Documente Academic
Documente Profesional
Documente Cultură
Fluid Power
Systems
Modelling and Analysis
March 2003
2nd Edition
Preface
Preface
The engineering effort in hydraulics is a problem of both educational and technical
nature, but of a restrictive influence on the extension and successful use of hydraulics.
The lack of well - established, systematic engineering methods to form the basic set-off
in analysis and design of hydraulic control systems has been obvious.
The employers within the branch - from which the majority started up without any
special hydraulic education - are so busy in designing that no time is left to put theory
behind the mainly empirical but often voluminous know-how gathered by many
hydraulic engineers.
No engineering discipline at university level, aimed at the professional engineering of
hydraulic control systems, has been established on a par with classical disciplines.
Furthermore, the theoretical tools needed for analysis of hydraulic problems are
complicated through the following reasons:
Many problems in hydraulic systems are of extremely dynamic nature and can only be
described by differential equations.
Hydraulic systems and components are inherently non-linear, making the describing
differential equations non-linear and thus limits the applicability of linear control
theory.
Many hydraulic systems are in fact distributed-parameter systems. It takes some skill to
make sound engineering judgements on when and how lumped-parameter approaches
can be made.
By the estimation of apparently simple hydraulic system-parameters a large number of
details must be taken into consideration, such as laminar or turbulent flow, viscosity-
temperature relations, friction factors, discharge coefficients, passage geometry,
cavitation, air content, and so forth.
Finally the development of valuable engineering theories very often must be based on
experimental verification of hypothesis.
It seems from above statements, that strong efforts in the elevation of the engineering
methods ought to be emphasised as one means to ensure continuous ability to compete.
Therefore it is rational efforts to seek the most intelligent ways of using hydraulic
control. This demands a continuous effort in research and development, and in
education of hydraulic engineers so their professional standing is adequate for dealing
with actual hydraulic control problems. Accordingly the necessary, engineering tools
have to be established and brought up to an operational level.
In Figure 1 is shown hydraulic control engineering as a special subject and the most
important contributing disciplines. “The-state-of the-art” is that 1, 2, 5, 6, and to some
extent 7, have been close allies of hydraulics.
1
11 CONTROL 2
PRODUCTION ENGINEERING GENERAL
TECHNOLOGY SYSTEM
DYNAMICS
3
DIGITAL
10 SIMULATION
TECHNICAL AND GENERAL
SALES & COMPUTING
CONSULTING TECHNIQUES
HYDRAULIC
9 CONTROL 4
DIGITAL LAB. METHODS
ENGINEERING & INSTRUMENTATION
SIGNAL CIRCUIT
DESIGN (ELECTRONICS)
5
8 APPLICATIONS
CONTINUUM (EX. BACKHOE -
MECHANICS 7
FLUID 6 LOADER)
(FEM)
MECHANICS MASCHINE &
& DYNAMICS MASCHINEPARTS
(CFD) DESIGN
° The physical system is not available. Often, simulations are used to determine
whether a projected system should ever be built.
° The cost of experimentation is too high. Often, simulations are used where real
experiments are too expensive. The necessary measurement tools may not be
available or are to expensive. The system might be used all the time and taking it
“off-line” would involve unacceptable cost.
° The time constants of the system are not compatible with those of the
experimenter. Often simulations are performed because the real experiment
executes so quickly that it can hardly be observed (for example the reaction time
for a spool valve towards neutral is important for safety reasons, but reaction times
around 200 ms can not be noticed by an operator).
° Control variables (disturbances), state variables, and/or system parameters may be
inaccessible. Often, simulations are performed because they allow us to access all
inputs ad all state variables, whereas in the real system, some inputs may not be
accessible to manipulation and some state variables may nor be accessible to
measurement. Simulation allows us to manipulate the model outside the feasible
range of the physical system.
° Suppression of disturbances. Often, simulations are performed because they allow
us to suppress disturbances that are unavoidable in the real system. This allows us
to isolate particular effects, and may lead to better insight into the system
behaviour.
° The number of parameters to adjust in the experiment are too large, giving a huge
number of possibilities and at best, result in a sub-optimal solution.
The most important strengths of simulation, but also its most serious drawbacks, are the
generality and ease of its applicability. It does not require much of a genius to be able to
utilize a simulation program. However, in order to use simulation intelligently, (having
the handle to make the real world behave the way we want it to), we must understand
what we are doing. Danger lies in forgetting, that the simulation model is not the real
world, but that it represent the world under a very limited set of experimental
conditions. Simulations are rarely enlightening. In fact, running simulations is very
similar to performing experiments in the lab. We usually need many experiments,
before we can draw legitimate conclusions. Correspondingly, we need many
simulations before we understand how our model behaves. While analytical techniques
(where they are applicable) often provide an understanding as how a model behaves
under arbitrary experimental conditions, one simulation run tells us only how the model
behaves under the one set of experimental conditions applied during the simulation run.
Therefore, while analytical techniques are generally more restricted (they have a much
smaller domain of applicability), they are more powerful where they apply. So,
whenever we have a valid alternative to simulation, we should, by all means, make use
of it. (for example, single-stage pressure control valves is applicable for analytical
analysis while two-stage valves are not).
However, as the level of analysis is elevated, as it for instance may happen in
connection to development of new components or new systems, a barrier will appear at
a certain level. The complexity which underlines the development of mechatronic
systems should be recognised.
Very few people are capable in making models of real mechatronic systems, and even
though simulation programs are appearing on the market that claims they can simulate
pretty much anything, there are still the number of parameters to adjust and the
programs can not design the system for you.
Contents
Chapter 1: Mathematical modelling
1.1 Introduction
1.2 Simple Models contra Complex Models
1.3 Systems Engineering and Component Design
1.4 Establishment of Mathematical Models
3.1 Introduction
3.2 Lumped Fluid Theory
3.3 Cylinders
3.4 Motors
3.5 Linear Characteristics of a Cylinder
4.1 Introduction
4.2 Flow Control Valves
4.3 Power Elements
5.1 Introduction
5.2 Hydraulic Load Holding Circuit
5.3 Load Sensing Directional Valve
5.4 Two Stage Relief Valve
1
Mathematical Modelling
1.1 Introduction................................………........................... 1
1.1 Introduction
The concept of models is probably one of the most fundamental concepts in science and
in the scientific approach to the solution of practical problems. In a broad sense, a
theory is an abstract model of, how things behave. To be of value, a model must come
close to reality, and to the engineer nothing is more practical than a good theory,
because it forms the basis on which realistic decisions are taken.
Claming a model to be both meaningful and operational is much the same as to require,
at the same time, accuracy and simplicity. As these properties usually are conflicting,
one has to compromise, to chose an acceptable degree of accuracy, a certain “level of
analysis”. The question generally turns out to be : Is it possible within acceptable time
and cost to set up a model of the specific system which is sufficiently simple to be
handled and still retains its meaning. If not, the engineering decision will, more or less,
be based on guess-work.
Generally, savings in work or cost are the essential reason why experiments are carried
out with models in stead of real systems. However, some situations occur in which no
alternative to model-experiments exist.
This is particularly the case when dealing with systems in which eventual instability
may initiate irreversably accelerating processes with hazardous and incalculable
consequences.
Finally, the model-concept and the associated simulation-techniques are very valuable
pedagogical tools which are brought into use whenever technical knowledge, on any
level, shall be transferred from one person to another in an efficient manner.
Like all design activities the modelling process has a recurrent nature, and obviously
many matters influence the level of analysis needed to make up a useful mathematical
model of a component. However, two extremes exist, determined by two characteristic
users of component-models.
The component-designer needs a detailed model, describing all internal effects with
significant influence on the total performance. This type of model is useful during
preliminary design of new components or when attempting to modify the performance
of a given design by parameter variation. But it is necessary to be careful with
complicated models, because they give the impression of a high level of analysis, while,
in reality, they might be a compilation of conventionally used mathematical equations,
poorly describing the virtual physical phenomena.
The indication is, that it is often difficult to verify the type of complex, mathematical
models established on a conventional empirical/theoretical analysis. The reason of this
discrepancy is undoubtedly, that it is difficult to establish generally applicable adequate
analytical expressions of flow-pressure-area relations, unless geometry is very simple
and flow conditions are very well-defined. Flow condition, separation- and cavitation-
effects must be taken into account as well as distributed parameters and several non-
linear effects are characteristics of hydraulic components. Care must be taken, particular
with flow-equations.
Basically two approaches exist: The analytical approach and the experimental
approach. In principle they are very different and have individual purposes. In fact they
are in many ways related and are complementary to one another: It is hardly possible to
plan rational experiments without having formed any analytical ideas of a model
(Hypothesis). Vice Versa, a complex analytical model without any experimental
verification will often be met with some doubtfulness.
Finally conventionally used flow-pressure-area equations, examplewise, usually contain
empirical ingredients as the conceptual discharge coefficient, etc. Hence it is not
relevant to talk about a pure analytical approach when modelling hydraulic systems.
What often is designated ‘the analytical approach’ is in fact a pseudo-analytical
approach with numerous empirical ingredients.
Agreement between simulated and measured over-all performance of model and
component, respectively, yields confidence in the validity of the model. But - depending
on the type of model wanted, which one of the two approaches should preferably be
chosen?
Also the structure of some flow-relations must often be laid down by experiments. This
means, that normally we cannot speak of a proper analytical approach to the
establishment of realistic models of hydraulic components even on a medium level of
analysis. Instead an analytical/experimental approach must be applied.
2
Steady State
Characteristics
2.1.1 Introduction
The main purpose of the hydraulic fluid is to transport energy from the pump to the
actuators. Secondary purposes involve the lubrication of the moving mechanical parts to
reduce wear, noise and frictional losses, protecting the hydraulic components against
corrosion and transporting heat away from its sources. The preferred working fluid in
most applications is mineral oil, although in certain applications there is a requirement
for water-based fluids. Water-based fluids and high water-based fluids provide fire
resistance at a lower cost and have the advantage of relative ease of oil storage and fluid
disposal. The recommended classification system is as follows:
HFA – dilute emulsions, i.e. oil-in-water emulsions, typically with 95% water content.
HFB – Invert emulsions, i.e. water-in-oil emulsions, typically with 40% water content.
HFC – Aqueous glycols, i.e. solutions of glycol and polyglycol in water, typically with
40% water content.
HFD – Synthetic fluids containing no water, such as silicone and silicote esters.
The selection of the appropriate fluid will require specialist advice from both the
component manufacture and the fluid manufacture.
As the most commonly used hydraulic fluid is mineral oil and in the following sections
it is the physical properties of commercial mineral oils that is discussed.
The purpose of this chapter is to define certain physical properties which will prove
useful and to discuss properties related to the nature of fluids. Because the fluid is the
medium of transmission of power in a hydraulic system, knowledge of its characteristics
is essential
The mass density, ρ , of a hydraulic fluid is defined as a given mass divided by its
volume, see Equation (2.1).
m
ρ= (2.1)
V
where
ρ mass density [kg / m 3 ]
m mass of the fluid [kg]
V volume of the fluid [m 3 ]
The mass density is both temperature and pressure dependent. It decreases with
increasing temperature but increases with increasing pressure.
A generally accepted empirical expression, the Dow and Fink equation, describes this:
(
ρ(t , p ) = ρ 0 ( t ) ⋅ 1.0 + A β (t ) ⋅ p − Bβ (t ) ⋅ p 2 ) (2.2)
where
ρ mass density [kg / m 3 ]
ρ0 mass density at atmospheric pressure [kg / m 3 ]
Aβ temperature dependant coefficient [bar −1 ]
Bβ temperature dependant coefficient [bar −2 ]
p pressure [bar ]
The density for a hydraulic fluid is normally (DIN 51757) given by the fluid
manufacturer as the density at 15 o C and atmospheric pressure. This reference density
lies between 0.85 and 0.91 g / cm 3 (850-910 kg / m 3 ) for commercial hydraulic fluids.
The reference mass density in Equation (2.2) may be determined by:
ρ15
ρ 0 (t) = (2.3)
(1 + α t ⋅ (t − 15))
where
ρ0 mass density at atmospheric pressure [kg / m 3 ]
ρ15 mass density at atmospheric pressure and 15 o C [kg / m 3 ]
αt thermal expansion coefficient [deg −1 ]
t temperature [ o C]
The two coefficients in Equation (2.2) are normally referred to as the Dow and Fink
coefficients. They have experimentally been found to:
(
A β = − 6.72 ⋅ 10 −4 ⋅ T 2 + 0.53 ⋅ T − 36.02 ⋅ 10 −6 ) (2.4)
Bβ = (2.84 ⋅ 10 −4
⋅ T − 0.24 ⋅ T + 57.17 ⋅ 10
2
) −9
(2.5)
where
Aβ temperature dependant coefficient [bar −1 ]
Bβ temperature dependant coefficient [bar −2 ]
T absolute temperature [K ]
The variation of the Dow and Fink coefficients with temperature is displayed
graphically in Figure 2.1
68 17
Aβ
66 16
64 15
62 14
60 13
58 Bβ 12
56 11
54 10
52 9
50 8
-20 0 20 40 60 80 100 120
t 0C[ ]
Fig. 2.1 The variation of the Dow and Fink coefficients with temperature
Inserting Equations (2.3)..(2.5) in Equation (2.2) means that the density can be
determined by calculations only (no measurements), for any pressure and temperature
combination, as long as the reference mass density, ρ15, is known. The variation of the
mass density with temperature and pressure is displayed graphically in Figure 2.2.
In Figure 2.2 the mass density is displayed relative to the reference mass density.
2.1.3 Viscosity
The most important of the physical properties of hydraulic fluids is the viscosity. It is a
measure of the resistance of the fluid towards laminar (shearing) motion, and is
normally specified to lie within a certain interval for hydraulic components in order to
obtain the expected performance and lifetime. The definition of viscosities is related to
the shearing stress that appear between adjacent layers, when forced to move relative
(laminarly) to each other. For a newtonian fluid this shearing stress is defined as:
dx&
τ xy = µ (2.6)
dy
where
τxy shearing stress in the fluid, [N/m2]
µ dynamic viscosity, [Ns/m2]
x& velocity of the fluid, [m/s]
y coordinate perpendicular to the fluid velocity, [m]
ρ
ρ 15
1.4
1.2
0 °C
1 20 °C
40 °C
0.8 60 °C
80 °C
100 °C
0.6
0.4
0.2
0
0 100 200 300 400 500 600 700 800
p [bar]
Fig. 2.2 The variation of the mass density with temperature and pressure
In Figure 2.3 the variables associated with the definition of the dynamic viscosity are
shown.
x = x(y) τxy
dy
τ xy dx
fluid
x
The usual units for the dynamic viscosity is P for Poise or cP for centipoise. Their
relation to the SI-units are as follows: 1 P = 100 cP = 0.1 Ns / m 2 .
For practical purposes, however, the dynamic viscosity is seldom used, as compared to
the kinematic viscosity that is defined as follows:
µ
ν= (2.7)
ρ
where
ν kinematic viscosity, [m2/s]
µ dynamic viscosity, [Ns/m2]
ρ density, [kg/m3]
The usual unit used for ν is centistoke, cSt, and it relates to the SI units as follows:
m2 mm 2
1 cSt = 10 −6 =1 .
s s
A low viscosity corresponds to a "thin" fluid and a high viscosity corresponds to a
"thick" fluid. The viscosity depends strongly on temperature and also on pressure. The
temperature dependency is complex and is normally, DIN51562 and DIN51563
described by the empirical Uddebuhle-Walther equation:
This dependency is normally shown in specially designed charts, where the kinematic
viscosity shown as function of the temperature becomes a straight line, see Figure 2.4.
ν [cSt]
5000
2000
1000
500
300
200
150
100
80
60
50
40
30
25
20
16
12 ISO VG 100
10
ISO VG 68
8
7 ISO VG 46
6 ISO VG 32
5
ISO VG 22
4
3.5
ISO VG 10
3
2.7
-50 -40 -30 -20 -10 0 10 20 30 40 50 60 70 80 90 100 120
t [°C]
Fig. 2.4 Uddebuhle-chart: The temperature dependency for some of the most commonly used
mineral oils. The ISO VG standard refers ν at 40°C
L−U
VI = ⋅ 100% (2.9)
L−H
where
VI viscosity index
L kinematic viscosity at 100°F for the temperature sensitive fluid
U kinematic viscosity at 100°F for the fluid to be indexed
H kinematic viscosity at 100°F for the temperature insensitive fluid
Different standards, e.g. DIN ISO 2909, offer a list of reference fluids with different
kinematic viscosities at 210°F to pick from. The method dates back to 1929 and the
improvement in mineral oil destillation and refining means that many hydraulic fluids
come out with an index above 100.
Beside the temperature dependency the viscosity also depends on pressure, especially at
higher levels. The general accepted expression is as follows:
Bη p
µ = µ0 ⋅ e (2.10)
where
µ dynamic viscosity, [Ns/m2]
µ0 dynamic viscosity at atmospheric pressure [Ns/m2]
Bη temperature dependant parameter, [bar-1]
p pressure, [bar]
The parameter Bη may, within temperature ranges from 20°C to 100°C, be determined
empirically as:
B η = 0.0026 − 10 5 ⋅ t (2.11)
where
Bη temperature dependant parameter, [bar-1]
t temperature, [°C]
µ 0 Bη p
ν= ⋅e (2.12)
ρ
where
ν kinematic viscosity, [m2/s]
In the above it should be remembered that the density increases with pressure, thereby
making the kinematic viscosity less sensitive to pressure rise.
2.1.4 Dissolvability
The capability of dissolving air (saturation point) varies strongly for hydraulic fluids
with pressure. For pressure levels up to approximately 300 bar, the Henry-Dalton
sentence applies:
pa
Va = α V ⋅ VF ⋅ (2.13)
p atm
where
Va volume of dissolved air in the oil, [m3]
αV Bunsen coefficient, approximately constant at 0.09
VF volume of the fluid at atmospheric pressure, [m3]
pa absolute pressure, [bar]
patm atmospheric pressure ≈ 1 bar, [bar]
The capability of hydraulic fluids to absorb air is a problem, because the subsequent
release of air at lower pressures leads to reduced fluid stiffness.
2.1.5 Stiffness
1 ∂ρ
κF = ⋅ (2.14)
ρ ∂p
where
κF compressibility of the fluid, [bar-1]
ρ mass density, [kg/m3]
p pressure, [bar]
1
βF = (2.15)
κF
where
κF compressibility, [bar-1]
βF bulk modulus, [bar]
Based on the above definition it can be shown that for fixed temperature the stiffness is
proportional to the pressure rise caused by a compression of the fluid:
β F ⋅ dV
dp = (2.16)
V0
where
dp increase in pressure, [bar]
βF bulk modulus of the fluid, [bar]
dV the compression, i.e., decrease in volume, [m3]
V0 the volume corresponding to the initial pressure, [m3]
Just like density the bulk modulus and the compressibility are functions of temperature
and pressure. Inserting Equation (2.2) in Equation (2.14) and Equation (2.15) leads to:
1.0 + A β (t ) ⋅ p − Bβ (t ) ⋅ p 2
β F (t , p ) = (2.17)
A β (t ) − 2 ⋅ Bβ (t ) ⋅ p
where
βF stiffness of the fluid, [bar]
Aβ temperature dependant coefficient, [bar-1]
p pressure, [bar]
Bβ a temperature dependant coefficient, [bar-2]
Where the temperature dependant coefficients can be determined from and. It should be
noted that Equation (2.17) implies that the fluid stiffness may be calculated for any
temperature and pressure combination regardless of the specific type of mineral oil. The
variation of the fluid stiffness with temperature and pressure is displayed graphically in
Figure 2.5.
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. 2.5 The variation of the fluid stiffness with temperature and pressure
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.
If the air, however, is dissolved in the fluid there is no significant effect on the
compressibility. Hence, it is the amount of free or entrapped air in the fluid that
markedly reduces the effective stiffness. Taking the presence of air into account the
effective stiffness of the fluid becomes:
1 1
β eff (t , p, ε A ) = ≈
1 1 1 1 εA (2.18)
+ ε A − +
βF βA βF βF βA
where
βeff effective stiffness of the fluid-air mixture, [bar]
εA the volumetric ratio of free air in the fluid
βF stiffness of the pure fluid according to, [bar]
βA the air stiffness according to, [bar]
p pressure, [bar]
VA
εA = (2.19)
VF + VA
where
εA volumetric ratio of free air in the fluid
VA the volume of air, [m3]
VF volume of the fluid, [m3]
Assuming adiabatic conditions the volume and stiffness of the air may be determined
as:
1
p cad
VA = VA 0 ⋅ atm (2.20)
pa
β A = c ad ⋅ p a (2.21)
where
VA volume of air, [m3]
VA0 volume of air at atmospheric pressure, [m3]
patm atmospheric pressure ≈ 1 bar, [bar]
pa absolute pressure, [bar]
cad adiabatic constant for air, 1.4
ρ 0 (t 0 )
VF (t , p ) = VF0 ⋅ (2.22)
ρ(t , p )
where
VF volume of the fluid, [m3]
VF0 volume of the fluid at atmospheric pressure and a reference
temperature, [m3]
ρ0 mass density at atmospheric pressure according to Equation
(2.3), [kg/m3]
ρ the mass density according to Equation (2.2), [kg/m3]
t0 reference temperature, [°C]
t temperature, [°C]
p pressure, [bar]
From Equation (2.20) and Equation (2.22) it is clear, that the volumetric ratio varies
with both temperature and pressure. A reference volumetric ratio at atmospheric
pressure is defined:
VA 0
ε A0 = (2.23)
VF 0 + VA 0
where
ε A0 the reference volumetric ratio of free air in the fluid at
atmospheric pressure
VA0 volume of air at atmospheric pressure, [m3]
VF0 volume of the fluid at atmospheric pressure and a reference
temperature, [m3]
Knowing this reference, volumetric ratio together with the reference temperature, t0,
may be rearranged to yield an expression for the volumetric ratio directly obtainable
from temperature and pressure:
1.0
ε A (t , p ) = −1
1.0 − ε A 0 ρ 0 (t 0 ) p atm c ad (2.24)
⋅ ⋅ + 1.0
ε A0 ρ(t , p ) pa
In Figure 2.6 the variation of the effective stiffness according to Equation (2.18) is
displayed. The variation of the stiffness is dramatic for small pressure levels. The curves
in Fig. 2.6 do not take into account the effect of the Henry-Dalton sentence, Equation
(2.13), according to which the free air should dissolve at a few bars pressure and
subsequently have no effect on the effective stiffness. The Henry-Dalton sentence,
however, is for static conditions and in a hydraulic system the pressure variations
outside the tank reservoirs are typically so fast, that the hydraulic fluid does not have
time to dissolve the free air. Naturally, some air is dissolved, meaning that the curves
shown in Fig. 2.6 represents worst case, i.e., instantaneously pressure build up.
Stiffness plays a central role w.r.t. to dynamic performance of hydraulic systems and
should be determined/predicted as precisely as possible. This is, however, not an easy
task. The sum VF0+VA0 in Equation (2.23) is relatively easily determined, whereas VF0
or VA0 are more elusive.
As a rule of thumb, the stiffness under working conditions used for modelling a system
should not be set above 10000 bar, unless verified by means of testing.
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]
Fig. 2.6 Variation of effective stiffness of fluid-air mixture with respect to pressure and
volume ratio of free air at atmospheric pressure. The temperature of the fluid is 40 °C and the
compression of the free air is assumed adiabatic
2.2.1 Introduction
Hydraulic control valves are devices that use mechanical motion to control a source of
fluid power. They vary in arrangement and complexity, depending on their function.
Because control valves are the mechanical to fluid interface in hydraulic systems, their
performance characteristics are essential. Although emphasis is placed on a principal
type of spool valve, the principles apply equal well to other valves, such as different
kinds of pressure valves and flow control valves.
The most common used control valve is the spool valve. Two typical spool valve
configurations are shown in Figure 2.7. One in the pilot stage and one in the main stage.
Spool valves can broadly be classified by the number of “ways” the flow can enter and
leave the valve, and the type of centre when the valve spool is in neutral position.
Main stage
Because all valves require a supply, a return, and at least one line to the load, valves are
either three-way or four-way (see Figure 2.7). Two-port valves are also available.
However, two-way valves cannot provide a reversal in the direction of flow.
If the width of the land is smaller than the port in the valve sleeve, the valve is said to
have an open centre or to be underlapped. A critical centre or zero lapped valve has a
land width identical to the port width and is a condition approached by practical
machining. Closed centre or overlapped valves have a land width greater than the port
width when the spool is at neutral.
The above examples serve the purpose of illustrating how flow paths may be created
using a variety of restrictions. The actual displacement of the spool which cause the
flow restriction is usually of such small value relative to port diameter that the
pressure/flow equations obey the Bernoulli equation.
The flow restrictions or orifices are a basic means for the control of fluid power. An
orifice is a sudden restriction of short length in a flow passage and may have a fixed or
variable area. In fluid power is it only inertia and viscous forces that matters.
Experience has shown that it is either the inertia forces or the viscous forces that
dominate, giving two types of flow regimes. Therefore, it is useful to define a quantity
which describes the relative significance of these two forces in a given flow situation.
The dimensionless ratio of inertia forces to viscous force is called Reynolds number and
defined by
ρ u dh
Re = (2.25)
µ
4 × flow area
dh = (2.26)
flow perimeter
p u2
+ = constant (2.27)
ρ 2
A0
A2
1 2 3
Since most orifice flow occur at high Reynolds numbers, this region is of great
importance. Experience has justified the use of Bernoulli’s equation in the region
Between point 1 and 2. The point along the jet where the area becomes a minimum is
called the vena contracta. The ratio between the area at vena contracta A 2 and the
orifice area A 0 defines the so called contraction coefficient C c .
Cc = A 2 / A0 (2.28)
After the fluid has passed the vena contracta there is turbulence and mixing of the jet
with the fluid in the downstream region. The kinetic energy is converted into heat. Since
the internal energy is not recovered the pressures p 2 and p 3 are approximately equal.
Now it is possible to use Bernoulli’s equation (2.27) to calculate the relation between
the upstream velocity u 1 to the velocity u 0 in vena contracta. Therefore
2
u 22 − u 12 = ( p1 − p 2 ) (2.29)
ρ
A1u 1 = A 2 u 2 = A 3 u 3 (2.30)
Combining Equation (2.29) and Equation (2.30) and solving for u 2 gives
−1 / 2
A 2 2
u 2 = 1 − 2 ⋅ (p1 − p 2 ) (2.31)
A 1 ρ
In the real world there will always be some viscous friction (and deviation from ideal
potential flow), and therefore an empirical factor C v is introduced to account for this
discrepancy. C v is typically around 0.98. Since Q = A 2 u 2 the flow rate at vena
contracta becomes, by using Equation (2.31).
CvA2 2
Q= ( p1 − p 2 ) (2.32)
1 − (A 2 / A1 ) 2 ρ
C vCc
Cd = (2.33)
1 − C c2 (A 0 / A 1 ) 2
Now, combining Equation (2.28), (2.32), and (2.33) the orifice equation (in Danish
blændeformlen) can be written
2
Q = Cd A 0 ( p1 − p 2 ) (2.34)
ρ
0.8
0.6
Cd
0.4
0.2
0
0 10 20 30 40 50
Re
Fig. 2.9 Plot of a discharge coefficient versus Reynolds number for an orifice
At low temperatures, low orifice pressure drop, and/or small orifice openings, the
Reynolds number may become sufficiently low to permit laminar flow. Although the
analysis leading to Equation (2.34) is not valid at low Reynolds numbers, it is often
used anyway by letting the discharge coefficient be a function of Reynolds number. For
Re < 10 experimental results shows that the discharge coefficient is directly
proportional to the square root of Reynolds number; that is C d = δ Re . A typical plot
of such a result is shown in Figure 2.9.
Supply
Return PS QS
PR QS
2 4
X
3 1
P2 QL QL P1
LOAD
Let the spool be given a positive displacement from the null or neutral position, that is
the position x = 0, which is chosen to be the symmetrical position of the spool in its
sleeve.
This allows the supply flow Q S to travel to the load as Q1 , the difference being only
leakage flow present, Q 4 , across the other land. The flow from the load returns as Q 3 ,
which with the possible addition of the leakage flow Q 2 , then forms the return flow.
Because we are only interested in the steady-state characteristics, the compressibility
flows are zero and the flow continuity equations for the valve chambers are
Q L = Q1 − Q 4
(2.35)
Q L = Q3 − Q 2
where Q L is the flow through the load. The load pressure differential is defined as the
pressure drop across the load.
p L = p1 − p 2 (2.36)
Flows through the orifices are described by the orifice Equation (2.34). Therefore
2 2
Q1 = C d A 1 ( p S − p1 ) Q2 = Cd A 2 (p S − p 2 )
ρ ρ
(2.37)
2 2
Q3 = Cd A3 p2 Q4 = Cd A 4 p1
ρ ρ
In the vast majority of cases the metering orifices are made so that they are matched and
symmetrical. Matched orifices require that
A1 = A 3 ; A2 = A4 (2.38)
A1 ( x ) = A 2 (− x ) ; A 3 ( x ) = A 4 (− x ) (2.39)
This means, that in neutral position (x = 0), all four orifice areas are equal ( ≈ A 0 ). If
further the orifice areas varies linear with the stroke, as is usually the case, the areas can
be described by only one parameter w, defining the width of the slot in the valve sleeve.
w is the area gradient.
A1 = A 3 = A 0 + x ⋅ w ; A2 = A4 = A0 − x ⋅ w (2.40)
The condition that the orifices are matched and symmetrical, gives that
Q1 = Q 3 ; Q2 = Q4 (2.41)
Substituting Equation (2.37) ( Q1 and Q 3 ), and Equation (2.38) into Equation (2.41)
yields
p S = p1 + p 2 (2.42)
Now, Equation (2.36) and Equation (2.42) can be solved simultaneously to obtain
pS + p L pS − p L
p1 = ; p2 = (2.43)
2 2
With the relations in Equation (2.41), and Equation (2.43) together with the equations in
Equation (2.35) it is possible to find an expression for the load flow as a function of the
load pressure.
1 1
Q L = C d A1 (p S − p L ) − C d A 2 (p S + p L ) (2.44)
ρ ρ
Equation (2.44) represent the general steady-state valve equation for a symmetric
matched four-way spool valve applied to a symmetric load.
2.3.1 Introduction
It will be found necessary in a dynamic analysis that the non-linear algebraic equation
which describe the pressure-flow curves to be linearised. From Equation (2.44) the load
flow can be written as a function of the spool position and the load flow
Q L = Q L ( x , p L ) . If x and p L only changes by a small amount about a operating point
( Q L 0 , p L 0 , x 0 ) the general expression for the load flow can be expressed by a Taylor’s
series. We only consider the first order terms, assuming that the higher order
infinitesimals are negligible small. Hence,
∂Q L ∂Q L
Q L = Q L0 + ∆x + ∆p L + ... (2.45)
∂x 0 ∂p L 0
The partials in Equation (2.45) defines the two most important parameters for a valve.
The flow gain is defined by
∂Q L
kq ≡ (2.46)
∂x
∂Q L
k qp ≡ (2.47)
∂p L
∂p L
kp ≡ (2.48)
∂x
∂p L ∂Q L / ∂x kq
= or k p = (2.49)
∂x ∂Q L / p L k qp
The three quantities k q , k qp , k p are called valve coefficients and are extremely
important in the dynamic analysis of valves in combination with actuators. k p express
the ability to of a valve-actuator combination to breakaway large friction loads. k qp has
a direct influence on the damping in the valve-actuator combination. k q directly affects
the open loop gain in a system and therefore has influence on system stability. The
valve coefficients evaluated in the neutral position of the valve (Q L 0 , p L 0 , x 0 ) = (0, 0, 0)
are called the null valve coefficients. This operating point is the most critical point from
a stability viewpoint, while the flow gain is largest, giving high system gain, and the
flow-pressure coefficient is smallest, giving a low damping.
Many valves are manufactured with a relative linear flow gain near null position,
meaning that A 0 = 0 in Equation (2.40). Assuming the valve to have ideal geometry the
leakage flows are zero. For such a valve the load flow can be expressed by
pS − p L
QL = Cd w ⋅ x ; x>0 (2.50)
ρ
pS + p L
Q L = −C d w ⋅ x ; x<0 (2.51)
ρ
1 x
Q L = Cd w ⋅ x (p S − p L ) (2.52)
ρ x
This is the general equation for the pressure-flow curves of an ideal critical centre valve
with matched and symmetrical orifices. The Equation (2.52) is plottet in Figure 2.11.
QL
-PS PL
PS
The valve coefficients for the important case of an ideal critical centre valve can be
obtained by differentiation of Equation (2.44), and are given below in Figure 2.12.
1 pS
kq Cd w (p S − p L 0 ) Cd w
ρ ρ
Cd w ⋅ x 0
k qp 0
2 ρ(p S − p L 0 )
kp 2(p S − p L 0 ) ∞
x0
Consider the steady-state flow through a spool valve as shown in Figure 2.13. When the
fluid is flowing through the valve there will be induced some forces acting on the valve.
X V1 Fluid
p2 V2
element
Θ
p
1
FR
Face b
Face a
L
Fig. 2.13 Flow forces on a spool valve due to flow leaving a valve chamber
r r r ∂ r r r r
F = FS + FB = ∫ Vρ dV + ∫ VρV ⋅ dA (2.53)
∂t CV CS
This equation states that the sum of all forces (surface and body forces) acting on a non-
accelerating control volume is equal to the sum of the rate of change of momentum
inside the control volume (CV) and the net rate of efflux of momentum through the
control surface (CS).
Since we are looking for the horizontal force the body force is zero, and the only surface
force in horizontal direction is the force FR , which is the force of the spool on the
control volume. Change of momentum inside the control volume occur when the spool
position is suddenly changed, say to the right, as shown in Figure 2.13. If the fluid
element is being accelerated, the pressure on the left side of the element must be greater
than the pressure on the right side. Therefore, the pressure on face a must be greater
than the pressure on face b. Thus the transient flow force is due to acceleration of the
fluid in the annular valve chamber. The direction of this force for the case shown in
Figure 2.13 is such that it tends to close the valve – however this is not the general rule.
A movement of the spool can also cause the fluid to be decelerated. Applying the
momentum equation in the horizontal direction gives
d (Q / A n )
F = FR = ρLA n + ρQV2 cos(θ) (2.54)
dt
where Q is the volumetric flow rate and A n is the annular area of the spool. The last
term in Equation (2.54) can be rewritten as
ρQ 2 ρQ 2
ρQV2 cos(θ) = = (2.55)
A2 CcA0
Where A 0 is the orifice area. The flow Q can be described by the orifice equation as
2 2
Q = Cd A 0 ( p1 − p 2 ) = C c C v A 0 ( p1 − p 2 ) (2.56)
ρ ρ
dx LC d wx d ( p1 − p 2 )
Ft = LC d w 2ρ(p1 − p 2 ) + (2.57)
dt (2 / ρ)(p1 − p 2 ) dt
The last term in Equation (2.57) is normally neglected. The velocity term is more
significant because it represents a damping force. The quantity L is the axial length
between incoming and outgoing flow and is called the damping length.
Inserting Equation (2.56) into Equation (2.55), the steady-state axial flow force acting
on the valve spool can be obtained as
For a spool with no radial clearance it is well known, and usually assumed, that the jet
leaves the control port at an angle of θ = 69 o . The sum of Equation (2.57) and Equation
(2.58) give the total flow force, steady-state and transient, opposing the spool motion,
while the force from the fluid on the spool is opposite FR , which is the external force
acting on the control volume.
3
Dynamic Modelling
3.1 Introduction....................................................................... 1
3.4 Motors...........................................................…………… 6
AaU ~ Forår 2003
3.5 Linear Characteristics of a Cylinder................................. 8
3.1 Introduction
Hydraulic control systems are used to control the position or speed of resisting loads.
The final drive is usually either a linear motion hydraulic cylinder or a rotary-motion
hydraulic motor. Figure 3.2 illustrates two simple hydraulic control systems. The
actuator develops its force or torque by receiving liquid from a positive displacement
pump at a relatively high pressure, and the actuator develops its motion by receiving
flow rate of liquid from the pump.
It is not sufficient for the designer of a hydraulic system to know that the system will
move. He also need to know how it will move. He needs to appreciate not only the
initial and final states of the responses, but also the time-domain path between these
states. He should know if the system response is stable, if it is fast enough or perhaps
too fast, if it is oscillatory, etc. A meaningful dynamic analysis cannot be made without
including the dynamics associated with the linear cylinder or the rotary motor.
One of the elements the designer must take into account is the natural frequency of the
system. From the laws of physics we know that the formula for undamped, natural
frequency is
ωn = c / m
where
Hydraulic Cylinder nm Tm
Fc Hydraulic Motor
Vc
150 bar
Hydraulic
pump
Filter Flow Control
Valve
np
Tp
Reservoir
The spring constant for a hydraulic system can be related directly to the oil volume
trapped between the controlling element, typical a valve, and the actuator. The natural
frequency determine how fast a load can be accelerated and decelerated without causing
instability and subsequent damage to the system. This frequency can be calculated
mathematically to tell us how fast this weight can be moved back and forth without
having the weight of the object directly oppose the input to the spring. For example, the
input to the spring-mass system (see Figure 3.1) could be someone’s hand moving the
spring-mass system up and down a certain distance. As long the spring is moved more
slowly than the natural frequency of the total spring-mass system, the weight will
follow the movement of the spring. There will be very little difference between the
movement of the spring and the weight.
The faster the input, or hand movement to the spring, the more the weight will lag. If
the input to the spring is at the same frequency as that of the total spring-mass system,
as one’s hand moves down, the weight moves up. Likewise, as the hand moves up, the
weight moves down. The weight would then be in direct opposition to the movement of
the spring. This would result in a system performing a function opposite of that
required. This is called instability, or resonance.
To put this concept in perspective with respect to a hydraulic system, the natural
frequency can be calculated and trying to accelerate or decelerate high inertia loads too
quickly is likely to cause a cylinder or motor to become unstable.
In the following sections the nonlinear equations of a hydraulic cylinder and rotary
motor will be derived, and the natural frequency and damping of a symmetrical
cylinder. But first some necessary fluid theory will be introduced.
The physical principle used to derive the continuity equation is conservation of mass. It
is intuitive that mass neither can be created nor destroyed; if the flow rate of mass into a
control volume exceeds the rate of flow out, mass will accumulate within the control
volume. Conservation of mass requires that the sum of the rate of change of mass within
the control volume (CV) and the net rate of mass outflow through the control surface
(CS) be zero.
∂ r r
0 = ∫ ρ dV + ∫ ρ V ⋅ dA (3.1)
∂t CV CS
In Equation (3.1) the first term represents the rate of change of mass within the control
volume; the second term represents the net rate of mass efflux through the control
surface. Now consider the control volume given in Figure 3.3 and the Equations in
(3.2).
pout
me
olu
nt r
ol
V
m
Co ∂
∂t ∫CV
ρout 0= ρ dV = m
& in − m
& out
ρin V Q out
Q ρ
m in p (3.2)
⇓
Q in
p ∂
in ρ in Q in − ρ out Q out = (ρV)
∂t
Fig. 3.3 Fluid volume
dV V dρ
Q in − Q out = + (3.3)
dt ρ dt
Equation (3.3) needs to be transformed to a more usable form. This can be achieved by
using Equation (2.14) and Equation (2.15). The latter defining the bulk modulus of the
fluid.
dρ dp
= (3.4)
ρ βF
dV V dp
Q in − Q out = + (3.5)
dt β F dt
The first term on the right side is the flow consumed by expansion of the control
volume; if the volume is fixed this term is zero. The second term is the compressibility
flow and describes the flow resulting from pressure changes.
Normally expansion of the control volume due to wall deformation in the different
components is assumed to be contained in the definition of the effective bulk modulus
β F . Equation (3.5) will be referred to as the flow continuity equation.
dv
(p in A in − p out A out ) − ∑ Fi = M (3.7)
dt
where A in is the input net cross-sectional area and A out is the output net cross-sectional
area. M is the mass of the fluid being accelerated. To change the velocity in for example
a pipe or a cylinder the fluid needs to be accelerated. The necessary force is generated
by a pressure difference.
The application of Equation (3.6) depends upon the component being considered. In the
following it is applied to a fluid line.
m
ρ A
From the flow continuity equation (3.5) we can write an expression for the pressure in
the pipe
V dp πd2
Q in − Q out = ; V = A L, A = (3.8)
β F dt 4
Where d is the internal diameter of the pipe and L its length. The momentum equation,
from Equation (3.7), becomes
dQ
(p in − p out )A − ∑ Fi = ρ L (3.9)
dt
The resisting force ∑F i is assumed to be entirely due to fluid viscosity effects. Other
effects are due to flow in a piping system may be required to pass through a variety of
fittings, bends, or abrupt changes in area. The loss can be expressed as a pressure loss,
meaning
l v2
∑ i
F = ∆p A ; where ∆p = λ ρ
d 2
(3.10)
Where v is the fluid velocity. The friction factor λ is determined experimentally, and
can be expressed as a function of Reynold’s number : Re = ρ v d / µ
64
For laminar flow (Re < 2300): λ=
Re
(3.11)
0.3164
For turbulent flow (Re > 2300): λ=
Re 0.25
3.3 Cylinders
Consider the asymmetric cylinder shown in Figure 3.6. We want to write up the
equations describing the dynamics of the cylinder and load.
V1 V2
FL
ML
p p
1 2
Q1 Q2
AP AR
Fig. 3.5 Asymmetric linear actuator
dv
ML = p1 A P − p 2 A R − FL (3.12)
dt
Consider next the flow continuity equation (3.5) applied to the actuator.
Extending Retracting
Extending Retracting
V1 dp1 V2 dp 2
Q1 = A P v + Q2 = AR v + (3.14)
β F dt β F dt
V dp 2 V dp
Q2 = AR v − 2 Q1 = A P v − 1 1
β F dt β F dt
If we are consistent about the notation, it is possible to reduce Equation (3.14). If the
flow directions shown in Figure 3.6 are termed positive and the extending velocity is
termed positive, then Equation (3.14) reduce to
dp1 β F dp 2 β F (3.15)
= (Q 1 − A P v) = (A R v − Q 2 )
dt V1 dt V2
A cylinder is a full stroke component and dynamically both volumes V1 and V2 vary
with piston motion. In linearised analysis we are often interested only in the transient
behaviour around some operating point. In this situation it is ok to neglect the variation
of the volumes. In a nonlinear simulation model, however, the variation must be taken
into account.
Also, other phenomena like viscous friction and spring terms can be added to the
momentum equation.
3.4 Motors
The rotary hydraulic motor (see Figure 3.6) is an important element in hydrostatic
transmissions. The motor transforms, as earlier described, hydraulic power to
mechanical power.
Newton’s 2nd law of motion for a rotating device states:
Now, let us look at the elements in Equation (3.16). The ideal generated torque is
Tg = D m (p1 − p 2 ) (3.17)
where D m , by definition is the flow through the motor ( Q L ) divided by the shaft speed
of the motor ( ω);
Dm ≡ QL / ω (3.18)
and is called the volumetric displacement (or simply displacement) of the motor,
[ cm 3 / rad ]. In Equation (3.17) the term p1 − p 2 is the pressure difference across the
motor.
Q 1 p1
TL
JL
ω
Q 2 p2
Fig. 3.6 Rotary motor with inertia load and load torque
However, there are at least three sources of resisting torque losses which detract from
the generated torque. The first one is viscous damping due to shearing in the fluid in the
tight clearances between the mechanical elements in relative motion. With B m as a
viscous damping coefficient, this damping torque can be written as
Td = B m ω (3.19)
Investigation into the movement and forces of each piston gives a friction force
opposing motion that is proportional to the pressure acting on the piston.
Tf = f (ω, p1 , p 2 ) (3.20)
This nonlinear term in Equation (3.20) has to be experimentally determined. Now, the
torque equation (3.16) can be written as
dω
JL = D m (p1 − p 2 ) − Td − Tf − TL (3.21)
dt
In the torque equation the load inertia J L and load torque TL are included, see Figure
3.6. Looking at the fixed displacement axial piston motor schematical represented in
Figure 3.7, we can write down the continuity equations for the two motor chambers.
Qel1
P1 Q1
Fluid
lines
P2 Q2 Q il
Qel2
Only two chambers is shown in Figure 3.7, but the leakage flows from all the pistons
are lumped at these two pistons.
From the figure we can see that there are at least two types of leakage flows; internal or
cross-port leakage between the pressure side and the return side, and external leakage
resulting from oil passing the pistons. Due to the very small clearances in hydraulic
motors the leakage flows are laminar and therefore proportional to the pressure drop.
Thus, the internal leakage can be written
Q il = C il (p1 − p 2 ) (3.22)
Q el1 = C el p1 ; Q el 2 = C el p 2 (3.23)
Q1 = D m ω + C il (p1 − p 2 ) + C el p1 (3.24)
Q 2 = D m ω + C il (p1 − p 2 ) − C el p 2 (3.25)
These two equations completely describe the flows in the motor. If leakage is zero, we
have that Q1 = Q 2 = D m ω (ideal motor).
In this section we will derive the dynamic characteristics of a cylinder. In Figure 3.8 is
schematically shown a double-acting cylinder and the related parameters that will be
used in the section.
α
x
F
m
Q2 Q1
P2 P1
A
The dynamic characteristics will be expressed by the natural frequency and damping
ratio. We consider a situation where the cylinder ports are blocked.
V1 dP1
Q1 = A x& − =0 (3.27)
β F dt
V2 dP2
Q 2 = A x& + =0 (3.28)
β F dt
where the chamber volumes V1 and V2 are functions of the piston displacement. The
equations (3.27) and (3.28) are non-linear, and we need to linearise the equations
around an operating point – the index o meaning evaluated in the operating point. To
linearise the product term we utilise Taylor expansion for functions of more than one
variabel (see also Equation (2.45)).
∂f ∂f
f ( x , y) = f ( x o , y o ) = (x − x o ) + (y − y o ) (3.29)
∂x o ∂y o
dp1 A β F
= x& (3.30)
dt V10
dp 2 A βF
=− x& (3.31)
dt V20
Equations (3.26), (3.30) and (3.31) represent the linear equations describing the
dynmics of the cylinder. After Laplace transforming the equations we have
A βF
p1 (s) = x& (s) (3.33)
V10 s
A βF
p 2 (s) = − x& (s) (3.34)
V20 s
1
s
x& (s) = m f (s)
A 2β F 1 (3.35)
α 1
s + s+
2
( + )
m m V10 V20
Comparing the coefficients in the denumerator with a standard second order system we
get the natural frequency and damping ratio as
βF 1 1 β V + V20
ωn = A ( + ) = A F ( 10 ) (3.36)
m V10 V20 m V10 V20
α V10 V20
ς= (3.37)
2A mβ F (V10 + V20 )
βF VT
ωn = A (3.38)
m V10 (VT − V10 )
Now, it is easy to verify that the minimum natural frequency, is when the piston is in its
mid position, meaning V10 = V20 = VT / 2 . In this position we obtain from Equation
(3.38) and Equation (3.39).
βF α VT
ωn = 2A and ς V / 2 = (3.40)
min
mVT 2
4A mβ F
EXAMPLE
rad
ωn min
= 2190 = 348 Hz ; ςV /2
= 0.2
sec T
3.6.1 Introduction
Over centre valves are useful in a variety of mobile fluid power applications. For
example they are often used in mobile cranes to ensure that a broken pipeline will not
cause the load to drop, see Figure 3.8.
Other applications are on vehicle and winch drives, to ensure that the load does not run
ahead of the pump flow, creating cavitation inside the motor. When the load is reversing
they allow normal operation using by-pass elements and has built in relief valve
protection. The valve also provides velocity control on descending load.
Over Center
Valve
Directional Control
Valve
T P
LS
Today, almost all systems including over centre valves, are based on a similar
configuration. The flow to the actuator is controlled by a directional control valve and
the overcenter valve is placed at the outlet of the actuator, pilot-operated from the
pressure in the inlet connection. Over centre valves, are closely related to pressure relief
valves and check valves opened by a pilot pressure. An over centre valve is, in effect, a
pilot opened pressure relief valve.
It is well known that systems equipped with over centre valves are prone to oscillations
in the load and can also become unstable. Another unwanted behaviour is the tendency
to abruptly stop when the speed of the load is retarded. In many cases, these types of
instability can lead to harzardous conditions.
Since safety reasons and legislations makes them a necessary element in many systems,
they are used despite the drawbacks of their dynamic behaviour, and the method used to
get a satisfactory operation is often trial end error. In particular, it is difficult to combine
this kind of valve with load sensing systems or systems with constant flow
characteristics (see section 5.2). In this section the over center valve will be modelled.
The over center valve considered in this section is shown in Figure 3.9. It consist of the
valve body (3), control plunger (4), check valve (2), and pilot piston (1).
2
1 Cylinder Port
When lifting the load, the fluid passes from the valve port V through the check valve
and port C to the consumer. During sinking, the flow directions is opposite – from C to
V . The check valve is closed and the flow to the output V is not possible until after the
control plunger is lifted due to the load pressure and the pilot pressure. The pilot, or
load lowering pressure acts on the pilot piston ( A P ), see Figure 3.10, and pushes open
the control plunger. Of course also the load induced pressure, which acts directly on the
control plunger ( A C ), tends to push the plunger in the open position.
Ap
AR
A v1 A v2
Ac
A high pilot ratio permits to lower the load with little pilot pressure, allowing a quicker
operation of the machine combined with energy saving. It is best suited for applications
where the kinematic motion of the structure maintains the load induced pressure
relatively constant.
A low pilot ratio requires a high pilot pressure in order to lower the load, but it permits
more precise and smooth control of motion. It is recommended for applications where
the load induced pessure varies during motion and can induce instability on the
machine.
In deriving the mathematical model we only consider the case where the load needs to
be counter balanced. In the other direction we assume that there is no loss when oil is
flowing through the check valve. It normally has a very soft spring and a large opening
area creating a negligable pressure drop.
Defining the load pressure as p C , the pilot pressure as p P , and the back pressure p V ,
we are able to describe the coverning equations for the over center valve.
The flow trough the valve from the load to downstream the over center valve can be
desribed by the orifice equation
2
Q o = C d A o (x o ) (p C − p V ) (3.42)
ρ
do
__
2 A o ( x o ) = π(d o x o sin α − x o2 sin 2 α cos α)
α
xo
Fig. 3.11 Discharge area for a seat valve
Normally the opening of such valves are relatively small, meaning that x o / d o is small.
This means that Equation (3.42) can be rewritten as
2
Q o = C d πd o x o sin α (p C − p V ) (3.43)
ρ
where Figure 3.10 defines the areas where pressure is acting, and
A V = A V1 − A V 2 + A R (3.45)
mo mass of plunger
FPL preloaded spring force on the plunger
FF flow force
K SR spring rate
Bo viscous damping coefficient
FC coulomb friction force
If A V is zero, we say that the over center valve is compensated to back pressure with
respect to the relief function. The back pressure still opposes the pilot piston, thus
increasing the pilot pressure needed to open the valve and to lower the load.
The preloaded spring force on the plunger is equal to
V1
Control volume
Ao
V2
FF FF
Using V1 >> V2 we have that the force opposing the spool motion can be written
FF = 2C d A o ( x o ) (p C − p V ) cos α (3.48)
where F is the value of the friction force, but this type of friction is of such complexity
that it mostly requires experimental investigation to find the friction characteristic.
----- oo0oo -----
4
Dynamic System
Analysis
4.1 Introduction....................................................................... 1
Fluid Power Systems
(Hydraulisk Komponentanalyse) 4.2 Flow Control Valves................................………….…… 1
4.3 Power Elements.................……………………………… 4
AaU ~ Forår 2003 Introduction • The electro-hydraulic valve • Valve-motor combination
• Valve-cylinder combination
4.1 Introduction
One of the major concerns of this note is that it attempts to reconcile linearised analysis
with dynamic simulation and measurements on a real system. Normally, linearised
analysis is extremely useful in understanding the behaviour of systems like those
mentioned above, because it shows how parameter variations will affect the system.
Dynamic simulation will then make it possible to examine the actual response of the
system including non-linearities. It is then possible to predict the behaviour of a system
after a design change, or in some other operating points. Finally, actual measurements
on a system similar to that studied by dynamic simulation will help improve the
simulation model, and make sure that it is realistic.
In Figure 4.1 below is given a restrictor type (two-way) pressure compensated flow
control valve.
x p1 Q 1
p2
Q2
p3 Q3
Flow control valves which are constructed in such a way that pressure difference across
control orifice is maintained constant, independently of pressure changes at the entry
and the outlet port of the valve. Because flow characteristics of these valves are
relatively insensitive to changes in supply or load pressure they are referred to as
”pressure compensated” flow control valves. The valves are used in systems which
require constant flow delivery to actuators, thus maintaining their constant speed
independently of load variations.
The dynamic equations that describe the dynamics of the valve can be stated as:
2
Q1 = C d ωx ( p1 − p 2 ) (4.1)
ρ
Flow equations
2
Q2 = Cd Ad (p 2 − p 3 ) (4.2)
ρ
dx V2 dp 2 (4.3)
Q 2 = Q1 + A s −
dt β dt
Continuity equations
dx V3 dp 3
Q3 = Q 2 − A s − (4.4)
dt β dt
d2x dx
M 2
= K o + K ( x o − x ) − A s (p 2 − p 3 ) − B Force equation (4.5)
dt dt
Where A s and A d are the areas of the spool end and the fixed orifice, respectively. K o
is the initial spring force. K is the spring constant, and B is a viscous damping
coefficient. ω is the area gradient. The maximum spring displacement is x o , and M is
the mass of the compensator spool. V1 and V3 are the volumes on both sides of the
restriction.
The valve in Figure 4.1 is a two-way type. When p1 = p 3 (= p 2 ) the flow through the
valve is zero. The initial spring force (K o ) will move the spool to the right ( x = x o ) . If
p1 is increased or p3 is decreased, there will be a flow through the valve, and p 2 > p 3 .
When A s (p 2 − p 3 ) ≥ K o the spool will move to the left and decrease the metering area
of the spool. If the pressure difference p1 − p 3 is increased further, then it can be seen
from the force equation that the pressure difference p 2 − p 3 will be proportional to the
spring force (in steady state)
A s (p 2 − p 3 ) = K o + K ( x o − x ) (4.6)
From the above equation it can be seen that by choosing the spring constant properly,
the flow will approximately be constant despite variations in the pressure difference
p1 − p 3 .
In linearising the set of equations it is assumed that p1 and A d are constant, - and
compressibility of oil is neglected. It is only the two flow equations that are non-linear.
Linearising these two equations around some operating point i, gives
∂Q1 2 ∂Q1 − C d ωx
K1 = = C d ω ( p1 − p 2 ) ; K2 = =
∂x i ρ i
∂p 2 i 2ρ(p1 − p 2 ) i
∂Q 2 Cd A d ∂Q1 − Cd A d
K3 = = ; K4 = =
∂p 2 i 2ρ(p 2 − p 3 ) ∂p 3 i 2ρ(p 2 − p 3 )
i i
Varying loads on the actuator will cause changes in the pressure drop across the valve
which in turn will alter the flow and, therefore, the speed of the actuator. To analyse this
effect we which to find the transfer function H(s) = q 3 (s) / p 3 (s) .
There are six variables x , p 2 , p 3 , q1 , q 2 and q 3 and the task is now to eliminate the
variables x , p 2 , q 1 and q 2 . Using in the following a notation without ( s) , one have
from Equation (4.16)
− A s (p 2 − p 3 )
x= (4.17)
Ms 2 + Bs + K
Equation (4.14) and Equation (4.15) gives that q 1 = q 3 . Using this and inserting
Equation (4.13) into Equation (4.14) gives
K 3 p 2 − A s sx = q 3 + K 3 p 3 (4.18)
K 2 p 2 − K1x = q 3 (4.19)
Solving the Equations (4.18) and Equation (4.19) for x and p 2 gives
(q 3 + K 3 p 3 )K 1 + A s sq 3 K 3 q 3 − K 2 (q 3 + K 3 p 3 )
p2 = ; x=
K 3 K1 + A s K 2s K 3 K1 + A s K 2s
The expressions for x and p 2 can now be put into Equation (4.17) and the transfer
function asked for can be found to
q 3 (s) M K 2 K 3 s 2 + (A s2 K 2 + B K 2 K 3 ) s + K K 2 K 3
H (s) = = (4.20)
p 3 (s) M (K 3 − K 2 ) s 2 + (B (K 3 − K 2 ) + A s2 ) s + B K 1 + K (K 3 − K 2 )
4.3.1 Introduction
x
TL
FL
JL ML
ω p2
p1 p2 p1 V2
V1 V2 V1 Q2
Q1 Q2 xv Q1 xv
ps ps
The purpose of this section is to define parameters and transfer functions of the valve-
actuator combinations. These analyses yield a description of the dynamic performance
of the power elements.
Return
LOAD
If we assume an ideal critical centre valve with matched and symmetrical orifices it was
shown in section 2.3 that the linearised valve flow could be written as Equation (2.45)
∆Q L = k q ∆x v − k qp ∆p L (4.21)
where Q L is the load average load flow and p L the load pressure difference. The
opening of the valve is described by the spool position x v . In Equation (4.21) the flow-
pressure coefficient is defined by k qp = −∂Q L / ∂p L , while ∂Q L / ∂p L is always
negative, making the flow-pressure coefficient always a positive number. An electro-
hydraulic actuated valve must also be treated as a dynamic system. Knowing the
parameters of the electrical activation it is possible to build a analytical model of the
valve, but often the dynamics is obtained by measurement using frequency-response
methods. From the frequency response a linear approximate model can be estimated. If
the frequency response is approximated with a second order system, then Equation
(4.21) can be Laplace transformed and written as (the ∆ is omitted in the following)
kq
Q L (s) = u (s) − k qp p L (s) (4.22)
(s / ωn ) + 2ς / ωn s + 1
2
Having described the dynamics of the valve, we need the actuator. In Figure 4.2 is
shown the two combinations considered in this text. Consider the left configuration.
Applying the continuity equation (3.5) to each motor chamber yields
V1 dp1
Q1 = D m ω + C il (p1 − p 2 ) + C el p1 + (4.23)
β F dt
V2 dp 2
Q 2 = D m ω + C il (p1 − p 2 ) − C el p 2 − (4.24)
β F dt
where the two volumes V1 and V2 includes the volumes in the valve, connecting line
and/or manifold, and volume in the motor. The other parameters are defined in section
3.4.
Define the average contained volume of each motor chamber V0 , and the total
contained volume of both chambers Vt . Then we can write
Vt = V1 + V2 = 2 V0 (4.25)
It is desirable to express the continuity equations (4.23) and (4.24) in terms of the load
flow, as done in Equation (4.24).
Using that Q L = (Q1 + Q 2 ) / 2 we get from Equation (4.23) and Equation (4.24)
C V d(p1 − p 2 )
Q L = D m ω + C il + el (p1 − p 2 ) + 0 (4.26)
2 2β F dt
Vt
Q L (s) = D m ω(s) + C tl p L (s) + s p L (s) (4.27)
4β F
Non-linear friction terms may also be present, but is neglected in the linear analysis.
The equations (4.22), (4.27), and (4.28) define the valve-motor combination. Normally,
the valve dynamics is much faster than the actuator and load dynamics, so it can be
neglected. Combining the above describing equations then yields
kq k qpt
x v (s) − 1 + Vt s TL (s)
Dm D 2m 4β F k qpt
ω(s) = (4.29)
Vt J L 2 k qpt J L B V B m k qpt
s + + m t2 s + 1 +
4β F D m
2 2
Dm 4β F D m D 2m
Where k qpt = k qp + C tl is the total flow pressure coefficient. Equation (4.29) describe
the dynamics of the valve-motor combination – the output being the motor shaft
velocity, due to the two inputs, valve spool position and the load torque. The term
k qpt / D 2m is usually much smaller than B m , meaning that the term B m k qpt / D 2m is small
compared to unity and can therefore be neglected.
Splitting up the transfer function in Equation (4.29) due to the two inputs we get
k qpt
kq
− 1 + Vt s
2
Dm D m 4β F k qpt ⋅ T (s)
ω(s) = ⋅ x v (s) and ω(s) = L
(4.30)
s2
2ς s 2 2ς
+ s +1 + s +1
ω 2n ω n ω 2n ω n
4β F D 2m k qpt βFJ L B Vt
ωn = and ς= + m (4.31)
Vt J L Dm Vt 4D m βFJ L
The response due to the valve input is often used in analysis and design and knowledge
of the involved parameters is essential.
It is worth mentioning that the flow-gain directly affects the open loop gain of the
system, and that the flow-pressure coefficient directly affects the damping. The
response of the system is determined by the natural frequency, which is dominated by
the inertia load and the trapped oil springs between valve and motor.
The approach describing the dynamics of the valve-cylinder combination shown to the
left in Figure 4.2 is the same as for the valve-motor combination.
Applying the continuity equation (3.5) to each of the piston chambers yields
V1 dp1
Q1 = A p x& + C tl (p1 − p 2 ) + (4.32)
β F dt
V2 dp 2
Q 2 = A p x& + C tl (p1 − p 2 ) − (4.33)
β F dt
where A p is the piston area, and C tl is the total leakage coefficient. The two volumes
V1 and V2 includes the volumes in the valve, connecting line and the cylinder
chambers. Actually the volumes is not constant but varies with the piston position as
where V10 and V20 are the initial volumes in the cylinder chambers. However, it will be
assumed that the piston is centred in the analysis. This is ok from a stability point of
view, while it was shown in section 3.5 that the minimum natural frequency is when the
piston is centred. As in the previous section we define the total volume Vt under
compression
Vt = V1 + V2 = V10 + V20 4-35
Vt
Q L (s) = A p x& (s) + C tl p L (s) + s p L (s) 4-36
4β F
Using Newton’s second law to the piston and load, the resulting force equation, Laplace
transformed is
where B p is a viscous damping coefficient of piston and load, and FL an arbitrary load
force. Equations (4.21), (4.36), and (4.37) are the basic equations describing the
dynamics of the valve-cylinder arrangement. Combining these equations we get
kq k qpt
1 + Vt s FL (s)
x v (s) −
Ap A 2p 4β F k qpt
x& (s) = 4-38
Vt M L 2 k qpt M L B p Vt B p k qpt
s + 1 +
s + +
4β F A p
2 A 2
4β A 2 A 2p
p F p
Where k qpt = k qp + C tl is the total flow pressure coefficient. Equation (4.38) describe
the dynamics of the valve-motor combination – and gives the response of the piston,
due to the two inputs, valve spool position and the load force.
The term k qpt / A 2p is usually much smaller than B p , meaning that the term B m k qpt / A 2p
is small compared to unity and can therefore be neglected.
Splitting up the transfer function in Equation (4.38) due to the two inputs we get
kq k qpt
− 1 + Vt s
2
Ap A p 4β F k qpt ⋅ F (s)
ω(s) = ⋅ x v (s) and ω(s) = L
4-39
s2
2ς s 2 2ς
+ s +1 + s +1
ωn ωn
2
ω 2n ωn
4β F A 2p k qpt βFM L Bp Vt
ωn = and ς= + 4-40
Vt M L Ap Vt 4A p βFM L
Comparing the transfer functions with those found for the valve-motor combination
they are very much alike – the displacement for the motor being replaced with the
piston area, and the inertia load replaced by the mass of the load.
The same conclusions can then be drawn, and care must be taken, especially
considering the natural frequency and damping ratio that varies a lot with different
operating points.
----- oo0oo -----
5
Advanced System and
Component Analysis
5.1 Introduction....................................................................... 1
Fluid Power Systems
(Hydraulisk Komponentanalyse) 5.2 Hydraulic Load Holding Circuit...............……….……... 1
Introduction • Mathematical model for the circuit • Transfer function
of the mathematical model • Stability analysis
5.1 Introduction
The object in this chapter is to relate the material presented in the previous chapters to
more in depth analysis of more complex components and systems.
The following studies are presented:
° Stability analysis of hydraulic load holding circuits. From this, stability criteria can
be obtained which can give guidelines on how the over centre valve should be
dimensioned to ensure stability.
A variety of modelling and control techniques are therefore presented by these studies.
5.2.1 Introduction
The purpose of this chapter is twofold. First of all it is to define parameters and
determine the transfer function of a basic hydraulic load holding circuit with varying
load. This analysis yields a description of the dynamic performance, knowledge of
which is absolutely essential in the rational design of load holding circuits. Second the
purpose is to illustrate a method used in analysing such systems, giving an
understanding of the problem.
The most important asset of a hydraulic system is stability, and therefore stability
should be based on hard quantities. Quantities, that can be easily identified, and
determined with fair precision and whose values remain relatively constant. For a more
in depth analysis performed, the performance index will more or less depend on soft
quantities, making stability and performance judgement more difficult, and computer
simulation becomes necessary. For example, the flow-pressure coefficients of orifices,
and most damping ratios are soft quantities.
The differential equations that describe hydraulic components and systems are non-
linear and, in most cases, of high order. Preliminary dynamic analysis is therefore
necessarily restricted to linear differential equations, because only they may be solved
without great difficulty. Also because general performance indices have only been
developed for linear systems, makes linear analysis perhaps the only tool.
The hydraulic circuit studied in this section is given by the schematic design in Figure
5.1. Note that the system is quite general. Nothing is said about the components in the
system, only how they are connected.
Ap Ar
Cylinder ratio (CR) = Ap/Ar
Pressure (Pb) Pressure (Pf )
Flow (Qf )
Volume (Vf)
Ao Ac Pressure compensated
T P flow control valve
Pilot ratio (PR) = Ac/Ao
LS
This kind of system is non-linear because of the flow-pressure relation for the orifice in
the over centre valve and friction both in the over centre valve and in the cylinder.
Though a linear analysis will not be especially helpful in designing a specific system, it
can still give guidelines to the behaviour caused by modifications of different
parameters of the system. Stability problems will obviously be present even in
completely friction-free systems.
The over centre valve illustrated in Figure 5.1 uses a spring-loaded poppet element to
control the flow of fluid from the load line to the tank line. When the plunger is moved
to the left by the application of the pressures Pf and Pr , the meter out and return port
are connected. The actual displacement of the plunger element, which cause the flow
restriction is usually of such small value relative to port diameter that the pressure/flow
equations obey the Bernoulli form:
2
Q r = C d wx 0 (Pr − Pb ) (5.1)
ρ
where
Qr return flow
x0 spool displacement from neutral
ρ density of oil
Cd flow coefficient
w area gradient
Q r = K q x 0 + K qp Pr (5.2)
∂Q r 2 ∂Q r C d wx 0
with K q = = Cd w Pr ; K qp = =
∂x 0 x ρ ∂Pr x 2ρPr
0 , Pr 0 , Pr
where
Kq valve flow gain
K qp valve flow-pressure coefficient
x 0 , Pr referring to operating conditions
Looking at the cylinder chambers, it is assumed that the pressure in each chamber is
everywhere the same and does not saturate or cavitate, line phenomena are absent, and
temperature and density are constant. Also, internal and external leakage of the piston
are neglected. Applying the continuity equation to each of the piston chambers yields
dVf Vf dPf
Qf = + (5.3)
dt β F dt
dVr Vr dPr
− Qr = + (5.4)
dt β F dt
where
Vf volume of forward chamber
(includes valve, connecting line and piston rod volume)
Vr volume of return chamber
(includes valve, connecting line, and piston volume)
βF effective bulk modulus of system
(includes oil, entrapped air, and mechanical compliance of chambers)
Vf = Vof + A r x p (5.5)
Vr = Vor − A p x p (5.6)
where
Ar area of piston at the rod side.
AP area of piston.
xP displacement of cylinder piston.
Vof initial volume of forward chamber.
Vor initial volume of return chamber.
Substituting Equation (5.5) and Equation (5.6) into Equation (5.3) and Equation (5.4),
we have
It is desirable and possible to express the continuity equations in a more useful form.
The displacement flow of the piston is defined by Q P (= A r ⋅ dx P / dt ) . The last terms
in Equation (5.7) and Equation (5.8) may be neglected by assuming that Vof >> A r x P ,
and Vor >> AP x P . By introducing the cylinder ratio CR = A P / A r , and the fluid
capacitances C f = Vof / β F , C r = Vor / β F , equations (5.7) and (5.8) can now be Laplace
transformed to give
C f sPf = Q f − Q P (5.9)
C r sPr = CR Q P − Q r (5.10)
The final equations arise by applying Newton's second law to the forces on the cylinder
piston and spool in the over centre valve. In considering the cylinder piston and the over
centre valve spool, friction forces and viscous damping is omitted. The resulting force
equation for the piston, Laplace transformed, is
M t s 2 x P = A r (Pf − CR Pr ) + FL (5.11)
where
Mt total mass of piston and load referred to piston.
FL arbitrary load force on piston.
Considering the over centre valve, and assuming that only the forces produced by
pressures, flow (the flow force is included in the spring constant), and spring are
meaningful terms of the static and dynamic behaviour of the valve, it is possible to
neglect the inertial forces of the spool and the friction force between the spool and
housing. Thus, the force equation describing the over centre valve takes the following
form
Pf A c + Pr A o = K o x o (5.12)
where
Ac area applied to pilot pressure.
Ao area applied to load pressure.
xo spool displacement.
Ko spring gradient.
Equations (5.2), (5.9), (5.1), (5.11), and (5.12) define the system given in Figure 5.1.
In this section the equations defined in the previous section 5.2.2. will be combined to
obtain physically interpretable results and an over-all transfer function. We seek the
cylinder response to a flow input. While the cylinder velocity is proportional to the
displacement flow Q P , the transfer function of interest is the one describing the relation
between Q f and Q P . Combining Equation (5.2) and Equation (5.12) gives
KqAo KqAc
Q r = + K qp Pr + Pf (5.13)
Ko Ko
Q r = (K qo + K qp )Pr + PR K qo Pf (5.14)
Usually force loads are omitted in system designs so that only the transfer function from
input is of interest. Using Equation (5.11), written as M t s ⋅ (sx P ) = A r (Pf − CR Pr ) ,
means that the displacement flow can be written as
A 2r
Q P = A r sx P = (Pf − CR Pr ) (5.15)
M ts
1 1 A 2r
H f (s) = (5.16) H r (s) = (5.17) G M (s) = (5.18)
Cf s Crs M ts
and inserting Equation (5.16) and Equation (5.17) into Equation (5.9) and Equation
(5.10) the basic equations, describing the dynamics of the system in Figure 5.1, are
For purpose of reducing the equations, they can conveniently be represented in a block
diagram, as shown in Figure 5.2.
Qf Pf + QP
Hf (s) GM (s)
+
CR
Kqo PR
Pr
Kqo+Kqp Hr (s)
+ + Qr +
CR
Closing the inner loop (and thereby defining H rq (s) ) gives the transfer function
1
H rq (s) =
1 (5.23)
+ K qo + K qp
H r (s)
By using H rq (s) , and closing the two remaining inner loops, the block diagram in
Figure 5.2 can further be reduced to the one shown in Figure 5.3. Block 1 consist of an
integrator and a gain, depending on the capacitance C f . Block 2 contains the feed
forward of Pf through the pilot line, and Block 3 includes the load, cylinder and over
centre valve, and the volume between the two.
Qf 1 QP
Hf (s) 1+ Kqo PR CR Hrq (s)
1
+ + CR 2 Hrq (s)
GM (s)
Go (s)
To allow for an analytical study of the system, we have to write out the terms in the
three blocks, shown in Figure 5.3. The transfer function of the first block is simply
G o1 (s) = H f (s) , and is given in Equation (5.16). Inserting Equation (5.17) into Equation
(5.23), H rq (s) can be written as
1
H rq (s) = (5.24)
C r s + K qo + K qp
with H rq (s) given as in Equation (5.24) the transfer function of block 2 can be
expressed as
1 C r s + K qo (1 + PR CR ) + K qp
G o 2 (s) = 1 + K qo PR CR = (5.25)
C r s + K qo + K qp C r s + K qo + K qp
1 + s / ω2 t
G o 2 (s) = K o 2 (5.26)
1 + s / ω2n
K qo (1 + PR CR ) + K qp K qo (1 + PR CR ) + K qp K qo + K qp
where K o 2 = ; ω2t = ; ω2n =
K qo + K qp Cr Cr
1 C r s + K qo + K qp
G o 3 (s) = =
Mt 1 M M (5.27)
s + CR 2 C r 2t s 2 + (K qo + K qp ) 2t s + CR 2
2
Ar C r s + K qo + K qp Ar Ar
1 + s / ω3 t
G o 3 (s) = K o 3 (5.28)
s / ω + 2ζ / ω 3 n s + 1
2 2
3n
K qo + K qp K qo + K qp CR A r 1 K qo + K qp Mt
where K o 3 = 2
; ω3t = ; ω3n = ; ζ=
CR Cr CrM t 2 CR A r Cr
Multiplying the three transfer functions G o1 (s) , G o 2 (s) , and G o 3 (s) , gives the open
loop gain function G o (s) . (notice that ω 2 n = ω3 t )
1 + s / ω2t
G o (s) = K (5.29)
s (s / ω32n + 2ζ / ω3n s + 1)
2
K qo (1 + PR CR ) + K qp β e K q A o (1 + PR CR ) + K o K qp β e
where K= =
C f CR 2 K o Vof CR 2
K qo (1 + PR CR ) + K qp β e K q A o (1 + PR CR ) + K o K qp β e
ω2t = =
Cr K o Vor
AP 1 K q A o + K o K qp βe M t
ω3n = ; ζ=
Vor 2 KoAP Vor
Mt
βe
From the mathematical description in Equation (5.29), stability and other performance
characteristics can be computed.
1 + s / ω2 t
1 + G o (s) = 1 + K =0 (5.30)
s(s / ω32n + 2ς / ω3n s + 1)
2
or
s3 2ς 2 K
+ s + (1 + )s + K = 0 (5.31)
ω3n ω3n
2
ω2t
Writing out the coefficients in Equation (5.31) using the terms defined in Equation
(5.29) we can write the characteristic equation in the general form
a 3s 3 + a 2 s 2 + a 1s + a 0 = 0 (5.32)
where a 3 = Cr Cf M t ; a 2 = C f M t (K qo + K qp )
a 1 = A 2r (C f CR 2 + C r ) ; a 0 = A 2r {K qo (1 + PR CR ) + K qp }
The system in Equation (5.32) is stable if all coefficients ( a 0 → a 3 ) are existing and
positive and if
a 2 a 1 − a 3a 0 > 0 (5.33)
Vf PR K qo
> (5.34)
Vr CR K qo + K qp
In many mobile system we often have high inertia loads combined with static loads.
This means that the pressure sensitivity of the over centre valve will be high and the
term in the brackets in Equation (5.34) will tend to unity. A lower pilot ratio will
increase stability, but from this simple stability result it is obvious that the stability
margin of the overall system will be very small as the hydraulic cylinder is a full stroke
component. The enclosed volumes Vf and Vr on the meter-in and the meter-out side,
respectively – influence the gain and the damping in the system. Hence, a big Vf and a
small Vr leads to increased stability, see Figure 5.4, and vice versa, see Figure 5.5.
250
Pr
200
Pressure [Bar]
150
Pf
100
50
0
0 1 2 3 4
Time [sec]
350
300
250
Pr
Pressure [Bar]
200
150
100
50 Pf
0
0 1 2 3 4
Time [sec]
Let us now make the stability analysis using a Bode diagram. The main advantage of
the Bode plot over other types of plots for frequency response is that the effects of
adding a pole or zero to a transfer function can be seen rather easily. For this reason,
Bode diagrams are very useful in designing control systems.
From the open loop gain function in Equation (5.29), the free s in the denominator
indicates an integration, so that the system is type 1 and has zero position error.
Variations in the gain constant K, the break frequency in the numerator ω 2 t , the
hydraulic natural frequency ω3n , and especially in the damping ratio ζ occur and cause
considerable shifting in the frequency response with different operating points. Hence
the frequency response is quite elastic and this must always be kept in mind during
design or in viewing test data. Though, the loop gain function is very complicated and
some simpler expressions still retaining information essential to stability, is desirable.
Examining Equation (5-29) it is possible to make a further reduction. First 2ω3n ζ can
be written as
βe
2ω 3 n ζ = (K q A o + K qp K o ) (5.35)
Vor K o
and
βe
ω2 t = (K q A o (1 + PR CR ) + K qp K o ) (5.36)
Vor K o
KqAo
(1 + PR CR ) + 1
K qp K o
ω 2 t = 2ζ ω3n (5.37)
KqAo
+1
K qp K o
In cranes the load is mostly high inertia loads combined with linear steady state loads.
This implies that the load pressure Pr can be high, sometimes close to the relief valve
setting of the over centre valve. Since the pressure drop over the valve will be high, the
opening of the valve must be very small if the load is to be moved slowly. This implies
that the pressure sensitivity K P will be very high, so that Equation (5.38) can be
approximated with the following expression
ω 2 t = 2ζ (1 + PR CR ) ω3n (5.39)
This equation indicates that ω 2 t will be equal or bigger that ω3n , even for a very low
damping ratio, and a small pilot ratio. The numerator in Equation (5.29) can then be
neglected because the phase from the quadratic factor decay much faster than the added
phase from the first order system – so the loop gain function, Equation (5.29), can be
approximated by.
K
G o (s) = (5.40)
s (s / ω + 2ζ / ω3n s + 1)
2 2
3n
Looking at Equation (5.40). From a Bode diagram (or Nyquist plot) it is possible to
determine closed loop stability. See Figure 5.6.
Go (jω)
-1 slope
ω 3n Log scale
1
1 ω rad/sec
K/2δ ω3n
K/ω 3n
-3 slope
If the resonant peak of the quadratic term rises above unity gain, then the system
becomes unstable because the critical point of the Nyquist diagram would be encircled.
From a Bode diagram the crossover frequency is approximately equal to K (velocity
constant). The gain level of the asymptotic diagram is K / ω3n at the frequency ω3n .
This level is amplified by the factor 1 / 2ζ , which is the amplification factor of the
quadratic, at resonance. Thus, the gain level at the resonant peak is K / 2ζω3n , and must
be less than unity for stability. Hence, the stability criterion is
K
<1 (5.41)
2ζω3n
Vor 1
K= ω2t (5.42)
Vof CR 2
and from Equation (5.39) an expression for 2ζω 3n can be found as a function of ω 2 t ,
meaning that Equation (5.41) can be expressed purely by hard quantities, as
Vor 1
(1 + CR PR ) < 1 (5.43)
Vof CR 2
For a given application - both quantities CR and PR are fixed by a cylinder and over
centre valve – Equation (5.43) then states that
where λ is some positive constant. From this simple stability result it is obvious that the
stability margin of the overall system will be very small as the total displacement of the
cylinder is used.
5.3.1 Introduction
The purpose of this section is to make a reduced order model of a pressure compensated
mobile proportional valve. Characterising a typical mobile machine, as a difference to
typical industrial applications, is that they are operated by a human being and that the
normally therefore do not have a well defined working cycle. The demand for
controllability and efficiency have led to an increasing use of load sensing (LS)
systems. Compared to conventional systems, LS systems are more efficient since flow
and pump pressure continuously are matched to the actual need, however an ideal LS-
system does not give any damping at all.
The valve considered in this section is a Sauer-Danfoss A/S a load-independent
proportional valve type PVG 32. A schematic drawing is shown in Figure 5.7.
PVG 32 is a hydraulic load sensing valve. The module system makes it possible to build
up a valve group to meet requirements precisely. In Figure 5.7. is shown a valve group
with 3 basic modules.
In Figure 5.8 is shown a sectional schematic drawing of a PVG with two basic modules.
The pressure adjustment spool is placed in the pump side module PVP (see Figure 5.7),
and the pressure adjustment spool and main spool is placed in the basic modules PVB.
When the pump is started and the main spools, in the individual basic modules are in the
neutral position, oil flows from the pump across the pressure adjustment spool to tank.
The oil flow led across the pressure adjustment spool determines the pump pressure.
When the main spools are in neutral this pressure is the stand-by pressure.
Main spool
Pressure compensator
RESERVOIR
When one or more of the main spools are actuated via the mechanical activation PVM
or the electrical activation PVE, the highest load pressure is sensed (load sensing!) and
fed through the shuttle valve circuit to the spring chamber behind the pressure
adjustment spool, and completely or partially closes the connection to tank. Pump
pressure is applied to the right-hand side of the pressure adjustment spool. This means
that the pump pressure is kept a certain value, defined by the preloaded spring, higher
that the highest load pressure.
In the basic modules the pressure compensator maintains a constant pressure drop
across the main spool – both when the load changes and when a neighbour module with
a higher load pressure is actuated. In other words – when several functions are activated
simultaneously the highest load pressure controls the pump pressure. Other modules
then get higher pressure drops since the difference between pump pressure and those
load pressures increases. In a non-pressure compensated valve, without a pressure
compensator, this will increase the flow to those loads which will cause interactions. In
a pressure compensated valve the increased pressure drop is compensated by the
In modelling the PVG 32 valve we will use the parameters defined in Figure 5.9. We
will only consider one basic module, while modelling other basic modules will be using
the same approach.
PLS
QV
A5 Q B1 MM
PK AH
A2
A4 AK
Pext MK V1
P1
A6 QB2
QV
V2 P2 P0 A1
Qext
PP
MP
AP
A3
P3 V3 XP
QP
QT
VP
RESERVOIR
Input to the valve model is a constant pump flow Q P , the opening area A H of the main
spool when actuated away from neutral by the electrical activation, the load pressure
p LS , an external load pressure p ext and external flow demand Q ext from other basic
modules. The output will be the flow Q V out of the valve.
The pump delivers the constant flow Q P . The flow that is not used by any of the
modules will be diverted back to tank by the spool area A 3 . The pump is looking into
the volume VP , which is composed of the volume in the hose and the volume in the
valve. Applying the continuity equation we get
dp P β F
= (Q P + Q B 2 − Q T − Q V − Q ext ) (5.45)
dt VP
The flow to tank Q T is a function of the area A 3 , which again is a function of the spool
position X P . Using the orifice equation gives
2
Q T = C d A 3 (X P ) (p P − p T ) (5.46)
ρ
Similar with the flow Q B 2 through the damping orifice with the area A 6 .
2
Q B2 = C d A 6 (max(p LS , p ext ) − p 3 ) (5.47)
ρ
Often the flow through damping orifices are small, meaning that the flow easily can be
laminar and not turbulent as when described by the orifice equation (2.34), or the flow
can shift between the laminar and turbulent regime. One way to solve this problem is to
make the discharge coefficient a function of Reynolds number C d = C d (Re) (see Figure
2.9). Applying Newton 2nd law to the pressure adjustment spool gives
d2XP
MP = (p 3 − p P )A P + FSP (X p ) + FFP (X P , p P , p T ) (5.48)
dt 2
FSP (X P ) is the spring force and include the preloaded spring force. FFP (X P , p P , p T ) is
the flow force trying to move the spool in the closing direction. From Equation (2.58) it
can be written as
To find the pressure p 3 in Equation (5.48) we apply the continuity equation to the
volume V3 .
dp 3 β F dX p
= Q B 2 − A p (5.50)
dt V3 dt
In the equations (5.45) and (5.50) the volumes VP and V3 are considered constant –
meaning that the displacement volume of the spool is negligible.
If it is assumed that the pressure build up in the volume V3 is very fast and the mass of
the spool is negligible the model can be reduced to second order. To summarise the
equations describing the dynamics of the pump side module then becomes:
dp P β F dX P Q B 2
= (Q P + Q B 2 − Q T − Q V − Q ext ) ; =
dt VP dt AP
2 2 (5.51)
Q T = C d A 3 (X P ) (p P − p T ) ; Q B2 = C d A 6 (max(p LS , p ext ) − p 3 )
ρ ρ
Looking at Figure 3.9 the flow Q V from the pump side module is entering the basic
module – passing through the compensator spool and the main spool and out to the
actuator. There is a small orifice in the compensator spool so that the pressure p 0 can
act on the right hand side of the spool. The pressure p 0 is a combination of the pump
pressure and the compensated/reduced pressure p K . The compensator spool will take a
position so that the pressure drop p K − p LS over the metering land A H will be constant.
The flow through the valve is therefore independent of the load pressure, giving a
consistent metering.
Applying the continuity equation to the spool end chamber volumes V1 and V2 , and
using the orifice equation for the damping orifices A 4 and A 5 gives:
dp1 β F dX K 2
= A K − Q A4 ; Q A4 = Cd A 4 ( p1 − p 0 ) (5.52)
dt V1 dt ρ
dp 2 β F dX K 2
= Q B1 − AK ; Q B1 = C d A 5 (p LS − p 2 ) (5.53)
dt V2 dt ρ
The volumes V1 and V2 is again assumed constant, neglecting the volume change
caused by spool movement.
Force equilibrium for the compensator spool becomes
d 2XK
MK = (p 2 − p1 )A K + FFK 2 (X K , p 0 , p K ) − FFK1 (X K , p P , p 0 ) + FSK (X K ) (5.54)
dt 2
FFK1 (X K , p p , p 0 ) and FFK 2 ( X K , p 0 , p K ) are the flow forces as a result of oil flowing
through the orifices A 1 and A 2 . Again from Equation (2.58) the flow forces can be
written as
As can be seen from the expressions for the flow forces we need to know the pressures
p 0 and p K . These can of cause be calculated by applying the continuity equation to the
volumes between the three orifices A 1 , A 2 , and A H . However, these volumes are very
small and the dynamics arising from the small volumes is so fast that it can be
neglected. We though still need an expression for the pressures in the volumes. A way
to cope with this problem is to consider the three orifices as hydraulic resistors in series.
We want to find the equivalent area as shown in Figure 5.10.
A1 A2 AH A eq
PP P0 PK PLS
QV QV
The flow through all the orifices is the same so we can write up the pressure drops as
functions of the areas and the flow.
2
2 ρ Q
Q V = C d A1 (p P − p 0 ) => p P − p 0 = V (5.57)
ρ 2 A 1C d
2
2 ρ Q
Q V = Cd A 2 (p 0 − p K ) => p 0 − p K = V (5.58)
ρ 2 A 2Cd
2
2 ρ QV
QV = Cd A H (p K − p LS ) => p K − p LS = (5.59)
ρ 2 A HCd
2
2 ρ QV
Q V = C d A eq (p P − p LS ) => p P − p LS = (5.60)
ρ 2 A eq C d
(p P − p1 ) + (p1 − p K ) + (p K − p LS ) = p P − p LS (5.61)
Using the right hand side of Equations (5.57) → (5.60) in Equation (5.61) yields
2 2
QV ρ 1 1 1 QV ρ 1
2 + 2 + 2 = 2
(5.62)
Cd 2 A1 A 2 A H Cd 2 A eq
or
1
2
A eq = (5.63)
1 / A + 1 / A 22 + 1 / A 2H
2
1
Now, combining Equation (5.57) and Equation (5.60), and using Equation (5.63) the
following result is obtained
2 2
1 Q ρ QV ρ
pP − p0 = 2 ⋅ V ; where = A eq
2
(p P − p LS )
A1 Cd 2 Cd 2
(5.64)
⇓
p P − p LS
pP − p0 =
A ⋅ (1 / A 12 + 1 / A 22 + 1 / A 2H )
2
1
In a similar manner it is possible to express the pressure drops in Equation (5.58) and
Equation (5.59) as functions of the areas and the total pressure drop over the basic
module
p P − p LS
p0 − pK = (5.65)
A ⋅ (1 / A 12 + 1 / A 22 + 1 / A 2H )
2
2
p P − p LS
p K − p LS = (5.66)
A ⋅ (1 / A 12 + 1 / A 22 + 1 / A 2H )
2
H
If we assume that the pressure build up in the volumes V1 and V2 are infinitely fast, the
mass of the spool is negligible, and the orifice A 5 dominates the damping of the spool
(meaning that there is no pressure drop over A 4 and p1 = p 0 ). Then the equations
describing the basic module can be represented as a first order system.
dX K Q B1 2
= ; where Q B1 = C d A 5 (p LS − p 2 )
dt AK ρ
p P − p LS p P − p LS
pP − p0 = ; p0 − pK = 2
A ⋅ (1 / A 1 + 1 / A 2 + 1 / A H )
2
1
2 2 2
A 2 ⋅ (1 / A 12 + 1 / A 22 + 1 / A 2H )
p P − p LS
p K − p LS = 2
A H ⋅ (1 / A 12 + 1 / A 22 + 1 / A 2H )
This rapport about modelling of pressure control valves, abbreviated PCV, concentrates
on that group of valves often indiscriminately named as relief valves, safety valves,
safety relief valves, overload protection valves, and so forth.
Typically the PCV is inserted into a system adjacent to a trifurcation, having one line
coming from a supply, another line feeding a load, and the third line conditionally
connecting to tank via the PCV, see Figure 5.11a. Let Q L 0 be the load flow at which the
pressure response PL from the load, according to some static load characteristic, see
Figure 5.11b, equals Pref . For Q S < Q L 0 the PCV is closed, the system pressure is not
controlled and equals the load pressure PL = PL (Q L ) .
QS , QV , QL Q L , Q V , PS
b) c) PS
a)
PL1
SUPPLY
QLO1 Q V1 QSO
PL2 Q V2
PS QLO
QS QLO2
QV PCV
PS t
Pref
Pref Q Q V , PS
d) e) O PS
QL
LOAD
TS
1
QV
δ
PS t
For Q S ≥ Q L 0 the system pressure PS should be kept constantly equal to Pref , whatever
the value of the valve flow Q V = Q S − Q L lower than the maximum rated flow of the
PCV.
This means, that the ideal PCV characteristics are the ones shown in Figures 5.11b and
5.11c. (The chosen graphical representation, using pressure as abscissa and flow as
ordinate, anticipates a later encountered advantage in the analytic modelling, as it
proves to be easier to express the PCV flow explicitly as a function of pressure while, in
reality, flow is considered the independent variable).
The static characteristic is ideally a vertical line. The ideal, dynamic performance is
characterised by an unaffected system pressure PS , despite the sudden change in valve
flow because of sudden changes in load characteristics e.g. from PL1 to PL 2 , see Figure
5.11b and Figure 5.11c. Virtually, the characteristics of a PCV are the ones shown in
Figure 5.11d and figure 5.11e.
The pressure build-up, having some value different from zero and equal to the inverted
slope δ of the static characteristic:
∂ PS
δ= (5.68)
∂Q V
The different static characteristic for increasing and decreasing valve flow (due to
friction effects). The overshoot ο in the system pressure PS in case of transient flow
steps. Settling-time TS in the transient response.
The magnitudes: δ , ο , and TS plus rated flow and pressure form a minimum set of
essential performance-figures of a PCV. However, the dynamic performance figures
may as well be expressed in frequency domain terms like resonant frequency ν r ,
M m − peak , bandwidth, etc.
The PCV is of the almost classical design as shown in Figure 5.12. The valve is
designed for a maximum rated flow of 200 l/min at a maximum pressure level of 250
bar.
QT
y
Q IT
Q
ST
PT
PS
QS PI
VS QI
AI VI
AS
Both the static and the dynamic characteristics for the valve also depend on the
connected hydraulic circuits. To obtain a simple situation the return line is simply
described by a return pressure Pr , and the upstream conduit is assumed to be a simple
fluid volume VS with negligible friction and mass force effects and with a stiffness
determined by the effective bulk modulus of the hydraulic fluid.
QI
QS QST QT
Furthermore, it shall be attempted to establish a detailed model of the valve, which can
be useful on component design level.
The first steps of the structural identification process suggest 17 basic equations
(Equation (5.69) to Equation (5.86)) by which the type of relief valve and its operation
is described. Some of the functional relationships, like in Equation (5.70), (5.76) etc.,
are given in a general form. Based on elementary fluid mechanics more specific
expressions are suggested such as equations (5.71), (5.79) etc.
Some of these relationships need experimental verification due to considerable
doubtfulness regarding the virtual flow conditions. Through experimental procedures
the relationships are determined as characteristics of “black-boxes”, representing
sectional phenomena of the valve. By experimental means the input-output relations,
often pressure-flow-area relations, are recorded. The finally accepted equations, appear
as those which combine reasonable simplicity and acceptable accuracy. In this way vital
system parameters are estimated and a total model is established. Finally, the total
model is checked through an overall experimental verification.
(Generally index P refers to parts of the pilot valve while index M refers to parts of the
main stage).
Q ci = Q in + Q di − Q out (5.69)
where
Q ci compression flow to the volume VI
Q in incoming flow = Q I
QI flow through the fixed throttle channel of the main spool
Q di main spool displacement flow
Q out outgoing flow = Q IT
Q I ≅ f I (PS − PI ) (5.70)
where f I is some flow-function. Assuming turbulent orifice flow the flow Q I can be
written as
Q I = C dI A O 2 / ρ (PS − PI ) (5.71)
where
C dI discharge coefficient
AO orifice area
ρ mass density of oil
Q I = k I (PS − PI ) (5.72)
dVI
Q di = = A I ⋅ x& (5.73)
dt
where
A I effective spool area facing the internal chamber
x position of the main spool
(= 0, when main-spool is seated)
VI − A I x &
Q ci = PI (5.74)
β
where
VI − A I x &
PI = Q I + A I x& − Q IT (5.75)
β
Equation (5.76) contain a ‘hard’ non-linearity. Further it is assumed that the return
pressure of the pilot valve is approximately equal to the return pressure PT , that the
upstream pressure of the pilot valve is approximately equal to the pressure PI , and that
the compressibility and mass effects in the up- and downstream channels of the pilot
valve can be neglected.
Q IT = C dP A P ( y) 2 / ρ(PI − PT ) (5.77)
By defining
k P = C dP πdsin(α) 2 / ρ (5.78)
Q IT = k P y PI − PT (5.79)
where
kP flow coefficient of the pilot valve
d pilot valve seat diameter
C dP discharge coefficient of the pilot valve
α half the conical angle of the poppet
Equation (5.79) is derived under the assumption that the ratio y / d is small, i.e. fore
relative small openings.
Equation (5.80) tacitly assumes, that coulomb friction forces on the poppet can be
neglected. Further, the first and the second term on the right hand side can be lumped
together into A P (PI − PT ) , where A P = A IP ≅ A TP .
The preloaded spring force on the poppet is equal to:
FP = y 0 k SP (5.81)
where
y0 initial compression of the pilot spring
The preset pressure level Pref of the valve is, by definition the value of ∆PP at which
the pilot valve is on the point to open. By definition y = 0 when the poppet is on the
seat. Consequently, P ref can be found from the steady-state solution to equation (5.80)
combined with equation (5.81) - thus Pref = y 0 k SP / A P . This magnitude is seen to be a
characteristic of the valve, while the opening pressure PO = Pref + PT is a characteristic
of both the valve and the actual operating situation of the return line.
When equation (5.80) to (5.82) are taken together, the force balance for the pilot poppet
becomes:
This simple form of the force balance equation can be used without complications if the
valve opens at e.g. t = 0 and remains open. However, a general validity of the equation
must provide for the phenomena occurring in the closure for the valve, that is when
y = 0 for t = t 0 ≥ 0. When this happens, an impact between the poppet and the seat
occurs, and y& and &y& may vary drastically, depending on the nature of this impact. For
the actual valve-configuration it is generally reasonable to assume a perfectly non-
elastic impact (with a coefficient of restitution equal to 0).
Q in = Q ds + Q cs + Q i + Q ST (5.84)
where
Q in = Q S = supply flow or load flow
Q ds main spool displacement flow
Q cs compression flow to the volume VS
Q ST flow through the main stage throttle
(from upstream to return line)
Consequently:
dVS
Q ds = = A S x& ( = Q di ) (5.85)
dt
VS + A S x & V
Q cs = PS ≅ S P& S (5.86)
β β
f ( x ,PS − PT ), x > 0
Q ST = ST (5.87)
0 , x = 0
where
f ST some flow function characterising the main flow path
Q ST = k S x PS − PT (5.88)
with
k S = C dS πd S sin( γ ) 2 / ρ (5.89)
where
kS flow coefficient of the main spool
dS main spool seat diameter
C dS discharge coefficient of the main spool
γ half the conical angle of the main spool
VS &
Q S = A S x& + PS + k S x PS − PT + k I PS − PI ; for x ≥ 0 (5.90)
β
FM = k M x 0 (5.92)
where
x0 initial compression of the spring
Combining Equation (5.91) to (3.93), the force balance equation of the main spool
becomes:
The 26 equations [(5.69) to (3.94)] form an overall dynamic model which, however,
must be considered a hypothesis because of approximations. An overall experimental
evaluation is required to verify the validity of the model.
geometry. Only in some special cases the flow and force characteristics can be predicted
analytically.
The various and complex valve geometry means that the above results may have a
rather limited use or may even be contradictory – hence a further analysis concerning
these characteristics may be needed.
It an be shown that poppet valves under certain conditions can be “bi-stable”, in that
either of two different stable flow patters could be induced under otherwise identical
conditions. this could give rise to differences and even a discontinuity in the flow
coefficient and the flow force. In Figure 5.15 these results are illustrated. also the
construction of the pilot poppet will have an effect of deflecting the jet radially and
thereby reducing the flow force influence. Thus in a simulation model the flow force
coefficient k fP and the discharge coefficient of the pilot valve C dP could be
implemented as functions of y according to Figure 5.15.
Cd A
B
B
Ff A
B
In Figure 5.16 is shown three different configuration of the main orifice. The first one
(Case 1) shows the standard configuration. In Case 2 the spool has no under cut and in
Case 3 the seat has a bigger fly cut. The jet patterns in Case 1 and Case 2 can be verified
by fluid dynamics simulation (CFD). In Case 3 the jet is assumed to follow the fact of
the poppet, and thereby giving a flow force corresponding to equation 5.93.
x x x
CASE 1 CASE 2 CASE 3
Flowforce Flowforce Flowforce
x x x
Representation of model I
Above analysis can be summarized into the following 5 fundamental equations.
VI − A I x &
PI = k I PS − PI + A I x& − Q IT (5.95)
β
Pilot flow Q IT :
Q IT = k P y PI − PT (5.96)
VS &
Q S = A S x& + PS + k S x PS − PT + k I PS − PI ; for x ≥ 0 (5.98)
β
Because the functions of the valve involves a number of non-linear effects, it has been
necessary to extend the linear block-diagram symbols with the special elements as
defined in Figure 5.17.
MULTI- e1 e0
e2 e0 =e1 e2
PLIER
DIVIDER e1 e0 e0 =e1 e2
e2
SQUARE e1 e0
ROOT e0= e1
The model shown in Figure 5.18 is particular suited for investigations of the ability of a
constant pressure valve to respond on flow disturbances.
It appears from the block-diagram that the function of the valve corresponds to that of a
controller with a fixed, but adjustable reference Pref . The output is the system pressure
PS , the flow Q S is the load, and the return pressure PT may be considered as a
disturbance. Regarding the performance of the valve, the most relevant feature is the
response in systems pressure PS in function of variations in the load flow Q S .
A compilation of all the parameters used in the model is shown in the Table 5.1 on the
following page. The way in which the individual parameters has been or should be
evaluated is indicated in the column to the right.
The need for partial experimental verification of the parameters as well the structural
identification of the valve will be discussed later.
2nd Edition
BP
_ + + +
Pref _ _ _ _ PS
+ 1 1 1 β 1
AP β kI _ _
+ _ MP s s + s VS s
kSP
+
kfP +
VI +
AI I-Volume
_
AS
PT _
kP A
+
PI
Fµ
AI BM Main Stage
x0 _
_ _ +
kM _ _
1 1
_ s
+ MM s
kM
AS k fM
kS
+
Fig. 5.18 Total block diagram of the two stage pressure control valve (Model I)
Chapter 5
Page 30 of 37
Chapter 5
It is to be expected, that the response speed of the pilot valve is superior to that of the
main stage spool both because of their intended relative functions and because of the
difference in the mechanical spring stiffness-to-mass ratio of the two elements.
However, also the flow-force component of the net spring stiffness should be taken into
account.
From Equation 5.97 it is seen, that the purely mechanical natural frequency ωnmP of the
pilot valve is
while the mechanical flow force natural frequency ωnfP , with a pressure difference
PT − PT equal to 100 bar becomes
while the mechanical-flow force natural frequency ω nfM , as derived from Equation
(5.9), becomes
with PS − PT = 150 bar. From these figures it can be found reasonable to establish a
second model without the pilot valve dynamics.
Through the next chapters only the reduced model II is considered. A block diagram of
this model is shown in Figure 5.19.
In this model equation (3.97), describing the pilot valve dynamics, is truncated, the
position y of the poppet becomes irrelevant, and the former used flow equation (5.79)
can be replaced with an expression for the static characteristic of the pilot valve. As
such a valve, being per se a single stage PCV, normally has rather linear characteristics,
the following linear relation is assumed:
VI +
AI I-Volume
_
AS
PT _
+
A
PI
Fµ
AI BM Main Stage
x0 _
_ _ +
kM _ _
1 1
_ s
+ MM s
kM
AS k fM
kS
+
Fig 5.19 Simplified block diagram of a two stage pressure control valve (Model II)
Page 32 of 37
Chapter 5
Chapter 5
Qpilot [l/min]
9
8
7
6
5
4
3
2
1
0
0 50 100 150 200
PI -PT[bar]
Reasonable PCV-performance requires, that PI does not exceed Pref drastically, i.e.
PI − Pref ≤ 10 bar should be a conservative estimate. Hence only the range of the curves
below approximately Qpilot < 1 l/min is relevant to the virtual operation of the total
valve, and for this range linear approximations according to equation (5.104) with
C P = 0.075 (l/min)/bar apply fairly well.
Assuming, for simplicity, that the return pressure PT is negligible and selecting a
specific operating point (PS0 ,Q S0 ) , the equations of the model, that is equations (5.95),
(5.104), (5.98), and (5.99), can be linearised and combined to yield the following three
linear equations:
VI − A I x
p& I = α1 (p S − p I ) + A I x& − C P p I (5.105)
β
VS
q S = A S x& + p& S + α 2 x + α 3 p S + α1 (p S − p I ) (5.106)
β
∂Q I
where α1 = (5.108a);
∂ (PS − PI ) 0
∂Q ST ∂Q ST
α2 = (5.108b); α3 = (5.108c);
∂x 0 ∂PS 0
∂FfM ∂FfM
α4 = (5.108d); α5 = (5.108e)
∂x 0 ∂PS 0
From computer simulation with model I, it has been found that spool position x 0 and
inner pressure PI 0 referring to the operating point (PS0 , Q S0 ) = (200 l/min, 242bar) are
approximately x 0 = 0.56 mm , PI 0 = 168 bar . Next VS is chosen equal to 2000 cm 3 as
a representative value for many practical systems. Then α1 to α 5 can be evaluated.
QS
1 + 1 PS
+ +
α1 VI
s + ( α1 + CP ) kI _
VS
s + ( α1 + α 3 )
+ β β
AI
AI s A S + α2
_
1
+ M M s2 + B M s + ( k M + α 4 )
A S _ α5
After considerable calculations it can be shown that the transfer function p S (s) / q S (s)
has the following form:
p S (s) a 3 s 3 + a 2 s 2 + a 1s + a 0
= (5.109)
q S (s) b 4 s 4 + b 3 s 3 + b 2 s 2 + b1s + b 0
where each of the coefficients a i and b i are determined by rather complex equations.
To manage the numerical calculations it is easiest to derive the transfer function via a
state space model. In state space notation the linear model becomes:
BM kM + α4 AI AS − α5
− − − & 0
&x& M M MM MM MM
x
x& 1 0 0 0 x 0
= βA I β(C P + α1 ) βα 1 ⋅ + 0 ⋅ qS (5.110)
p& I V 0 −
VI VI pI β
I
p& S A Sβ α 2β αβ β(α1 + α 3 ) p S V
− − − 1 − S
VS VS VS VS
After having calculated the elements of the system matrix and the distribution matrix,
which obviously is much easier than to calculate the coefficients of equation (5.109),
the factorised form of the transfer function was found to be:
s 2 2ζ
G (1 + τ a s) + b s + 1
PS (s) ω b ωb
= (5.111)
Q S (s) s 2 2ζ s 2 2ζ
+ c
s + 1 + d s + 1
ωc ωc ωd ωd
where
bar
G = 5.7 ⋅ 10 − 2 (= δ = pressure build up)
l/min
s rad
τ a = 0.0132 ⇒ ωa = 75.7 ⇒ ωa = 12 Hz
rad s
rad
ω b = 11304 ⇒ ν b = 1799Hz
s
ζ b = 0.068
rad
ωc = 251 ⇒ ν c = 40Hz
s
ζ c = 0.343
rad
ωd = 11309 ⇒ ν d = 1800Hz
s
ζ d = 0.065
Neglecting the static gain, a Bode-plot of Equation (5.111) is shown in Figure 5.22.
It appears from the figures above that the two second order systems indicated by b and d
in the numerator and denominator, respectively, are very much alike numerically.
However, a reduction is, unfortunately, not generally applicable.
Gain dB
0
Nonlinear model
-20
Linearized model
-40
-60 0 1 2 3 4
10 10 10 10 10
Phase deg Frequency (rad/sec)
30
Linear model
0
-30
Nonlinear model
-60
-90
0 1 2 3 4
10 10 10 10 10
Frequency (rad/sec)
To check the accuracy of the linear model against that of model I, block diagram Figure
5.18, a curve derived from computer simulations with model I has been drawn on
Figure 5.22 together with the linear frequency transfer function. The two curves exhibit
a reasonable agreement. They describe the ability of an open, constant pressure valve to
respond on flow disturbances. The high-frequency asymptote of the frequency
characteristics are -20 dB/dec and -90 degrees, respectively, because, at higher
frequencies the system is essentially an integrator, integrating the input flow Q S in the
volume VS to yield the output-pressure PS .
VI s rad
τI = ≅ 9.0 ⋅ 10 − 4 ⇒ ω I = 1110 ⇒ ν I = 176 Hz
β( α 1 + C P ) rad s
VS s rad
τI = ≅ 0.029 ⇒ ω I = 35 ⇒ ν I = 5.5 Hz
β( α 1 + α 3 ) rad s
The main-spool has a mechanical natural frequency which earlier (Equation (5.102))
has been found to be:
ν nmM ≅ 28 Hz
Taking into account the flow forces on the main spool, the natural frequency will
increase with increasing pressure, while the damping ration will decrease.
βA 2I rad
ω nhm = = 10892 ⇒ ν nhm = 1733 Hz
M m VI s
Comparing these figures with the numerical values of the parameters of the transfer
function, Equation (5.111), it seems reasonable to conclude, that the time constant τ a of
the numerator stems from the s-volume, and that the two numerically identical second
order terms, appearing both in the numerator and the denominator are identical with the
mechanical-hydraulic second order term of the main spool. The other second order term
of the denominator might correspond with the mechanical natural frequency and
damping ratio of the main spool.
The resonance peak, characterized by a resonance frequency ν r and the maximum
amplitude ratio M m , is a characteristic feature of the dynamic performance of the PCV.
Mm νr [Hz]
[dB]
Pref = 150 bar
20 80
Qs = 200 l/min
70
15 60
Mm
50
10 40
30 νr
5 20
10
Vs
0
0 1000 2000 3000 4000
[cm3 ]
Mm νr [Hz] Mm νr [Hz]
[dB] [dB]
20 60 20 60 Mm
Mm
15 45 15 45
νr
νr
10 30 10 30
P Qs
ref
0 0
0 37.5 75 112.5 150 0 50 100 150 200
[bar] [l/min]
For the specific operating situation referred to in Figure 5.22 it appears, that flow rate
oscillations in the 30-40 Hz range will result in pressure oscillations, amplified by
M m ≈ 15 dB or a factor of 5, compared with the value determined by the static pressure
build-up coefficient solely. Based on computer simulations, the curves in Figure 5.23
gives a survey over the variations of ν r and M m with some of the most important
operating-state parameters like the upstream volume VS , the preset reference pressure
level Pref , and the amplitude q s 0 of the sinusoidal input flow rate.
----- oo0oo -----
A
Flow Force
Compensation
A1 Introduction...................................................................... 1
Fluid Power Systems
(Hydraulisk Komponentanalyse) A2 Flow Force Compensation
3
Methods.........………….……
Radial hole orifices • Angled hole orifices • Pressure drop
AaU ~ Forår 2003 compensation • Recirculation lands • Profiled spool lands • Scalloping
techniques
A1 Introduction
A0
A2
1 2 3
Since most orifice flows occur at high Reynolds numbers, this region is of greater
importance. The flow between points 1 and 2 in Figure A1 is potential flow and
experience justifies the use of Bernoulli’s equation in this region. Applying Bernoulli’s
equation between points 1 and 2 gives the well known equation
2
Q = Cd A0 ( p1 − p 2 ) (A.1)
ρ
where C d is called the discharge coefficient. Although the analysis leading to Equation
(A.1) is not valid at low Reynolds numbers, the equation can be extended to the laminar
region by plotting discharge coefficient as a function of Reynolds number.
Another important factor in control valves is the axial force necessary to operate the
spool. This force has three principal components: inertial force, friction force, and a
usually far larger force which is produced by the flow through the metering orifices.
With the assumption that the flow is two-dimensional, irrotational, non-viscous, and
incompressible, the solution for this last mentioned force has been found by von Mises,
for the configuration shown in Figure A2.
Valve body
b θ
a
e d c
f i
Spool
g h
x
Fig. A2 Square-Land Configuration
The axial force on the piston equals the axial component of the net rate of efflux of
momentum through the boundary a-b-c-d-c-f-g-h-i-a, where a-b is the vena contracta of
the jet. In an actual valve the area of a-b is much smaller than that of d-e where the fluid
enters the upstream chamber. Since the velocities are inversely proportional to the
areas, the influx of momentum through d-e is negligibly small compared with the efflux
at a-b, which is equal to ρQV , where Q is the total rate of flow, V the velocity of jet
at vena contracta, and ρ the density of fluid. The net axial force can then be written as
By use of Bernoulli’s equation, Equation (A.2) can be transformed into the more useful
form
F = 2C d wx∆p cos(θ) (A.3)
where w is the peripheral width of orifice, x the axial length of orifice, and ∆p the
pressure difference between the upstream and downstream chamber.
Equation (A.1) and Equation (A.3) are the most used equations today when estimating
the flow through valves and the actuation force, but the effect of geometrical deviations
from the ideal ones in Figure A1 and Figure A2 is difficult to calculate theoretically. In
relation to the global variables some general methods though have been developed to
influence the flow-force and avoid cavitation, as discussed in the next sections.
Over the past forty years, a considerable amount of effort has been spent on creating
and developing compensation methods. Consequently an extensive range of techniques
have been proposed and documented, some of which are more effective than others.
However, the majority of methods tend to use one of three general principles to achieve
compensation, as identified by Backé. These are as follows:
1. Influence the jet angle such that all mass flows entering and exiting the spool
chamber control volume are as perpendicular as possible to the valve axis. This
minimises the net change in momentum of the fluid in an axial direction and
hence the reactionary axial forces acting on the valve spool.
2. Generate a pressure drop within the valve to counteract the reactive force of the
flow.
3. Balance the axial momentum components of the flow entering and exiting the
spool chamber control volume. This generally requires alteration of the influx
and/or efflux jet angles and can involve partial reversal of the flow by means of
flow recirculation within the valve chamber.
This method utilises holes drilled radially into the valve sleeve to meter the flow from
the valve chamber, as shown in the upper Figure A3. Another configuration is to drill
the holes in a hollow spool. The holes drilled in the sleeve is placed in diametrically
opposite positions. This ensures that there is no lateral force on the spool, i.e. the
momentum components ρQVsinθ cancel out. The method can be applied equally
effectively to both meter-in and meter-out configurations. The flow-force characteristic
of a single hole was investigated experimentally by Clark and is shown in the lower
Figure A3. Initially as the holes are being uncovered the jet angle is 69 deg, but as the
opening increases the angle increases until at full opening it becomes 90 deg. At this
point the net axial rate of efflux is zero and there is no flow force. However, Clark
points out that the abrupt change in the force gradient can lead to stability problems.
The problems with the abrupt change of the force gradient and the limited compensation
of the overall flow force can be eliminated/improved by replacing the individual holes
by a series of smaller holes, as illustrated in the upper Figure A4. Arranging the holes
such that the overlap, s, between successive holes is held to 0.18 ⋅ d , an approximately
linear area/displacement characteristic is created with a corresponding flow force, as
shown in the lower Figure A4. The choice of hole sizes and distribution is, of course,
dependent upon the metering characteristics required and the permissible pressure
losses.
Flow-force
e e
ific ific
Or Or
a ted ate
d
s s
en le en
mp l ho mp
co dia co
Un h ra Un
Wit Total force
Force from each hole
0 20 40 60 80 100 Flow
Percent Diameter Orifice Open
Flow forces exist where there is a net rate of efflux of momentum. If the inlet is angled
as shown in Figure A5, an axial component of inlet momentum will exist and this will
reduce the net rate of efflux momentum. The flow force is then represented by the
following formula:
where α is the inlet angle, and θ the exit angle, both relative to the spool axis. To
obtain maximum advantage from this method the inlet angle α must be as small as
possible and the inlet velocity V1 must be as high as possible. These requirements
create problems, of which the inlet velocity is the most significant since it implies
increased pressure loss at the inlet, and controlling dimensions, which determine the
inlet and outlet velocities, become critical. The inlet design conditions must be carefully
determined, the implication here being that for maximum advantage and performance,
spools will have to be designed for limited flow and pressure range.
If the first hole in the stagged radial hole compensation method, shown in Fig. A.6, is
angled, a further reduction in flow-force magnitude can be realised. The efflux jet
impinge upon the spool land, generating an opening reacting force. The qualitative
effect of this configuration is shown in the lower Fig. A.6, and tends to reduce the initial
“hump” in the flow-force curve.
V1
Flow-force
e ce
ific ifi
d Or Or
ate d
ns s ate
pe en
co
m
Increasing α mp
Un nco
U
Radial hole compensation
Flow Flow
In the analysis of momentum effects frictional forces in the valve chamber were
neglected. However, if the annular space between spool and sleeve is small, velocity in
the valve chamber will be high and a friction loss, proportional to the square of the fluid
velocity in the valve chamber, will occur. Referring to Figure A7 this will result in the
pressure in the inlet side being greater than that of the exhaust side.
V v e
val
ed
Flow-force
a t
p e ns
c om
un decreasing spool
shank diameter
Exhaust
Flow
This gives an opening force which opposes the closing force. The smaller the area
between spool and sleeve, the greater the opening force generated. However, this cannot
be carried too far since it causes choking of the flow. Typical flow-force curves for this
type of valve is shown to the right in Figure A7. Pressure drops associated with the
thick spool method are large. Furthermore, perfect flow-force compensation cannot be
achieved using this scheme as the steady-state flow-force varies linearily with the flow-
rate, while the force used to compensate it varies with the square of the flow-rate. For
low capacity valves the pressure loss will be a linear law and the transition to the square
law may be unpredictable.
Exhaust Curved
exhaust Exhaust
chamber h
α3
Inlet
x
f θ1
θ3
θ2
a
e α2
g b
β α1 c
β = 70 0
Increasing radial
clearance
Flow-force
Flow-force
Uncompensated valve
β = 50
0
β = 30
Flow Displacement x
Another compensation method, presented as negative force port, were proposed by Lee
and Blackburn.
A schematic representation of the complete valve chamber profile used is shown in the
upper Figure A9. The stem is shaped to look like a turbine bucket and the recess in the
sleeve makes the down-stream chamber larger. The smooth profile ensures minimal
loss in the momentum of the fluid in its passage through the valve chamber. The jet
enters at angle θ1 and leaves at angle θ 2 and hence if the cross section of the flow
remains constant, the closing force would be
Obvious, if cos(θ 2 ) > cos(θ1 ) , i.e. θ 2 < θ1 , the force will tend to open the valve. In
addition to this an eddy is formed in the chamber causing an additional flow to enter the
chamber and angle θ3 . This also tends to open the valve. The presence of the eddy
makes the total force almost impossible to predict analytically. Experimental results,
shown qualitatively in the lower Figure A9, verifies the predicted negative
characteristic but also reveals the presence of small positive “hump” at low flows, the
magnitude dependent on valve radial clearance, and the metering edge radius.
Unfortunately there is no method of calculating the dimensions of the various features
illustrated in the figure.
They made an experimentally study of the effect of the geometric parameters, shown in
Figure A10, on the magnitude and trend of the flow force characteristic. From their
work the following results were drawn:
w1
lve
R R va
ed
Flow-force
β2 β1 te
n sa
pe
m
β4 β3 n co
U
R R
D1 d
B
D
d1 w2
Flow
The presence of blend radii, R, have a beneficial effect upon compensation. Rounded
corners reduced the magnitude of the flow force with 35 %.
For the range 69 0 ≤ β 1 ≤ 90 0 , compensation is not significantly affected.
Chamber width, ω 2 , made less than sleeve chamber width, ω 1 , has a negative effect
upon compensation. Reduction in D1 improves the compensation effect, but may lead to
instability if carried to far. Increasing d 1 gives small improvement in compensation.
No clear guidelines were given on the influence of the other parameters. Flow force
characteristic for the most effective and stable choice of parameters is shown to the
right in Figure A10. The flow force is approximately reduced to 28 % of that measured
on an equivalent sized conventional square-edged valve.
A2.5 Profiled spool lands
This arrangement differs from those previously considered in that the metering
characteristic is generated between the spool and sleeve edge, as shown in upper Figure
A11. The throttling zone is formed tangentially between the two components and
intrudes into the spool chamber. Under meter-in conditions, as well as the flow force
trying to shut the valve there exist an additional force component, acting in an opening
directing. This is a result of a pressure imbalance on the spool. The portion of the spool
land from a-b is located in the sleeve chamber and so is subjected to approximately
supply pressure conditions. The downstream land, d-e, is exposed merely to the return
pressure. The portion, b-c, is acted upon by a pressure which is a function of the flow
velocity. These pressures combine to generate a compensating force. This compensation
technique was studied by Lang and Fassbender, who produced a single metering edge
device with a profile similar to that discussed. The effectiveness of the compensating
force present in the meter-in configuration is demonstrated in lower Figure A11. The
meter-out characteristics represent purely the reactionary force to the flow as no
compensation force is generated under this condition.
A modification of this method (Lang and Fassbender) is illustrated in the upper Figure
A12. In addition to the flow-force and pressure force described above, flow between the
closely adjacent spool land and sleeve gives rise to a friction force acting on both
components. With a meter-in configuration, this force acts to open the valve.
Sleeve Sleeve
a
e
b
d c Spool Spool
Meter-out
Flow-force
Flow-force
Meter-out
Meter-in
Meter-in
0
Flow Flow
Fig. A11 Profiled spool land Fig. A12 Profiled spool land
Utilisation of friction and pressure forces allows the flow force to be balanced more
evenly, since they do not depend on the same variables. Experimental results for the
profile, shown in the lower Figure A12 demonstrate effective compensation.
Scalloping refers to the technique of cutting notches into the spool shoulder used for
metering the flow, as shown in Figure A13.
As partial, rather than full annulus opening is provided, all metered flow is guided
through the scallop. By careful design of the scallop geometry, the direction of the
metered jet can be controlled and the resultant flow-force characteristic influenced. In
addition, scallops give the designer control over the flow versus spool displacement
characteristic and increase valve stroke which reduces the criticality of alignment
between the spool to sleeve metering edges. Notching the spool shoulder also gives rise
to slower metering effect (i.e. a reduced valve gain) and provides a smoother action for
the valve. An added consequence is that valve leakage around the null position is
greatly reduced, due to the greater overlap of the spool in the sleeve.
Exhaust Exhaust
Inlet Inlet
Inlet Exhaust
γ
γ=78
0
Flow-force
γ=71
γ=30
Flow
As illustrated in the right Figure A18, the flow exit angle, γ , had a significant effect on
flow-force if made less than 70°.
----- oo 0 oo -----