the spool at some parts
are as small as a few micrometers. This valve and its control units are illustrated in Figure 2.3. The
electro hydraulic control system is the part on the right side of the spool, and the mechanical lever is
on the opposite side.
2.2. THE INTERNAL VALVES AND CIRCUIT DIAGRAM OF THE PVG 32 5

QL QV Electric control
Mechanical lever

Pp
MM

AH
AH
To tank To tank

Figure 2.3: This figure illustrates the main spool of the PVG 32, which acts as a remote controlled
valve.

The electrical control unit uses an input signal, to displace the main spool, through a secondary hy-
draulic system, with electromagnetic valves. Even if there is a change in pressure inside the valve
under operation, the position of the spool is kept constant for the same input signal. In reality the po-
sition will never be constant, but the electrical control unit will continuously make the spool converge
to the desired position. This is done by a secondary hydraulic system, which is managed through four
electromagnetic valves. The electromagnetic valves are working in pairs with a switching frequency
of 100-1000 Hz[Danfoss, 2000]; when two valves are open, two will be closed. By using an uneven
up-time, down-time period, one side of the spool will have a higher pressure than the other. Because
of the high switching frequency, the pressure changes will happen at a very short time. This will cause
the spool to move to the desired position. A hydraulic circuit diagram of the secondary hydraulic
system is given in Figure 2.4.

Electromagnetic
valve NC 1 NC 2

Main spool
Pump

NO 1 Checkball valves NO 2

Reservoir

Figure 2.4: Basic hydraulic circiut diagram of the secondary hydraulic system used to move the main-
spool.

As it is evident in the basic hydraulic circuit diagram, the electromagnetic valves are basically on-off
valves. All the valves has an electromagnetic component to manage the state by a duty cycle, and the
pump side valves also has a spring attached. The function of the springs, are to change the initial
position, when there are no input signals. While the pump side valves initially are blocking for fluid
passage (off), the reservoir side valves are open for passage (on). The electromagnetic valves can
be divided in pairs. The valves on the left-hand-side, NC 1 and N 01, are connected in series, with
6 CHAPTER 2. PROBLEM ANALYSIS

the pump in one end and the tank in the other. Between the valves one side of the main spool is
connected. All the valves are working in pairs. The valves on each side of the main spool gets the
same signal, but they are working inversely, since one of the valves are normally open, and the other
is normally closed, hence, whenever one is open for fluid passage, the other will block Danfoss [2000]

The pressure compensator valves

The two pressure relief valves are controlled internally, which means the pilot pressure that controls
the compensator spools runs internally in the PVG. If the pressure at the load rises, the rising pressure
will be led to the compensator spool and the two internally controlled valves will compensate when-
ever the pressure at the load changes. To implement this feature, the compensator valves are using
a feedback system of the load pressure and a preloaded spring attached to the spool. The existence
of the preloaded spring forces F spr i ng means that the spool will yield resistance when trying to open
for the fluid flow. The spools are initially blocking for passage, and only when a force F is present,
which is opposite and greater than the initial spring force, the spool will move and open for flow. This
is illustrated in Figure 2.5. The mentioned force F , is generated by the pump-side pressure P pump
working on the area of the spool A spool . The force F , has to overcome an initial spring force called
F cr ack , this will move the spool a distance of x cr ack . This displacement x cr ack serves to secure some
pressure drop and to block for leakage. The longer the x cr ack distance is, the less flow will run through
the cavities between the spool and the valve-cylinder.

Fspring

Fspring X Aspool
Fcrack
Mspool F
Qv Ppump
Xcrack X

Figure 2.5: This figure illustrates the principle idea behind a preloaded spring attached to a spool
inside a directional control valve. The graph on the left shows that in order to compress the spring so
that fluid can pass, a force, high enough to do so, is needed.

The main reason why the preloaded springforces initially are blocking for fluid passage, is not to block
the fluidflow, but to force the pressure inside the valve to rise in order to overcome the springforces.
This serves to block the fluidflow when the pressure inside the valve is too low, and open up again
when the pressure has increased. The pressure needed to establish equilibrium between springforce
and pressure force, for the compensator spools, are for the one in the PVB module 5 B ar and for
the one in the PVP module 15 B ar [Pedersen, 2009] higher than the payload pressure. The payload
pressure is implemented in the force-equilibrium of the spool, by leading a small amount of fluid at
the payload to the spring side of the compensator valve, so the pressure P LS works together with the
spring. This approach is called feedback, and this is the reason why the internal pressure is contin-
iously adjusted according to the payload pressure. An equilibrium diagram of the forces observed by
the spool is shown in Figure 2.6.
2.3. THE TOTAL STRUCTURE AND THE MODULES OF THE PVG 32 7

x Aspool

Fspring

Pf eedback Ppump

Figure 2.6: This figure illustrates the forces and pressures observed by the two compensator spools.
The pressure on both sides have an effect on an area of the same size

2.3 The total structure and the modules of the PVG 32

Sauer Danfoss has chosen a modular based architecture in order to build the PVG 32, by respecting the
basic principles behind the valve and to keep a flexible structure. This gives the possibility to connect
multiple PVG 32 in parallel to drive multiple actuators, typically pistons. The modules in the PVG 32
are respectively PVE (Electro hydraulic module), PVP (Pump side module) and PVB ( Compensator-
and Main spool module) modules. Each module is treated separately to give an idea of their basic
features.

• PVE module
The primary function of the PVE module is to move the main spool, and thereby changing the
passage area and direction of the fluid. By changing the passage area, simultaniously, the flow
through the PVG 32 to the payload is controlled. In order to move the main spool, the PVE
module transforms an input signal to a fixed spool displacement through a secondary hydro-
dynamic amplifier system. To assure that the position of the spool has reached the desired
position, the PVE module constantly measures the position of the spool. If needed the PVE
module will compensate for a misposition through its internal control system. This feature of
the PVE module is based on a closed loop feedback system, as seen in Figure 2.7. The input sig-
nal is constantly compared to the measured signal from the sensor. Whenever they are equal,
the system will balance the forces on both sides of the valve, to stop further displacement of the
spool.

Measured System
refference System output
error
+ Controller
input
System
-

Measured output
Sensor

Figure 2.7: The figure illustrates a simple closed loop feedback system.Wikipedia [n.d.a]

• PVP module
The PVP module has three ports. The pump is connected to one of the ports, the reservoir is
connected to another and the last port is letting the fluid flow through to the PVB module.
8 CHAPTER 2. PROBLEM ANALYSIS

The main purpose of the PVP module is to adjust the pump pressure to a level of about 15bar
above the load pressure of the highest consumer. Because the motor driving the pump often has
a fixed operating point, the flow into the valve will be constant. If this flow where just dropped to
tank at the highest possible working pressure, the effect loss would be large. Therefore the PVP
module lowers the pressure so that it is at the minimum needed level for the most demanding
actuator. If for example two cylinders where to be controlled, the one working at the highest
pressure will be the one determing the pump pressure. It is therefore very ineffective to controle
an actuator with a high pressure demand together with an actuator with a high flow demand,
or for example controlling an actuator with a high pressure demand using very little flow over
long time, because the rest of the flow will then be lead to tank at a high pressure, thus wasting
energy.

• PVB module

The last module is the PVB module. This module contains the main spool and a compensator
spool. The function of the main spool is to direct the correct amount of flow to an actuator.
Defined by the spool position. In order to do so it changes the area of the passage leading to
the two output ports, while it moves. The flow is assumed turbulent and therefore given by the
orifice equation (Equation 2.1).

s
2
Q = Cd · A · · ∆P (2.1)
ρ

The purpose of the compensator spool is to keep the pressure drop across the main spool con-
stant so that the position of the spool, and hereby the area, is the only thing determing the flow.

2.4 Exploring the PVG 32 under action

The position of the main spool either widens or narrows the passage area for the fluid. The fluid is
typically oil. In order to understand how the valve works, it is necessary to describe the process from
a working condition where the PVG 32 doesn’t lead any fluid through, to a working condition where
the fluid passage through the PVG 32 is open. This is illustrated in Figure 2.8.
Initially the main spool is blocking for fluid-passage, this is equivalent with an OFF-stage. The off
stage of the valve means, there is no fluid passage from one end of the valve to the other. The main
spool is, simply, blocking for fluid passage. Even when the passage is blocked, the pump will keep
running and pumping fluid. The fluid will fill the chambers until it meets resistance at the main
spool, and build up pressure. As the pressure rises, eventually, the spring connected to the pressure
adjustment spool will give in and open the passage to the reservoir. Because the spring is preloaded
with a force, the pressure has to rise untill the spring starts to give in. Whenever the pressure exceeds
this level, the fluid will start flowing into the reservoir. This is analogous to a motor running idle; all
the work done by the pump on the fluid is going to waste.
The picture changes when the main spool is moved so that it opens for fluid passage. This is equiva-
lent with an ON-stage. Immediately after this process, the pressure will fall inside the valve, and the
pressure adjustment spool will automatically narrow the passage to the reservoir. The fluid runs into
the actuator and it is now possible to transfer some power through the fluid. If the feedback pressure
makes the spools block for passage, the pump side pressure will just rise untill the forces is balanced
and the spool once again is in equilibrium.
2.5. CONCLUSION ON THE PROBLEM ANALYSIS 9

QV

MM MM

ON-stage OFF-stage

Figure 2.8: The picture illustrates the difference in the ON and OFF stage, which is determined by the
position of the spool.

2.5 Conclusion on the Problem Analysis

The initial problem was:


How does the PVG32 proportional valve work?
Due to the problem analysis it is now possible to, partly, answer the question. Partly; because only the
steady state operation of the valve has been analysed, although some dynamics have been introduced
in order to describe the steady state behaviour, a more extensive analysis needs to be carried out in
order to fully answer the question.
To help focussing on the right part of the problem a problem statement has been defined. This prob-
lem statement narrows down the initial problem, so that the time and effort now only concerns the
problem statement and not the broader initial problem. The statement is shown in the succeeding
chapter.

2
Chapter 3

Problem statement

3.1 Problem statement

Main problem

How to model and understand the dynamics of a Sauer Danfoss PVG 32 valve
To answer the Problem statement, the following items will be determined:

• Make a describtion of the valve which also takes the dynamics into account.

• Create a simulink model which simulate the real nature of a PVG 32 valve.

• Perform test to use for validation of the model.

• Validation of the dynamic model. The relative error of the response are not allowed to exceed
30% in order for the model to be considered valid at the given point of operation. If the relative
error is less than 10% the model are describing the reality in a good and reliable manner, at that
specific set of parameters and test conditions.

• Simplify the advanced dynamical model to a linear model. This is considered achieved if the
model can be used as a tool to better understand the valve, and the control of it.

• Understand how to use the developed tools to optimize the valve. This criteria is considered as
fullfilled if the two models can be used in the development of a controller.

11
Chapter 4

Problem solving

4.1 Introduction

This part of the report are focussing on deriving models describing the valve, and at the end validat-
ing them, so they may be used af a tool in understanding af the pvg works. This way the problem
statement will be fullfilled. A unlinear dynamic model describing allmost all parts of the PVG are de-
veloped and validated, also linear models of subsystems in the PVG are developed. Including a linear
model discribing the dynamics from the controle input to the position of the main spool. First the
unlinear dynamic model will be discribed, starting with the PVP-Module.

4.2 Modeling the pumpside module

QB2 A6
Ap QV
Pp
kc

Mp
Pp P3
Vp θ V3
A3

Xp Qp QT
PT

Figure 4.1: Illustration of the pump side module with the pressure adjustment spool

The PVP-module, or pump side module, is usually connected to a pump in a hydraulic system. The
function of this module is to adjust the pressure difference through the PVB module to a pre-defined
pressure level. As illustrated in Figure 4.1, the spool will relief the pressure if it exceeds a certain level

13
14 CHAPTER 4. PROBLEM SOLVING

defined by the spring. What happes is that the pressure will rise in chamber V3 , the spool will move
to the right and lead the fluid to tank.
As seen in the figure, there are several parameters that needs to be determined in order to establish
a dynamic model. The parameters can be split into measureable quantities and variables. The mea-
sureable quantities that needs to be determined are listed in Table 4.1.
£ 2¤
A3 Area of the opening A 3 m
£ 3¤
V3 Volume of spring chamber V3 m
£ 3¤
Vp Volume of covering from the pump piping to compensator spool chamber m
£ 2¤
A6 Cross-sectional area of the orifice A 6 m
£ 2¤
Ap Cross-sectional area of the pump side spool m
£N ¤
kp Spring constant of the spring located on the pump side spool m

Table 4.1: Measureable quantities to be used in the pump side module

The area of the spool-opening A 3 needs to be determined as a function of spool-displacement x p .


This function is derived later in subsection 4.2.1.

Initially the spool displacement is defined as x p = 0mm. This means that the spool is sticked to the
left wall. The volume of the chamber V3 is calculated based on the biggest possible chamber size.
This value is estimated to be V3 = 8cm 3 . Because the volume depends on the spool displacement, the
volume will decrease as seen in Equation 4.1.

V3 (x p ) = V3 − A p · x p (4.1)

The orifice in the pump side spool is neglected, because it is assumed that the pressure buid up in
such a small chamber will happen very fast. The volume of Vp represents the whole volume covered
by the fluid from the pump to the compensator spool, including piping and spool chambers as seen
partly in Figure 4.1 and Figure 4.7. The size of this volume are given in equation 4.2 and how the
quantity is found are explained in section B.6.

Vp = 2.7L (4.2)

The volume Vp dependens, in reality, on the travel of the spool, and should increase with a positive
spool displacement x p , but is neglected as it yields a very small change compared to size of the vol-
ume.
The mass of pump side spool M p was determined using a scale. The mass of the pump side spool was
determined to be:
M p = 70 · g m (4.3)

The pump side spring constant k p , is found in an experiment, which is described in section B.1. The
value of the spring constant is repeated here:
N
k p = 15000 (4.4)
m
4.2. MODELING THE PUMPSIDE MODULE 15

The pressure in the tank P T is determined to be 0 B ar relative or 1B ar absolute, because the tank is
open to the atmosphere.
Cross sectional area of the pump side spool A p is determined from the diameter of the pump side
spool, d p = 18mm. Thus the cross sectional area of the pump side spool is.

A p = π · r 2 = 2.54cm 2 (4.5)

4.2.1 Determination of the PVP spool opening area

The pump side compensator spool opening area is designed as a cone with different slopes as indi-
cated in 4.2a. The area function of the opening area is dependent on the displacement x p . The shape
of the spool yields a function of the resulting area. This function can be derived by measuring the
diameter of the spool at the points where the magnitude of the slope changes.
The slope φ6 has an inclination of 11.45◦ and φ7 41.82◦ . The diameter of the slopes at the different
inclinations are; r 1 = 9.00mm, r 2 = 8.84mm and r 3 = 8.15mm. These values have been used to make
two 6 point vectors of the resulting area function. These vectors can be seen as a plot in 4.2b.

-4
x 10
1.8

1.6

1.4

1.2

6 φ6 1
Area [m3]

6 φ7
0.8

0.6

0.4

PVP spool
0.2

0
r1
r2
r3

0 0.002 0.004 0.006 0.008 0.01 0.012


Displacement [m]

(a) PVP spool slopes. (b) Plot of the opening area function of the PVP module
spool.

Figure 4.2

4.2.2 Force equilibrium on the pump side spool

To determine the motion of the spool, Newton’s 2. law will be utilized. The forces that acts on the
spool is shown in Figure 4.3.
16 CHAPTER 4. PROBLEM SOLVING

FFR-P

FS
FPp
Mp
FP3

FFL3
xp A3

Figure 4.3: Forces acting on pump side spool

From figur 4.3, Newtons 2. can be utilized to describe the motion and the forces acting on the spool.

M p · ẍ p = F P p − F P 3 − F SP − F F L3 − F F R−P (4.6)

£m ¤
ẍ p Resulting acceleration of the mass s2

F SP Force of the spring at the pumpside spool [N ]

F F L3 Flow force due to the opening A 3 to tank [N ]

FP p Force due to the pump pressure on the left side [N ]

FP 3 Force due to the pressure on the right side [N ]

F F R−P Friction force between the spool and the cylinder [N ]

As explained previously, the small orifice seperating the flow to the left chamber of the pumpside
spool is neglected. By doing that, it is assumed that there won’t be any pressure build-up in this
chamber. The pump pressure P p will therefore be observed here as instantaneously. As seen from
equation 4.6, there are only two forces as a result of the pressure that needs to be determined, respec-
tively F P p and F P 3 .

FP p = P p · A 3 (4.7)
FP 3 = P 3 · A 3 (4.8)

4.2.3 The pump pressure P p and the spring-side pressure P 3

To be able to express the pump pressure P p in known terms, the flows in and out of the chamber
needs to be taken into account. The relationship is given in the continuity equation. An explanation
of the continuity equation can be found in Equation A.7).
4.2. MODELING THE PUMPSIDE MODULE 17

dV V dP
Q i n −Q out = + ·
dt β dt
dV V dP
Q p −Q v −Q T = + · (4.9)
dt β dt
V dP
Q p −Q v −Q T +Q b2 = · (4.10)
β dt

There are three flows through Vp . The flow from the pump Q P is dependent on the size and type of
the pump. According to the datasheet, this PVG 32 can handle flows up to 130l /mi n[Danfoss, 2000].
The flow to the PVB module Q v , is dependent on the main spool position. This will be explored in
the next chapter about modelling the main spool. A look on Figure 4.1 reveals that a movement of
the spool will give cause to a change in volume as stated in equation 4.9. This change in volume will
have an effect on both the left and right chamber of the spool. In order to take this volume change
into account, the change in volume is replaced with the flow Q B 2 , which is the flow that runs through
the pilot channel into the spring chamber. This is visible in equation 4.10. This replacement assumes
that the flow into the spring chamber equals the volume change of Vp , which is an assumptions which
require the pressure build up in the spring chamber to be much faster than the pressure build up in
the pump chamber. The flow Q B 2 is determined using the orifice equation.

s
2
Q B 2 = C d · A 6 (x p ) · (P 2 − P 3 ) (4.11)
ρ

Here ρ is the density of the oil. The density is dependent on the temperature and normally it varies
kg kg
from 850 − 950 m 3 , but is assumed to be a constant at 900 m 3 . The flow discharge coefficient is defined
to be C d = 0.6. P 2 is the pilot pressure in the spring chamber of the PVB modul. This will be evident
in the next chapter about the compensator spool modulation.
The last unknown term in equation 4.10 is the Q T , which is the flow to the tank. This is, once again,
calculated using the orifice equation:

s
2
QT = Cd · A3 · · (P p − P T ) (4.12)
ρ

Now that the flows in- and out of the spool is determined, an expression for the pressure P p can be
d Pp
derived by solving the continuity equation with respect to dt :

d Pp β
= (Q p −Q v −Q T +Q B 2 ) · (4.13)
dt Vp

Where β is the Bulk Modulus, which is determined to be 7000B ar as stated in section A.5.
To determine the pressure in the spring chamber P 3 , the continuity equation is used, and in the same
way as it was when finding P p :

V3 − A p · x p d P 3
Q B 2 = −A p ẋ p + (4.14)
β dt
d P3
By solving equation 4.14 with respect to dt the following expression is found:

d P3 β
= (Q B 2 + A p · ẋ p ) · (4.15)
dt V3 − A p · x p
18 CHAPTER 4. PROBLEM SOLVING

4.2.4 The flow and spring forces at the pumpspool

The flow force F F L3 is a variable force that will always work in the direction closing the valve. The flow
force is determined by Equation 4.16.

F F L3 = 2 ·C d · A 3 · (P p − P T ) · cosθ (4.16)

Where θ is the angle of the flow which is shown in figur 4.1 on page 13. Since there are no slot’s in the
spool, the angle of the flow is determined to be 69◦ [Andersen & Hansen, 2007b].
The force of the spring F SP is a function of the spring constant k p and the displacement x p . The
spring is always decreased in length, which means there will be a constant spring force, even when
pump spool hasn’t been moved and x p is zero. The constant spring force F pr el oad −p is calculated in
section B.1 with a distance x pr el oad −p = 20mm to be:

F pr el oad −p = 300N (4.17)

By adding the preload to the spring force, a function for the overall spring force is obtained:

F SP = k p · x p + F pr el oad −p (4.18)
3
F SP = 15 · 10 N · x p + 300N (4.19)

4.2.5 The viscous and coulomb friction forces at the pump spool

In section A.3 the friction force is explained in details. This chapter divides the friction force into
three different friction forces; striction, coulomb and viscous friction. Measuring the friction forces is
complicated, and the division in the three types of friction force is advanced. Available data in the lit-
terature divides the friction force in a simpler model with two quantities, respectively a static friction
force and a kinetic friction force. Both of these quantities coefficients are usually given as a con-
stant. Therefore these quantities has similarities to the more advanced model of the friction forces.
The static friction force corresponds to striction and coulomb combined, and the kinetic friction cor-
responds to the coulomb friction force. The static friction force will only have an effect between
the spool and the cylinder, when the spool is at rest, and because the spool is almost never at rest,
the static friction force will be neglected. The kinetic friction force will therefore be the same as the
coulomb friction force. In order to look up the kinetic k k friction coefficients, the shape and material
needs to be examined. It is assumed that the spools and the cylinders are made of steel, and there
are some leakage oil in between the objects. For this reason the friction coefficients are found for
lubricated steel. The kinetic friction coefficient differ, depending on the source. The sources agree on
a value from 0.05 to 0.1 for the kinetic friction coefficient, Based on these observations the coefficient
for kinetic friction between two lubricated steel plates are:

k k = 0.05 (4.20)

The friction force depending on this coefficient yields a constant value to the total friction force. This
friction force is dependent on the normal force between the cylinder and the spool. Because it is
assumed that the spool is placed horizontal, the normal force for a spool in motion will be the grativy
force multiplied with the mass. Eventhough there is high pressure fluid inside, these won’t have an
effect on the kinetic friction, because the pressure will be observed on both sides of the spool as seen
in figure 4.4.
4.2. MODELING THE PUMPSIDE MODULE 19

Pp
Ft
Fsk

Figure 4.4: The kinetic friction force between the spool and cylinder

The kinetic friction force are of the following size:

m
F sk = k k · F t = k k · g · m p = 0.05 · 9.82 · 0.070kg = 0.0344N (4.21)
s2

The edge effect on the spools

Another important contribute to the friction force is, the fact that the edge of the spool will hit the
cylinder as seen in figure 4.5 and the fact that there aren’t much fluid around these edges. However,
this is just an illustration and these conditions are heavily exaggerated.

Ft Pp
Fse

Figure 4.5: The kinetic friction force between the spool and cylinder because of the edges

This will cause the oil in between the cylinder and the housing to be minimized, which makes a kinetic
friction coefficient for dry steel probable. This friction coefficient, once again, vary with the source
and is defined to be:

µke = 0.57

The normal force in this case will be given by the pressure observed by the cylinder. Because each
opposite edge of the spool hits opposite sides of the cylinder, only the pressure working on one half
of the cylinder with the length l 1/2 will contribute to the normal force for each edge. A non uniform
shape of the spool with respectively a diameter of D 1 = 18mm and D 2 = 9mm yields the following
total surface area of the spool.

D1 D2
A SP = · π · l1 + · π · l2
2 2
= 1289.88mm 2 (4.22)
20 CHAPTER 4. PROBLEM SOLVING

Aactive
D1 D2

Figure 4.6: The active area of the cylinder which on which the pressure works as the normal force

Noticing that only the contribution from the pressure working on the upper part of the cylinder will
have an effect, the active area reduces likewise as seen in the previous figure to one half again and the
active area are calculated based on this equation:
1
A ac t i ve = · A Sur f ace
4

The normal force given by a pressure working on the active area, when the gravity force is neglected
is given by:
F n = P f f · A ac t i ve

As an example, for a pressure of P f f = 100B ar and an active area of A ac t i ve = 0.001m 2 yields the fol-
lowing result, which makes it obvious why the gravity force and the kinetic friction force for lubricated
steel calculated in Equation 4.21 are neglected:

F n = P f f · A ac t i ve = 100 · 105 P a · 1 · 10−4 m 2 = 10kN

It should be noticed that this way of finding the friction force, will only give a very rough starting point
for the model, it could be far from the friction seen experienced in reality.
The friction force based on the edges will work on both sides of the cylinder. In order to take this into
account the force needs to be multiplied by 2. The friction force, because of this effect, will give a
constant contribution to the total friction force:

F f c = 2 · k ke · F n = 2 · k ke · A ac t i ve · P f f (4.23)

This parameter can vary with the pressure and the actual active area, but because it is much larger
than the friction force caused by the gravity, the total coulomb friction force will be determined by this
edge effect. The pressure inside the pump module is assumed to be the pump pressure P p . However,
besides the variables that has an impact on this edge effect, there is another important and unknown
factor; How often will this edge effect occur? This can not be known, and the coulomb friction will be
a parameter which is adjusted in order to tune the model.

The viscous friction coefficient

The viscous friction force is another contribution to the total friction, which is a term in the advanced
friction model as explained previoulsy. This friction is dependent on the viscosity of oil, the distance
between the spool and cylinder, and the velocity of the spool, as seen in the shear stress equation:
du
F f v = τoi l = µ · (4.24)
dy

According to the datasheet of the PVG 32 the kinematic viscosity of oil is given to be in the range from
νoi l = 12−75mm 2 /s. Using a value of νoi l = 32mm 2 /s to find the shear stress, yields a transformation
4.2. MODELING THE PUMPSIDE MODULE 21

to a dynamic viscosity given by equation 4.25. The velocity gradient can be calculated in the manner
of the velocity of the spool and the smallest perpendicular distance between the housing walls and
the spool as seen in equation 4.26. Due to the precision of the materials, this perpendicular length is
assumed to be l = 5µm. The velocity is zero at the housing, and at the spool it is the velocity of the
spool.

µoi l = νoi l · ρ oi l = 32 · 10−6 · 900 = 0.029 (4.25)


du 1
= · ẋ p = 2 · 105 · ẋ p (4.26)
dy 5 · 10−6

The total friction force is given as the coulomb and the viscous force contribution as seen in Equa-
tion 4.27. Because the values in this equation is based on some parameters that can vary a lot, a more
general form of the friction force, with a coulomb and a viscoius friction coefficient, is derived.

F F R−P = Ff v +Ff c

νoi l · ρ oi l
F F R−P = · ẋ p + 2 · k ke · P f f · A ac t i ve
dy

F F R−P = c v · ẋ p + c c (4.27)

The coefficients to the viscous friction force and the coefficient for the coloumb friction force is sum-
marized to viscous coefficient c v and a coloumb coefficient c c as stated in Equation 4.28.

νoi l · ρ oi l
cv = dy
(4.28)
cc = 2 · k ke · P f f · A ac t i ve

Inserting the value calculated in this chapter yields the values of the friction coefficient for the pump-
side spool. Both the viscous friction coefficient c v and the coulomb friction coefficient c cp can be
adjusted because of numerous unknown parameters.

kg
cv = 5760 [ s ]

cc p = 2 · 0.57 · 322.47 · P P [N]

4.2.6 Conclusion

The dynamics in the PVP module have been described mathematically, and Newton’s second law
have been utilized to described the motion of the spool. By applying the forces found in the previous
chapter, Newton’s second law, decribing the motion of the spool, is as decribed in Equation 4.29 The
final model with all the terms derived of the main spool dynamic and these expressions are inserted
in equation 4.29.
¡ ¢
M p · ẍ p = (P p − P 3 ) · A p − k · x p − 2 ·C d · A 3 · (P p − P T ) · cosθ − c v · ẋ p + c cp (4.29)

To model the dynamics of the PVP module the following assumptions have been made:
22 CHAPTER 4. PROBLEM SOLVING

• Constant flow from the pump Q p .

• Constant flow angle θ.

• Instantaneous pressure P p in the pressure chamber, hence neglecting the orifice.

• Pressure in the reservoir is kept constant at 0bar relative.

• The coulomb friction is neclected for now.

4.3 Modelling the compensator spool

The compensator spool, is located in between the main spool and the pressure compensator spool.
The function of this spool is to keep the pressure a certain amount above the load pressure. This way
the spool will compensate for a fall or rise in pressure when ever needed, in order to keep a constant
pressure drop from the compensator spool to the load. Another feature of the compensator spool is
to block backwards fluid flow, if the pressure from the pump should fall below the load pressure.

QB1 A5
PK
Qv

A2
Ak kc2
c
kcc1
A4
V1 Mk Q
B2 A6
P1 A1 P0 P2
Vp φ1 V2

xk PP
Qv

Figure 4.7: A sketch of the compensator spool.

In Figure 4.7, the constants and variables that needs to be determined to model the compensator
spool, can be seen. Before determining equations for the variables, the constants will be determined.
Table 4.2 shows as list of the constants and variables that will be determined in this section.
Determination of the volume of the spring-chamber V2 and the volume of the pressure chamber V1
is changing the volume of the cavities before movement of the spool, to the volume emerged or con-
tracted as an impact of change in x k . The original volumes are determined based on a zero point
when compensator spool is at the most left position, as seen by the x k mark in Figure 4.7. The initial
volume V1 is actually 0, but in order to make the model work, a small volume has been addded.
4.3. MODELLING THE COMPENSATOR SPOOL 23

£ 3¤
V1 Volume of pressure chamber V1 m
£ 3¤
V2 Volume of spring chamber V2 m
£ 2¤
A1 Area of the spool opening A 1 m
£ 2¤
A2 Area of the spool opening A 2 m
£ ¤
Mk Mass of compensator spool kg
£ 2¤
Ak Cross-sectional area of the compensator spool m
£N ¤
k c1 Spring constant of the spring located on the compensator spool m
£N ¤
k c2 Spring constant of the spring located on the compensator spool m
£ 2¤
A4 Cross-sectional area of the orifice A 4 m
£ 2¤
A5 Cross-sectional area of the orifice A 5 m
£ 2¤
A6 Cross-sectional area of the orifice A 6 m

Table 4.2: Measurable constants, necessary to model the compensator spool

V1 x p = 0.5 · cm 3 + A k · x k
¡ ¢
(4.30)
V2 x p = 1 · cm 3 − A k · x k
¡ ¢
(4.31)

Cross sectional area of the compensator spool A k is determined from the diameter of the compen-
sator spool, d = 18 · 10−3 m. Thus the cross sectional area is:

A k = π · r 2 = π · (9 · mm)2 = 2.54 · cm 2 (4.32)

The mass of the compensator spool was determined by using a scale, which revealed a mass of

m k = 66g (4.33)

There are two springs in the compensator spool, which leaves two spring constants to find. There is a
thorough description on how the spring constants was determined in section B.2.

N
k c1 = 9945 (4.34)
m
N
k c2 = 35000 (4.35)
m

The longest spring that is active all the time, has the smallest spring constant k c1 , while the short
spring has a larger spring constant. This comes in sight when looking at the combined spring constant
k c2 . The reason for this is to avoid that the spool travels too far in the positive direction x k .

4.3.1 The areas in the compensator module

The orifice A 4 , that is located in the compensator spool, which leads fluid to the pressure chamber V2
on the left side of the spool is determined to have a diameter D 4 = 1.5mm which yield a cross sectional
24 CHAPTER 4. PROBLEM SOLVING

area as stated in equation 4.36. The orifice areas A 5 and A 6 , which are located in the pilot-lines has
the same diameter D 5 6 = 1mm and this yields a cross sectional area given in equation 4.37.

A 4 = 3.5 · mm 2 (4.36)
2
A 5 = A 6 = 1 · mm (4.37)

There are two discharge areas between the compensator spool and the housing from which the fluid
flow Q v can pass through. These discharge areas are created two different places in the spool. The first
discharge area A 1 will decrease with a positive spool displacement x k as a result of a higher pressure
drop between the pump and load pressure. Contrary, the opening area A 2 will increase with a positive
displacement x k . The spool opening that creates the discharge area A 2 has a cylinder shape with a
uniform diameter D A2 = 18mm. Whenever there is a gap between the spool and the housing, fluid
is free to flow through this gap.The gap determines the discharge area for A 2 . This disharge area is
calculated in equation 4.38 as a function of the displacement x k .

A 2 = π · D · x k = π · 18 · mm · x k (4.38)

The part of the spool that creates the discharge area A 1 is cylinder shaped as well. This part of the
spool is not uniform, instead numerous cavities are drilled in it at the edges. With these cavities there
doesn’t have to be a gap between the spool and the housing before fluid can flow. Whenever the spool
is near the edge, the fluid is free to flow through these cavities. Because of this, the discharge area
is no longer cylinder shaped. The shape of these cavities are based on two different slots, and there
are four of each these types of slots at the edge of the spool. This arrangement of the slots makes the
discharge area non-uniform. Using the surfaces of a cylinder, but only where the slots are placed will
give one result, while using the perpendicular area of the slots will give another. The discharge area
for a fluid through an orifice, is based on an opening from which it is possible for the fluid to flow
through. Because of this fact, if there are two discharge ares, the fluid flow will depend mostly on
the smallest. Using the program AMOC [Power, 1999], it is possible to determine the smallest cross
sectional discharge area A 1 as a function of the spool displacement x k . The slot arrangement and the
shape of the 2 different cavities in the compensator spool are illustrated in Figure 4.8:

(a) Slot arrangement (b) Disc 1. (c) Disc 2.

Figure 4.8: Slot arrangement and slot geometry


4.3. MODELLING THE COMPENSATOR SPOOL 25

As seen in 4.8a, there are 8 slots; 4 smaller and 4 bigger slots. Disc 1 is the name of the bigger slots
which is illustrated in 4.8b, while Disc 2 are the smaller slot and is illustrated in 4.8c.

(a) Horizontal discharge area. (b) Vertical discharge area. (c) Discharge area

Figure 4.9: Discharge area

In 4.9c, graphs of the discharge area as a function of the spool displacement x k is shown. The figure
contains plots for the discharge area from respectively the four disc 1 slots and four disc 2 slots and
a total discharge area as a combination of all of the slots. It can be seen, that the discharge area
for Disc1 is allmost constant when the spool displacement exceed approximatly 4.5mm, which is
because the smallest discharge area changes from a horizontal surface area as shown in 4.9a, to be
the vertical cross area as shown in 4.9b. After defining the slots in AMOC, the program generates a
[2x100] matrix, that contains a vector describing the total opening area in the compensator spool, and
a vector describing the spool travel. To get a better resolution, Linear interpolation is used to achieve
10000 points in the table instead of 100.

4.3.2 Force equilibrium on the compensator spool

In order to determine the dynamics of the compensator spool, Newtons 2nd law will be utilized. The
forces acting on the spool are shown in Figure 4.10:
26 CHAPTER 4. PROBLEM SOLVING

FFF2 FFR

A2
FSC
FP1
FP2
Mk
A1
FFF1

X XK

Figure 4.10: Forces acting on the compensator spool

After determining the forces that acts on the compensator spool shown in Figure 4.10, Newton’s 2nd.
law can be utilized for determining the motion of the compensator spool:

M k · ẍ k = F P 1 + F F F 1 − F P 2 − F F F 2 − F F RC − F SC (4.39)

£m ¤
ẍ k Resulting acceleration of the mass s2

FP 1 Force due to the pressure in the left chamber [N ]

FP 2 Force due to the pressure in the right spring chamber [N ]

F F RC Friction force between the spool and the cylinder [N ]

F SC Force of the spring [N ]

FF F 1 Flow force due to the opening A 1 [N ]

FF F 2 Flow force due to the opening A 2 [N ]

In order to calculate the forces involved in this equation, the pressures having an effect on the com-
pensator spool needs to be found. The pressures in each end of the spool, respectively F P 1 and F P 2
have a direct effect on the spool, while the pressures effecting the flow has an impact on the flow
forces. These quantities will be derived in the following section.

4.3.3 Pressures inside the compensator spool

The pressure P p is already determined in connection with the pumpside module by Equation 4.13.
As it was evident in the previous section, it is only needed to derive an expression for the pressures
P 1 , P 2 , P 0 and P k in order to determine the pressure forces and flow forces.

To determine pressure drops through the compensator spool, it is necessary to determine the flow
through the compensator spool QV , which is dependent on the load pressure P LS . So in order to
achieve this, the main spool discharge area A H needs to be brought into the equation (the derivation
of A H will be discussed in section 4.4). By utilizing the orifice equation, a series of equations that
4.3. MODELLING THE COMPENSATOR SPOOL 27

describes the flow QV can be set up.

s
Qv 2 1 ρ
µ ¶
2
Q v = Cd · A1 · · (P p − P 0 ) ⇔ P p − P 0 = · 2· (4.40)
ρ Cd A1 2
s
Qv 2 1 ρ
µ ¶
2
Q v = Cd · A2 · · (P 0 − P k ) ⇔ P 0 − P k = · 2· (4.41)
ρ Cd A2 2
s ¶2
Qv 1 ρ
µ
2
Qv = Cd · AH · · (P k − P l s ) ⇔ P k − P l s = · 2 · (4.42)
ρ Cd AH 2

For simplification these equations is merged to one equation with all the orifice areas merged to an
equivalent area. Summarizing all the pressure drops and putting them equal to the pressure drop of
the equivalent orifice yields:

Qv 2 ρ Qv 2 ρ
³ ´ ³ ´ ³ ´
1 1 1
Cd ·2 · A 21
+ A 22
+ A 2H
= Cd ·2 · A12
eq

m
1 1 1 1
A 21
+ A 22
+ A 2H
= A 2eq

m
r
1
A eq = 1
+ 1
+ 1
A2 A2 A2
1 2 H

Lastly the orifice equation for the equivalent orifice:

s
Qv 2 1 ρ
µ ¶
2
Q v = C d · A eq · · (P p − P l s ) ⇔ P p − P l s = · 2 · (4.43)
ρ Cd A eq 2

For this equivalent orifice equation to be valid, the volumes between the orifices must be small and
the flow in each orifice must be the same. Allthough a very small amount of fluid will leak to the spring
chamber due to the pressure difference and a tiny opening between the spool and the surrounding
walls in the housing, this leakage can be neclected, because the cavity in the spring chamber is very
small and the larger fluid flow contribution to this spring chamber from the pilot line. It is therefore
legit to assume that the same flow will occur in all the mentioned orifices. As regards to the volumes
in between the orifices, not only are the chambers very small, the fluid flow Q v is large during op-
eration. If these chambers wasn’t neglected, the pressure build up in these chambers would happen
allmost instantaniously because of the large flow. Another important fact to point out, is that if all the
orifices observe the same flow, the flow in and out of all these cavities will be zero. In addition, if the
volume can’t change, there wont be any change in pressures as a result of the fluid flow. Therefore
it is also reasonable to neglect the volumes in between the mentioned orifices. The impact of these
assumptions, is that a change in pump pressure, will be faster for the main spool, than it would be in
reality.

Solving Equation 4.40, 4.41, 4.42 and 4.43 for P P , P 0 , P k and P l s respectively, gives these equations:
28 CHAPTER 4. PROBLEM SOLVING

¶2
Qv 1 ρ
µ
P p = P0 + (4.44)
Cd A 21 2
Qv 2 1 ρ
µ ¶
P0 = Pk + (4.45)
C d A 22 2
Qv 2 1 ρ
µ ¶
Pk = Pl s + (4.46)
C d A 2H 2
Qv 2 1 ρ
µ ¶
Pl s = P p − (4.47)
C d A 2eq 2

The flow Q B 3 through the orifice with the discharge area A 4 needs to be found in order to determine
the pressure P 1 . By utilizing the orifice equation, the following occur:

s
2
QB 3 = Cd · A4 · · (P 0 − P 1 ) (4.48)
ρ

To determine the pressure P 1 , the continuity equation is utilized:

V1 + A K · x k d P 1
Q B 3 = A K · ẋ c + · (4.49)
β dt

d P1
By solving Equation 4.49 with respect to dt :

d P1 β
= (Q B 3 − A K ẋ c ) (4.50)
dt V1 + A K x k

The pressure P 2 in the spring chamber V2 , is determined by the flow into the chamber (Q B 1 -Q B 2 ). Q B 2
was determined by Equation 4.11, so whats left to be found is Q B 1:

s
2
QB 1 = Cd · A5 (P LS − P 2 ) (4.51)
ρ

Now that there is derived an expression for both Q B 1 and Q B 2 , the pressure in the spring chamber P 2
can be determined by utilizing the continuity equation for the spring chamber:

V2 − A K · x k d P 2
Q B 1 −Q B 2 = −A K · ẋ k + (4.52)
β dt

d P2
By solving equationEquation 4.52 with respect to dt :

d P2 β
= (Q B 1 −Q B 2 − A k ẋ k ) (4.53)
dt V2 − A k · x k

As a result of the areas and the found pressures the pressure forces observed at both ends of the
compensator spool, can be calculated using the relationship between force, pressure and area:

FP 1 = P1 · Ak
FP 2 = P2 · Ak
4.3. MODELLING THE COMPENSATOR SPOOL 29

Because the area on both ends of the compensator spool are the same, these forces can be combined
to one equivalent equation where the direction of the force are taken into account, as seen in Equa-
tion 4.54.

F P 12 = F P 1 − F P 1 = (P 1 − P 2 ) · A k (4.54)

4.3.4 Flow forces in the compensator spool

As seen from Figure 4.10, there are two flow forces acting on the spool F F F 1 and F F F 2 . Flow forces
will always work in the direction to close the opening where the flow occurs. This means that the two
flow forces involved in the compensator spool works in opposite directions. The flow force F F F 1 that
occures at the spool opening A 1 can be described as shown in Equation 4.55:

F F F 1 = 2 ·C d · A 1 · (P p − P 0 ) · cos(φ1 ) (4.55)

φ1 The angle of the flow as shown in Figure 4.7. [r ad ]

Pp The pressure on the pumpside of the compensator spool [P a]

P0 The pressure between the two openings A 1 and A 2 . [P a]


£ 2¤
A1 Discharge area as a function of the displacement x k . m

The flow force F F F 2 occurs at the spool opening A 2 , and is given by Equation 4.56:

F F F 2 = 2 ·C d · A 2 (P 0 − P k ) · cos(φ2 ) (4.56)

φ2 The angle of the flow through the orifice with the area A 2 [r ad ]

Pk The pressure in the chamber between the main and compensator spool [P a]
£ 2¤
A2 Discharge area as a function of the displacement x k . m

4.3.5 The spring forces in the compensator spool

There are two springs in the compensator spool which are being turned in to a single equivalent
spring force F SC . The first of the springs has a preload, and the other spring isn’t activated before the
spool displacement reaches 5.2mm. This force depends on the spring constant k c1 and the displace-
ment x k . Because the spring is preloaded, the spring is allready loaded at the chosen zero point for
the displacement vector x p as seen in Figure 4.7. This initial deflection of the spring is estimated to
be x spr i ng = 10.57mm. Using this value, the preload of this spring F pr el oad is calculated in section B.2
and repeated here:

F pr el oad = 105.12N (4.57)

The size of x spr i ng is the minimum compression of the spring, which also is the distance where x k is
defined to have it’s zero point x k = 0. By adding the preload to the spring force, a function F SC −1 for
the spring that is always active is found:
30 CHAPTER 4. PROBLEM SOLVING

F SC −1 = k c1 · x k + F pr el oad −k
N
F SC −1 = 9945 · x k + 105.12N (4.58)
m

When the spool displacement exeeds 5.2mm, both springs will be compressed. The spring force will
then be a function of a new preload F pr el oad −k2 to secure that in the transition both functions will
have the same value. The spool displacement x k and a combination of the two springs which is de-
fined to be k c 2 will also be a part of the function, which is expressed by:

F SC −2 = k c2 · x k + F pr el oad −k2

At a deflection of x k = 5.2mm the spring force F SC −2 out to have the same value as F SC −1 . The spring
force F SC −1 for the longest spring at the given point is:

N
F SC −1 (5.2mm) = 9945 · 5 · mm + 105N = 155N
m

Based on this observation, the following calculations, will determine the second preload value:

F SC −2 (5.2mm) = F SC −1 (5.2mm)
3N −3
35 · 10 · 5.2 · 10 m + F pr el oad −k2 = 155N
m
N
F pr el oad −k2 = 155N − 35000 · 5.2 · mm
m
= 155N − 175N (4.59)
= −20N (4.60)

This yields a spring force model for the part where both the springs are active, which is expressed in
Equation 4.61

F SC −2 = k c2 · x k + F pr el oad −k2
N
F SC −2 = 35000 · x k − 20N (4.61)
m

The two expressions determined for the springs is combined in one equation, which is given in Equa-
tion 4.3.5.


 9945 N · x k + 105N if 0mm < x k ≤ 5.2mm
m
F SC =
N
35000 m · x k − 20N if x k > 5.2mm

4.3.6 The friction forces in the compensator spool

The friction-force F F RC depends on the velocity of the spool, and will always work against the moving
direction of the spool. The friction force for the compensator spool is in aggrement with the friction
force derived for the pressure adjustment spool in Equation 4.27. The coefficients for the viscous
4.3. MODELLING THE COMPENSATOR SPOOL 31

friction coefficient c v will be the same, if assumed the proporty of oil doesn’t change from one module
to the other. The coulomb friction coefficient c c is given in Equation 4.27. As the active area and the
pressure are the only values that are different, calculating these will yield a value for the coloumb
friction coefficient. The pressure inside this module is different for different parts of the spool, but
is assumed to have the pump pressure P F F = P p . The spool has a diameter of D 1 = 18mm and D 2 =
9mm, which yields a surface area of:

D1 D2
A SC = · π · l1 + · π · l2
2 2
1.8 0.9
= · π · 2.9 + · π · 3.7
2 2
= 13cm 2 (4.62)

And the active area will become a quarter of this size as explained in the previous deriviation.

1
A ac t i ve−c = A SC
4
= 3.3cm 2 (4.63)

The coulomb friction coefficient will now be calculated. As stated previously the friction coefficient
c v can be adjusted because of numerous unknown parameters.

kg
cv = 5760 [ s ]

c cc = 3.7 · 10−4 · P P [N]

4.3.7 Conclusion

After inserting all the terms that are derived previously in this chapter, the model for the compensator
spool will be as stated in equation 4.64 for a spool travel x k ≤ 5.2mm:

M k · ẍ k = (P 1 − P 2 ) · A k + 2 ·C d · A 1 · (P p − P 0 ) · cos(φ1 )
−(k c1 · x k + F pr el oad −k ) − (c v ẋ p + c cc ) − 2 ·C d · A 2 (P 0 − P k ) · cos(φ2 ) (4.64)

And in equation 4.65 for a spool travel x k ≥ 5.2mm:

M k · ẍ k = (P 1 − P 2 ) · A k + 2 ·C d · A 1 · (P p − P 0 ) · cos(φ1 )
−(k c2 · x k + F pr el oad −k2 ) − (c v ẋ p + c cc ) − 2 ·C d · A 2 (P 0 − P k ) · cos(φ2 ) (4.65)

This is all the forces acting on the compensator spool.

To model the dynamics of the compensator spool the following assumptions have been made:

• Constant flow angles

• Instantaneous pressure build up in the chambers containing P k and P 0 .

• The coulomb friction is neclected for now.

This concludes the modelling of the compensator spool, and the focus will now be directed to the
main spool.
32 CHAPTER 4. PROBLEM SOLVING

4.4 Modeling the main spool

QL QV
Side B
Side A
Pbp PLS
LVDT
MM
φ3 φ3 P5

A5 AH
P4 AH
PK xM
To tank To tank

Figure 4.11: Main spool

4.4.1 Calculating the pressures in the main spool module

The pressure drop over the main spool are calculated as:

¶2
Qv 1 ρ
µ
Pk − Pl s = (4.66)
Cd A 2H 2

The pressures P 4 and P 5 is determined by the electro-hydraulic actuation and this will be described
in details in section 4.5. The model of the main spool consist of a number of measurable constant.
The mass of the main spool was determined, using a scale to be:

M m = 185g (4.67)

4.4.2 Calculating the areas in the main spool module

The area A H , of the main spool opening is determined as a function of the displacement x m . x m is
0 when the main spool is in the middle position, i.e. when no oil is flowing. In 4.14a a plot of the
opening area of the main spool in terms of spool displacement can be seen. It is clear that there is a
deadband of 1.5mm in both directions. At maxmimum displacement x m = 7mm the opening area is
69.549mm 2 . It is also seen that the area function is symmetric about the point x m = 0.
The displacement of the main spool is controlled either by the manual lever connected directly to
the spool or by the electro-hydraulic actuation as mentioned in section 2.3. The opening area in the
mainspool A H as a function of x m is, 2like for the compensator spool, determined using the program
AMOC ™. 4.12a shows the arrangement of the slots. There is a total of 4 slots at the edge of the main
spool, and in figure 4.12c on the facing page, the geometry of one of the slots is illustrated, seen from
above.
4.4. MODELING THE MAIN SPOOL 33

(a) Slot arrangement (b) Angle of the slot. (c) Slot geometry.

Figure 4.12: Slot arrangement and slot geometry

As seen in 4.12b, unlike the slots in the compensator spool, the slots in the main spool is angled
which will have an impact on the flow angle, which will be determined later on.
As stated earlier, the main spool has 4 slots per edge, which all have the same shape. The geometry of
the slots yields a total discharge area as a function of the spool displacement, as shown in 4.13b

(a) Angle of discharge area. (b) Total discharge.

Figure 4.13: Total discharge

The discharge area is always the smallest possible cross sectional area, 4.13a shows how the area is
determined. Unlike the function of the discharge area for the compensator spool, the discharge area
34 CHAPTER 4. PROBLEM SOLVING

for the main spool, as shown in 4.13b is more uniform, because the smallest cross sectional area fol-
lows the angle of the slots.

-5
x 10
7

5
Discharge area [m2]

2 l
φ3
h
1

Main spool
0
-8 -6 -4 -2 0 2 4 6 8
Spool displacement [m] x 10
-3

(a) Opening area in terms of spool displacement. (b) Slope of slot in main spool

Figure 4.14

As explained previously in section 4.2 AMOC creates a matrix, describing the total discharge area of
the mainspool at a given displacement x m . The discharge area of the main spool opening A H is given
in 4.14b. The main spool can travel in both directions, changing the flow of the fluid to two different
ports, which means the model has to compensate for the change in flow direction. If x m is positive
the main spool in figure Figure 4.11 is displaced to the right, which changes fluid flow from port A to
port B. In order to keep consistency in the terms for the pressure, P l s follows the flow to port B, and
reversed for P bs . Solving this problem can be done by having the model monitor the value of x m and
changing the direction of the fluid flow through the load. In praxis this would have an impact in the
resistance of the load as it is less likely that the resistance would be the same for both directions.
As seen in 4.12b, the slots in the main spool are angled, which will make that the flow follow the slope
of the slots.

The angle of the flow φ3 through the spool opening A H is dependent on the angle of the slope in the
slots. The slope is illustrated in 4.13b, and the angle φ3 is determined by Equation 4.68

h
µ ¶
−1
φ3 = tan (4.68)
l
From 4.12b the height of the slot h is determined to be (3.115mm − 0.15mm) and the lenght of the
slot is determined to be 5.64mm which yields Equation 4.69
µ ¶
−1 3.115mm − 0.15mm
φ3 = tan = 27.73◦ (4.69)
5.64mm

The angle of the slot needs to be described in [r ad ] to be implemented in simulink, which yields
Equation 4.70:
4.4. MODELING THE MAIN SPOOL 35

³ π ´
φ3 = 27.73◦ · = 0.48 [r ad ] (4.70)
180

A m is the cross-sectioal area of the main spool. The Diameter D m of the main spool was measured
with a precise vernier caliper and determined to be D m = 18mm. The calculation of A m is as follows.

2
Dm 0.0182
Am = π · = π· = 2.54 · cm 2 (4.71)
4 4

The area of the orifice A 5 placed in the pilot line, was not to be found in any datasheets. As such, it was
necessary to determine the area based upon common sence. Making the area too small would mean
that the pressure build up would be too slow. Knowing that the flow in the pilot line is very small the
area of the orifice can be set equally small. For this it is estimated that an orifice cross-sectional area
of the following size is reasonable:

A 5 = 1mm 2 (4.72)

The spring forces in the main spool module

In section B.4 the experiment to determine the spring constant of the main spool k m can be seen. The
spring constant is defined to be:

N
k m = 3695 (4.73)
m

In the PVE module their is a small spring attached to the LVDT sensor k l vd t yielding work on the main
spool, as section B.3 outlines. The size of the spring constant is:

N
k l vd t = 190 (4.74)
m

The spring constant for the LVDT sensor in comparison with the spring constant attached to the main
spool is very small, thus the spring connected to the LVDT can be neglected as it will have a very small
effect on the dynamics of the main spool.

4.4.3 Forces acting on the main spool

Newton’s 2nd law will be utilized to determine the motion of the main spool. The dynamics behind
the motion of the main spool will be explored by utilizing newtons 2. law. The forces acting on the
spool for a negative spool displacement are shown in figur Figure 4.15.
36 CHAPTER 4. PROBLEM SOLVING

−xM x0

FP 4 FP 5

FSM MM

FF R−M

FF F H2
FF F H1

Figure 4.15: Forces acting on the main spool

After determing the forces that acts on the compensator spool shown in Figure 4.10, Newton’s 2nd.
law can be utilized in determining the motion of the compensator spool as seen in Equation 4.75.

M m · ẍ m = F P 4 − F P 5 − F SM − F F R − F F F H 1 − F F F H 2 (4.75)
£m ¤
ẍ m Resulting acceleration of the mass s2

FF F H 1 Flow force 1 at the area A H from port B to load [N ]

FF F H 2 Flow force 2 at the area A H from load to port A [N ]

F F R−M The friction force between the spool and the walls of the housing [N ]

FP 4 The pressure force in the left chamber working on the spool [N ]

FP 5 The pressure force in the right chamber working on the spool [N ]

F SM Double acting spring force for the main spool [N ]

Inertia of the Lever

The main spool is directly coupled to a mechanical lever. As this is exerting a force to the main spool
whenever it is subject to an acceleration, it is necessary to determine the mass of it.

A complete model of the lever was designed in Solidworks ™, see Figure 4.16. After that, the "mass
properties" tool was used to determine the moment of inertia, of the lever, around the axis of revoulu-
tion, ie. where the lever is atached to the PVG 32.
The moment of inertia is 62732 g · mm 2 . As this is a rotational term and the mainspool is moving
£ ¤

along a linear axis, the rotational force has to be transformed into linear force. As seen in Figure 4.17
the lever is coupled to the main spool via a 20mm long rod welded to the axel wich the lever rotates
about.

The equation for the moment of inertia is a follows:

τ = I ·α (4.76)
4.4. MODELING THE MAIN SPOOL 37

Figure 4.16: An illustration of the lever that controls the main spool.

7mm 7mm

MM
20mm
Main spool
he
i n g t mm
ac l1
i spl spoo
D in Axis of revolution
1mm ma

6 φ

20mm

Figure 4.17: Working principle of the lever. The lever is coupled to the mainspoll via a 20mm long
rod.

Where τ is the torque, I is the moment of inertia and α is the angular acceleration. These components
are the rotational inertia analogues to linear inertia equation:

F = m·a (4.77)

Here, F is the force, m is mass and a is the acceleration. To convert the rotational inertia in the
lever to translatoric inertia in the main spool, it is necessary to convert the linear acceleration of the
main spool, ẍ M to rotational acceleration. This is done by expressing a main spool displacement as a
corresponding rotation around the axis of revolution.
So, if the main spool is displaced by 1mm the corresponding rotation of the lever is:
µ ¶
−1 1
sin = 0.050021 r ad (4.78)
20

Furthermore, a displacement of 1m of the main spool corresponds to a rotation of 50.021 r ad . Thus


38 CHAPTER 4. PROBLEM SOLVING

h i
m
an acceleration of the main spool of y 2 can be converted to a rotational acceleration by this ratio:
s

m r ad
· ¸ · ¸
y 2
= 50.021 · y 2
(4.79)
s s
Inserting this, and the value of I into the moment of inertia equation.

r ad
· ¸
2
τ = 62732 g · mm · 50.021 · a
£ ¤
2
(4.80)
s
However, for this expression to be valid for a rotational equivalent to the linear, it is necassary to
convert the torque [N m] produced by the lever to a force [N ].
h i
2 r ad
£ ¤
τ 62732 g · mm · 50.021 · a 2
s
F= = (4.81)
m 20 [mm]

Here m is the distance from the axis of revolution to the end of the rod. The final calculations are
carried out and the expression is as follows.

F = 1.5690 · a (4.82)

Where a is the acceleration of the main spool. This expression is a way of expressing how the rota-
tional torque from the lever exerts a linear force on the main spool, and 1.5690 is the equivalent of
£ ¤
the "rotational motion mass" transformed into "linear motion mass" and has the unit of kg . These
calculations is only valid for small φ4 . Because in reality, if the main spool is displaced an infinite
long distance x m i n f i nt e , then ∠φ4 goes towards 90◦ . For this expression to be valid, it is assumed
that ∠φ4 is linearly dependent on x m and as such, φ4 will grow as x m grows. However, the maximum
displacement of the main spool is x m max = 7mm, and thus the maximum error is ≈ 0.5%. Therefore
the unlinear behavior of φ4 is neglected.

As Equation 4.82 is an equivalent to the mass of the main spool this can just be added to the original
mass of the main spool.

4.4.4 The flow forces at the main spool

The flow forces F F F H 1 and F F F H 2 acting on the main spool is located at the left hand side of the main
spool if the displacement x m is negative as seen in figure 4.15. F F F H 1 is allways located in the middle
of the main spool and F F F H 2 is either on the left or the right side of F F F H 1 depending on how the
main spool is displaced. However, F F F H 1 and F F F H are always acting in the same direction. F F F H 1
and F F F H 2 are calculated using the flow force equation. The first flow force F F F H 1 depends on the
area A H , the pressure drop over the main spool from P K to P LS , while the second flow force depends
on the area A H and the pressure drop from the returning load pressure P B P to the reservoir pressure
P t . These flow forces are derived in equation 4.83.

= 2 ·C d · A H · (P K − P LS ) · cos φ3
¡ ¢
FF F H 1 (4.83)
= 2 ·C d · A H · (P B P − P T ) · cos φ3
¡ ¢
FF F H 2 (4.84)

In steady state when the external load is in balance, the flow out of Port B to the load will be the
same as the flow from the load to the resevoir through Port A, when the displacement to the spool
is negative in figure Figure 4.11. Because both ports have the same discharge area A H , the orifice
equation requires the pressure drop to be the same over both orifices. These observations leaves the
4.4. MODELING THE MAIN SPOOL 39

flow force to be modelled as one, with twice the magnitude, which is given in Equation 4.85. This
is also valid for a positive spool displacement, because both flow forces will work in the opposite
direction.

F F F H = 4 ·C d · A H · (P K − P LS ) · cos φ3
¡ ¢
(4.85)

4.4.5 The pressure forces and spring forces

The force acting on the main spool in terms of pressure difference from the electro-hydraulic actua-
tion is calculated by the pressure difference between P 4 and P 5 .

F P 45 = F P 4 − F P 5 = (P 4 − P 5 ) · A M (4.86)

The spring in the main spool is a double acting spring preloaded with a predefined force. This is
done to prevent the spool from oscillating when small pressure differences occur. Preloading the
spring means the pressure difference has to build up to overcome the preloaded force and hence
move the main spool. According to the experiment in section B.4 the spring is initialy deflected with
x pr el oad −m ≈ 18mm. The calculations shows that this deflection adds a preload to the main spool of
the following size:

F pr el oad −m = 65.05N (4.87)

Further calculation of the resulting spring force are as follows.

F SM = k m · x M + F pr el oad −m
= 3596N · X M + 65.05N (4.88)

4.4.6 The friction forces in the main spool

The friction-force F F R M is given by Equation 4.27 with a coloumb friction coefficient c c and a viscous
friction coefficient c v . The viscous effect is dependent on the viscosity of the oil, and because this
doesn’t change, the size of the coefficient is the same as for previous spools. The viscous friction
coefficient c v depends on the active area which is given by the surface area. The dimensions of the
spool are seen in Figure 4.18
18

111,480

Figure 4.18: The dimensions of the main spool used to calculate the surface area
40 CHAPTER 4. PROBLEM SOLVING

The main spool have been fully modeled in Solidworks ™. The "measure" tool was used to determine
the surface area to be 5173mm 2 The biggest diameter is D 1 = 18mm and has a length of l 1 = 104mm,
while the smallest diameter is D 2 = 12mm and has a length of l 2 = 7mm. This gives a surface area of:

D1 D2
A SM = · π · l1 + · π · l2
2 2
1.8 1.2
= · π · 10.4 + · π · 0.7
2 2
= 31cm 2 (4.89)

The active area based on the explanation following Figure 4.4 becomes:

1
A ac t i ve−m = · A SM = 12.93mm 2 (4.90)
4

Inserting this into the friction coefficient equations will yield a value for the friction forces of the
following size. Remark that the friction coefficients can be adjusted because of numerous unknown
parameters as explained previously.

kg
cv = 5760 [ s ]

c cm = 8.8 · 10−4 · P P [N]

4.4.7 The final model of the main spool

M m · ẍ m = (P 4 − P 5 ) · A m − k m · x m − (c v · ẋ m + c cm ) − 4 ·C d · A H · (P K − P LS ) · cos φ3
¡ ¢
(4.91)

This concludes the modeling of the main spool. To be able to establish a dynamical model it was
necessary to determine all of the forces acting on the spool. Furthermore it was assumed that the
their was the same flow rate in both orifices allthough this is only somehow true in steady state. Also
the coulomb fiction is neglected. The assumptions are listed in Equation 4.4.7.
To model the dynamics of the main spool the following assumptions have been made:

• Constant flow angles

• The same flow force exists in both the output port and the input port of the main spool.

• The coulomb friction is neclected for now.

Next, the modeling of the PVE module will be covered.

4.5 Modelling the PVE module

This section will derive and explain a dynamic model of the PVE module. The PVE module contains
an electric control unit and a hydraulic system. These units are ment to displace the main spool,
as explained in section 2.3. A model of the PVE module will be derived by examining the governing
equations. With the purpose of deriving a model, all the measured constants for this module will be
studied and used as the proper coefficients. Those quantities are listed in Table 4.3.
4.5. MODELLING THE PVE MODULE 41

£ 2¤
A checkv al ve Area of the checkvalves A and B m
£ 2¤
A d r ai n The area of orifice leading to the reservoir m
£ 2¤
A suppl y The area of the supply orifice m
£ 3¤
V4 Volume of the Port A chamber m
£ 3¤
V5 Volume of the Port B chamber m

P pi l ot The pilot pressure at the supply line [P a]

P r eser voi r The reservoir pressure at the drain line [P a]

Table 4.3: Measureable quantities to be used in the PVE module.

Because the PVE module contains both an electric and a hydraulic part, these parts will be treated
seperatly. First, the hydraulic circuit with the diagram shown in Figure 4.19, will be studied.

Ppilot
Qsupply
Pupper

Vupper
NC1 NC3

Port B Port A
P5 P4

V5 V4

NO2
b Acheckball NO4
a

Vlower

Plower Qdrain
Ptank
Figure 4.19: Hydraulic circuit diagram in the PVE module to displace the main spool.

4.5.1 The areas of the orifices

The two, spring loaded checkball valves has the same diameter D seat = 1.7mm. Whenever there is a
partial vacuum in port A or Port B, the fluid will flow through the checkball valve from the lower cavity
to the Port through a discharge area A checkv al ve .

D 2seat
A checkv al ve = π · = 2.3mm 2
4

Pupper
42 CHAPTER 4. PROBLEM SOLVING

The drain orifice Q d r ai n , is the orifice between the lower cavity and the tank. It has a diameter of
D d r ai n = 1mm, and this gives the following area:

D d2 r ai n
A d r ai n = π · = 0.79mm 2
4

The supply orifice leading the fluid from the pilot relief valve to the upper cavity is determined to have
a diameter of D suppl y = 4mm which yields:

D 2NC 1
A suppl y = π · = 12.6mm 2
4

4.5.2 Chambers and piping between the components

A positive displacement of the main spool will change the volume in Port A and port B. The volume
in Port A will increase and port B will decrease. The size of the main spool determines the change
in volume in the chambers. The spool, which is cylinder formed, has a diameter of D = 18mm and
therefore a cross sectional area of:

A m = 254mm 2 (4.92)

The position of the spool determines the volume of the chambers at port A and port B. In an experi-
ment which is described in section B.5 the size of the initial cavities are determined to be:

V4 = 13cm 3
V5 = 27cm 3

The volume of these cavities changes with the displacement of the spool. In order to account for this,
the direction of the spool displacement x m yields the following connection:

V4 (x m ) = V4 + x m · A m = 13cm 3 + x m · 254mm 2 (4.93)


3 2
V5 (x m ) = V5 − x m · A m = 28cm − x m · 254mm (4.94)

To model the fluid in the pipes between the relief pilot valve and the solenoid valves, a cavity with the
name upper cavity is added. Likewise there is added a cavity with the name lower cavity at the other
end. These volumes are assumed to be:

Vl ower = 1000mm 3 (4.95)


3
Vupper = 2000mm (4.96)

4.5.3 Pressures inside the PVE module

Inside the PVE module there are four nodes which contains a volume. To model the dynamics of
main spool movement, the pressures inside these chambers P l ower , P upper , P 5 and P 4 needs to be
determined. The flows running in and out of each node is used in the continuity equation to calculate
the pressures. These continuity equations are given in equation 4.100. Notice that the change in
volume is neglected in the upper and lower chamber, because these only contains piping.
4.5. MODELLING THE PVE MODULE 43

Vupper d P upper
Q suppl y −Q NC 1 −Q NC 3 = · (4.97)
β dt
V5 − x m · A m d P5
Q NC 1 +Q check−B −Q NO2 = −A m · ẋ m + · (4.98)
β dt
V4 − x m · A m d P4
Q NC 2 +Q check−A −Q NO4 = −A m · ẋ m + · (4.99)
β dt
Vl ower d P l ower
Q NO2 +Q NO4 −Q check−A −Q check−B −Q d r ai n = · (4.100)
β dt

The pressure drop over the main spool causes the spool to move. The continuity equations will be
solved with respect to the pressures in order to determine the pressure drop across the main spool:

β
Z
¡ ¢
P upper = · Q suppl y −Q NC 1 −Q NC 3 d t (4.101)
Vupper
β
Z
P5 = · (Q NC 1 +Q check−A −Q NO2 + A m · ẋ m ) d t (4.102)
V − xm · A m
Z 5
β
P4 = · (Q NC 3 +Q check−B −Q NO4 − A m · ẋ m ) d t (4.103)
V + xm · A m
Z 4
β
P l ower = · (Q NO2 +Q NO4 −Q check−A −Q check−B −Q d r ai n ) d t (4.104)
Vl ower

The pressures inside the ports depends on the position of the spool. This is because the volume
changes with the movement of the spool in Port A and Port B. The pressure inside Port B is given
in equation 4.102 and for Port A in equation 4.103. These continuity equations dependens on flow
through different components. The flow through the checkball valve is modelled as a rectified orifice.
This means that the fluid only can flow in one direction. The area of this equivalent orifice is still the
value found for A checkv al ve , and using this in the orifice equation yields:

 q
 C d · A checkv al ve · 2 (P l ower − P 5 )

if P l ower > P 5
ρ
Q check−B =
 0 if P l ower ≤ P 5

h i
m3
Q check−B Flow to Port B from the lower cavity s
£ 2¤
A checkv al ve Area of checkball-seat equivalent to a orifice area m

P5 Pressure in port B [P a]

P l ower Pressure in the lower cavity [P a]

The drain flow to the tank, and the supply flow through the pilot relief valve are dependent on the
pilot pressure and the pressure in the tank. The pilot pressure depends on the flow through the pilot
line and the amount of electric actuation. For continuous electric actuation this flow is measured to
be Q pi l ot = [0.7 : 1.0]l /mi n. As seen in Figure 4.20 the pilot pressure for this flow is approximately
P pi l ot = [13 : 14]B ar [Danfoss, 2000].
44 CHAPTER 4. PROBLEM SOLVING

bar
20

15
Max
(Elect Act)

Min.
10 (Elect Act)

0
0 1 2 3 4 5 l/min
Figure 4.20: The pilot line pressure as a function of the flow through the pilot line.[Danfoss, 2000]

Assuming a tank pressure of P t ank = 0B ar and a pilot pressure of P pi l ot = 13.5B ar , the values can be
used to calculate the flow through the drain and supply orifice, given in the following equation.

s
2
Q d r ai n = C d · A d r ai n · (P l ower − P r eser voi r ) (4.105)
ρ
s
2 ¡ ¢
Q suppl y = C d · A suppl y · P pi l ot − P upper (4.106)
ρ

The flow through the solenoid valves NC and NO are determined by a 4D model explained in the
succeeding subsection, which covers all the flows through the hydraulic circuit in the PVE module.

4.5.4 Flow through the solenoid valves; NO and NC

The solenoid valves are used to build up a variable pressure inside port A and port B. To implement
this, there are used two types of solenoid valves, respectively a normally open NO and a normally
closed NC valve. The NO valves are connected to the tank side, while the NC to the pilot relief side.
The flow through each solenoid valve depends on the dutycycle, which controls the on-off dynamic of
the valves. To take the dynamic properties of the fluid and the on-off dynamics of the solenoid valve
into account, a 4D model is created to determine the flow as a function of 4 parameters [Danfoss,
n.d.a]. This is seen in 4.21. T
T Orifice 1

Tfluid 4D-model
4-D T(u)
NC1, NC3, NO2 & NO4
u1 NO2 Fow_ripple
T
Vs u2
Q
D u3

Pvalve u4

Figure 4.21: The 4D models of the solenoid valves; One for the normally closed and one for the nor-
mally open valves.
4.5. MODELLING THE PVE MODULE 45

As seen in the figure, four different parameteres determines the value of the flow, and these parame-
teres are listed here. An explanation of why these parameters have an effect on the flow through the
solenoid valves is convenient for understanding the model. These are included in the table

• The temperature of the fluid: The temperature of the oil has an effect, as it changes the viscos-
ity. The effects of the viscosity will be explained later on.

• The supply voltage to the PVE module: The supply voltage has an effect on the dynamic pro-
porties of the solenoid valve. A higher supply voltage, will increase the magnetic field causing
the solenoid valve to have a faster reaction time.

• The duty cycle to the solenoid valve in question: The duty cycle describes how long, during a
period T p the valve has to switch between on and off. A higher dutycycle for a NC valve means,
that longer time during a period, the valve will be ON and fluid can flow through it. During on
time, the inductor in solenoid valves will have a voltage drop enforcing the current to rise. This
inductor convert electric energy to mechanical through a magnetic field, by moving the valve
connected to the inductor and opening or closing for flow.

• The pressure drop across the valve: The pressure difference has a direct effect on how much
flow that can pass through the valve. The higher pressure difference, the more flow.

Changing the viscosity yields a change in the density of the fluid, which will force the pressure to
change. This is seen in Equation 4.107. All the other parameters in this equation are constant [Ander-
sen & Hansen, 2007a].
µ
ν = · e (0.0026−10 · T ) · P
5
(4.107)
ρ
h 2i
m
ν Kinematic viscosity of the fluid s
£ Ns ¤
µ0 Dynamic viscosity at atmospheric pressure 2
hm i
kg
ρ The density of the fluid m3

T The temperature of the fluid [◦ C ]

P The pressure of the fluid [B ar ]

Changing the density has an effect on the stiffness, also known as the compressibility of the fluid κ,
which is the inverse of the Bulk modulus.

1 1 δP
βF = = =ρ· (4.108)
κF 1 δρ δρ
ρ · δP

β Bulk modulus [B ar ]

κ B ar −1
£ ¤
Compresibility

δρ kg /m 3
£ ¤
The density change of the fluid

δP The pressure change of the fluid [B ar ]

Bulk modulus is used in the continuity equation and is more thoroughly explained in Equation A.7.
Changing the temperature and thereby indirectly changing the Bulk modulus, effects the importance
of the pressure changes inside the continuity equation [Andersen & Hansen, 2007a].
The 4D-model is based on various different conditions, by changing the voltage, the fluid tempera-
ture, the duty cycle and the pressure drop across the valve. To plot a graph of how the fluid changes
46 CHAPTER 4. PROBLEM SOLVING

with varying duty cycle, some of the parameters must be held constant. If assumed that the fluid
temperature is T = 50◦C , and the supply voltage is Vs = 13V , a graph of the flow as a function of the
duty cycle can be plottet for different pressuredrops. The graph at figure 4.22 the fluid flow through
orifice NC 1 and NO2 as a function of the duty cycle.

−5 −5
x 10 x 10
3 5
2 Bar 2 Bar
4 Bar 4.5 4 Bar
6 Bar 6 Bar
2.5
4 8 Bar
8 Bar
10 Bar 10 Bar
3.5 12 Bar
2 12 Bar
14 Bar

Flow Q [m3/s]
14 Bar
Flow Q [m /s]

3
16 Bar
3

16 Bar
1.5 2.5

2
1
1.5

1
0.5
0.5

0 0
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Duty cycle DT [%] Duty cycle DT [%]

Figure 4.22: The graph in the left represents the flow for a NC 1 valve as a function of the duty cycle
for different pressures, while the graph in the right represents the flow for NO2. The one with the
biggest flow is measured at 16B ar , and the lowest flow at 2B ar

The flow through a NC valve increases with the duty cycle, while the flow through an NO valve will
decrease. Furthermore the NC valve has a more linear characteristic, while the NO valve has a larger
constant region, causing the flow through each valve to be different for the samme presure drop. A
pressure drop ∆P = 2B ar over both valves, yields an actual flow to each chamber as illustrated in
Figure 4.23. When the duty cycle is DT = 80%, this will yield more flow into than out off port B.
Besides, this yields a lower duty cycle to port A seen in Figure 4.28, resulting in more flow out of Port
A than in. This consequently will cause a pressure difference between the chambers in both ports,
resulting in a force on the main spool.

−5
x 10
1
2 Bar

0.5
Flow Qin−Qout [m3/s]

−0.5

−1

−1.5

−2
0 10 20 30 40 50 60 70 80 90 100
Duty cycle DT [%]

Figure 4.23: The actual flow in and out of the chamber at the ports.
4.5. MODELLING THE PVE MODULE 47

4.5.5 The model of the electric actuator

The electro-hydraulic unit serves to move the spool according to a desired position. In order to assure,
that the spool has moved to the desired location, the electric unit uses a feedback system. This system
is build using the principle of closed loop feedback control theory as illustrated in Figure 4.24. The er-
ror in position turns into a duty cycle which defines a different flow pattern for each port through the
electromagnectic valves. The two electromagnetic valves connected to each port builds up a pressure
both at Port A and Port B. The pressure inside these ports are dependent of the spool velocity, posi-
tion and the flow pattern of the connecting valves. The pressure difference at each port are forcing
the main spool to move.

Spool postion
Spool velocity
DTNC1 Q NC1 Port B
X err DTNC1 Actual spool
Defined
+ P5
-
spool position position
Q5
Error in pos DTNO2 Q NO2
X set X err DTNO2 -
+ +
Spool System
Vrel - DTNC3
X err
Q NC3
DTNC3
Port A
+ P4
DTNO4 Q NO4 - Q4
X err DTNO4

Measured spool
position
LVDT sensor

Figure 4.24: The closed loop feedback control of the pve module.

To sum up and explore the entire process from an input signal to a displacement of the spool, the
process needs to be broken into parts. The parts that are involved in this process, includes a sensor
to measure the actual position of the spool, an input signal that defines the position of the spool and
an output signal to each electromagnetic valve. The difference between the input and the measured
signal is used to calculate two different PWM signals; one for port A cavity and one for port B cavity.

The input signal defining the position

The input signal defines the position of the main spool. How this is implemented in the PVE module is
basically using the voltage source as a full scale resolution, and the input signal as a varying parameter
used to compare with the voltage source. This is done by dividing the input voltage Vi nput with the
supply voltage Vs , and using this relative value to specify a main spool displacement value x m .

Vi nput
Vr el = (4.109)
Vs

Because the spool can travel ±7mm, a relative value of Vr el = 0.5 is defined to be the mid position
of spool travel x m = 0mm. Furthermore, the maximum displacement in both directions is defined to
be Vr el = 0.25 in negative travel direction, and Vr el = 0.75 in positive. This gives a linear relationship
between the spool displacement and the relative input signal as shown in Figure 4.25.
48 CHAPTER 4. PROBLEM SOLVING

Main spool position xset [mm] 4

−2

−4

−6

−8
0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75
Relative input voltage Vref [.]

Figure 4.25: The ouput for the transfer function.

The relation between the relative input signal Vr el and the defined settling point for spool travels
is given by Equation 4.110. This equation is used to transfer the input signal to a position, so it is
possible to compare with the measured position.

x set = 0.028 ·Vr el − 0.014 (4.110)

Position Sensor

There are two common categories of position sensors; A linear sensor to measure a linear movement,
and an angular sensor to measure an angular movement. The one inside the PVG 32 is a linear sensor
of the type linear variable differential transformer (LVDT). Figure 4.26 illustrates the principle idea
behind a LVDT sensor.
Vin
Moveable Primary coil
core

Ls Lp Ls
+
Lp
Vin Ls Vout
Lp
-
Secondary coil

+ Vout -
Figure 4.26: The figure on the left side shows the actual composition of an LVDT sensor, while the
figure on the right side shows an electric circuit diagram of a LVDT sensor

The advantage of this sensor is the low maintenance requirements because the sensor doesn’t have
any contact with the moving parts. This provides a longer lifetime of the sensor because of less wear
on the components. The moving part of the sensor is an iron core, which is directly connected to the
main spool, so a spool movement yields a movement of the iron core used in the sensor. An electric
conductor is wound around the iron core and connected to an AC voltage source. This is called the
primary winding. The windings will create a magnetic field, this field will be magnified in a direction
perpendicular to the windings as shown in Figure 4.27.
4.5. MODELLING THE PVE MODULE 49

iron core

+ I in O
Vin
-
primary winding
Figure 4.27: Illustration of how the flux is magnified through the iron core using an alternating cur-
rent.

By placing an iron core inside the windings, nearly all the generated flux is passed through the iron
core and not the air. This happens because of the high relative permability µr of the iron core rela-
tive to the low relative permability of air at µr = 1. A high relative permability gives a low magnetic
reluctance, which is the equivalent of electric resistance. The relation is given by Equation 4.111. This
equation shows that the magnetic reluctance of the iron is much smaller than the air reluctance, and
therefore the main part of the magnetic flux will pass through the iron core.

l
ℜ= (4.111)
µ0 · µr · A

At
£ ¤
ℜ Magnetic Reluctance Wb

l Lenght of the element [m]


£ H ¤
µ0 Permability of free space m −1
Henvisninger
H
µr
£ ¤
Relative permability of the material m −1
£ 2¤
A The cross sectional area of the magnetic circuit m

Two other sets of windings, called the secondary windings, is connected to the same iron core as seen
in Figure 4.26. Depending on the spool position, one of the secondary windings will cover the iron
core more than the other. As a result of this, the mutual inductance between the primary inductor
and the two secondary inductors will vary as a function of the position of the iron core [Wikipedia,
n.d.b]. The total flux going through the iron core is given by the flux density which is dependent on the
current I i n , and the number of turns of the windings n. This is given in the following Equation 4.112.

φ = n · Ii n (4.112)

Depending on the size of the iron core, and the number of turns in each inductor, the relation between
the input voltage and the output voltage differ [Wikipedia, n.d.c]. According to the model made by
Sauer Danfoss [Danfoss, 2000], the transfer function for the LVDT sensor in the PVEH module is given
in Equation 4.113, where the x m variable is the actual position and the x m 0 variable is the measured.

xm0 0.945
= (4.113)
xm 0.0059 · s + 1

This first order transfer function of the LVDT sensor has a time constant of τ = 0.0059s. Additionally,
it is seen that the steady state gain is lower than 1.
50 CHAPTER 4. PROBLEM SOLVING

The duty cycle for each port depending on position error

The error in position defines the duty cycle of each port, and the duty cycle defines how much flow
the electromagnetic valves pass through to each port. The error in position is calculated taking the
difference between the defined spool position and the actual spool position as seen in Equation 4.114.

x er r or = x set − x m (4.114)

The two electrocmagnetic valve types NC and NO are both recieving the same duty cycle. The only
difference of the dutycycle for these valves are in the implementation. The pilot side valves NC 1 and
NC 3 has a time delay of 20ms and the reservoir side valves has a time delay of 3ms. The duty cycle
signals for both valves in each port has an in-between time delay of 17ms. The function of the time
delay is to imitate the physical proporties of a solenoid valve. This time delay is implemented in the
control. The reason for this is to avoid a shortcut, which means that the valves connected to the same
port won’t open for passage at the same time. Whenever there is a hightime, first the NO valve will
close and 17ms later the NC valve will open. This also serves to imitate the physical behaviour of the
valves. Since the NC valves work against the pressure they are somehow slower. The error in position
defines a duty-cycle which can vary between 0 and 100 percent. Figure 4.28 illustrates the dutycycle
as a function of the error in position.

100
Port B
90 Port A

80

70
Duty cycle DT [%]

60

50

40

30

20

10

0
−1.5 −1 −0.5 0 0.5 1 1.5
Error in position [mm]

Figure 4.28: The duty cycle output for the port A cavity and port B cavity side valves as a function of
the error in position.

The duty-cycle as a function of the error in position for both port A and port B valves are given at
Equation 4.115. There is an outer boundary for the output values in these equation ranging from 0%
to 100%.

DT A = −70 · x er r or + 28.4
(4.115)
DTB = 70 · x er r or + 28.4
4.6. MODELING THE LOAD 51

4.6 Modeling the load

In order to validate the model of the PVG 32, the valve has to be added to a hydraulic circuit with
an external load. In addition, the load will have flowmeters and pressure gauges around the load,
so it is possible to compare the measurements with the simulations. This load has the function to
switch the load between using one piping line or two. One of the piping lines will allways be active,
while the other can be opened if desired. Consequently this means that the flow Q v has to be split up
into the two lines and therefore several equations for determination of flows and pressure have to be
established.

4.6.1 Flows and pressures in the load

Pf Pressure in the front of the load [P a]

Pm Pressure in the mid section of the load [P a]

Pb Pressure in the back of the load [P a]


£ 2¤
A L1 Area of the orifice A L1 m
£ 2¤
A L2 Area of the orifice A L2 m
£ 2¤
A L3 Area of the orifice A L3 m

Table 4.4: Constants necessary to model the load

Vm
AL1 Pm
LOAD QL1

QL3 QL2

AL3
Vb Vf
Pb Qb Qf
Pf

QL QV

Pbp PLS

Figure 4.29: The working load of the system


52 CHAPTER 4. PROBLEM SOLVING

The pressure in the load are calculated utilizing the continuity equation wich takes the following form.
V f ∂P f β
Z
Q f −Q L2 −Q L3 = · ⇔ Pf = dt (4.116)
β ∂t (Q v −Q L3 −Q L2 ) ·V f
Vm ∂P m β
Z
Q L2 −Q L1 = · ⇔ Pm = dt (4.117)
β ∂t (Q L2 −Q L1 ) ·Vm
Vb ∂P b β
Z
Q L1 +Q L3 −Q b = · ⇔ Pb = dt (4.118)
β ∂t (Q L1 −Q L3 −Q b ) ·Vb
Now the Flows throughout the load i.e. the flow through the orifices Q L1 , Q L2 and Q L3 can be calcu-
lated using the orifice equation.
s
2
Q 1 = C d · A L1 · · (P m − P b ) (4.119)
ρ
s
2 ¡ ¢
Q 2 = C d · A L1 · · P f − Pm (4.120)
ρ
s
2 ¡ ¢
Q 3 = C d · A L1 · · P f − Pb (4.121)
ρ

The orifice equation assumes that the flow is turbulent, which will be investigated further when the
test data is analysed.

4.7 Validation of unlinear advanced model

This section takes results from the test section B.7, and the model in order to verify the model. To do
so, the area of the load orifices are to be determined.

4.7.1 Introduction to the test

In the laboratory test the PVG32 was controlling the flow through two orifices of variable contraction
area. Refer to section B.7 for more details about the test. Beside the two orifices the test arrangement
also included a solenoid valve which would block the flow to one of the orifices. The system was tested
with three parameters varied. The parameters where, the contraction area of the two load orifices, the
magnet valve, and the control voltage on the PVG, which in steady state leads to a certain flow. The
solenoid valve was either opening or closing during the tests. The classification of the test series is
illustrated in Figure 4.30. Each section of the block symbolizes a single test and each test has a unique
set of parameters.

Opening
Closing
Smal flow
l area
Large
Larg
e are l flow
a Smal

Figure 4.30: Classification of test series


4.7. VALIDATION OF UNLINEAR ADVANCED MODEL 53

As described in the test report, the things being measured includes, the steady state flow into the load
Qv, and the flow in the branch containing the solenoid valve L2 and the orifice L1. The flow in this
is denoted Q L2 . An illustration of the load is seen on Appendix D. Beside flows, also pressures was
measured. Three pressure transducers were applied to the the load, measuring the input pressure P f ,
the intermediate pressure P m , and the output pressure P b . In contrast to the flow measurements, the
pressure measurements were logged over time, and not just one single steady state value.

4.7.2 Calculating areas of the load

In order to compare the results from the model with the test results, it is necessary to find the areas
of the load during the test. As no direct measurement of the orifices have been carried out, it is only
possible to find the areas through calculations, based on pressure and flow meaurements. The test
was carried out with two different opening areas for each of the two orifices, one large denoted b and
one small denoted s.
The relationship between flow, pressure and area of an orifice is described in the orifice equation. For
more details, see Equation A.1. This orifice equation solved for area is shown in Equation 4.122.

Q
A0 = q (4.122)
2
Cd ρ (P i n − P out )

Using this equation it is possible to set up the following equations describing the areas of the load
orifices from the measured data.

Q L2
A L1 = q (4.123)
2
Cd ρ (P m − P b )
Q L2
A L2 = q (4.124)
2
Cd ρ (P f − Pm )
Q f −Q L2
A L3 = q (4.125)
C d ρ2 (P f − P b )

Applying these expressions to each of the 12 tests, returns the areas given in Table 4.5.
54 CHAPTER 4. PROBLEM SOLVING

Test no. A L1 [mm 2 ] A L2 [mm 2 ] A L3 [mm 2 ]

1 - - 3.319

2 2.174 5.800 2.429

3 - - 3.422

4 - - 2.290

5 1.553 3.879 1.565

6 - - 2.308

7 - - 4.635

8 2.585 5.623 3.629

9 - - 4.671

10 - - 3.422

11 1.933 3.626 2.637

12 - - 3.397

Table 4.5: Areas of the load orifices based on data from the specific test

As you can see, the results yields a great amount of variety, so in order to find the best suitable area the
Reynolds numbers are calculated. The reynolds number is an indicator of which forces is dominant
in the current flow. If the inertia forces are dominant the Reynolds number is high and the flow is
turbulent. On the other hand, if the Reynolds number is low the viscous forces are dominant and this
kind of flow are called laminar.
For the flows in this particular test, Equation 4.126 is used to calculate the Reynolds number. The
equation is for a smooth pipe, but it’s assumed that the parameters of the equation yields the same
effect to the flow through an orifice. The transition from laminar to turbulent will however happen at
a lower Reynolds number because of the more disturbing geometry of an orifice.

V ·D
Re = (4.126)
ν

Re Reynolds number [·]


£m ¤
V Velocity of the fluid s

D Diameter of the pipe [m]


h 2i
m
ν Kinematic Viscosity s

The kinematic viscosity is unknown, but the group has been informed that it was a common mineral
oil that where used in the tests. The kinematic viscosity for some common mineral oils are listed
in Figure A.2 in section A.1. As there is no further information available the kinematic viscosity is
assumed to be ν = 32cSt . The velocity and the diameter, used in the formula, are calculated using the
following two equations.
4.7. VALIDATION OF UNLINEAR ADVANCED MODEL 55

r
4
D= · A0 (4.127)
π

Q
V= (4.128)
A0
The reynolds number for each flow are listed in Table 4.6

Test no. L1 L2 L3

1 - - 3527

2 2062 1263 2328

3 - - 3568

4 - - 2246

5 1316 833 1423

6 - - 2237

7 - - 2991

8 1587 1075 2070

9 - - 2990

10 - - 2083

11 1065 777 1409

12 - - 2040

Table 4.6: Reynolds number of the flows in the load orifices

The transition point from laminar to turbulent flow isn’t well defined, but as a rule of thumb it hap-
pens at 2300 for hydraulic hoses and pipes [Andersen & Hansen, 2007a]. In this case it’s an orifice and
because of the disturbing effects of the rough geometry it will likely happen at a much lower value.
Based only on the Reynolds numbers all the tabulated flows seems turbulent, some of course more
than other. The flow which will determine the area used in the orifice equation of the load, is the
one with the highest Reynolds number, hence this flow is the most turbulent. The elected areas are
tabulated in Table 4.7.

A L1 [mm 2 ] A L2 [mm 2 ] A L3 [mm 2 ]

Small 2.174 5.800 3.422

Large 2.585 5.800 4.635

Table 4.7: Areas of the load orifices. Asuming turbulent flow.

Turbulent flow through an orifice is determined by the orifice equation, which states that the flow rate
is related to the square root of the pressure drop. In contrast; laminar flow has a linear relationship
determined by Equation 4.129, the so called Hagen-Poiseuille equation.
56 CHAPTER 4. PROBLEM SOLVING

∆P Pressure drop [P a]
£ Ns ¤
µ Dynamic viscosity m2

D Diameter of the pipe [m]


h 3i
m
Q Flow rate s

L Length of the pipe [m]

128 · µ · L ·Q
∆P = (4.129)
π · D4

Which can also be written as in Equation 4.130.

A2
Q= · ∆P (4.130)
π·8·µ·L

Flow characteristics of orifice AL3 in load


17.5

15
Flow through orifice [l/min]

12.5

10

7.5

5
Small opening GR522 test
2.5 Small opening GR520 test
Large opening GR522 test
Large opening GR520 test
0
0 5 10 15 20 25 30 35 40 45 50 55 60
Pressure drop across orifice [Bar]

Figure 4.31: Plot of steady state flow and pressure for four different areas of the same variable orrifice
A L3 in the load

Remember that the areas calculated with the orifice equation yields great difference. Therefore it
could be interesting to see if the flows really form the graph of a square root if plotted. To assist
a conclusion, data from another test series performed by another group of students, with the exact
same equipment, three days before the measurements done by our group, is also included. The figure
showing the four plots and their corresponding linear fits are shown in Figure 4.31. The linear fits are
made so they contains origo, because when there is no pressure drop there is no flow.
The figure shows the magnitude of the flow rate, as the pressure drop across the orrifice A L 3 is varied.
All 4 plots shows linear behaviour. This is definitely a surprising result. One should expect to see a
4.7. VALIDATION OF UNLINEAR ADVANCED MODEL 57

square root characteristics because of the high pressure drop, but this is not the case. Let’s consider
the sources of error. As stated in section B.7, the tolerance of the flow meter as low as 1 mil n , and
for the pressure transducers it can differ as much as 1B ar [Pedersen, 2009], so when pressure drop is
considered it will be up to 2B ar total.
The first graph to consider is the one which would have the highest Reynolds number, since if this is
a laminar flow, it is reasonable to assume the other flows are aswell. The one considered is the lowest
graph, "Small opening GR522 test", wich is the blue one. Imagine that each of the 3 points making
up the blue graph is allowed to move in a square 2 bars wide and 1 l/min high, will it be possible
to fit this graph to a square root function? Even if the sources of errors are considered to be all in
the favour for a turbulent flow, it is simply impossible to fit these data to a square root function, as
the curvature of the plot are all greater or equal to zero. If it are to reach zero flow at zero pressure
it can’t have a negative curvature in all of it’s regime, which all real square root functions must have
(see Equation 4.131). For the graphs to have a negative curvature and still reach zero it must have it’s
highest point below the linear fit, and the lowest point above. The curvature must also be largest at
the smallest pressures, which require that the slope between the start point and the midpoint should
be significant larger than the slope after the midpoint to the endpoint, which is impossible to achieve
while still reaching origo, even though the sources of error are considered. The same is, by the way,
also true for all of the 4 graphs in the figure. So it is reasonable to assume the flows in all of the orifices
are laminar, as the flow through the other two orifices of the load happens at a much lower Reynolds
number.

d2 p −1
2
x= (4.131)
dx 4 · x 1.5

In order to still use the data to test the model, it’s necessary to change the load block of the model, to
assume linear flow (see Equations 4.132).

Q L1 = K L1 · (P m − P b )
Q L2 = K L2 · (P f − P m )
Q L3 = K L3 · (P f − P b ) (4.132)

The constants in the expressions is the slope of the linear fits of the measured data. In order to find
the constants for the last to expressions two more plots are needed. The plot of the solenoid valve and
its linear fit is seen in Figure 4.32.
As the area is the same for the solenoid valve, the measurements for both groups can be utilized to
form the graph. The error tolerances are assumed to be the same absolute value. This means that the
measuring error compared to the magnitude of the measurements can be very large, but as this graph
has 8 data points, it is reasonable to assume that the fit on the graph would explain the relationship
between flow and pressure for the solenoid valve.
58 CHAPTER 4. PROBLEM SOLVING

Flow characteristics of orifice AL2 in load


7.5

6
Flow through orifice [l/min]

4.5

1.5

Solenoid valve GR522 and GR520


0
0 1 2 3 4 5
Pressure drop across orifice [Bar]

Figure 4.32: Plot of steady state flow and pressure for the solenoid valve A L2 in the load

Flow characteristics of orifice AL1 in load


7.5
Flow through orifice [l/min]

2.5

Small opening GR522 test


Small opening GR520 test
Large opening GR522 test
Large opening GR520 test
0
0 5 10 15 20 25 30 35
Pressure drop across orifice [Bar]

Figure 4.33: Plot of steady state flow and pressure for four different areas of the same variable orrifice
A L1 in the load
4.7. VALIDATION OF UNLINEAR ADVANCED MODEL 59

The constants for the last orifice is the thoughest one to capture, as there is very little data for disposal.
Only two data points for each orifice area besides zero. The linear fits are made so they fit zero and the
data point at the highest pressure, and the fits actually allmost corresponds to the midpoint aswell,
eventhough there is relative high errors in this domain.
The slopes of all the orifices, based on the data that this group collected, are tabulated in Table 4.8.

L L L
£ ¤ £ ¤ £ ¤
K L1 B ar · mi n K L2 B ar · mi n K L3 B ar · mi n

Small 0.2105 1.6111 0.2203

Large 0.3544 1.6111 0.4533

Table 4.8: Flow constants of the load orifices. Asuming laminar flow.

At this point all the necessary parameters of the load are known, so it is now possible to compare the
test results with the results from the model.

4.7.3 Comparison

The model are to be verified. This is done by comparing the results with those from the above-
mentioned test, but before this is possible it is necesary to define some uncertain parameters, which
are to be alteret, based on the results from the comparison to the test results. The parameters and
their initial values are defined in Table 4.9.

Parameter Symbol Value Unit

Flow rate of the pump Qv 17 L/min

Supply voltage bat _vol 12 V

Bulk modulus β 7000 Bar

Viscous Fric. PVP comp. cv p 9000 Ns/m

Viscous Fric. PVB comp. c vc 9000 Ns/m

Viscous Fric. Main. c vm 25 Ns/m

Coulomb Fric. PVP comp. c cp 0 N

Coulomb Fric. PVB comp. c cc 0 N

Table 4.9: Uncertain parameters. Will be alteret to fit test results.

Steady State Flow

The first thing to do, will be to find out whether the model has the right steady state respons. To
evaluate this, the results from the second part of the test described in section B.7 will be used. The
model is provided with the same input voltage as measured in the test, and the steady state value of
Q v is logged. The result is seen in Figure 4.34.
60 CHAPTER 4. PROBLEM SOLVING

Steady State Qv Comparison


18
Simulation
16 Test
14
Flow rate [L/min]

12

10

0
3.5 4 4.5 5 5.5 6
Input voltage [V]

Steady State Qv Comparison


18
Simulation
16 Test
14
Flow rate [L/min]

12

10

0
−5 −4.5 −4 −3.5 −3 −2.5 −2 −1.5 −1 −0.5 0
Spool travel [mm]

Figure 4.34: Plot of steady state flow with varied input voltage, and corresponding calculated spool
travel

The two responses show the same tendencies, but the simulated result is higher in magnitude. The
simulated graph clearly saturate at a point between the two inputs 4.45V and 4.21V . The test are
showing the same, but it is not fully saturated until somewhere after the input has reached 4.21. If the
graph is assumed to follow a linear slope until it saturates, the lowest input value can be captured as
the point where the line reaches the maximum value, this is also the flow rate of the pump, which is
read to be 16.3L/mi n. A small zoom of the plot, illustrating the procedure, can be seen in Figure 4.35.
The minimum input value, before saturation, is determined to be Umi n = 4.19V . The actual flow will
follow a quadratic slope but with such a steep slope at such a short range, it could be approximated
by a straight line.
By changing the flow rate of the pump, the system is changed, but it is assumed that it will only affect
the maximum value of the two last datapoints, so there is no need to re-simulate all 7 situations. As
the flow out of the valve is determined only by two parameters; the opening area of the main spool
4.7. VALIDATION OF UNLINEAR ADVANCED MODEL 61

and the pressure drop across it, the error must be found here. The pressure drop across the main
spool is graphed in Figure 4.36.

Steady State Qv Comparison


17 Simulation
Test
16

15
Flow rate [L/min]

14

13

12

11

10

3.9 4 4.1 4.2 4.3 4.4


Input voltage [V]

Figure 4.35: Procedure to determine minimum input voltage before saturation, and pump flow dur-
ing the test

Steady State ∆Pm Comparison


14

12

10
Pressure [Bar]

0
3.5 4 4.5 5 5.5 6
Voltage [V]

Figure 4.36: Plot of steady state pressure drop, in the main spool, with varied input voltage

The whole purpose of a proportional valve is the ability to control the flow out of the two ports pre-
62 CHAPTER 4. PROBLEM SOLVING

cisely. This is achieved through implementation of the compensator spool in the PVG. It is seen from
the graph that the spool can’t keep the pressure drop constant when the system saturates, nor when
the main spool blocks the flow. In all other situations the compensator is doing what it should; keep-
ing the pressure drop constant. This is also supported by more simulations than presented, at this
point, in the report. The only way that the pressure drop could lead to errors, is, if the pressure drop
across the main spool was less than the 5B ar in the test. This value is mainly determined by the
preload of the spring in the compensator spool, and the area characteristics of the A 1 opening of the
compensator spool. Even though this can affect the flow, the change from 5bar to i.e. 4B ar , only
changes the output 10.6% because when the flow is calculated it’s the square root of the pressure that
enters. So this can’t be the only reason to the difference in flow. The reason have to be the area of the
mainspool, either directly or by an error in the input voltage, which will have a great effect in the flow.
If the measurements where 0.3V too low, the model actually imitate the flow allmost perfectly accord-
ing to Figure 4.34. The effect of the input voltage U on the spool travel, is also highly dependent on
the supply voltage, as the actual input to the modulator is the ratio of the input voltage and the sup-
ply voltage, not the input voltage directly. If the area characteristics of the main spool corresponds
to the flow characteristics from the data sheet, the difference in the flow, can only be explained by
measurement errors and errors in the supply voltage.

Comparison of the Area Characteristics of Main Spool


80
LABTEST1
70 LABTEST2
AMOC
DATA SHEET E
60 DATA SHEET D
DATA SHEET C
Area AH [mm2]

50

40

30

20

10

0
−7 −6 −5 −4 −3 −2 −1 0
Spool travel [mm]

Figure 4.37: Area of main spool opening, based on data from data sheet, geometric measurements
and labtests.

The Figure 4.37 contains 6 plots, all describing the area A H as a function of spool travel x m . Three
of them are based on flow and pressure data from the data sheet Danfoss [2000], and two of them
are based on data from the test in section B.7. All graphs are formed using the orifice equation and
the assumption of a pressure drop of 5bar across the main spool. The last graph is made solely on
4.7. VALIDATION OF UNLINEAR ADVANCED MODEL 63

geometric measurements, using a vernier caliper, and the computer program AMOC Power [1999],
and this is the one used in the model. Judging from the figure, there is no doubt that the spool of
which the measurements where done, was a spool of type E. The data from the two tests doesn’t
fit another type of spool. Therefore it is reasonable to assume that the model calculates the flow
correctly, and the reason why the tests doesn’t correspond to the output from the model, are errors in
the measurements and in the adjustment of the supply voltage. To further investigate the effect of a
small change in the supply voltage the following calculations are made (see Equation 4.133).

U U
µ ¶ µ ¶
14mm 14mm
E (U ) = − 0.5 · − − 0.5 ·
U assumed 0.5V U ac t ual 0.5V

µ ¶
1 1
E (U ) = 28 · − ·U (4.133)
U assumed U ac t ual

E (U ) Steady state error in the position of the main spool. [mm]

The number illustrates how much more the spool travel than it actually does.

The sign convention follows that of xm.

U assumed The assumed supply voltage [V ]

U ac t ual The actual supply voltage [V ]

U The input voltage [V ]

The effect of having a supply voltage of 11.50V instead of 12V is plottet in Figure 4.38. The graph
illustrates the error in spool travel, as a function of the input voltage. The error is larger at the higher
signals, and smaller at the lower, an average error of about −0.42mm is present in the interval from
3.5V..5V , which corresponds to the plot which contains the comparison of the two steady states flows,
see Figure 4.34. The simulation is done 7 times again, and each time, the steady state value of the flow
is tabulated. The result is shown in Figure 4.39.

3 3.5 4 4.5 5

− 0.2

E( U )− 0.4

− 0.6

− 0.8

Figure 4.38: The error in main spool travel x m by assuming 12V power supply, and actually having
11.50V
64 CHAPTER 4. PROBLEM SOLVING

Steady State Qv Comparison


18
Simulation
16 Test

14
Flow rate [L/min]

12

10

0
3.5 4 4.5 5 5.5 6
Input voltage [V]

Steady State Qv Comparison


18
Simulation
16 Test

14
Flow rate [L/min]

12

10

0
−5 −4 −3 −2 −1 0 1
Spool travel [mm]

Figure 4.39: The steady state flow rate Q v from the model compared to the test results. The model is
simulated at at supply voltage of 11.50V as this is assumed to be the actual voltage for test aswell.

The primary error are considered to be the error in the supply voltage, as the data now completely
corresponds to the original. The only change from Figure 4.34 to Figure 4.39 is the battery voltage,
and the flow rate of the pump flow. In order to use the data to verify the rest of the model, the sup-
ply voltage of the model is kept at bat _vol = 11.50V , and the flow rate of the pump is changed to
Q v = 16.3L/mi n. Before proceeding to test other parameters, the steady state flow of the first part
of the test, described in section B.7, will be compared to the simulation. The result is seen in figure
Figure 4.40. The data that is now plotted is different in the way that, the solenoid valve has been ac-
tivated, either closing og opening, in all of the tests. A total of 8 steady state values are captured from
both the simulation and the test. Each input voltage have corresponding flows. One for the opening
of the solenoid valve, resulting in a steady state measurement of the flow, where the oil flows in both
branches of the load, and a closing where the oil only flows in one of them, when the steady state
measurement is done.
4.7. VALIDATION OF UNLINEAR ADVANCED MODEL 65

Steady State Qv Comparison


16
Simulation
15 Test

14

13
Flow rate [L/min]

12

11

10

6
4.2 4.25 4.3 4.35 4.4 4.45 4.5 4.55 4.6 4.65 4.7
Input voltage [V]

Figure 4.40: The steady state flow of the first part of the test described in section B.7. Only the tests
where the solenoid valve is activated are included in the graph, which make up a total of 8 test and 8
simulation points.

If compared to the graph of the other data series (Figure 4.39) this graph looks a bit surprising. The
values still lies inside the tolerances of the measurement equipment, but shows a larger difference
in the datapoint at the top. This could mean that the valve is a bit more saturated in reality than the
model assumes. The data plottet at the top point is for the test series numbered 2 and 3, where 2 is
the upper one. Besides beeing about 1 l/min of, the data shows the same strange jump in flow. The
flow is a bit higher when the solenoid valve is open than it is when it’s closed, but only at this point. At
the other points the flow is exactly the same, both in the test and in the simulation separately, which
means that in steady state, a input signal gives the same output flow independent of the load. It could
be interesting to analyse wheter the reason for the error at the top point is a result of saturation or
simply error in measurements. It’s most likely the former, because if the reson is the error in mea-
surements it isn’t very likely that the only place where the error should occur is at that point, all the
other points, from both parts of the test, have much smaller errors. As the model describe the flow in
the other points sufficient, the text will move on and consider the steady state pressures instead.

Steady State Pressure

The steady state pressure is only affected by the flow into the load, Q v , and the areas of the orifices
in the load, or in this case the equivalent linear constants, K L1 , K L2 and K L3 , listed in Table 4.8. As
the steady state flow is validated, and therefore fit the test data, the comparison of steady state pres-
sure will investigate those constants. First the constant K L3 is considered and therefore only the tests
where the solenoid valve is closed in the end of the test, are considered.
66 CHAPTER 4. PROBLEM SOLVING

Test no. ∆P (SI M ) [B ar ] ∆P (T E ST ) [B ar ] Q v (SI M ) [L/mi n] Q v (T E ST ) [L/mi n]

6 31.720 34.07 6.988 7.23

9 30.599 30.08 13.944 13.75

12 16.210 19.25 7.354 8.00

Table 4.10: Steady state pressure drop and flow through load orifice L 3

In Table 4.10 the simulated and the tested pressure drop over the orifice are tabulated, along with the
corresponding flow rate. Only one data point for each input is available, that makes up a total of 4
datapoints to describe 2 different areas, which is 2 per area, there is only one data point to describe
the small area of the orifice, because the flow isn’t calculated correctly at one of the inputs. It would
be possible to use the data from all the tests, but as it wouldn’t change the result the time and effort
are placed elsewhere.

L L
£ ¤ £ ¤
Test no. K L3 (SI M ) B ar · mi n K L3 (T E ST ) B ar · mi n

6 0.2203 0.2122

9 0.4544 0.4571

12 0.4532 0.4156

Table 4.11: Flow constants of the load orifice L 3

The pressure drop from the simulation is allmost the same as the one from the test. It varies as much
as 3bar at test no. 12, but as the flow in that point is equaly high, the values of K L3 are considered to
be correct. The value of K L3 for each test and simulation are tabulated in Table 4.11, and calculated
using Equation 4.134 The most reliable values are the one where the error in flow is smallest, which
will be test no. 9, and as there is only one value for the small orifice, the result from test no. 6 is used
for comparison. And the values corresponds to the test values.

Q
K= (4.134)
∆P

The same procedure are utilized in the evaluation of the other two constants K L1 and K L2 , this time
of course other tests are used. The ones considered appears together with the pressure drop in Ta-
ble 4.12.

Test no. ∆P 1(SI M ) [B ar ] ∆P 1(T E ST ) [B ar ] ∆P 2(SI M ) [B ar ] ∆P 2(T E ST ) [B ar ]

5 15.201 17.530 1.986 2.81

8 15.355 15.32 3.371 3.23

11 8.110 9.22 1.784 2.62

Table 4.12: Steady state pressure drop and flow through load orifice L 1 and L 2
4.7. VALIDATION OF UNLINEAR ADVANCED MODEL 67

The pressure drop over the two orifices also look much alike. Of course such small measurements will
contain errors, so even though there is a relative high error in ∆P 2, it doesn’t mean that the constant
is wrong in the model.

L L L L
£ ¤ £ ¤ £ ¤ £ ¤
Test no. K L1 (SI M ) B ar · mi n K L1 (T E ST ) B ar · mi n K L2 (SI M ) B ar · mi n K L2 (T E ST ) B ar · mi n

5 0.204 0.199 1.198 1.242

8 0.354 0.354 1.611 1.681

11 0.331 0.342 1.194 1.202

Table 4.13: Flow constants of the load orifices L 1 and L 2

As for L 3 the constants are calculated and listed. They can be seen in Table 4.13. Test no. 8 yields great
results for both orifices, test no. 11 has an error in L 2 , but that could be due to measurement errors,
test no. 5 doesn’t yield as good results as the others. As the values of K is relatively far from each other
in some of the tests, it can be necessary to alter the values to fit that particular test series while other
parameters are considered.
The steady state response has now been analysed and documented, it is thus time to concentrate on
the dynamics of the model. As a small summary, the supply voltage was changed from 12V to 11.5V
and the pump flow was changed from 17l /mi n to 16.3l /mi n. The error in the pressure drops and the
flows varies for the different tests, but are all tolerable. The steady state flow response for test no. 2
and 3 are too high when the solenoid valve is closed, hence also the pressure drop over L 3 , this should
be kept in mind when analysing the dynmaics. The result of test no. 2 and 3 will only be used as a
help in determining the right parameters, not as a goal for the simulation to fit.

Dynamics

The analysis of the dynamics starts with a comparison of all the tests to the equivalent simulation. The
compared values are the pressure drop across the load orifice L 3 . The results are seen in Figure 4.41.
68 CHAPTER 4. PROBLEM SOLVING

Test number 2 Test number 3


80 80

60 60
Pressure [Bar]

Pressure [Bar]
40 40

20 20

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Time [s] Time [s]
Test number 5 Test number 6
80 80

60 60
Pressure [Bar]

Pressure [Bar]
40 40

20 20

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Time [s] Time [s]
Test number 8 Test number 9
80 80

60 60
Pressure [Bar]

Pressure [Bar]

40 40

20 20

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Time [s] Time [s]
Test number 11 Test number 12
80 80

60 60
Pressure [Bar]

Pressure [Bar]

40 40

20 20

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Time [s] Time [s]

Figure 4.41: The simulated (GREEN) and tested (BLUE), uncorrected, pressure drop over load orrifce
L 3 , at 8 different settings

The graphs have all been edited to fit the step time of the solenoid valve. The axes are the same in all
of the plots in order to illustrate the absolute error. The tests in the left column are the tests where
the solenoid valve opens, and the ones to the right are when it closes. The four plots in the top are for
the small orifice openings, and the four at the bottom are for large orifice openings. All the graphs fits
more or less in steady state, except the two in the top and the reason why is explained in section 4.7.3.
If the dynamics of the graphs are considered, they all seem to have a downward slope that is to steep
compared to the test. The effect is seen very clearly in the graph of test no. 2. What this means is that
when the solenoid valve opens, the pressure at the output of the valve (P l s ) drops too much before it
is limitted, compared to how much it drops in the test. What happens is that the pressure drops at
4.7. VALIDATION OF UNLINEAR ADVANCED MODEL 69

the output, thus there also is a pressure drop at P k , P 0 and P p aswell. When the pressure P p drops in
the simulation, a direct drop in force will take place aswell, since the pressure in the spring chamber
of the pump side spool is damped and the pressure in the pressure chamber is not. This results in a
closing motion of the spool leading more of the contracted fluid to the load instead of the tank. This
could infact be a good solution in reality, but the main purpose of the pump side spool is to make sure
that the pressure P p is about 15 bar higher than P l s, and therefore it should in fact open to the tank
when P l s drops. The reson why it doesn’t is that the small orifice in the pump side spool has been
neglected. The orifice and the pressure build up in the pressure chamber is inserted in the model,
the same way as it was for the compensator spool, and the simulations are made again. The orifice is
assumed to be A 7 = 0.3mm in diameter. Figure 4.42 shows the result.

Test number 2 Test number 3


80 80

70 70

60 60
Pressure [Bar]

50 Pressure [Bar] 50

40 40

30 30

20 20

10 10

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Time [s] Time [s]

Test number 5 Test number 6


80 80

70 70

60 60
Pressure [Bar]

Pressure [Bar]

50 50

40 40

30 30

20 20

10 10

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Time [s] Time [s]

Figure 4.42: The simulated (GREEN) and tested (BLUE), pressure drop over load orrifce L 3 . The
model has been changed to take care of the pressure build up in the pressure chamber

The figure now shows the right motion in test no. 2, and the velocity of the main spool are opening
as it should. But the motion is much more oscilatory. It is especially clear in test no. 5 and 6. As
the coulomb friction were neglected at the first approach, this is now introduced. When one look
at damping in linear systems, the constant added friction of a coulomb friction, doesn’t add more
damping, but in reality it will, since it will pull energy out of the motion that is not restored. The
Figure 4.43 shows the same simulations, but this time with an added coulomb friction of 100N for
70 CHAPTER 4. PROBLEM SOLVING

each compensator spool.

Test number 2 Test number 3


80 80

70 70

60 60

50 50
Pressure [Bar]

Pressure [Bar]
40 40

30 30

20 20

10 10

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Time [s] Time [s]

Test number 5 Test number 6


80 80

70 70

60 60

50 50
Pressure [Bar]

Pressure [Bar]

40 40

30 30

20 20

10 10

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Time [s] Time [s]

Figure 4.43: The simulated (GREEN) and tested (BLUE), pressure drop over load orrifce L 3 . The
model has been changed to take care of the pressure build up in the pressure chamber and the
coulomb friction.

Now the system behaves more like it should, at least the oscillatory motion that was too big is gone,
but the dynamics of the system still isn’t satisfying. The steady state error at the two top graphs have
increased, and the frequency of which test 5 and 6 oscilliates is also wrong. These coulomb frictions
therefore have an effect on these parameters. As the system is very unlinear, it is hard to say exactly
what a change in a parameter will result in. Therefore some linearizations is done and some method-
ical trial and error. This will help in the understanding of the system. First the coulomb friction is
tuned. As the steady state error have gotten bigger and the dominant natural frequency of the system
also have increased, the tuning have been focused on lowering the coulomb friction until the system
gets unstable, and then increasing it again, in order to decrease the coulomb friction as much as pos-
sible. Both the coulomb friction of the pump side spool and the PVB spool have been considered, and
the system is most sensitive to the coulomb friction of the PVB spool, so this one is kept at 50N, and
the coulomb friction of the pump side spool is removed again. The process also involved the viscous
friction. It was considered if the coulomb friction would be unnecessary if the viscous friction was
4.7. VALIDATION OF UNLINEAR ADVANCED MODEL 71

just high enough. That isn’t the case, the coulomb friction is mandatory, at least for the compensator
spool, in order to avoid the oscilation at the current areas of the different damping orifices. The orifice
in the newly installed orifice in the pump side compensator was also changed 0.2mm in both direc-
tions, resulting in worse results in either cases. Changing the viscous friction of the two spools results
in a strange behaviour compared to what the transfer function of the system predicts. The various
transfer functions derivation appear in section C.1. The change in the viscous friction, changes the
natural frequency of the response. The viscous friction doesn’t directly affect the natural frequency of
any of the spools according to the linear transfer functions. But the change in damping will change
the filter effect of the spools, so that it could amplify a broader or a more narrow spectrum. Maybe
the frequency of another system like the pressure build up frequency of P p and P l s will be amplified,
at least this is the only way the linear model can explain the phenomenom the unlinear model shows.
The effect is seen in Figure 4.44.

Test number 6
80
cvk=10000
70 cvk=5000

60

50
Pressure [Bar]

40

30

20

10

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time [s]

Figure 4.44: Two simulations of the system. Both displaying the pressure drop across the load orifice,
but with two different viscous friction coefficients.

Raising the viscous friction, changes the natural frequency, and lowering the coulomb friction yields
a faster pressure build up. Which is seen in test no. 6 particullary. The optimized result is seen in
Figure 4.45.
72 CHAPTER 4. PROBLEM SOLVING

Test number 2 Test number 3


80 80

70 70

60 60
Pressure [Bar]

Pressure [Bar]
50 50

40 40

30 30

20 20

10 10

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Time [s] Time [s]

Test number 5 Test number 6


80 80

70 70

60 60
Pressure [Bar]

Pressure [Bar]

50 50

40 40

30 30

20 20

10 10

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Time [s] Time [s]

Figure 4.45: The simulated (GREEN) and tested (BLUE), pressure drop across load orrifce L 3 . The
model has been changed to take care of the pressure build up in the pressure chamber and the
coulomb friction. The viscous friction is also changed in order to optimize the result.

The tests; 5 and 6, now fits quite well. The slope in test 5 is still a bit to steep, but it looks like the fre-
quency fits. The test number 3 also looks okay. Keep in mind that the steady state flow and therefore
also the pressure will be higher in the simulation than in the reality (see section 4.7.3). If they where
to hit the same pressure level, it looks like they could have the same settling time. Test number two
also looks good, except for the drop in pressure where the test actually shows the opposite. The spool
travel of the pump side spool (see Figure 4.46), reveals the reason.
Because of the deadband of the pump side area, it will first open to tank when it reaches 2mm. When
the unwanted pressure drop occur, the spool has just reached the 2mm point, and is moving down-
ward again as the pressure drops. But because it has some stored kinetic energy, it will take some
time. This effect is not seen when the system isn’t saturated, because in that case the spool will not
have to travel that far before it get’s damped by the falling pressure, hence also containing less kinetic
4.7. VALIDATION OF UNLINEAR ADVANCED MODEL 73

energy and therefore stops faster. If the damping was changed, it would be possible to make it fit the
curve of the test, but it will ruin the dynamics of the other tests. The reason is, that it is supposed
to hit the 2mm point a bit faster, thus lowering the pressure drop just like the test result. As there
is saturation involved this is just noticed and not changed, because the test and the simulation isn’t
exposed to the same flow Q v and therefore the other tests are considered to be more accurate and
therefore also more relevant. To prove that the lowering of the deadband will result in an elimination
of the bump, two simulations are made, and the result are given in Figure 4.47.

Time Series Plot of x_p


-3
x 10
3

Spool travel xp

2.5

1.5

0.5
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Time (seconds)

Figure 4.46: The spool travel of pump side spool during test no. 2

Test number 2
80
A3 0.5mm deadband
70 A3 2 mm deadband

60

50
Pressure [Bar]

40

30

20

10

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time [s]

Figure 4.47: Two simulations, describing what a change in the deadband of the pump spool will look
like in test number two.

All the graphs still shows very steep drop in pressure at the time of the step. This could be do to the fact
that the model assumes a laminar flow in the orifices, and maybe the flow at that point is turbulent.
The result have also gotten more oscillatory and therefore the settling time has increased, and this is
the reason for the inclination on the slope at the beginning of test no. 2 and 4. The rest of the tests
are plottet using the new found friction values c v p = 35000N s/m, c vc = 20000N s/m, c cp = 0N and
c cc = 50N , the result is seen in Figure 4.48.
74 CHAPTER 4. PROBLEM SOLVING

Test number 2 Test number 3


80 80
Pressure [Bar]

Pressure [Bar]
60 60
40 40
20 20
0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Time [s] Time [s]
Test number 5 Test number 6
40 40
Pressure [Bar]

Pressure [Bar]
20 20

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Time [s] Time [s]
Test number 8 Test number 9
40 40
Pressure [Bar]

Pressure [Bar]

20 20

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Time [s] Time [s]
Test number 11 Test number 12
40 40
Pressure [Bar]

Pressure [Bar]

20 20

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Time [s] Time [s]

Figure 4.48: The simulated (GREEN) and tested (BLUE), pressure drop over load orrifice L 3 . All 8 tests
compared with the simulation from a model, with updated frictions including the orifice in the pump
spool.

Parameter Symbol Value Unit

Flow rate of the pump Qv 16.3 L/min

Supply voltage bat _vol 11.5 V

Viscous Fric. PVP comp. cv p 35000 Ns/m

Viscous Fric. PVB comp. c vc 20000 Ns/m

Viscous Fric. Main. c vm 25 Ns/m

Coulomb Fric. PVP comp. c cp 0 N

Coulomb Fric. PVB comp. c cc 50 N

Diameter orifice PVP comp. D7 0.3 mm

Table 4.14: Alteret parameters of the model


4.7. VALIDATION OF UNLINEAR ADVANCED MODEL 75

4.7.4 Conclusion

The model could be optimized in many ways if there were put some more time into it, but as for
now the model describes the reality sufficient, so that it can be used as a tool to understand and
optimize the valve. Therefore the time and effort will be spent on this instead. The final result using
the parameters listed in Table 4.14 is shown in Figure 4.48, and it is seen that the model fits all the tests
but 2. It fits test 8 and 9 particullary well. To show that the model fullfill the requirement from the
problem statement, stating; no more than 30% relative error, and 10% will be considered as a good
result, the relative error of each test are plottet in Figure 4.49.

Test number 2 Test number 3


30 30

25 25
Relative error %

Relative error %
20 20

15 15

10 10

5 5

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Time [s] Time [s]
Test number 5 Test number 6
30 30

25 25
Relative error %

Relative error %

20 20

15 15

10 10

5 5

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Time [s] Time [s]
Test number 8 Test number 9
30 30

25 25
Relative error %

Relative error %

20 20

15 15

10 10

5 5

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Time [s] Time [s]
Test number 11 Test number 12
30 30

25 25
Relative error %

Relative error %

20 20

15 15

10 10

5 5

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Time [s] Time [s]

Figure 4.49: The error of the 8 different simulations relative to the corresponding tests.
76 CHAPTER 4. PROBLEM SOLVING

The data are smooth using moving average filter, in order to remove the little peak in error, due to the
small misalignment at the time where the step occur.

4.8 Linear model

To get a deeper understanding between the parameters interacting when trying to control the sys-
tem, a linear model will be developed. This model will lack some accuracy compared to the unlinear
model, but the simplicity makes it easier to tune a controller, and also helps in the understanding of
the system. Therefore it is evaluated to be within the boarders of the problem statement. Because
the linear model is an approximation around a given point of work, the model can’t give better results
than the accuracy of the linearizations.

4.8.1 Initial thoughts and assumptions

The unlinear, advanced model is validated in the previous section. This is therefore a usefull tool
when considering simplifications and assumptions when designing the linear model.

The compensator spool is designet to keep a constant pressure drop over the main spool. In order
to reveal this, the advanced model is simulated. There is a step in the simulation in the main spool
position at time t = 0.1s and a step in the external load at t = 0.8s. The plot seen in Figure 4.50 shows
the same characteristic for the load pressure P l s , the pressure before the main spool P k and the pump
pressure P p in the volume Vp . This must therefore result in a constant pressure drop over both the
compensator spool and the main spool.
6
x 10
10

7
Pp
6
Pressure [Pa]

Pk
5 Pls

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
Time t [s]

Figure 4.50: The pressure measured at the load P l s , before the main spool P k and at the pump side
volume P p

This observation makes it possible to model the pressure drop as a constant given in Equation 4.135,
making the flow rate controlable by the position of the main spool.

K P kl s = P k − P l s (4.135)

As the linear model will involve a force equilibrium of the main spool, the flowforces are investigated,
to see whether they can be ignored. To assist in this decision the unlinear model are utilized. All the
4.8. LINEAR MODEL 77

simulated flowforces of the valve are illustrated along with the force due to pressure, of the pump
side spool, in Figure 4.51. As the flow forces are much smaller than the pressure force, they are all
neglectable. The reason is the small flow demand in all the tests. If the valve was exposed to a higher
flow, the flow force will increase quadratic.
The different functions in the model has to be linearized in order to derive a transfer function. The
first order taylor series is used to expand the functions about a working point. Depending on the
order of the functions, in worst case scenario, merely a small interval around the working point x 0 is
represented by the linear model. The linearization of the PVE module follows.

Multiple Time Series


0.2
A2 Flow force
0.1

0
3
A1 flow force
2

0
4
AH Flow force

0
20
A3 flow force
10

0
1000
Force due to Pump pressure

500

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
Time (seconds)
Figure 4.51: The simulated flowforces in the valve during a simulation with parameters equivalent to
test number 2 explained in section B.7.

4.8.2 The PVE module

The model of the PVE module is derived in section 4.5. The block diagram in figure 4.52 illustrates the
starting point of the PVE module. The input signal is transfered to a position and the feedback signal,
through the LVDT is subtracted. The difference between these positions gives the error in position.
78 CHAPTER 4. PROBLEM SOLVING

X set
Vrel
Vrel
+ X error
Xm LVDT
-
Figure 4.52: Block diagram for the starting point of the PVE module, from an input signal to an error
in position.

The block diagram illustrates how the error signal is calculated (see equation 4.136).

x er r or = ∆x set − x m 0 (4.136)

The input signal to a defined position x set is calculated in Equation 4.137 using taylor expansion.

∂x set ¯¯
¯
x set ≈ x0 + · (Vr el − V0 ) = x 0 + K i nput · (Vr el − V0 ) (4.137)
∂Vr el ¯V0

The working point is defined as V0 . From Equation 4.138 it is seen that there is no dynamics in this
part of the model, hence only a constant gain will be the transfer function.

x set − x 0 = K i nput · (Vr el − V0 )



∆x set = K i nput · ∆Vr el

∆x set
= K i nput (4.138)
∆Vr el

The LVDT transfer function is of first order, and has dynamics represented by a time constant τl vd t .
Furthermore there is a steady state gain between the actual position x m and the measured x m 0 which
is represented by K l vd t . The LVDT sensor model is derived in Equation 4.113 and repeated in equation
4.139.
xm0 K l vd t
= = G l vd t (4.139)
xm τl vd t · s + 1

To compensate for the error in position x er r or , the solenoid valves will try to build up pressure and
move the spool in the correct direction. Ultimately, the NC and NO valves will open for fluid flow as
seen in Figure 4.53.

DTNC1 Q NC1
X err DTNC1
+ Port B
X err
Q NO2 - system
DTNO2
X err DTNO2
Figure 4.53: Block diagram from the error in position to a flow through one port chamber at the PVE
module.
4.8. LINEAR MODEL 79

To reach a given flow, first the error in position is transformed into a duty cycle, which are different
for each port as seen in Figure 4.28. The first order transfer function describing this proces is given
in equation 4.115. Apparent from previous calculations, a taylor expansion yields the result given in
4.140 for port B and in 4.141 for port A.

DT5
= K DT −5 (4.140)
x er r or
DT4
= K DT −4 (4.141)
x er r or

The saturation of the duty cycles in the unlinear model are not implemented, in these linear equa-
tions, hence the duty cycle for both ports will be the same with opposite signs, leading to same fluid
flow to both chambers. To avoid this, the linear model will focus on one equivalent port obliging this
problem. When the duty cycle is negativ, the pressure in the chamber will be negative, and hereby
modelling a negative pressure drop as seen illustrated in Figure 4.54.
The transfer function for the duty cycles will hereby change to the
ones given in equation 4.142.
Ppilot
Qsupply
Pupper
DT4
= K DT (4.142)
x er r or Vupper
NC1 NC3

The pressure drop over the NO and NC valves will change as the
pressure inside the chamber increase and decrease as a resultPort of theB Port A
variations in the flow as seen in figure 4.22. Initially, when P the
5 cham- P4
bers starts filling up, the pressure over a NC valve can reach the pilot
V5 V4
pressure. The linearization of the flow through the NC valve will be
calculated based on a pressure drop at P pi l ot , while
NO2for the NO valve
will be calculated based on pressure less than half than P pi l ot . The b Acheckball NO4
a
tank side valves needs to close before the pilot side valves open in
order to avoid shortcut. Here the linear model falls short in imple-
Vlower
menting the time constants. Implementing this in the linear model
a first order system with a time constant for both valves
Plower are intro- Qdrain
duced, as seen in equation 4.144. Ptank

Figure 4.54: The figure illus-


trates which part of the PVE
Q NC K NC
= τNOC · s+1 (4.143) modules hydraulic that are
DT NC modelled.
Q NO K NO
= τNOC · s+1 (4.144)
DT NO

The pilot pressure is previously defined to be P pi l ot = 13.5B ar ,


hence the linearization of the flow as a function of the dutycyle, will
be based on a pressure drop over the NC valve at P NC = 13.5B ar
and for the NO valve P NO = 4B ar . A linear approximation of these
graphs are given in figure 4.55.
Pupper
80 CHAPTER 4. PROBLEM SOLVING

−5 −6
x 10 x 10
3 20
2 Bar 2 Bar
13.5 Bar Linfit2
2.5 y = − 1.8e−007*x + 1.6e−005 4 Bar
Linfit13.5
15 Linfit4
Linfit2 y = 3e−007*x − 3.7e−006
2
[m3/s]

Flow Qin−Qout [m /s]


3
10
1.5
out
Flow Q −Q
in

1
5
y = 1.1e−007*x − 2.4e−006 y = − 1.7e−007*x + 1.3e−005
0.5
0
0

−0.5 −5
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Duty cycle DT NC [%] Duty cycle DT NO [%]

Figure 4.55: Linear fit of the flow through an NC valve with a pressure drop of 13.5B ar and a NO
valve at 6B ar

As the pressure increases in the the port, the pressure drop over the upper valve, NC , will decrease,
while the lower valve NO will experience a higher pressure drop. This will effect the flow through each
valve. Because a transfer function with a single input is wanted, the change in pressure can only be
implemented using feedback. The transfer fucntion and the linearization is calculated for a duty cycle
at D = 70% in Figure 4.22 for both valves. These linearizations are given in figure 4.56 with equations.
−5 −6
x 10 x 10
2 12
∆ Q (PB) for DT=70% ∆ Q(Pb) for D=80%
1.8 y = 1.2e−011*x + 1.5e−006 y = 8.3e−012*x − 1.9e−006
Linfit 10 Linfit
1.6

1.4 8
∆ QNC [m /s]

[m /s]

1.2
6
3

1
NO

4
∆Q

0.8

0.6 2

0.4
0
0.2

0 −2
0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14
Pressure PB [Pa] 5
x 10 Pressure PB [Pa] 5
x 10

Figure 4.56: Linearization of the change in flow as a function of the pressure inside the chamber P 4
for the NC valve in the left and the NO valve in the right side graph

This transfer function is given in equation 4.146 with the gain as the slope of the graphs.
dQ NC
= K NC −QP (4.145)
P4
dQ NO
= K NO−QP (4.146)
P4

The flows through each valve, including the change in flow, will be summarized with the direction
of flow in mind. The reason why the change in flow for the NO valve is added, is because a higher
pressure inside the equivalent chamber results in a higher pressure drop over the NO valve which
results in a higher flow rate.

dQ NC dQ NO
µ ¶ µ ¶
∆Q 4 = Q NC − − Q NO +
P4 P4
4.8. LINEAR MODEL 81

The flows yields a pressure build up in P 4 . The block diagram showing the step from flow through the
valves to a pressure difference over the main spool are seen in figure 4.57.

Port B
P5 Actual spool
Q5 Q5 position
- Spool System
Port A
P4
+
Q4 Q4
Figure 4.57: Block diagram from the flow through each valve to a pressure build up in the chamber

The continuity equations describing the pressure build up inside the chamber are given in equation
4.103 for the equivalent port. Notice that the change in volume are neglected, and only the com-
bined flow through the NO and NC valves are kept in the calculations, because the flow through the
checkball valve is assumed to be very small.

d P4 β
= · ∆Q 4
dt V4

β
s · P 4 (s) = · ∆Q 4 (s)
V4
m
P4 β
= (4.147)
∆Q 4 s ·V4

Internal feedback implementation

At this point the linearization falls short. The pressure inside the chamber P 4 will keep rising as the
flow keeps running in, and eventually P 4 will be greater than the pilot pressure. To regulate the flow
as a result of the change in P 4 , a feedback will be implemented. Since the duty cycle controls the flow
through each valve, the duty cycle needs to be adjusted as a result of a higher pressure. Ultimately,
when P 4 is equal to the pilot pressure the flow should stop from flowing through the NC valve, by
adjusting the duty cycle to zero. Equation 4.148 states the transfer function for the feedback.

DT p−d t 100 − 0
= = K P DT (4.148)
d P4 P pi l ot − P r eser voi r

The implementations made to compensate for changes in flow as a result of changed pressure are
illustrated in figure 4.58.
82 CHAPTER 4. PROBLEM SOLVING

∆QN C
PB

-
QN C
+
xerror DTB DTB
+ + PB PB
xerror
- - Q
QN C
+
DTB
+

∆QN O
PB

∆DTB
Pb

Figure 4.58: Block diagram from the error in position to a pressure drop between main spool

The block from an error in position to the pressures in the chambers in each port, will be calculated in
the following equations for the pressure in the equivalent chamber. First new constant are introduced
to ease the calculations:
K NC
K LX P 1 =
τNOC · s + 1
K LX P 2 = K NC −QP
β
K LX P 3 =
V4 · s
K LX P 4 = K NO−QP
K NO
K LX P 5 =
τNOC · s + 1

Using the new constans, the closed loop feedback can be reduced for the pressure as a function of
the error in position given in Equation 4.149 with a new constant introduced to reduce the term for
succeeding calculations.

P4 = (x er r or · K LX P 1 − K LX P 2 · P 4 ) · K LX P 3 − (x er r or · K LX P 5 + K LX P 4 · P 4 ) · K LX P 3
m
x er r or · K LX P 3 · (K LX P 1 − K LX P 5 ) = P 4 · (1 + K LX P 2 · K LX P 3 − K LX P 4 · K LX P 3 )
m
P4 K LX P 3 · K LX P 1 − K LX P 3 · K LX P 5
=
x er r or 1 + K LX P 2 · K LX P 3 − K LX P 4 · K LX P 3
m
P4
= K Peq (4.149)
x er r or

Multiplying the pressure drop over the main spool with the area of the main spool A m yields a force
which are used to move the main spool. In order to compensate for other forces that has an impact
on the main spool dynamic behaviour, these forces needs to be examined.
4.8. LINEAR MODEL 83

4.8.3 Dynamics of main spool

The model of the main spool was derived in a previous section and is stated in equation 4.91. The
input parameters to this model has been linearized in the previous sections. A laplace transform will
be applied to derive a linear model of the main spool, with initial conditions assumed to be zero:

s 2 · x m (s) · M m = −s · c vm · x m (s) − k m · x m (s) + K Peq · x er r (s) · A m


m
2
s · x m (s) · M m = x m (s) · (−k m − s · c vm ) + x er r 0r (s) · K Peq · A m
m
³ ´
x m (s) · s 2 · M m + k m + s · c vm = x er r (s) · K Peq · A m
m
1
x m (s) · = x er r or (s) ·G pve
G spool

The superior transfer function is shortened into smaller transfer function in parts that can be related
to the psysical system; The spool, the pve module and the LVDT sensor which are derived in Equa-
tion 4.139. These smaller transfer functions are listed in Equation 4.152.

1
G spool = (4.150)
M m + k m + s · c vm
G pve = K Peq · A m (4.151)
K l vd t
G l vd t = (4.152)
τl vd t · s + 1

Inserting equation 4.138, 4.139 and 4.136 in the final expression for the transfer function for the pve
module and the main spool yields a closed loop given in equation 4.153.

x m (s) G pve ·G spool


= (4.153)
x set (s) 1 +G pve ·G spool

4.8.4 Closed loop control system

The closed loop transfer function of the system is now ready to be derived. The input signal is de-
rived in equation 4.138 and the spool position is measured by the LVDT sensor which was derived in
equation 4.139. The error in spool position is calculated as the difference between the desired and the
measured position. Figure 4.59 illustrates the closed loop system with a plant G pl ant to be controlled,
which symbolises the linear model of the PVE module and the main spool. The controller is named
G c and will be adjusted according to the response and the defined requirements of the controller.

xset xerror xm
X set
Vrel + Gc Gplant
-x
m'

LVDT sensor

Figure 4.59: The model of the PVE module and the main spool in a closed loop control system named
the plant G pl ant
84 CHAPTER 4. PROBLEM SOLVING

The coefficients for the linear model are calculated based on the values introduced through the sec-
tions describing the linerization and the measured constants for the PVG 32. Using these coefficients,
the transfer function for each individual part of the linear model can be derived as stated in Table 4.15

1
G spool = M m · s 2 +c vm · s+k m
1
= 1.7540 · s 2 +25 · s+3695
1
= 1.7540 · s 2 +25 · s+3695

G pve = K Peq · A m
K LX P 3 · (K LX P 1 −K LX P 5 )
= 1+K LX³P 3 · (K LX P 2 −K LX P 4 ) ´
β
V ·s
4
· τ K NC· s+1 − τ K NO· s+1
NOC NOC
= β
1+ V
4 ·s
· (K NC³ −QP −K NO−QP ) ´
7000 · 105 3 · 10−7 1.7 · 10−7

= 1.29 · −5 · s · 0.003 · s+1 − 0.003 · s+1


1+ 7000 ·−5105
· s · (8.8 · 10 −8.3 · 10 )
−12 −12
1.29 ·

= (s+0.33 · 102.4 · 10 9
3 )(s+0.027 · 103 )

G i nput = K i nput = 0.0028


K l vd t 0.945
G l vd t τl vd t · s+1 = 0.0059 · s+1

Table 4.15: How to calculate the coefficients in the linear model of the main spool

Using these coefficients gives an unstable step response. Figure 4.60 shows a plot of the closed loop
transfer functions poles and zeros.
Pole−Zero Map
500
0.5 0.38 0.28 0.17 0.08
400 0.64 400

300 300
0.8
200 200

0.94 100
100
Imaginary Axis

−100
0.94 100

−200
200
0.8
−300
300

−400 0.64
400
0.5 0.38 0.28 0.17 0.08
−500
−350 −300 −250 −200 −150 −100 −50 5000 50 100 150
Real Axis

Figure 4.60: A plot of the poles and zeros in the closed loop transfer function G cl

One conjugated pole pair in the right hand plane, is the reason why the transfer function is unstable.
The first parameter to analyse is the damping ratio ζ. For a second order linear system the general
transfer function is given by:
C (s) K ω2n
= 2 (4.154)
R(s) s + 2ζωn · s + ω2n

Basically there are two systems that has an effect on the response of the transfer function, respectively
4.8. LINEAR MODEL 85

G pve and G spool . Expanding these functions for s, the parameters determining the damping will be
revealed. The damping ratio depends on the coefficients multiplied with the s term.

β (K NC + K NO )
G pve = (4.155)
(τNOC · s + 1) V4 · s + K NC −QP · β − K NO−QP · β
¡ ¢

β (K NC + K NO )
=
s · τNOC ·V4 + s · τNOC · K NC −QP · β − τNOC · K NO−QP · β + V4 + K NC −QP · β − K NO−QP · β
¡ ¢
2

(4.156)

For the PVE module the expanded equation 4.156 shows that the damping ratio depends on the bulk
modulus β, the volume V4 and the transfer functions regulating the flow as a function of the pressure
build up (K NC −QP and K NO−QP ). Adjusting all the parameters in an reasonable range in order to make
the damping ratio larger, does seems to move the poles in the right half plane closer to the origine,
but no further. Equation 4.150 is allready in an expanded form, and the equation reveals that only
the viscous friction coefficient has an effect on the damping ratio. Enlarging this coefficient from
c vm = 25 to c vm = 2500 is the crucial factor for damping the transfer function as seen in Figure 4.61.

Pole−Zero Map
400
0.955 0.91 0.84 0.74 0.56 0.3

300 0.98

200
0.995
100
Imaginary Axis

1.4e+003
0 1.2e+003 1e+003 800 600 400 200

−100
0.995
−200

−300 0.98

0.955 0.91 0.84 0.74 0.56 0.3


−400
−1500 −1000 −500 0
Real Axis

Figure 4.61: A plot of the poles and zeros for the transfer function G cl after the damping ratio is
adjusted for the spool

To compare the step response of the linear model and the unlinear model, a plot of both models are
made using a step input of x set = 3.5mm. The step resonses are shown in figure 4.62
86 CHAPTER 4. PROBLEM SOLVING

−3
x 10
4.5
Unlinear model
Linear model
4

3.5

3
Step response

2.5

1.5

0.5

0
0.1 0.15 0.2 0.25 0.3 0.35
Time [s]
Figure 4.62: The figure shows the steprespons of the linear and the unlinear model

The dynamics of the linear model seems to match quite well, and the steady state error are not far
from the unlinear. One remarkable fact, is the difference in the rise time. This is because of satura-
tions in the unlinear model, especially in the duty cycle, the pressure inside the ports are limited to
a range between the pilot pressure P pi l ot and the tank pressure P t ank . In addition, a more accurate
implementation of the flow in and out of Port A and Port B chambers in the unlinear model, does also
play an important role in deviation between the linear and unlinear model. Even though the linear
model contains a simple implementation of the change in flow, the pressure in the chambers still ex-
ceeds the limits, resulting in faster step respons time. As for the unlinear model the model can also
be optimized and tuned further, but as a tool for understanding the valve, it is suffficient as it is.

4.9 Controlling the valve

This section will show a small analysis of the system using the linear model and frequency analysis. It
is meant as a small demonstration of how the models can be used in controller designs.
The step response for the linear system in Figure 4.62 shows some overshoot O, a steady state error
e ss and a rise time t r . These quantities are measured to be:

• RiseTime: t r = 0.0079s

• SettlingTime: t s = 0.0619s

• Overshoot: O = 27.7916%

• Steady state error: e ss = lims→0 s ·G ol (s) = 4%

The rise of the system is very small, causing a somehow large overshoot. The purpose of the controller
could be to reduce the overshoot by increasing the damping and increasing the rise time. The tools
used to optimize a step response are the bodeplot and the rootlocus given in respectively Figure 4.63
and Figure 4.65.
4.9. CONTROLLING THE VALVE 87

Bode Diagram
Gm = 9.91 dB (at 214 rad/sec) , Pm = 49.7 deg (at 87.5 rad/sec)
50

Magnitude (dB)
−50

−100

−150

−200

−250
0

−90
Phase (deg)

−180

−270

−360

−450
−1 0 1 2 3 4 5
10 10 10 10 10 10 10
Frequency (rad/sec)

Figure 4.63: Bodeplot of the linear system

The bodeplot shows the following gain and phase margins:

• At ωg = 214r ad /s the gain margin is G m = 9.91d B

• At ωp = 87.5r ad /s the phase margin is P m = 49.7◦

The PVE module already contains a closed loop controller, that controls the position of the main spool
through a LVDT sensor. The data from the bodeplot shows that the controller are very well tuned and
only a small margin is left for optimization. A reduction in proportional gain will cause both the phase
and gain margin to increase and result in a more stable system with more damping on behalf of how
fast the system behaves as seen in figure Figure 4.64.
88 CHAPTER 4. PROBLEM SOLVING

Step Response
1.4

1.2 Gcl

Gcl-damped
0.8
Amplitude

0.6

0.4

0.2

0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
Time (sec)

Figure 4.64: Step response with a proportional controller G p = 0.3 (green) and step respons of the
linear system

As seen in the Figure 4.65, a larger gain will cause the complex conjugated poles to move to the right
half plane and make the system unstable. Because of this, the system is called conditional stable.
Using a P controller will change the amplitude characteristics also changing the bandwidth BW of
the system. As seen in equation 4.157 the rise time t r will increase when the bandwith is reduced.
Consequently, this will also have an impact on the settling time.

K
tr = (4.157)
BW
4.9. CONTROLLING THE VALVE 89

Root Locus

1500

1000

500
Imaginary Axis

−500

−1000

−1500

−2000 −1500 −1000 −500 0 500 1000


Real Axis

Figure 4.65: Root locus of the linear system

After analysing the system with the two methods, a controller could be designed. To validate the con-
troller the unlinear model could be used, as this takes saturation and other unlinearities into account.
This small section shows that it is possible to use the models as a tool in controller design, utilizing
bodeplot and root locus design methods.
Chapter 5

Conclusion

How to model and control a Sauer Danfoss PVG 32 valve?


The purpose of this project has been to thoroughly understand the principles and dynamics of the
Sauer-Danfoss PVG 32 proportional valve. To accommodate this, the following items were listed as
goals in the problem statement:

• Make a dynamical mathematical description of the valve.

When developing the mathematical model, a deep understanding of the equations used to describe
hydraulic systems where achieved. In order to determine the quantities used in the equations, it was
necessary to dismantle the PVG 32 valve. Several lab tests have been performed to find the precise
physical properties of these parts.

• Create a simulink model which simulates the real nature of a PVG valve.

The mathematical model was modelled in Matlab Simulink ™, in order to solve the equations in
a clear and structured manner. The system was recognized as a stiff system with different states,
therefore a suitable solver utilizing variable time step was found.

• Perform tests to use for validation of the model.

A test arrangement was established in the laboratory. Several measurements of different load situa-
tions were measured. It was posible to log the pressures in the test rig with Labview™and the flows
were manually read out. The process of making the test, gave a better relationsip to how the valve ac-
tually acts, and what physical apperance the different modules of the hydraulic diagram looked like
in reality.

• Validation of the dynamical model.

The tests were analysed and because of a good understanding of the characteristics describing the two
flowtypes, the load orifices were recognized as laminar. Also, the supply voltage of the test seemed
lower than informed, which also shows good knowledge of how the flow is affected by the supply
voltage. Upon simulating the different tests in the model, the uncertain parameters were tuned so
the dynamics of the simulation were describing the reality more accurate. In order to figure out the
right parameters to tune, and how to do it, a methodical trial and error approach was used, along with
derivations of linear models of interesting subsystems. At last a comparison of the model to all the

91
92 CHAPTER 5. CONCLUSION

tests were caried out, and the relative error was all less than 30% as required, and the relative error
of two of the tests showed particularry good results, which where under 10% in all of the scope, and
less than 5 % in most of it. This was considered a tolerable error due to measurement errors, and
unmeasureable quantities.

• Simplify the advanced dynamical model to a linear model.

The part of the dynamic unlinear model, concerning control of the main spool position was simplified
to a linear model which made it easier to determine the various parameters affecting the dynamics of
the valve. To be able to linearize the model, a point of operation was determined, and certain forces
and dynamics were neglected, based upon information recieved from simulations using the unlinear
validated model.

• Understand how to use the developed tools to optimize the valve.

The last section of the report showed a small demonstration of how the bodeplot and root locus de-
sign methods could be applied to the linear model, after a brief discription of step response design
criterias.
Chapter 6

Future work

When modelling the PVG 32, some parts of the valve was neglected or considered constant. This was
done to simplify and make the model faster, but if a more detailed simulation is needed, functions
could be derived for; oil density ρ, Bulk modulus β or the flow angles φ.

A better controller could be developed to control the main valve, and an external system, like a crane.
This could be done using the two validated models.

Also if more tests where executed it would be possible to better determine the parameters of the
model, and to test whether the model of the turbulent load is also valid.

93
Bibliography

Andersen, Torben Ole, & Hansen, Michael Rygaard. 2007a (June). Fluid Power Circuits - System
Design and Analysis. Tech. rept. 3.ed. Institute of Energy Technology and Institute of Mechanical
Engineering at Aalborg University.

Andersen, Torben Ole, & Hansen, Michael Rygaard. 2007b (June). Fluid Power Circuits - System
Design and Analysis (section 3.2 - The flow force equation). Tech. rept. 3.ed. Institute of Energy
Technology and Institute of Mechanical Engineering at Aalborg University.

Danfoss. Simulink model of the PVE module. CD/model/PVE2009/PVE_CM_191109.mdl.

Danfoss. Tryktransmitter til industriel anvendelse Type MBS 32 og MBS 33.

Danfoss, Sauer. 2000. Technical Information - PVE Series 4 for PVG 32, PVG 100 and PVG 120.

Danfoss, Sauer. 2006. PVG32 product brochure. june.

Parker. 2000. Measurement technology for flow volume in hydraulics.

Pedersen, Henrik Clemmensen. 2009. Conversations during the project period.

Power, Danfoss Fluid. 1999. AMOC (Adaptive Orifice Configuration) Version 1.0 Beta.

Wikipedia. Control Theory. http://en.wikipedia.org/wiki/Control_theory.

Wikipedia, the free encyclopedia. Linear variable differential transformer.


http://en.wikipedia.org/wiki/Linear_variable_differential_transformer.

Wikipedia, the free encyclopedia. Transformer. http://en.wikipedia.org/wiki/Transformer.

B1
Part I

Appendix

B3
Appendix A

Govering Equations

A.1 Orifice equation

In hydraulics, orifices are used to control the flow, This is equivalent to using resistors in electrical
circuits. In electronics, the relationship between voltage and currrent is given by Ohm’s law. In hy-
draulics, the relationship between pressure and flow is given by the orifice equation (Equation A.1).
Almost every valve is designed to follow this equation. However, this is not always the main design
criteria.

P1 A0
Q P2

Figure A.1: Turbulent flow through an orifice.

The discharge coeffient is a function of the Reynolds number, which again is a function of the ve-
locity, vicosity, density and hydraulic diameter of the orifice. The Reynolds number is proportional
to the velocity and the hydraulic diameter of the pipe, hence flows with low velocity and a small ori-
fice opening will have a small Reynolds number. The viscosity changes highly with temperature and
pressure. A plot of the viscosity for commonly used mineral oils are shown in Figure A.2.

B5
ν kinematic viscosity, [cSt]
Cν,mν C v A 2 for the specific
constants 2 fluid
ta Q = absolute temperature, p1 − p 2 )
([K] (1.26)
1 − (A 2 / A1 ) 2 ρ
B6 This dependency is normally shown in specially designed charts,A.where
APPENDIX the kinematic
GOVERING EQUATIONS
viscosity shown as function of the temperature becomes a straight line, see Figure A4.
Defining the discharge coefficient C d in Equation 1.26 it is possible to express the
ν [cSt]
orifice flow by the orifice area.
5000

2000
C vCc
1000 Cd = (1.27)
500 1 − C c2 (A 0 / A 1 ) 2
300
200
150
Now, combining Equation 1.22, 1.26, and 1.27 the orifice equation (in Danish
100
80
blændeformlen) can be written 60
50
40
30
25
2
20
Q = Cd A 0 ( p1 − p 2 ) (1.28)
ρ
16
12 ISO VG 100
10
ISO VG 68
8
Normally A 0 is much smaller than A 1 and since C v ≈ 1 , the ISO
7 discharge
VG 46 coefficient is
6 ISO VG 32
approximately equal5 to the contraction coefficient. Different ISO VG 22 theoretical and
4
experimental investigations
3.5
has shown that a discharge coefficient of C d ≈ 0.6 is often
ISO VG 10
assumed for all spool2.73orifices.
-50 -40 -30 -20 -10 0 10 20 30 40 50 60 70 80 90 100 120
t [°C]
At low temperatures, low orifice
Fig. A4 Uddebuhle-chart: pressure
The temperature drop, for
dependency and/or
some ofsmall
the most orifice
commonly openings,
used the
Figure A.2: Uddebuhle-chart: Viscosity of commonly used mineral oils. The ISO VG standard refers ν
Reynolds number may become mineral sufficiently
oils. The ISO VGlow to refers
standard permitν at 40°C
laminar flow. Although the
at 40°C. Andersen & Hansen [2007a]
analysis leading to Equation 1.28 is not valid at low Reynolds numbers, it is often used
The vertical axis of an Uddebuhle chart is a mapping of log log(ν+0.8), i.e.,
anyway by letting the discharge coefficient be a function of Reynolds number. For Re <
approximately a double logarithmic axis (especially at higher values of ν). The
10 experimental
The orifice results
equation
horizontal axis isshow
is only thatofthe
a theoretically
mapping discharge
valid
logT, for coefficient
i.e.,non-laminar
a logarithmic flows A is
axis.(flows directly
with high
hydraulic proportional
is, in num- to
Reynolds
fluid
ber), butgeneral,
when the discharge coefficient is made dependant on the Reynolds number, it can be em-
the square root of referred
Reynolds
ployed on any kind of flow,
to by number;
its kinematic viscosity
that is C d = δ
at 40°C. Re . A typical plot of such a result
p provided that the characteristic of the discharge coefficient is available. A
is shown inplot
typical Figure
of C d 1.20.
versus Re is given in Figure A.3.
Hydraulic Fluids Page 5 of 11
0.8

0.6
Cd
0.4

0.2

0
0 10 20 30 40 50
Re
Figure 1.20 Plot of a discharge coefficient versus Reynolds number for an orifice
Figure A.3: Discharge coefficient as a function of the squareroot of the Reynolds number. Andersen
& Hansen [2007a]

Experiments have shown that when the reynolds number is below Re = 100 a linear relationship exist
Hydraulic Principles Page 19 of 21
A.2. THE ORIFICE EQUATION IN MODELS B7

p
between the reynoldsnumber and the discharge factor C d = δ Re.

s
2
Q = Cd A0 (P i n − P out ) (A.1)
ρ

h i
m3
Q Flow out of orifice s
£ 2¤
A0 Area of the orifice m

Cd Flow factor (often assumed to be 0.6) [·]


h i
kg
ρ Density of the fluid m3

Pi n Up stream pressure. The pressure before the orifice. [P a]

P out Down stream pressure. The pressure after the orifice [P a]

The orifice equation states that the flow through an orifice is proportional to the square root of the
pressure drop across it, hence the pressure drop across an orifice is proportional to the flow squared.
The relationchip is k eq which is a equivalent constant defined in Equation A.3.

∆p = k eq ·Q 2 (A.2)

ρ 1
k eq = ·p (A.3)
2 Cd · A0

h i
m3
Q Flow out of orifice s
h i
kg
k eq Equivalent discharge factor m3

∆p Pressuredrop across the orifice [P a]


£ 2¤
A0 Area of the orifice m

Cd Flow factor (often assumed to be 0.6) [·]


h i
kg
ρ Density of the fluid m3

If the manufacturer of a valve provides a pressure vs. flow curve, it is possible to obtain k eq .

A.2 The orifice equation in models

For fluid flow, an orifice works in both directions as long as the direction of the fluid follow the pres-
sure drop. A positive flow value illustrates a correct modelled flow direction, while a negative flow
value should illustrate an opposite flow direction. The original equation doesn’t take this basic fea-
ture into account. Therefore it is assumed, that all the orifice equations are working in both directions
if nothing else is stated. This means that the orifce equation will be changed to the following Equa-
tion A.4, so the correct flow direction is given by a positive value and a negative value means opposite
flow direction.
B8 APPENDIX A. GOVERING EQUATIONS

s
2
Q = Cd · A · |P 1 − P 2 | · si g n(P 1 − P 2 ) (A.4)
ρ

The absolute value of the pressure difference will allways give a positive number, while the si g n func-
tion take care of the direction by the following method:





 1 if P 1 ≥ P 2

si g n(P 1 − P 2 ) = 0 if P 1 = P 2



 −1 if P 1 < P 2

A.3 Translatoric friction

Consider the free body diagram of Figure A.4. This model contains a mass m, which is moving with
the speed v, and a friction force F acting in the opposit direction of the motion. The model isn’t con-
sidering gravity directly hence no normal force. Indirectly the gravity is included in the friction force.
The friction force F acting on the body is composed by 3 types of friction forces: "Striction, Coulomb
and Viscous". The different frictions is best described with an illustration. Consider Figure A.5. The
figure illustrates the three friction forces, and the resulting friction force which is the sum of them.

M
V

Figure A.4: A free body diagram illustrating the friction force, the velocity and the mass of an arbitrary
mass. Static friction Coulomb friction
F F
M
V

F
A.4. FLOW FORCE B9

Striction Coulomb friction


F F

V V

Viscous friction Striction + Coulomb + Viscous


F F

V V

Figure A.5: An illustration of the different friction forces composing a resulting friction, which is also
illustrated

The equation describing the resulting friction as a function of velocity is given in Equation A.5. The
equation is quite intuitive when the graphs are considered, hence no further explanation is given.

|ẋ|
· µ ¶¸
F f (ẋ) = σẋ + si g n(ẋ) · F co + F so exp − (A.5)
Cs

£m ¤
ẋ Velocity of the spool s
h i
kg
σ Viscous friction factor s

F co Constant Coulomb friction [N ]

F so Stribeck friction [N ]

Cs A factor that determines the decay of the stribeck effect as velocity increases. [·]

A.4 Flow force

When fluid flows across an orifice, it will try to close it. The acting force is called a flow force. If the
orifice is fixed, this force is of little interest, because it is often small compared to the forces due to the
pressure. When the orifice is variable, this force can lead to errors, because the force required to give
a specific discharge area is underestimated. The principle of conservation of momentum is used to
describe this phenomenon.

∂ ∂~
v
~=
F v ) = ṁ ·~
(m ·~ v +m
∂t ∂t
B10 APPENDIX A. GOVERING EQUATIONS

If the flow is set to pass through a control volume ( ∀ ) and assumed to be in steady state, the velocity
is constant, and the equation can be rewritten. This way only the stationary flow forces are described.
This assumption is valid because, in this equation, the change in flow is the dominating force.

Q
~ = ṁ ·~
F v = ρ ·Q ·~
v = ρ ·Q · vˆ
·~
Cd · A0

The discharge coefficient (C d ) is assumed equal to the contraction coeffiecient (C c ), so C d is used


in the expression. The terms Q, A 0 and ρ are, the flow out of the control volume, the smallest cross
sectional area and the density of the fluid, respectively. The term ~ vˆ is a unit vector pointing in the
direction of the flow. The force of interest is the force acting on the moving part. If the angle between
the flow and the line of movement for the moving part is denoted θ, the flow force can be determined
by the following eguation.

Q2
FF S = ρ · · cos(θ)
Cd · A0

The orifice equation A.1 on page B7 is utilized, so that the equation can be written in terms of pressure
instead of flow.

³ q ´2
C d A 0 ρ2 (P i n − P out )
FF S = ρ · · cos(θ)
Cd · A0

C d2 · A 20 · ρ2 · (P i n − P out )
FF S = ρ · · cos(θ)
Cd · A0

F F S = 2C d A 0 (P i n − P out )cos(θ) (A.6)

FF S Flow force. Will always point in the direction closing the valve. [N ]

Cd Discharge coefficient [·]


£ 2¤
A0 The smallest cross sectional area of the opening m

Pi n Upstream pressure. Pressure before the opening. [P a]

P out Downstream pressure. Pressure after the opening. [P a]

θ Angle of the flow [r ad ]

A.5 Continuity equation

The so called continuity equation is an equation used to describe the change of pressure in a cavity
of a hydraulic system.
A.5. CONTINUITY EQUATION B11

P
Qin Qout
A

Figure A.6: An arbitrary control volume (∀) with flows and pressure.

In the Figure A.6 it is seen that if the two flows aren’t exactly the same there must be a pressure change
in the control volume, unless the volume is changed. This is described in the continuity equation A.7.

d∀ ∀ dp
Q i n −Q out = + (A.7)
dt β dt

h i
m3
Qi n Flow into the cavity s
h i
m3
Q out Flow out of the cavity s
£ 3¤
∀ Volume of the cavity m

β Bulk modulus of the fluid (Stiffnes of the oil) [P a]

p Pressure in the cavity [P a]

The stiffnes of the oil β isn’t a constant. It is highly temperature and pressure dependent. In [Andersen
& Hansen, 2007a], is a figure (see figure A.7) describing the change in stiffness as the pressure changes,
at different temperatures. At an operating point of 200 bar, it is seen that a temperature change from
20◦ to 40◦ results in a change of about 1000 bar in stiffnes (18500bar − 17500bar ). Which is a change
of about 5.5%. In some applications this is a small change, and therefore the bulk modulus can be as-
sumed temperature independent. If the temperature of the oil graphed in figure A.7 is 20◦ the change
in pressure from 0 bar to 200 bar result in a change of stiffnes of 1750 bar (18500bar − 16750bar ),
which is a change of about 10%. In some applications this is also a small change, and therefore the
bulk modulus can be assumed to be pressure independant, however, this is not allways the case, es-
pecially when there is air dissolved into the fluid, which is almost always the case in real systems.
Bβ B a temperature dependant coefficient, [bar-2]

Where the temperature dependant coefficients can be determined from Equation (A.4)
and Equation (A.5). It should be noted that Equation (A.17) implies that the fluid
stiffness may be calculated for any temperature and pressure combination regardless of
B12 APPENDIX
the specific type of mineral oil. The variation of the fluid A. temperature
stiffness with GOVERINGandEQUATIONS
pressure is displayed graphically in Figure A5.

β F [bar ]

32000
30000 0 °C
28000
20 °C
26000
24000 40 °C
22000 60 °C
80 °C
20000 100 °C
18000
16000
14000
12000
0 100 200 300 400 500 600 700 800
p [bar]
Fig. A5 The variation of the fluid stiffness with temperature and pressure
Figure A.7: The variation of the fluid stiffness with temperature and pressure. Andersen & Hansen
[2007a] In real systems air will be present in the fluid. The volume percentage at atmospheric
pressure will go as high as 20 %. As air is much more compressible than the pure fluid
it has, potentially, a strong influence on the effective stiffness of the air containing fluid.

Hydraulic Fluids Page 8 of 11

When air is dissolved into the fluid the stiffness of the oil is dramatically lowered, especially at low
pressures. The Figure A.8 illustrates how the effective stiffness of an oil changes with pressure for
different volumetric ratios of air entrapped in the fluid. The definition of the ratio is given in Equa-
tion A.8.

V A0
² A0 = (A.8)
VF 0 + V A0

² A0 The reference volumetric ratio of free air in the fluid at atmospheric pressure [·]
£ 3¤
V A0 Volume of air at atmospheric pressure m
£ 3¤
VF 0 Volume of the fluid at atmospheric pressure and at the reference temperature m
A.5. CONTINUITY EQUATION B13

β eff [ bar ]
20000

18000
ε A0 = 0
16000

14000

12000
ε A0 = 0.1
10000
ε A0 = 0.05
8000
ε A0 = 0.02
ε A0 = 0.01
6000 ε A0 = 0.005

4000

2000

0
0 20 40 60 80 100
p [bar]

Figure A.8: Plots ofFig.


howA6the Variation of effective
effective stiffness
stiffness ofofanfluid-air mixture with
oil changes respect
with to pressure
pressure anddifferent volumetric
for
volume ratio of free air at atmospheric pressure. The temperature of the fluid is 40 °C and the
ratios of air entrapped in the fluid. [Andersen
compression of the&free
Hansen, 2007a]
air is assumed adiabatic

-----oo0oo-----

Experiments have shown that in most hydraulic systems the bulk modulus is allmost never more than
10.000 bar Andersen & Hansen [2007a].

Hydraulic Fluids Page 11 of 11


Appendix B

Test Reports

B.1 Compensator spool in PVP module spring

Purpose of the experiment

The purpose of this experiment is to determine the spring constant of the spring, attached to the
pressure sompensator spool in the PVP module, and the preload of this spring. The spring constant
was determined by applying a force to the spring, and measure the displacement.

Materials used for the experiment

• Avery (machine for determination of spring constant)

Theory

The theory behind springs is approximated by Hooke’s law


of elasticity, and states that the spring force is directly pro- Force kg Travel mm
portional with the displacement. The constant that defines
0 0
this proportionality is called the force constant or spring con-
stant. This coherence is described by the following equation: 4.6 3

F = kp · x 9 6
In order to determine the spring constant it is necessary to 13.4 9
record a wide range of force-deflection measurements. By in-
22.4 15
serting these values in a (displacement x, force F)-graph and
making a linear approximation, it will be possible to deter- 24.15 16

mine the spring constant, from the slope of the graph. 30.2 20
In theory, one measurement should be sufficient to calcu- 34.2 30
late the constant of the spring. But to increase accuracy, nu-
merous measurements is performed and then linearized to Figure B.1: Measurements from the
an equation containing information about the spring con- experiment
stant. Numerous approaches can be used to do the experi-
ment, and to record the displacement of the spring under a
load force. One approach is, simply, to measure the displacement with a ruler and use a Newton-
meter to determine the load force. By varying the load-force it is possible to get a wide range of
measurements. This practice is used by a test machine, which can give a variable load force and si-
multaneously measure the deflection with a precise ruler that worked as a vernier caliper.
The arrangement of the experiment can be seen in figure B.2a and a table with the measuments from

B15
B16 APPENDIX B. TEST REPORTS

the experiment can be seen in Figure B.1. Matlab was used to make a graph of the data from the ex-
periment and the Basic fitting tool to derive an expression. The graph of both the derived expression
and the data set can be seen in B.2b.

Spring from PVP module, deflection under different forces


350
Measurement points
y1= 14688⋅ x − 0.41704
300

250
Force F [N]

200

150

100

50

0
0 0.005 0.01 0.015 0.02 0.025
Displacement x [m]

(a) Arrangement of the experiment. (b) Graph from the experiment. Blue line: linearization. Red dots: ac-
tual data from the experiment.

Figure B.2

Determination of preload

The spring, attached to the pressure compensator spool is preloaded, and the force of this will now
be determined.
To determine the preload, the deformation of the spring when the spool is pushed as far away from
the spring as it can be, has to be determined. The equation to determine the preload is expressed in
Equation B.1
F pr el oad = (L 0 − L 1 ) · k p (B.1)
Where k p is the spring constant, L 0 is the lenght of the spring before the deformation and L 1 is the
lenght when being mounted in the PVP module. The length of the spring, when being applied no
force, is measured to be:x unl oad ed = 45.9mm and when it is mounted in the PVP module the lenght is
measured to be x l oad ed = 28.9mm. the deformation can then be calculated in Equation B.3

F pr el oad −p = (x unl oad ed − x l oad ed ) · k p (B.2)


N
= (45.9mm − 25.9mm) · 10−3 · 15000 = 300N (B.3)
m

Sources of error

The most significant source of error is from the reading of the displacement. Other sources of error is
from the scale built in in the machine.
B.2. COMPENSATOR SPOOL IN PVB MODULE SPRING B17

Conclusion

The almost linear behavior of the data set indicates precise and useable measurements and the spring
N
constant was determined to be ≈ 15000 m .

B.2 Compensator spool in PVB Module spring

Purpose of the experiment

The purpose of this experiment is to determine the spring constant of the spring attached to the
compensator spool in the PVB module. Determination of the spring constant will make it possible to
calculate the spring force as a function of spool movement. Because two springs is attached to the
compensator spool the graph of the experiment will show an unlinear behavior.

Materials used for the experiment

• Avery (machine for determination of spring constant)

Theory

The theory behind springs is approximated by


Hooke’s law of elasticity, and states that the spring Force kg Travel m Force kg Travel m
force is directly proportional with the displace-
0 0.081 9 0.0719
ment. The constant that defines this proportional-
ity is called the force constant or spring constant. 0.4 0.0805 10 0.0709

This coherence is described by the following equa- 0.8 0.08 11 0.0698


tion: 1.2 0.0795 12 0.0688
F = kc · x
1.6 0.0793 13 0.068
In order to determine the spring constant it is nec-
essary to record a wide range of force-deflection 2.0 0.0788 14 0.067

measurements. By inserting these values in a (dis- 2.6 0.0784 15 0.0664


placement x, force F)-graph and making a linear 3.2 0.0776 16 0.066
approximation, it will be possible to determine the
3.8 0.0769 18 0.0654
spring constant.
4.4 0.0766 20 0.0649
In theory, one measurement should be sufficient
to calculate the constant of the spring. But to in- 5 0.0759 22 0.0643
crease accuracy, numerous measurements is per- 6 0.0749 24 0.0637
formed and then linearized to an equation contain-
7 0.0738 26 0.0633
ing information about the spring constant. Numer-
8 0.0728 27 0.0630
ous approaches can be used to do the experiment,
and to record the displacement of the spring under
Figure B.3: Measurements from the experi-
a load force. One approach is, simply, to measure
ment
the displacement with a ruler and use a Newton-
meter to determine the load force. By varying the
load-force it is possible to get a wide range of measurements. This practice is used by a test machine,
which can give a variable load force and simultaneously measure the deflection with a precise ruler
that worked as a vernier caliper.
The arrangement of the experiment can be seen in figure B.4a and a table with the measuments from
the experiment can be seen in Figure B.3.
B18 APPENDIX B. TEST REPORTS

Matlab was used to make a graph of the data from the experiment and the Basic fitting tool to derive
an expression. The graph of both the derived expression and the data set can be seen in B.4b.

Pressure compensator springs deflection under different forces


400

350
Measurement points
y1= 9942,9⋅ x - 1.8024
300
y2= 35113⋅ x - 368.75

250
Force F [N]

200

150

100

50

0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Displacement x [m]

(a) Arrangement of experiment. (b) Graph from the experiment. Blue lines: linearization. Red dots: actual
data from the experiment.

Figure B.4

Determination of preload

The spring in the compensator spool is preloaded and the force of the preload will now be deter-
mined.
To determine the preload, the deformation of the spring when the spool is pushed as far away from
the spring as it can be, has to be determined. The equation to determine the preload is expressed in
Equation B.4
F pr el oad = (L 4 − L 5 ) · k c (B.4)

Where k c is the spring constant, L 4 is the lenght of the spring before the deformation and L 5 is the
lenght when being mounted in the PVB module. The length of the spring when it is untense is mea-
sured to be:x unl oad ed = 36.74mm and when it is mounted in the PVB module the lenght is measured
to be x l oad ed = 26.17mm. the deformation can then be calculated inEquation B.5

F pr el oad −c = (x unl oad ed − x l oad ed ) · k m


N
= (36.74mm − 26.17mm) · 10−3 · 10000 = 105.12N (B.5)
m

Sources of error

The most significant source of error is from the analogue read out of the displacement. Other sources
of error is from the scale built in in the machine.
B.3. LVDT SPRING B19

Conclusion

The almost linear behavior of the data set indicates precise and useable measurements and the spring
N
constant was determined to be ≈ 10000 m for the first spring with a force of gravity set to 9.82N . And
N
≈ 35000 m when both springs is compressed.

B.3 LVDT Spring

Purpose of the experiment

The purpose of this experiment is to determine the spring constant of the small spring attached to the
LVDT located in the PVE module. Because of the low tolerances of the machine available to determine
the constant, it is decided to make a more precise machine. A precise scale is used to measure the
force exerted to the spring and a bench drill with a mm gauge to measure the travel of the spring.

Materials used for the experiment

• Scale.

• Bench drill with mm gauge.

• Drill.

Theory

First, the spring is placed on the weight and then the tip of
the bench drill is lowered so that it touches the spring. The Force kg Travel m Force kg Travel m
weight is then reset. For the measurements to be precise, it
0 0 0.151 0.008
is necessary to read out the force exerted- and the travel of
the spring in small steps. It is decided to measure in steps 0.015 0.001 0.151 0.009

of 5mm. The arrangement of the experiment can be seen in 0.035 0.002 0.171 0.010
figure B.6a and a table with the measurements from the ex- 0.054 0.003 0.194 0.011
periment can be seen in Figure B.5. Matlab was used to make
0.073 0.004 0.200 0.012
a graph of the data from the experiment and the Basic fitting
tool to derive an expression. The graph of both the derived 0.098 0.005 0.227 0.013

expression and the data set can be seen in B.6b. 0.115 0.006 0.251 0.014

0.136 0.007

Figure B.5: Measurements from the


experiment
B20 APPENDIX B. TEST REPORTS

Spring from PVE module, deflection under different forces


4

3.5
Measurement points
y1= 190⋅ x - 0.023
3

2.5

Force F [N]
2

1.5

0.5

-0.5
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02
Displacement x [m]

(a) Arrangement of the experiment. (b) Graph from the experiment. Blue line: linearization. Red dots: ac-
tual data from the experiment.

Figure B.6

Sources of error

The most significant source of error is from the analogue read


out of the displacement. Other sources of error is from the
scale built in in the machine.

Conclusion

The almost linear behavior of the data set indicates precise


N
and useable measurements and the spring constant was determined to be 190 m .

B.4 Main Spool Spring

Purpose of the experiment

The purpose of this experiment is to determine the spring constant of the spring located on the main
spool. Determination of the spring constant will make it possible to calculate the spring force as a
function of spool movement. After having determined the springforce, the preload of the spring will
be determined.

Materials used for the experiment

• Machine for determination of spring constant


B.4. MAIN SPOOL SPRING B21

Data processing and theory

The arrangement of the experiment can be seen in


figure B.8a and a table with the measuments from Force kg Travel m Force kg Travel m
the experiment can be seen in Figure B.7.
1 39.4 5 38.5
Matlab was used to make a graph of the data from
2 39.25 6 38.25
the experiment and the Basic fitting tool to derive
3 38.95 8 37.8
an expression. The graph of both the derived ex-
pression and the data set can be seen in B.8b. 4 38.75 10 36.9

Figure B.7: Measurements from the experi-


ment

Spring from main spool, deflection under different forces


90

80
Measurement points
y1= 3596⋅ x + 4.791
70

60
Force F [N]

50

40

30

20

10

0
0 0.005 0.01 0.015 0.02 0.025
Displacement x [m]

(a) Arrangement of experiment. (b) Graph from the experiment. Blue lines: linearization. Red dots: actual
data from the experiment.

Figure B.8

Determination of preload

The spring in the main spool is preloaded and the


force of the preload will now be determined.
To determine the preload, the deformation of the
spring when it is tightened together in the main
spool. The equation to determine the preload is expressed in Equation B.4

F pr el oad −m = (L 2 − L 3 ) · k m (B.6)

Where k m is the spring constant, L 2 is the lenght of the spring before the deformation and L 3 is the
lenght when being mounted on the main spool. The length of the spring when it is untense is mea-
sured to be:x unl oad ed = 41.82mm and when it is mounted in the PVB module the lenght is measured
to be x l oad ed = 23.73mm. the deformation can then be calculated inEquation B.5
B22 APPENDIX B. TEST REPORTS

F pr el oad −m = (x unl oad ed − x l oad ed ) · k m


N
= (41.82mm − 23.73mm) · 10−3 · 3596 = 65.05N (B.7)
m

Sources of error

The most significant source of error is from the analogue read out of the displacement. Other sources
of error is from the scale built in in the machine.

Conclusion

The almost linear behavior of the data set indicates precise and useable measurements and the spring
N
constant was determined to be ≈ 3596 m with a force of gravity set to 9.82N . The preload of the spring
was determined to be 65.05N

B.5 Cavities of Main Spool

Purpose of the experiment

The purpose of this experiment is to determine the size of the cavities at both side of the main spool.
This is the initial cavities, when the main spool is at the position x m = 0mm. The volume at port B is
called V5 and the volume at port A is called V4 .

Materials used for the experiment

• Water at room temperature: Approximately 20◦C

• A precision weight measured in grams

Data processing and theory

Because this is a very basic experiment, there wont be a distinct theory section. The weight is mea-
suring in grams, so the only thing needed to know is the conversion rate from grams g to mm 3 for
water, at room temperature. The density of water at 20◦C is found in table to be ρ = 998.2071kg /m 3 ≈
1000kg /m 3 . The lower the temperature is the higher the density will be, and the opposite when the
temperature rises, but the changes are so small that an approximate value for the density is chosen.
The weight of the electric module m el ec and the mechanical handle part m mech with and without
water in the cavities are measured. The data can be found in the table.

Team sheet

m el ec The electric component without water 0.8364kg

m el ec+w at er The electric component with water 0.8493kg

m mech The mechanical component without water 0.2632kg

m mech+w at er The mechanical component with water 0.2887kg


The weight of the water in the cavities is found by substraction the two matching values a the com-
ponents with and without water
B.6. VOLUMES OF THE HOSES AND PIPES B23

m w at er @4 = m el ec+w at er − m el ec = 0.8493kg − 0.8364kg = 12.9g


m w at er @5 = m mech+w at er − m mech = 0.2887kg − 0.2632kg = 25.5g

Further more a hollow cylinder is connected to port B side cavity of the main spool. This cylinder
has a lenght of l = 0.1m and a diameter of D = 0.004m. Adding this to the port B side, and using the
weight of the water will give the values for the cavities:

V4 = 0.0129kg ∗ 0.001m 3 /kg = 12.90 · 10−6 m 3 (B.8)


2
(0.004m)
V5 = 0.0255kg ∗ 0.001m 3 /kg + · 0.1m = 26.76 · 10−6 m 3 (B.9)
4

B.6 Volumes of the Hoses and Pipes

Purpose of the experiment

The purpose of the experiment was to determine the volumes pipes connecting the pump with the
test arrangement and the hoses interconnecting the orifices, flowmeters etc.

Materials used for the experiment

• Test rig for hydraulic components.

• Vernier calliper

Data processing and theory

The data of the hoses and pipes is found according to manufacturer data. And as such there is no
theory and data processing.

Determination of the volumes of the hydraulic hoses

To obtain higher accuracy of the experiment it is necesary to calcutale the volumes of the hoses that
connect the hydraulic components. The hose that is ued is Din 20023 4SP/EN 856 4SP with an outer
diameter D out er of 25mm. This type of hydraulic hose has an inner diameter D i nner of 12.7mm.
Hose 1,4,5,6,7 and 8 is a flexible hose, type; Din 20023 4SP/EN 856 4SP: and hose 1 and 2 is made of
precision steel piping, type; EN10305-4. As outlined in Figure B.9, it is necesary to find 8 different
volumes of the test arrangement.
The volume of hose 1 is calculated as:

π·d2 π · 0.01272
Vhose1 = · l eng t h = · 1.5 = 0.00019m 3 (B.10)
4 4

Where d is the inner diameter of hose 1.


And the rest of the volumes is calculated by the same principle.
B24 APPENDIX B. TEST REPORTS

Flowmeter Orifice QL1


Magnet valve
Flowmeter Hose 7
Hose 8
Pressure gauge
PVG32
Hose 6 Pressure gauge

Pressure gauge
Hose 5
Hose 1 Orifice QL3
Hose 4

LOAD
AL1
QL1
Hose 3 Hose 7

Hose 8
Pump
QL3 QL2
Hose 2

AL3 Hose 6
Hose 5
Qb Qf

QL QV

Figure B.9: An illustration of the test arrangement with numbers for the different hoses.

Hose nr. Type D i nner Length Volume

1 Din 20023 4SP/EN 856 4SP 12.7mm 1.5m 0.00019m 3

2 EN10305-4 20mm 8m 0.002513m 3

3 EN10305-4 20mm 8m 0.002513m 3

4 Din 20023 4SP/EN 856 4SP 12.7mm 1.5m 0.00019m 3

5 Din 20023 4SP/EN 856 4SP 12.7mm 1m 0.000127m 3

6 Din 20023 4SP/EN 856 4SP 12.7mm 1m 0.000127m 3

7 Din 20023 4SP/EN 856 4SP 12.7mm 0.5m 0.000063m 3

8 Din 20023 4SP/EN 856 4SP 12.7mm 1m 0.000127m 3


B.7. APPLICATION IN HYDRAULIC SYSTEM B25

B.7 Application in Hydraulic System

Purpose of the experiment

The purpose of the experiment is to compare the dynamic model of the PVG 32 valve, modelled in
Simulink™to an actual PVG 32. The test will reveal information about the steady state flows and
pressures, and the dynamics of the valve.

The pump used to deliver the oil flow is not big enough to fully load the valve. And as such the flow
will saturate at its maximum operating point wich is approximately 17 mil n . As a result, all the data
from the experiment will be between 0 mil n and 17 mil n .

Materials used for the experiment

• Test rig for hydraulic components.

• Measuring tools e.g. a computer with Labview.

• Flow meter.

• Pressure gauge.

As it can be seen in Figure B.10 the test rig consist of 2 variable orifices L 1 and L 3 in parallel and a
solenoid valve, which can open or close the flow to one of the orifices. There are two flow meters
in the test rig Q L2 and Q f . Q L2 is located directly after the PVG 32 so the total flow into the two
orifices can be measured, Q f is located after the solenoid valve so the flow through the orifice L 3 is
measurable.
To determine pressures in the load, 3 pressure gauges are mounted. P f is located at the point where
the flow branches into 2, one leading to the solenoid valve which leads to the orifice L 3 , and the
other branch leads to orifice L 1 . P m is located between the solenoid valve and the orifice L 3 . The last
pressure gauge P b , is located near the PVG 32 valve on the return hose from the orifices.

PVG32

Pf
Pump
QL2
Qf

Pm
Pb
Reservoir

L1 L3

Figure B.10: An illustration of the test arrangement with flow meters, valves and pressure gauges.
B26 APPENDIX B. TEST REPORTS

Data processing and theory

The 3 pressure gauges, makes it possible to determine the pressure drops across the individual ori-
fices, the measured pressures, was logged by using Labview™, which makes it possible to compare
the dynamic pressures in the load, with the dynamics in the simulink model. The pressure transmit-
ters is of the type M B S32 made by Danfoss™, which have an accuracy of ±0.8%[Danfoss, n.d.b]
The flows was determined in steady state, so the measurements was done by manually read out the
flow, from a digital flow meter, of the type, SCQ 60 made by Parker™, with an accuracy of ±2%[Parker,
2000]. Since it is unknown when the pressure transmitter and the flow meter was last calibrated, it is
reasonable to believe that the accuracy of the pressure transmitter is 1B ar and the flowmeters accu-
racy can be set to 1 mil n [Pedersen, 2009]

The arrangement is set to operate at steady state with a given main spool dispacement and a given
orifice area at both orifices. In the beginning, no signal will be given to the solenoid valve, which
means the flow will only run through the orifice L 1 . Then a step input is given in terms of applying a
signal to the solenoid valve, so the oil also will be able to flow through the orifice L 3 . The steady state
pressures and flows throghout the setup is carefully monitored to establish how the system behaves.

Test no. L1 L3 (NC) QL2 [L/min] Qf [L/min] U [V] Pf [Bar] Pm [Bar] Pb [Bar]

1 Small Small Closed SS 13.67 0 4.24 66.32 8.67 7.42

2 Small Small Opening 14.19 6.47 4.24 43.01 38.69 7.96

3 Small Small Closing 14.04 0 4.24 65.94 8.71 7.51

4 Small Small Closed SS 7.23 0 4.62 43.04 9.24 8.42

5 Small Small Opening 7.28 3.49 4.62 28.60 25.79 8.26

6 Small Small Closing 7.23 0 4.62 42.45 9.19 8.38

7 Large Large Closed SS 13.70 0 4.29 37.95 8.70 7.62

8 Large Large Opening 13.82 5.43 4.29 26.27 23.04 7.72

9 Large Large Closing 13.75 0 4.29 37.66 8.68 7.58

10 Large Large Closed SS 8.20 0 4.60 27.99 8.89 8.06

11 Large Large Opening 8.02 3.15 4.60 20.00 17.38 8.16

12 Large Large Closing 8.0 0 4.60 27.32 8.88 8.07

The second part of the test determines the behavior of the main spool as the displacement is changed
i.e. the supply voltage is changed. 6V is the neutral position and then the voltage is gradually lowered.
However, because of the low maximum flow output of the pump, the flow saturated at about 16.30 mil n
or in voltage terms at 3.94V supply voltage.

Test No. Pf (avg) [Bar] Pm (avg) [Bar] Pb (avg) [Bar] QL2 [l/min] U [V]

1 9.119 9.156 8.683 0 6

2 8.259 8.321 7.810 0 5.5

3 15.143 9.894 9.388 1.98 4.93

4 23.580 8.784 8.079 6.74 4.65

5 29.316 8.561 7.707 9.85 4.45

6 38.680 8.350 7.227 15.70 4.21

7 39.773 5.778 4.796 16.30 3.94


B.7. APPLICATION IN HYDRAULIC SYSTEM B27

Sources of error

There is a measuring error on the flow meter of approximately ±2%, and ±0.8% on the pressure gauge.
These are the main sources of error. A measuring error of this magnitude can cause errors of up to
50% as the oil flow in one of the testsd is only 1.98 mil n

Conclusion

The experiment provided useable results to compare with the results from the model. However, be-
cause of the large measuring errors it is necessary to be cautious when relying to much on a single
test. Especially at low flow rates.
Appendix C

Transfer functions

C.1 Notes when deriving transfer functions, used in tuning the model

C.1.1 Pump spool dynamics

FFR-P

FS
FPp
Mp
FP3

FFL3
xp A3

Figure C.1: Forces acting on pump side spool

m p · s 2 · x p = A p · (P 6 − P 3 ) − ν · s · x p − k f j p + k F F · x p
¡ ¢
(C.1)

Ap
xp = · (P 6 − P 3 ) (C.2)
mp ·s +ν+k
2
f jp + kF F

Ap
G (s) = (C.3)
m p · s 2 + ν + k f j p + kF F

The subsection C.1.3 shows the derivation of G 2 (s). The relationship between the pressure in chamber
P 6 and the pump pressure P p .

xp = (G (s) · P 6 ) − (G (s) · P 3 ) (C.4)


= G (s) ·G 2 (s) · P p → P 3 = const ant (C.5)

B29
B30 APPENDIX C. TRANSFER FUNCTIONS

s
k f j p + KF F
ωn = (C.6)
mp

ν
ζ= (C.7)
m p · 2 · ωn

Ap
k= (C.8)
m p · ω2o

C.1.2 Flowforce

Q t2
FF F = ρ· · cos θ (C.9)
cd · A 3
cd 2 · A 23 · ∆P
= ρ· (C.10)
cd · A 3
= ρ · cd cos θ · Ak 3 · P p · x p (C.11)
= kF F · x p (C.12)





 0 if 0mm ≤ x p ≤ 2mm

2
Ak 3 = 180mm
 3mm · x p if 2mm ≤ x p ≤ 5mm


 180mm 2

if 5mm ≤ x p ≤ 10mm

p
Q t = cd · A 3 ∆P (C.13)
∆P = P p ← P t = 0 (C.14)

C.1.3 Pressure build up in chamber of pump side module

Q4
V6 P6

Figure C.2: Illustration of the pump side module with the pressure adjustment spool
C.1. NOTES WHEN DERIVING TRANSFER FUNCTIONS, USED IN TUNING THE MODEL B31

β β ¡ ¢
s · P6 = ·Q = · k · P p − P6 (C.15)
V6 V6
β β
µ ¶
P6 · s + k · = · k · Pp (C.16)
V6 V0
β
P6 V6 ·k
= β
(C.17)
Pp s + V6 · k
1
= V0
(C.18)
β·k ·s +1
P6 = G 2 (s) · P p (C.19)
(C.20)

C.1.4 Load Pressure Build Up

Vm
AL1 Pm
LOAD QL1

QL3 QL2

AL3
Vb Vf
Pb Qb Qf
Pf

QL QV

Pbp PLS

Figure C.3: The working load of the system


B32 APPENDIX C. TRANSFER FUNCTIONS

d P Ls β
= (Q v −Q L3 ) (C.21)
dt V ¡ ¢
QL3 = K L3 P Ls − P bp (C.22)
β ¡ ¢
s · P Ls = · (Q v − k L3 ) P Ls − P bp (C.23)
V
β ¡ ¢
P Ls = Q v − k L3 · P Ls + k L3 · P bp (C.24)
V ·s
β β ¡ ¢
P Ls + · k L3 · P Ls = Q v + k L3 · P bp (C.25)
µV · s V ·s
β β ¡

¢
P Ls · 1 + · k Ls = Q v + k k3 · P bp (C.26)
V ·s V ·s
β k L3 · β
V ·s V ·s
∆P Ls = β
· ∆Q v + β
· ∆P bp (C.27)
1 + V · s · k L3 1 + V · s · k L3
1
k L3 1
∆P Ls = V
∆Q v + V
∆P bp (C.28)
β · k L3 ·s +1 β · k L3 ·s +1
(C.29)

V
τ= (C.30)
β · k L3

β · k L3
ωbw = (C.31)
V

1
k ss = (C.32)
k L3
Appendix D

Foldout page with overview of the PVG 32

B33

Das könnte Ihnen auch gefallen