Anda di halaman 1dari 25

SP-1

DESIGNING
QUIETER HYDRAULIC
SYSTEMSSOME RECENT DEVELOPMENTS
AND CONTRIBUTIONS

Kevin
Centre

for Power
Department

EDGE

Transmission
of Mechanical
University
Bath,

BA2

and Motion
Engineering

Control

of Bath
7AY,

UK

(e-mail:K.A.Edge@bath.ac.uk)

ABSTRACT

This paper reviews the considerable


progress that has been made over the last 25 years in the understanding
of noise in
hydraulic systems.
Sources of fluid borne noise, structure borne noise and the resultant airborne are discussed in detail
and methods of measuring,
analysing
and predicting
behaviour
are presented.
This leads on to a discussion
of
component
and system design issues. The paper concludes
by identifying
areas where further work is required.

KEYWORDS
Airborne

noise, Fluid

borne noise,

Structure

NOMENCLATURE
B:
F:

bulk modulus
frequency-dependent

Fe:

piston force exerted on swashplate

coefficient

V:

volume

of fluid

x:

distance

along

Zo:

characteristic

Zs:

source

Fe,sum: total piston force exerted on swashplate


H:
frequency-dependent coefficient
j:
imaginary operatorl
: length of transmission line
Me:

borne noise, Design

in passageway
pipe

from

source

impedance

of

the

pipe

impedance

swashplate

angle

propagation

coefficient

: frequency

1.

INTRODUCTION

moment exerted on swashplate by single piston


The

Me , sum: total moment on swashplate


P:
pressure ripple at a distance x from the source
P1, P2: pressure ripple at ends of a transmission line
Q: flow
ripple at a distance x from the source
Q1, Q2: flow ripple at ends of a transmission line
Qs:
source flow ripple

Fluid Power. Forth JHPS International

problem

systems
for

long

heating

high

well

loss,
lead

The

health

to

impaired

safety

levels

generated

Exposure

can

worker
and

noise

known.

periods

also

Symposium (c)

of

is

result

to
in

performance
fatigue
issues

and

hydraulic

noise

levels

clinically-significant
of

workers

associated

relating

1999 JHPS. ISBN 4-931070-04-3

by

high

to

and

may

carelessness.
noise

have

been

recognised for many years and legislation is now


placing clear demands on manufacturers to reduce noise
levels. In Europe, for example, noise abatement is
principally covered by the Machinery Directive [1] and
the Noise Directive [2]. The Machinery Directive
requires that "Machinery must be so designed and
constructed that risk resulting from the emission of airborne noise are reduced to the lowest level taking
account of technical progress and availability of means
of reducing noise, in particular at source". Despite the
introduction of legislation and the growing demands of
customers for quieter machines, noise considerations are
rarely an integral part of design process. Yet, in the
context of hydraulic machinery, the vibration of
components and piping systems is not only the principal
cause if air-borne noise but can also be responsible for
the loosening of pipe fittings (and hence leakage),
fatigue and reduced reliability.
It should not be
surprising, therefore, to see system designers consider
and indeed, adopt, an alternative form of power
transmission medium.
In 1976 a major project aimed at reducing hydraulic
system noise was undertaken in the United Kingdom,
over a four year period, by four contractors: The
National Engineering Laboratory; BHRA Fluid
Engineering (now the BHR Group); the University of
Bath; and the University of Aston. Each contractor
focussed on a number of particular aspects of noise
generation and transmission. The results of these
studies were reported in two Seminars organised by the
UK Institution of Mechanical Engineers [3, 4]. At the
time these reports were restricted to conference
delegates only and as a consequence were not generally
available to the world-wide research community.
Nonetheless, the work reported brought together a
number of related issues and provided. considerable
impetus to subsequent work in the field.
The design of quieter systems can only be tackled
through a proper appreciation of the mechanisms
involved in the generation and transmission of noise.
As this paper will show, through a review of previous
work, the issue is a complex one and this has
undoubtedly made it difficult to achieve effective
knowledge transfer from researchers to practising
engineers. Firstly, it is essential to have an appreciation
of the three types of noise present in all hydraulic
circuits and the inter-relation between them: fluid borne
noise (FBN), structure borne noise (SBN) and airborne
noise (ABN).
The term FBN describes the unsteady component of the
pressure and flow in the liquid in the circuit, sometimes
termed 'hydroacoustic' energy. The principal source of
pressure and flow pulsations (or "ripples") is the
hydraulic pump, but motors and oscillating parts of
control valves are also potentially important sources.
All designs of positive displacement pump create an
unsteady flow which is superimposed on the mean flow

rate. For operation at a constant speed, the unsteady


ow ripple is periodic in nature. This flow fl ripple
interacts with the fluid in the attached circuit to create a
pressure fluctuation
which propagates through the fluid
at the local acoustic velocity. At any point where there
is change in pipe cross-sectional area, a pipe-junction or
component a partial reflection occurs, leading to a wave
travelling back to the pump where another partial
reflection occurs. The result of the multiple travelling
waves arising from multiple reflections is a complex
standing wave in the fluid
in the pipe work;
the
pressure ripple waveform at one location in the circuit
can be quite different to the waveform at another
location.
These pressure ripples create fluctuating
forces on the pipes which lead to vibration of the
pipework, attached components and the supporting
structure. The motion of these, in turn, creates ABN.
Unsteady flow
drawn in by motors creates identical
phenomena; the system can also be excited in a similar
manner as a consequence of the fluctuating motion of
components in valves.
The pumping dynamics of most designs of positive
displacement pump result in structural excitation of the
pump casing and as a consequence, the supporting
structure and attached pipework. This SBN can also be
a significant source of ABN.
The generation and
transmission process for pumps is summarised in Fig. 1.
Identical behaviour occurs in motors and thus is another
potential contributor to overall system noise.
ABN is also created as a consequence of turbulent flow
in valves and cavitation in pumps, motors and valves.

Figure

1 Noise transmission
paths from the pump
and prime mover

This paper provides a detailed review of world-wide


research on hydraulic system noise with particular
reference to the production of quieter systems. Despite
the strong inter-relations between FBN, SBN and ABN,
it is convenient to consider these individually and this
forms the structure of this paper.
2. FLUID BORNE NOISE
Any discussion on fluid borne noise generation must
start with the source, which is usually taken to be the
pump

(but

can

also

be

a motor

or

some

other

component). Sources may be (partially) characterised in


terms of their flow
ripple.
The nature of wave
propagation in most high pressure hydraulic circuits is
such that linear behaviour can be assumed and this
means that pressure ripple levels anywhere in the circuit
are directly proportional to the source flow
ripple.
Hence, from the pump design point of view it is clearly
essential that the pumping mechanism produces as low a
level of flow ripple as possible. From the system design
viewpoint, the designer should aim to select a low flow
ripple pump.
2.1 Generation

the best compromise


over the required range of
operating conditions. It is known that a well-designed
relief groove, machined in the portplate ahead of the
main kidney slot, can also improve the flow ripple
characteristics:
a good discussion on relief groove
design issues has been presented by Palmberg [6]. The
design of swash-plate type pumps for low fluid borne
noise has been assisted by the use of simulation
software for the prediction of pumping dynamics.
Development of such software has been reported by,
inter alios, Foster and Hannan [7], Kelsey et al [81,
Edge and Darling [9], Palmberg [6] and Harris [10].
Most axial piston pump studies have been concerned
with mineral oil as the working fluid, although Kelsey et
al [8] examined experimentally the effect of a variety of
uids on pump FBN and found that significantfl
differences were encountered when using a high water
content fluid. They concluded that the differences were
sufficiently large that they would prevent the design of a
"universal pump" for both mineral oil and high -water-

of FBN

This section will review FBN issues arising from the


design of pumps, motors and valves. Greatest emphasis
will be given to pump design.
The highest levels of FBN are generally created by
piston pumps and the lowest levels by screw pumps and
internal gear pumps. In between these lies external gear
and vane pumps.
Methods of establishing the flow
ripple characteristics of pumps experimentally will be
discussed later.

A detailed explanation of the behaviour is given by,


inter alios, Edge & Liu [5]. The dynamic component is
usually the dominant of the two and is highly dependent
on the timing of the portplate, the delivery pressure,
speed and, in the case of a variable capacity pump, the
fraction of maximum displacement. In an ideally-timed
pump, the portplate is designed so that the fluid is
precompressed to the delivery pressure before each
piston opens to the delivery port. Unfortunately, it is
only possible to achieve this with a conventional

content fluids.
With the recent re-emergence of water
hydraulics, this is an important conclusion.
Most of the modelling and simulation developments
relating to swashplate type pumps should be applicable
to bent axis pumps and be readily extended to the design
and prediction of performance of motors (although it is
notable that much less work on motors is reported in the
literature).
Various novel methods of modifying piston pumps to
reduce FBN have been reported. A good summary of
these techniques and their limitations is provided by
Harrison. [11]. Most of the proposals are unlikely to be
employed in practice because of the high cost or the
impact on efficiency or life. The pre-compression filter
volume described by Pettersson et al [12] from
Linkoping University, Sweden is an elegantly simple
solution to the problem of FBN reduction.
In the
authors' design, a small chamber (the "filter volume") is
located such that each cylinder communicates with the
chamber before it opens to the delivery port. The fluid
in the chamber is at a higher pressure than that in the
cylinder.
Hence the flow compressing the cylinder
contents is provided by the chamber rather than the
delivery line. As a consequence when the cylinder
opens to the delivery port the difference between
cylinder and delivery pressures is much smaller than
would otherwise be the case and the consequent
"reverse flow"
is significantly
diminished
.
The
chamber has to be "recharged" from the delivery line
before the arrival of the next cylinder. However, this
can be done in a controlled manner over a sufficiently
long period to provide a smoother flow
ripple. The
authors report a valuable reduction in flow ripple over a
range of pressures and speeds.
However,
their
investigations were confined to a fixed displacement

portplate at one particular operating condition, and


hence it is necessary to produce a design which obtains

pump and it is not clear how effective the scheme would


be for variable capacity machines

2.1.1 Piston pumps


A typical swashplate-type
piston pump flow
ripple
characteristic is shown schematically in Fig. 2. The
waveform is principally comprised of a kinematic
component,(corresponding
to the summation of the
ows created by the motion of the individual pistons)
fl
and a dynamic component (arising from the effect of the
compressibility of the fluid).

Figure

2 An example

of axial piston pump flow

ripple

Another

equally-simple

delayed

delivery

valve,

Fig.

port

concept
in combination

involves
with

the

use

of

groove".
Harrison
range of operating

a non-return

3.

opening

of

normal.

the delivery

With

this

damped

valves

design

is given

FBN

and

Figure

and Edge
conditions

ABN

found
that to cover a wide
it is beneficial
to delay the

port

to a point

arrangement,

are required.
in Fig.

4.

two

much
or

An example

three

later

than

heavily-

of a two-valve

Valuable

reductions

non-return

valves

in both

are reported.

4 Heavily
in axial

damped
piston

pump

mounted

portplate

Plunger-type piston pumps employ poppet valves, rather


than sliding plate valves and in most designs these are
simply pressure-actuated
rather than mechanically
driven. Pressure-actuated
valves open and close
according to the differential pressure present.
A
delivery valve, for example, will not open until the
cylinder contents have been compared to a pressure
level slightly above the delivery pressure (as a
consequence of the valve return-spring preload). Hence,
unlike swash-plate type pumps, near-perfect timing can
be achieved.
This tends to result in a much lower
dynamic flow
ripple component
and hence the
kinematic component is usually dominant.
However,
because plunger-type pumps are commonly designed
with one, three or five cylinders the kinematic
component is large, relative to the mean flow
rate.
Reduction of the flow ripple of plunger pumps has
received very little attention in the literature, although
Johnston et al [17] demonstrated that optimising the
stiffness of valve return springs can help to reduce flow
ripple. Changes to return-spring stiffness can also result
in a direct reduction in ABN as a consequence of the
reduced impact velocity of a valve as it reseats on
closure.
Plunger pump system pressure ripple is
normally dealt with through the use of FBN attenuators

Figure 3 Axial piston pump portplate with delayed


delivery port opening combined
with a non-return valve

When the motion of the piston has raised the cylinder


pressure to a level equal to the delivery pressure, the
non-return valve opens allowing communication with
the delivery port.
In ideal circumstances this
arrangement achieves perfect pre-compression of the
cylinder contents regardless of speed and delivery
pressure. Helgestad et al [13] outlined this scheme and
stated that it was found in experimental studies to give a
considerable reduction in noise. However, no results
were presented in the paper. Grahl [14] also conducted
experiments using the same idea but concluded that
whilst noise reduction was effective at low delivery
pressures, it actually increased noise levels at pressures
over 100 bar. A solution which avoided moving parts
was devised by Weddfelt et al [15]. This essentially
involved replacing the non-return valve with a vortex
diode; unfortunately, the dynamic response of the diode
was found to be too slow for it to work effectively. The
non-return valve idea has recently been revisited by
Harrison and Edge [16] who have overcome many of
the limitations of the original scheme. They propose
essentially the same concept except that the non-return
valve is heavily damped. High damping is achieved by
means of a chamber behind the valve which
communicates to the delivery port via a laminar-flow
damping orifice.
As a consequence there is no
requirement for the valve to open and close at pumping
frequency; it simply acts as a variable geometry "relief

(silencers).
Such pumps are not widely used in fluid
power applications although they are adopted for use
with water and high-water-content fluids
(for example
in the steel industry and in mining) and in some
automotive systems [18, for example].
Wobble plate pumps also use self-acting valves and
hence should have the same desirable characteristics of
near perfect timing as plunger pumps. The author is
not aware of any detailed FBN investigations on such

designs of pump.
However, experimental studies
conducted at Bath in 1984 [19] established that a
specific design of machine generated substantial levels
of FBN. This was considered to be a consequence of
highly oscillatory motion of the self-acting valves.
Recent studies on a single-piston plunger pump [18]
using computational fluid dynamics have demonstrated
that delivery valve instabilities can, indeed, occur and
that these are a significant source of noise.

Fortunately, the fundamental frequency component of


the flow ripple generated by an external gear pump is
dominant and is usually the most common source of air
borne noise. Nonetheless, it is possible that a system
which has a strong structural resonance at say the
second harmonic could be noisier with a split-gear
pump design than with a conventional design.
Relief of the volume of fluid trapped as the gears mesh
is usually achieved through grooves machined in the
side plates. From an FBN viewpoint the design of these
grooves is less critical than piston pump relief groove
design and has received much less attention in the
literature. Over 20 years ago Duke and Dransfield [30]
reported on a method of improving groove design with
the specific aim of reducing FBN.
A significant
reduction in FBN was reported. However, the nature of
the approach, which introduced an additional leakage

2.1.2 External Gear Pumps


Because of the relatively gradual increase in pressure of
the fluid in the inter-tooth spaces, gear pumps are not
subject to the same rapid pressure changes that occur in
swash plate pumps. As a consequence the flow ripple is
largely dictated by geometrical aspects of the design,
and there is relatively little dependency on mean
delivery pressure. A very comprehensive theoretical
assessment of fluid displacement characteristics of gear
pumps was undertaken by Yudin [20]. A typical flow
ripple waveform for an external gear pump is shown in
Fig. 5. The absence of backflow means that the
waveform is comprised of a small number of harmonic
components.

path,
must have reduced the volumetric efficiency
appreciably.
More recently, Ishibashi et al [25]
presented a theoretical basis for groove design focussed
on helical gear pumps (although it is stated that the
technique is suitable for other gear forms). The method
proposed is shown to improve volumetric efficiency but
also appears to have the merit of reducing flow ripple
(curiously, the authors make no claims about the
improved FBN characteristics of their design).
Another approach to the design of low FBN external
gear pumps is described by Edge & Lipscombe [31].
The authors propose the introduction of small radial
piston assembly into the pump in which the pistons are
driven by a cam mounted on the drive shaft, Fig. 6. The
cam is designed such that the motion of the pistons

Figure

5 An example

of external

gear

pump flow

produces a flow fluctuation


which cancels the pump
ow ripple.
The arrangement intrinsically fl and
simultaneously cancels the flow ripple at the pump inlet.

ripple

Methods of reducing gear pump flow


ripple include
increasing the number of teeth [21], modifying tooth
profile [22, 23, 24], improving relief grove design [25]
and eliminating backlash [21, 26]. The relative merits
of some of these solutions are discussed by Molton [21].
The "zero-backlash" solution was evaluated by Molton
[21] in 1986 but it is only very recently that it has been
possible to manufacture such designs cost-effectively
[27]. The use of a split gear arrangement is also an
effective method of reducing flow ripple [28, 29]. This
scheme essentially consists of two gear pumps which
are mounted on a common drive shaft such that the
uctuating components
of the two flows
are antifl
phased. In practice this can be constructed with both
sets of gears mounted in common housing. The solution
is elegant and effective but adds a significant cost
premium for a given swept volume.
Moreover an
analysis of the interactions between the two flow ripples
reveals that whilst near-perfect cancellation of the oddnumbered-harmonics
of flow ripple is achieved, the
even-numbered-harmonics
are doubled in magnitude.

Figure 6 Multi-lobe cam mechanism


for the
reduction of gear pump flow
ripple

Experimental studies demonstrated that the flow ripple


could be substantially reduced with a corresponding
reduction in system ABN. The authors also suggested
that such a scheme might be adopted for piston pumps
although it would be necessary to introduce a threedimensional cam in order to account for the influence of
mean delivery pressure on flow ripple. The cost of
manufacture has probably been the biggest deterrent in
exploiting the solution.

significant contributor to the FBN level. Nagata and


co-workers establish that cam ring deformation can take
place under load and take account of this in their model
through the use of finite element modelling of the ring.
Both of these studies consider pressure, rather than
ow, fluctuations
in the assessment of their model
fl
which can introduce uncertainties
because of the
measurement of pump-system interaction (this issue will
be discussed at greater length in a later section). From a
design viewpoint, Nagata et al [38] concluded that FBN
can be reduced by reducing vane thickness and vanerotor clearance. This was not confirmed by Dickenson
et al [34], however, who reported that tests on a thin
vane pump produced larger flow ripple than a thickvane pump.

2.1.3 Internal Gear Pumps and Screw Pumps


Yudin [20] has shown that the internal gear pump has an
inherently low flow ripple in comparison to an external
gear pump. This was confirmed experimentally by
McCandlish et al [32]. Despite their superior FBN
performance, internal gear pumps are not widely
adopted largely as a consequence of their cost, larger
installation volume and generally lower maximum
pressure rating relative to external gear pumps. Similar
limitations apply to screw pumps. As a consequence of
the design of the screw pump, which creates a steady
transfer of fluid from the suction to the delivery it
should, ideally, be ripple-free. However, as Weddfelt
[33] has pointed out, the fluid is transported in the form
of closed volumes which are subjected to rapid pressure
changes.
This results in a dynamic flow ripple.
Weddfelt confirmed the existence of a flow ripple
experimentally and its strong dependence on mean
delivery pressure although the levels are undoubtedly
low in comparison to other types of pump.
2.1.4 Vane Pumps

Figure

The pumping dynamics of a vane pump are similar to a


piston pump, in that there is a strong dynamic
component as well as a kinematic component. The
dynamic component can be moderated through the use
of a well-designed relief groove but in addition there is
scope to design the cam ring geometry such that the
ow ripple is reduced. A theoretical and experimental
fl
investigation into vane pump FBN is reported by
Dickenson et al [34]. The authors developed a
mathematical model of the pumping dynamics which
included the radial motion of the vanes and accounting
for the possibility of vane bounce (which can occur if
the forces acting on a vane are insufficient to maintain
contact with the cam ring). A comparison of the
predicted and measured behaviour is presented in Fig. 7.
Explanations for differences between measurement and
prediction are given by Dickenson et al [34].
Investigations into the loss of contact between vanes
and the ring have been reported by Nishiumi and Maeda
[35] and Matten and Lang [36]. Other modelling work
is reported by Ueno et al [37] and Nagata et al [38].
The former reports investigations on a variable-capacity
pump and identifies vibration of the cam ring as a

7 An example

predicted

of measured

vane pump flow

and

ripple

Kojima et al [39] concluded, from an experimental


investigation, that vane pump relief grooves are only
effective in reducing the higher harmonics of FBN
(above the third harmonic). They also established the
presence of a low frequency component of pressure
ripple which was composed of harmonics of shaft
frequency. This was found to be a consequence of
uctuation in leakage with rotor angle. The authors
fl
state that such irregularities were not found in tests on a
piston pump and gear pump. There are, in fact,
harmonics of shaft frequency present in most pump
FBN waveforms, although these are generally much
weaker than the harmonics of pumping frequency. The
presence of fluctuations at shaft frequency and its
harmonics are usually ignored as their contribution to
ABN is considered to be minimal (except perhaps with
very high speed pumps).
The presence of vane bounce has been linked as a
significant contributor to pump ABN as a consequence
of the impact of the vane on the cam ring following loss
of contact [36]. Skaistis [40] refers to this as

"pounding" .

Skaistis also points out that a similar


mechanism can occur in axial piston pumps following
loss of contact between the piston slipper and the swash
plate at the start of the suction stroke. This is usually a
consequence of poor design or inadequate boost
conditions and should seldom be encountered in
practice.
2.1.5 Pump Suction Lines

harmonic frequencies the suction ripple was


substantially larger than the delivery ripple. One factor
contributing to this characteristic was the absence of a
relief groove at the start of the suction port. The
relatively high level of fluid aeration was a further
contributory factor.
At very high pump speeds it is difficult to avoid
cavitation even at high levels of boost pressure. Highspeed piston pumps, in particular, are prone to
cavitation with a consequent reduction in volumetric
efficiency, erosion damage and increased noise. The
latter problem arises not just from the familiar highfrequency cavitation noise (associated with bubble
collapse), but also as a consequence of the reduced fluid
bulk modulus in the cylinders leading to increased
dynamic flow ripple at the pump delivery. Harris et al
[43] attempted to deal with this problem by evaluating a
variation of the "Pre-compression Filter Volume" used
by Petterson et al [12]. In the original design the
objective was to reduce the magnitude of the reverse
ow into a cylinder as it opened to the delivery
fl line.
Harris and Edge proposed a "Pre-Expansion Volume"
which could be located at the start of the portplate
suction port, Fig. 9. The objective was to reduce the
magnitude of the sudden outflow from the cylinder into
the suction port at the point where the cylinder
communicates with the portplate. Simulation studies
predicted that this arrangement would help to reduce
transient cavitation. It would also help reduce suction
port flow ripple.

The potential contribution of the pump suction line to


overall system noise is often overlooked. In systems
where the pump is unboosted and a high-compliance
suction hose is employed, the potential for noise
generation is likely to be low. Problems are more likely
to be encountered when a pump is boosted or when a
rigid pipe or low-compliance hose is used for the
suction line. One of the earliest studies on suction line
behaviour was reported by Edge & Freitas [41]. The
authors established that a significant level of flow
fluctuation could be generated at the inlet port of both
gear and piston pumps, and that this was strongly
dependent on the inlet pressure, Fig. 8. At low inlet
pressures, the flow fluctuation was relatively small, but
an increase in the mean inlet pressure of just a few bar
was sufficient to create substantial flow fluctuations. It
was also found that the system ABN level increased
appreciably with the same small increase in inlet
pressure. The authors concluded that air released from
the fluid in the pump inlet passage tended to suppress
ow fluctuations. This behaviour was more pronounced
fl
at the lower inlet pressures. It was also concluded that
for low noise levels from the suction line, it is desirable
to keep the inlet pressure as low as possible, consistent
with avoidance of cavitation problems.

Figure 9 Pre-expansion volume (PEV) geometry for the


reduction of piston pump suction flow ripple. (After
[43] with permission of Research Studies Press)
Figure

8 Effect
pump

of mean
suction

inlet

port flow

pressure

on gear

2.1.6

ripple

Water Hydraulic

Pumps

The renewed interest in water hydraulics has led, in


recent years, to new designs of piston pump suitable for
use with raw water and sea water. Claims for low ABN
have been made [44] but the author is unaware of any
FBN investigations on new pump designs. Trostmann
[45] identifies the potential for water hammer problems
with "high transient
pressure
peaks, noise and
resonance in the system" but does not address the issue
of FBN.

More recently, work on suction flow ripple has been


undertaken by Jones et al [42]. The authors focussed on
the behaviour of a vane pump used in a vehicle power
steering system and examined the effects of various
parameters on performance.
They identified a strong
influence of the mean delivery pressure on suction flow
ripple and found that the magnitude of the suction flow
ripple was comparable to that of the delivery. Indeed, a
frequency domain analysis revealed that at some

2.1.7

2.1.8

Motors

Because most designs of hydraulic control valve have


free-moving internal parts it is possible for a valve to
become a source of FBN. Oscillation of internal parts
of valves can be self-excited, as discussed by Shorin and
Sverbilov [49] and Maeda [50] for example, or can be
created by the presence of another FBN generator in the
circuit. The nature of such oscillations tends to result in
a periodic excitation of the system. Various methods of
analysis have been presented for those self-excited
oscillations which arise from interactions between the
valve and the circuit.
The most common practical
solution to the problem is simply to use a valve of a
different design or to move the valve to a different point
in the circuit. Whilst crude, this is often effective.
The other main sources of valve noise are flow-induced.
Cavitation noise is broad-band and influences FBN,
SBN and hence ABN. Heron [51] points out that very
little of the cavitation sound energy passes back into the
high pressure supply line and largely propagates in
return lines. The main problem is generally the direct

The nature of motor flow ripple, in general, has received


limited attention in the literature. An early evaluation of
piston motor inlet flow ripple was reported by Tilley
and Butler [46]. They conducted tests on a machine
running firstly as a pump and subsequently as a motor
and demonstrated that a lower level of FBN was
obtained in the motoring mode. When a motor cylinder
opens to the inlet port, the volume of fluid in the
chamber is smaller than the volume when a pump
cylinder opens to the delivery port which has a direct
impact on the dynamic flow component. Kojima and
Shinada [47] developed a mathematical model of the
ow ripple of a motor and used it, in combination with
fl a
pump flow ripple model, to predict pressure levels in a
hydrostatic transmission. Good agreement was obtained
with experimental measurements.
Detailed experimental measurements of motor flow
ripple have recently been described by Ichiyanagi et al
[48]. The authors measured the flow ripple of a bent
axis motor over a range of pressures, speeds, and
displacements. A typical result is presented in Fig. 10.
It is clear from the results of the investigation that
motors are capable of generating pressure pulsation
levels similar in magnitude and frequency content to
those generated by pumps. Ichiyanagi et al [48] also
measured the flow ripple at the motor outlet port which
may lead to problems similar to those experienced with
pump suction lines. Indeed, in closed circuit hydrostatic
transmission systems, the return line is boosted and the
motor outlet flow ripple will undoubtedly make an
additional contribution to overall ABN levels. There is
also anecdotal evidence that motor return line dynamics
can have a deleterious impact on motoring dynamics,
particularly in low-speed high-torque machines. This is
an area which deserves further attention.

Figure

10 Example

of bent

axis

piston

motor flow

Valves

generation of SBN in the valve casing.


2.2

Transmission

of FBN

The transmission of FBN is a complex phenomenon and


has been the subject of considerable research effort. A
significant amount of the work published has addressed
the mathematical modelling of pressure and flow
transients in circuits. This involves the solution of the
wave equations, with the piping treated as a distributed
parameter system known as a transmission line.
Strictly, two-dimensional (radial, as well as axial)
distributions should be modelled although it is possible
to apply one-dimensional plane wave theory with
corrections to account for radial variations arising from
uid viscosity effects. A good overview of the issues
fl
involved and modelling approaches available is
provided by Stecki and Davis [52]. In the context of
FBN studies it is usually convenient to adopt a
frequency domain, rather than a time domain,
interpretation of behaviour.
This relies on the
uctuating components of pressure and flow to flbe
sufficiently small to allow non-linear effects to be
neglected. In such an approach the FBN source (usually
pump or motor flow ripple) is treated as a periodic
excitation signal which can be described in terms of its
Fourier components. Each Fourier component can be
studied in isolation and the total effect of all harmonic
components is established by superposition. In the case
of a pump the flow ripple can be characterised, in its
simplest form, by harmonics of pumping frequency
(corresponding to the product of the shaft speed and the
number of pumping elements). A good approximation
of the flow ripple waveform of a gear pump can
typically be obtained from around 5 harmonics. Piston

ripple

10

pumps and vane pumps usually require a larger number


of harmonics, ranging from 10 to 15 or more, depending
on the design, in order to obtain a good representation.
A more precise definition of a waveform in the
frequency
domain
necessitates
consideration
of
harmonics of the shaft frequency rather than pumping
frequency. This allows variations from pumping cycle
to pumping cycle, arising from manufacturing variations
for example, to he taken into account.
The application of plane wave theory to describe wave

At higher frequencies more complex behaviour is


observed, as shown in Fig. 11, for example.
The
passageway can behave like a uniform blocked-end
transmission line [54, 55] a helmholz resonator [55], or
a multiple resonator [48] depending on geometry.
Pumps with relatively low volumetric efficiency can
deviate
from capacitive
behaviour
at very low
frequencies
where resistance (leakage) effects can
become dominant.
It should be noted that the
representation of a pump in terms of a flow ripple in

propagation
in "rigid pipe" hydraulic circuits is
summarised by, inter alias, Edge and Tilley [53]. The
term "rigid" pipe is not strictly correct because due
allowance for pipewall elasticity can be taken into
account by appropriate adjustment of the fluid
bulk
modulus.
However, the terminology
conveniently
distinguishes
such circuits from those containing
exible pipe (hose).
fl
In order to design quieter systems an understanding of
the factors influencing the transmission of pressure and
ow ripple in circuits is essential.
Because of flthe
complex nature of the interactions between components,
even in very simple circuits, it is usually necessary to
use computer simulations to evaluate the impact of
design changes.
Edge and Tilley [53] demonstrate that in a simple

parallel with a source impedance, as used here, is


directly analogous to the Norton model used in
electrical network analysis. This is the most commonlyadopted approach. It is also possible however to use a
Thdvenin model comprised of a pressure source in
series with the same source impedance.

pump-pipe-loading valve circuit, the pressure and flow


ripples at any location in the pipe can be described by
the equations:
(1)

(2)
where F and H are frequency-dependent
complex
coefficients which are both directly proportional to the
pump (source) flow ripple but also are functions of the
source impedance, the characteristic impedance of the
pipe and the impedance of the termination (loading
valve).
Impedance, in an FBN context, is the complex ratio of
the pressure ripple to the flow ripple. For a restrictor
valve, for example,
the impedance
is closely
approximated
by the slope of the pressure-flow
characteristic
of the valve at the operating point
(although there may be some deviation from this at
higher frequencies).
The impedance
of other
components is generally more complex and will be
discussed later.
The source impedance of a pump
describes the nature of the fluid
mechanics in the
discharge
passageway.
At low frequencies
the
behaviour is principally capacitive and can be described
by the equation:

Figure 11 Experimentally-determined
source impedance
of an external gear pump (numerically-smoothed)
The impedance of many designs of pump remains
largely independent of operating conditions since the
principal factors affecting behaviour are the geometry
and fluid
properties.
However, variable capacity
machines with a servo-actuation system fed from the
delivery
passageway
may exhibit
variations
in
impedance with operating condition [48].
Motor
impedance is very similar in nature to pump impedance
and again depends primarily on the passageway
geometry [48, 54].
In order to gain an appreciation of the factors affecting
FBN consider the case of the pump-pipe-valve system
previously mentioned. For a given pump, length of line,
and load valve setting, pressure ripple levels can be
reduced by increasing pipe diameter.
However, the
impact of this on SBN and ABN is not necessarily
beneficial since any unbalanced axial forces are the
product of pressure ripple and pipe cross-sectional area.

(3)

11

A change in pump speed will create variations in the


harmonic frequencies which can move one or more
harmonic components away from a resonant condition
but can equally move others into a resonant condition.
The speed change will also change the mean flow rate.
Even if the load valve setting is adjusted to maintain the
same delivery pressure, there will be a change in the
termination impedance, with a further impact on
pressure ripple levels. For a given flow rate, a change
in the load valve setting will alter the delivery pressure
which in turn will affect flow ripple levels. In gear
pumps, this effect is generally small but for piston and
vane pumps the effect is substantial.
It is clear from the above example that the system
designer is presented with significant challenges in
analysing even the simplest of circuits. This serves to
emphasise the importance of developing software tools
to aid in the design of quieter systems and to assist in
the interpretation of the resultant predictions.
2.2.1

Prediction

of FBN

it is essential to have accurate information about the


characteristics
of the components
attached to the
transmission lines. Some progress has been made in
developing
mathematical
models
for
certain
components.
For example the flow ripple of axial
piston pumps can now be predicted with a high degree
of confidence [11, 60], and such computer models are
likely to be readily adapted to model axial piston
motors.
Good progress has also been made in the
development of vane pump models [34, 37, 39 for
example]. Even relatively simple models can be
effective for the flow ripple prediction of some gear
pumps [61]. However, a good source impedance model
has proved to be more elusive.
As previously explained, various structural models for
source impedance have been established. However, the
prediction of impedance using geometric data still
presents problems.
Firstly, it is difficult ab initio to
select the correct structural model.
Secondly, the
parameters required to give good agreement with
experimental results are often significantly different to
those anticipated from geometrical dimensions [54].
One possible explanation for these differences is the

in circuits

The approach to the prediction of pressure and flow


ripples summarised by equations (1) and (2) for a
simple circuit can be extended to handle more
complicated circuits as described by Tilley and Butler
[46]. for example.
However, a matrix formulation is
more generally preferred, both from the viewpoint of
conciseness
of
description
and
computational
convenience.
Transmission
lines can be described as two-port
components where the pressures and flows at each end
of the line are related by the expression:

presence of mechanical compliance, as well as fluid


compliance, which is not directly accounted for in the
model. At the present time, it appears necessary to use
experimentally-determined
source
impedance
characteristics in simulation models if high accuracy is
required over a wide range of frequencies.
The impedance of bladder-type accumulators can be
predicted with reasonably good accuracy (Fig. 12) from
a knowledge of the geometry and the pre-charge
pressure [62, 63]. Models of restrictor and flow control
valve impedance have been developed by Johnston and
Edge [64]. The authors found that these valves were
reasonably straightforward to model, although in the
case of restrictor valves the effect of vibration and
downstream conditions could introduce complications.

(4)

For
given

the

"rigid

pipe"

case,

the

matrix

coefficients

are

by:

(5)
(6)
Boundary conditions are defined according to the FBN
characteristics of the components which are attached to
the transmission
line.
These may be lumped
impedances,
such as a restrictor
valve or an
accumulator, or distributed impedances
such as a
silencer or a flexible hose.
Various software packages have been developed for the
prediction of pressure pulsation levels in circuits:
Johnston and Edge in the UK [56]; Lecerf in France
[57]; Kojima et al in Japan [58] and Wacker in Germany
[59], for example. In order to use any of these packages

Figure

12 Predicted
characteristics

12

and measured

impedance

for an accumulator

The use of computational fluid

dynamics

spiral-wound tubular inserts. These 'tuner hoses' or


'resonator hoses' are commonly adopted for use in

(CFD) to

predict valve impedance has been reported by Lau [65]


who obtained good agreement between predicted and
measured behaviour for a variety of different orifice

automotive power steering systems. Studies of these


special hoses have been reported, inter alios, by Drew
[74], and Kojima et al [75].
Taken overall, there still remains a significant amount
of research effort required in the area of impedance
modelling. At the present time, for a confident
prediction of FBN levels in all but the simplest of
circuits, it remains necessary to use component data
which has been established experimentally.

geometries. Relief valves in general have been found to


be difficult to model accurately [63].
Silencer modelling is quite well developed, probably as
a consequence of the technology also being used in
other application domains.
For most designs of
reactive-type silencer it is generally possible to apply
plane wave propagation theory to the wave motion
inside the device. High accuracy predictions have been
achieved for a wide range of designs by, inter alios,
Whitson [66]. Some interesting variations on classical
designs have recently been developed (Fig. 13) and
tested by Kojima and Ichiyanagi [67]. These can be
designed such that a number of discrete frequencies can
be substantially attenuated and are aimed at constant
pump-speed applications where the most troublesome
harmonics can be targeted.

2.3 Measurement of FBN Characteristics


2.3.1 Pumps
Considerable effort has been devoted to the
measurement of pump FBN characteristics. Much of
the early work was devoted to the development of test
methods which would provide a "figure of merit" for
comparative purposes. A useful review of some of the
early attempts is given by Bowns et al [76]. The
greatest difficulty in achieving the goal of a simple
"figure of merit" is that FBN behaviour in a real circuit
is a result of a complex interaction between the pump
and circuit characteristics. Nonetheless, it has been
recognised that circuit designers do need some measure
of performance which allows one pump to be compared
with another. Over the years various proposals have
come forward, including the use of a reflectionless
circuit as means of loading the pump [77, 78 for
example]. A popular solution is the adoption of the
"blocked acoustic pressure" as a suitable figure of merit .
This is, theoretically, the pressure ripple level that
would be generated at the outlet of pump which is
subject to a load of infinite impedance. An infinite
impedance is unrealisable in practice, but provided that
the load is sufficiently high (relative to the source
impedance at each frequency of interest) the
measurement error is negligible. The resultant pressure
ripple is a product of the source flow and source
impedance (Qs Zs) and hence is solely dependent on
pump characteristics alone. Moreover the test is
potentially simple to conduct as it requires only one
pressure transducer which is mounted at the pump
outlet. However, the design of a circuit which achieves
the objective of a sufficiently high load impedance has
proved
problematic.
al [76]involved
developed
the
"high impedance
pipe"Bowns
methodetwhich
the use

Figure 13 Example of variable resonance mode side


branch (reproduced from [67] with permission of
Research Studies Press)
Hose modelling is the subject of significant research
effort at the present time. Hoses can be used to reduce
FBN and also provide a means of reducing structural
transmission.
The level of FBN reduction that can be
achieved is dependent on hose construction and high
pressure hoses are usually too rigid to be beneficial
except at the higher frequencies or when very long
lengths are employed.
For lower pressure operation,
more expandable hoses can be employed and these can
be very effective in reducing FBN. Wave propagation
in hoses differs from that in rigid pipes in two main
respects: the low wall stiffness of a hose results in a
lower wave speed in the fluid;
in addition, waves
propagate in the hose wall and these are coupled to
waves in the fluid.
Various models have been
developed, some relatively simple [68] and others of
greater sophistication [69, 70, 71, 72 for example].
Although very good progress has been made, some of
the parametric data required can only be established
experimentally [73]. Some hoses are designed with a

of a long small-diameter pipe. Whilst this was found to


be effective for many designs of pump it was later found
that it was sometimes necessary to use a range of pipe
lengths, selected to cover subsets of the full range of
harmonics of interest.
This detracted from the
simplicity of the original scheme. This procedure was,
at one time, adopted as a British Standard but has
subsequently been superseded. A draft ISO Standard

13

has recently been developed [79] based on refinements


of a procedure described by Unruh [80]. This has the
merit of simplicity albeit at the loss of some accuracy
(-3dB/+ 1dB) compared with the high impedance pipe
method (-0.9dB/+0.8dB).
Although the blocked acoustic pressure is potentially
useful for comparative purposes, it is of no direct value
for prediction purposes. In addition, it is no direct value
to pump designers who are aiming to reduce flow ripple
(which is the principal system forcing function). These
objectives can only be addressed by development of
test procedures which allow the measurement of both
source flow ripple and source impedance. At the
present time, pump flow ripple cannot be accurately
measured directly (except, perhaps, using highly
specialist instrumentation under laboratory conditions).
It is notable, however, that good progress is being made
in the area of flow ripple sensor development. Esser
and Backe [81] and Murrenhoff and Ristic [82], for
example, have developed prototype devices which show
considerable promise - although the technology is
probably not yet sufficiently well-developed to be used
confidently as part of a general purpose test procedure.
In the absence of appropriate sensor technology flow
ripple has to be inferred from pressure ripple
measurements.
Various procedures have been
developed and employed by researchers but all involve
the measurement of pressure ripples at two or three
locations in specially-configured test circuits [33, 83,
84, 85, for example]. All procedures are based on a
model of the wave propagation behaviour in the test
circuit and essentially use the measured pressure ripple
levels data in the relevant model in order to infer the
forcing function (source flow ripple). Data analysis is
conducted in the frequency domain and results in a flow
ripple spectrum from which the time domain
characteristics can be synthesised. The relative merits
of the use of three transducers over two is still the
subject of some debate, although such a method overparameterises the problem. As a consequence it appears
to be less sensitive to measurement errors and
circumvents the problem of solving ill-conditioned
equations under some conditions of operation. It also
has the added benefit of allowing the acoustic velocity
in the fluid to be evaluated [86] from which the
effective bulk modulus of the fluid in the circuit can be
calculated. The bulk modulus has to be used in the data
analysis and it is generally preferable to use data
relating to the prevailing conditions rather than to rely
on manufacturer's data (if available) which relates to
idealised conditions. The main disadvantages of the
three-transducer method are the additional cost of the
third transducer (and associated signal conditioning) and
more complex data processing (although since this is
usually accomplished very rapidly using computer
software, it is of minor consequence). Whatever test
method is used, considerable care needs to be taken in

the processing of the test data. Some of the important


issues involved are discussed in a valuable paper by
Weddfelt [87].
An
ISO
Standard
for
establishing
the
FBN
characteristics of pumps (ISO 10761-1) has recently
been published [88]. This permits the use of either two
pressure transducers or three.
The three-transducer
approach (Fig. 14) is based on the "secondary source"
method described by Edge and Johnson [83], which has
been shown to be effective in determining
the
characteristics of many different type of pump [61].
Refinements
to the data reduction process of the
secondary source method which are not included in ISO
10767-1 have recently been published by Johnston and
Drew [55]. Individual measurement of source flow and
source impedance permits the issue of a figure of merit
for pump FBN to be re-examined in a more flexible
manner. Johnston and Longmore [89] provide a useful
assessment of a number of different alternatives and
conclude that a rating based on acoustic power has
much merit.

Figure

14 Circuit for the determination


of pump flow
ripple using the secondary source method
2.3.2

Motors

and

other

components

The secondary source method is readily extended to


allow the characteristics
of other components to be
measured. Flow ripple and impedance characteristics of
a hydraulic motor has been measured using a variation
of the approach for pumps by Ichiyanagi et al [48] and
the procedure is described in a draft ISO Standard [90].
The characteristics
of "passive" components
(those
which do not contain a source of hydro-acoustic energy)
can be completely described by their impedance. Single
port devices, such as accumulators and some types of
valve can again be measured using a two-transducer or
three-transducer
approach [57, 63, 64, for example].
Strictly, all valves are two-port devices but where the
valve flow is returned to the reservoir with minimal
back-pressure,
the downstream effects are generally
negligible and the valve can be treated as a one-port
device.
As previously mentioned, it is possible for self-excited
oscillations to occur inside a valve containing freemoving parts.
Measurement
of valve impedance
characteristics under such conditions has received very
little attention although some studies have been reported
by Lau [91]. He clearly established experimentally that

14

to those previously outline for pump FBN testing. Both


methods have been successfully applied in the testing of
a range of different components, including silencers [94,
95], filters
[95],
and
hoses
[75,
93]
and
hydromechanical components [96].
As previously mentioned, in vehicle power steering
systems it is common practice to employ hoses with a
exible "tuner" inserts [97] which essentially act as afl
silencer. The characteristics of these "components" can
again be established using the multi-pressure transducer
test methods.

there is a marked difference


between impedance
characteristics in the stable and unstable states.
Initial efforts on the characterisation
of two-port
components were focussed on silencers. It is common
practice to characterise silencers in terms of their
Characteristic
Transmission
Loss.
This relates to
acoustic power associated with the transmitted wave to
the acoustic power associated with the incident wave,
under relfectionless
conditions
downstream
of the
device. Such a rating provides a performance measure
which is simply a function of frequency.
However, it
can be misleading as it does not give a true indication of
behaviour in a real system: the limitations of the rating
are clearly identified by Bais and Met [92]. Once again
the issue is clouded because of component-system
interaction.
For a complete description of a two-port
component
it is necessary
to adopt a matrix
representation.
The impedance matrix formulation has
previously
been presented
in equation (4).
An
alternative is the transfer matrix formulation:

2.4 Attenuation

Most commercial designs of silencer work on the


principle of reflecting hydroacoustic energy back to the
source.
These are known as 'reactive' or 'passive'
types. As previously mentioned significant effort has
been devoted to the design and analysis of these devices
and in many circumstances they can be very effective in
reducing FBN.
However, there does seem to be
reluctance on the part of system designers to include a
silencer in their circuit designs.
This is largely
attributable to the high cost. Skaistis [40] suggests that
the attenuation characteristics of commercial devices are
much higher than is really necessary and that simpler
and cheaper devices could be constructed by machine
builders themselves. This is a debatable point: Skaistis
bases
his argument
on high transmission
loss
characteristics which, as previously explained, are not a
true measure of performance in real systems. Indeed,
this raises the question of how circuit designers should
select a silencer for a particular duty. Until silencer
manufacturers begin to present the performance of their
devices in the form of impedance matrices (which can
then be used in a FBN simulation of the circuit)
designers have only two choices:
use published
transmission
loss characteristics
or measure
the
impedance matrices themselves.
A further issue is
where to locate a silencer in the circuit. The widelyadopted approach is to locate the silencer as close as
possible to the source. This would appear to have been
established empirically
and is intuitively sensible.
However, no one appears to have published any
analytical studies on the effect of silencer location on
FBN and the consequent SBN and ABN. This is worthy
of further study.
More sophisticated methods of attenuating FBN adopt
active control techniques.
High-bandwidth
devices
have been reported by Yokota et al [98], Yamaoka et al
[99] and Brennan et al [100] amongst others.
The
techniques adopted are undoubtedly highly effective and
will have a role to play in specialist applications where
cost is a secondary issue (such as in submarines).
It is
doubtful, however, that active devices will be employed
in industrial or mobile applications unless the cost can
be reduced significantly.

(7)

Behaviour of two-port components under particular


boundary conditions can be calculated from this matrix
description. In the case of a silencer for example, it is
possible to evaluate
characteristic transmission loss
from either the transmission matrix or the impedance
matrix.
The
experimental
determination
of the
matrix
coefficients requires more complex procedures those
used for single-port components.
Johnston et al [93]
describe a test method which involves the measurement
of pressure ripples at three locations upstream of the
component under test and at a further three locations
downstream, Fig. 15.

Figure

15 Circuit

for the measurement


transfer matrix

of a component

Kojima and Edge [94] on the other hand employ two


transducers upstream and two downstream. Once again
the use of the larger number of transducers overparameterises

of FBN

the problem and has similar pros and cons

15

2.5

Computational
circuits
the

tools

[e.g.56,

opportunity

stage
available).

for
58,

to

examine

Steps
in

However,

to

there

is

of

provide

the

FBN

for

can

response

prediction

59]

data

still

FBN

levels

at

the

to

review

outcome

of

simulation

to

in
with

design

components

taken

much

levels

designer

individual

be
the

pump and prime mover [101] which can


additional source of SBN and associated
nature of the internal excitation is dependent
of pump and a good review of the basics
Skaistis [40].

issues

the

57,

(provided

design

Design

be

is

the

3.1.1 Piston Pumps

circuit
studies.

done

to

aid

In the case of swash plate pumps, the main sources of


excitation are the fluctuating forces and moments which
arise as a result of the internal changes in pressure:- the
contents of a cylinder are approximately at suction

this

process:

EIt

is

not

Low

simply

ABN

between

sufficient
is

the

FBN

to

focus

ultimate

and

ABN

is

solely

on

FBN.

target

and

the

complex

(as

discussed

of

prediction

pressure for half the pumping cycle and at delivery


pressure for the other half of the pumping cycle. The
summation of the contributions
of the individual
cylinders gives rise to a fluctuating force and moment
on the swash plate which, in turn, are transmitted to the
casing through the swash plate support.
In his
explanation of behaviour, Skaistis [40] assumes that the

link

later).
EThe visualisation
data

and

is difficult.

the

FBN

For

levels

need

locations

in

operating

parameter

levels.

the

How

over

a range

The

selection

largely
required

perhaps)
mind.

and
the

one
as

seems

to

than

in the

The

emergence
may

Cost

case

trade-offs

are

How

selection

of

the

time

3.

with

pump
although

and

after

it

is

is

be

possible

in

EQuietening
and
and

against

the

components

their

selection?

NOISE

deal

to

desig

difficult

with

separately
not

inserts).

here.

noise

to

low

construction

BORNE

important

play

balanced

involved

plate [102]. Typical time-domain characteristics are


presented in Fig. 16 which show the significance of
swash angle on behaviour.
The importance of pump
timing on the level of excitation was recognised by
Helgestad et al [103] who established that it is possible
to reduce noise by reducing the rate of pressure rise in
the cylinders. Wicke et al [102] have established that
the timing of the suction port also plays an important
role in noise generation potential.

(other

valve

judge.

cost/lower

SBN

of

tuner

notoriously

this

effort

motor)
it

to

be

issue

with

to

in
for

functionality.

a general
role

STRUCTURE

FBN
(and

as

motors,

a component

systems

higher

is
is

performance
of a valve,

the
of

hoses

can

FBN

and

basic

difficult

can

low

Guidance

addressing

CFD

noisy

expensive.

its

important

commissioning

As

achieve

E pumps

function

an

unacceptably

and

to

process.

of flexible

have

port-plate timing is ideal and that the compression and


decompression are instantaneous.
Whilst this aids the
explanation
of behaviour,
it over-simplifies
the
situation. In order to obtain a better assessment it is
necessary to predict the pressures in the cylinders over
the pumping cycle.
The instantaneous forces and
moments can then be synthesised from a knowledge of
the geometry of the machine and the manner in which
the cylinder pressure forces are applied to the swash

FBN

behaviour

process.

of

be

of

one

the

trade-off

than

a secondary

many

just

conditions?

selection

a by-product

at
in

change

designer

error

is

condition

change

are not designed


with
FBN
The
impedance
characteristics

FBN

tool

the

(other

example,
No

examined

generally

components

trial
aid

be
A

operating

of

to

operating

to

will

does

Components

given

system.

of

interpretation
one

also be an
ABN. The
on the type
is given by

the

from

lose

issues

circuit

sight

of

their

of

issues
inter-

relationships.

3.1

This

section

considered
of

will
apply

the

casing

consequence
occur
secondary

Pumps

focus

of
of

on

equally
a

the

inherently
consideration

and

pumps
well

since

to

pump
rapid

in

Motors

arises,
changes

the

the

motors.

the

pressure
mechanism.

coupling

to

be

excitation

primarily,
in

pumping
is

factors
The

between

as

which

Figure

16a

Predicted
for a range

the

16

axial

piston

of swashplate

pump

piston

angles

forces

axial piston and vane pumps. A good review of the


SBN generating mechanisms involved is given by
McCandlish and Linney [108]. These authors draw
attention, in particular, to: the non-uniform increase in
pressure in the tooth spaces as the fluid is transferred
from suction to delivery; the possibility of high
transient pressures generated in the meshing region;
cyclic loading of the pinion bearings and side (wear)
plates; and vibration of components in the running
clearances. The nature of the pressure rise in the tooth
spaces has been measured by Kojima and Shinada [109]
and Manco and Nervegna [110], and also predicted by
Manco and Nervegna [111].
3.1.4 Pump casing SBN
The ABN generated by the casing of a pump, its
mounting and attachments is dependent on the structural
vibration of the casing in response to excitation by the
pumping dynamics. Unsurprisingly both the stiffness
and damping characteristics of the casing play an
important role in the overall response.
Early
experimental studies on piston pumps are reported by
Helgestad et al [103] and Dunlop [114
The latter
study was undertaken in conjunction with one of the
rst Finite Element Method (FEM) studies of pump
fi
casing behaviour in order to predict natural frequencies
and mode shapes. Discrepancies between predicted and
measured behaviour were accounted for by the
difficulty in correctly specifying some of the boundary
conditions and also because of a sensitivity to bolt
tightness which could not be accounted for in the FEM
model. Improvements in the accuracy of FEM analysis
have been achieved in recent years with, in particular,
greater attention being given to mounting conditions
(e.g. Kojima et al [113]). With recent improvements in
FEM software coupled
with the continuing
enhancements in computer processing power, the
potential of this technique is now beginning to be
recognised by the industry. Nonetheless, the effort
involved in mesh generation can still be quite
substantial.
An approach which circumvents the need for FEM has
been described by Pettersson et at [107]. The authors
took a specific design of pump and measured the
acceleration of the pump casing, at particular locations,
in response to simulated swash plate forces and
moments. This allowed frequency-domain transfer
functions to be established and used to predict, amongst
other things, the effect of new port-plate designs on
vibration levels and ABN (Fig. 17). This is a valuable
contribution and has a useful role to play in the design
of lower noise machines. Indeed, the authors draw
attention to the scope for design optimisation.

Figure 16b Predicted axial piston pump piston


swashplate moment for a range of swashplate angles

A further design issue is the choice of an even number


of cylinders against the much more common odd
number [7]. Skaistis [40] points out that under perfect
conditions
(instantaneous
compression
and
decompression)
the force fluctuation
with an even
number of cylinders is zero, and therefore ideal.
However, this cannot be achieved in practice. Studies
on this design issue have been reported, inter alios, by
Haarhaus [104], Bergemann [105], Masuda and Ohuchi
[106], and Pettersson et al [107]. Masuda and Ohuchi
[106] and Pettersson et al [107] acknowledge the
importance of the fluctuating
moment on the swashplate.
Masuda and Ohuchi take account of the
contribution of the moment by an ingenious scheme
which generates a compensating
moment from the
swash plate servo-cylinder.
Whilst this is shown to
provide a valuable reduction in pump ABN, it adds
significant complexity to the design.
3.1.2 Vane pumps
The generation of fluctuating forces and moments in
vane pumps has received less attention in the literature
than piston pumps. Skaistis [40] gives an overview of
the mechanisms involved for both unbalanced and
balanced types of design, again adopting the concept of
"idealised -timing" to aid explanation . A more accurate
assessment of SBN should be possible using computer
simulation software for the prediction of instantaneous
chamber pressures [34, 39, for example] although the
author is unaware of any publications discussing such a
study.
3.1.3

Gear pumps

In gear pumps the transfer of fluid from suction to


delivery is significantly different to that taking place in

17

Poaibon

Poaibon

Poaibbn

Sound

presaura

Figure 17 Calculated (-) and measured (- -) vibration and sound pressure of an axial piston pump
(reproduced from [107] with permission of Research Studies Press)

However, it is constrained to a particular design of


pump casing. A more general approach necessitates the
coupling of an FEM model to a pumping dynamics
model. Progress has already been made in this area by
Bruer-Stercken and Jochum [ 114], albeit with a
simplified model of the excitation mechanism. The
FEM model is again notable for the inclusion of the
pump mounting arrangement which was neglected by
Dunlop [112].
A "fully-coupled" model has been recently presented by
Kojima et al [115]. The authors use simulated forces to
excite the FEM model. Because the point of application
(as well as the magnitude) of the net axial force moves
during the pumping cycle [ 107], Kojima et al replace
the single point force with some equivalent fixedlocation forces because this is more readily handled by
the FEM model. Good agreement between predicted
and measured vibration levels have been achieved. This
is clearly an important step forward in the development
of pump design tools.
Most of the published FEM studies to date have
focussed on piston machines, although there has been
some work published on gear pumps. Fiebig and Heisel
[116], for example, describe an FEM investigation of an
external gear pump. They draw attention to the
dynamic forces in bearings being the principal source of
structure borne noise.
An FEM model of a vane pump has been reported by
Nagata et al [38] in the context of modelling the
pumping dynamics and FBN.
The model could
undoubtedly be extended to cover SBN although the
authors' objective was to establish deformation rather
than dynamic behaviour. No other publications on vane
pump FEM modelling are known to the author.
An important practical issue is the manner in which
pumps (and motors) are mounted. Heron and May
[117] demonstrated that pump bell housing mounts can
generate high levels of noise and transmit vibration to
the prime mover and supporting structure. Through the
use of appropriately-designed isolation elements
between the pump and mount, a significant reduction in
noise can be achieved. This is a relatively simple and
inexpensive measure although care must be taken over
static deflections, torque reactions and behaviour at
resonance conditions.

3.2 Valve SBN


The throttling action of many types of control valve can
induce cavitation and create the familiar broad-based
noise associated with it. Various novel methods have
been adopted to reduce or eliminate cavitation in valves,
either through the introduction of modifications to the
internal geometry [51] or by arranging for abrupt
pressure drops to be replaced by a distributed pressure
loss [40, 118]. The latter approach can be particularly
effective but tends to result in a component with a much
larger installation volume and higher cost.
Flow-induced noise in a valve has been reported by
Weber et al [ 119] who investigated a high frequency
noise (dominantly at a frequency of over 5kHz)
generated by a counterbalance valve.
From a
combination of experiments and numerical modelling it
was concluded that shear-layer instability was
responsible. Changes to the geometry of the valve
enable the noise to be eliminated.
3.3 Circuit SBN
It is frequently found in practice that the majority of the
hydraulic system ABN is attributable to the vibration of
the piping and attached components, rather than
vibration of the pump (or motor). Pipework vibration
arises as a consequence of mechanical excitation and
FBN. The former can in general be minimised through
the use of a short length of hose between the pump and
the circuit. However, FBN, as previously explained,
propagates through the pipes to all components. When
uid-structural coupling is ignored, a single length
fl of
pipe can be modelled using a two-port representation
(Equation 4). With fluid-structure coupling the problem
becomes significantly more complex. At each end of
the pipe, in addition to pressure and flow, it is necessary
to account for forces in three dimensions (3D), moments
in 3D, deflections in 3D and rotations in 3D. Hence a
14 x 14 matrix is required to model each length of
transmission line. Pipe deflections are dealt with
through the use of Timoshenko beam theory coupled
with plane wave theory for the fluid. Initial studies in
this area by Wiggert et al [ 120] and Brown and
Tentarelli [121] have been developed and enhanced by

18

Kwong and Edge [122, 123]. Similar work has been


reported by de Jong [124].
As a consequence it is now possible to predict the
vibration levels of 3D circuits (albeit of relatively
simple geometry).
From the design point of view the
mathematical model can be used to locate pipe clamps
such that SBN (and hence ABN) is minimised at
strategic locations. Fig. 18 shows a comparison of the

3.4 Measurement
At

Figure
power

18 Measured
spectral

density,

pipe acceleration,

time

there

are

no

Standard

test

quantities. The former of these is an exact parallel of


the problem of quantifying
pump FBN:
direct
measurements in an arbitrary circuit only characterise a
pump-system interaction; measurements in a speciallydesigned circuit may not be representative of behaviour
in a real system.
As with the characterisation
of FBN sources it is
possible
to adopt Norton or Thevenin
network
representations of SBN. However, the source has to be
described in multiple dimensions and as a consequence
test procedures are by no means straightforward and the
measurement effort is considerable.
An alternative,
which is applicable to pumps and motors is based on the
concept of pseudo-forces [127]. In this approach, the
internal dynamics are represented by a number of
equivalent pseudo-forces (and pseudo-moments) on the
casing of the machine. For a rigid body six forces and
moments are required but for flexible
bodies an
additional force and moment is strictly necessary for
each vibrational mode of interest.
In practice, this
requirement can often be relaxed. The pseudo-forces
(and moments)
can often be established
from
experimental measurements in situ. It is also possible to
use the pseudo-forces, once measured, to predict sound
pressure levels (providing the transfer function matrix
relating sound pressure to force has been measured). It
is notable that the work of Pettersson et al [ 107],

pumping frequency, whilst 5dB and 2.5dB reductions


were achieved respectively at the second and third
harmonics.

best damping

present

procedures
for characterising
SBN generated
by
hydraulic fluid power machines. A useful summary of
some of the problems associated with quantifying SBN
is given by Hiibner [ 126]. Particular emphasis is given
to the problems of structures connected to the machine
under test and their impact on the vibration levels, and
the difficulty in selecting appropriate measurement

power spectral density of acceleration, measured at one


critical location of a 3D piping system, for particular
combinations of pipe clamp locations.
A genetic
algorithm coupled to the vibration prediction software
established 'best' and 'worst' clamp locations. As can be
seen from Fig. 18, there is a considerable difference
between the two systems.
This was reflected by the
corresponding change in airborne noise levels.
The
'best' system was 16dB quieter than the 'worst' at the

worst damping

the

of SBN

locations

locations

in terms of

for best and worst pipe clamp


locations

previously mentioned, has certain parallels with this


technique. The application of the pseudo-force method
to the characterisation of hydraulic machinery has not,
apparently, received much attention. There is certainly
scope for further work in this area.

In order to adopt an optimisation methodology, it is


necessary to select an object function which has to be
minimised.
For their work on optimising pipe clamp
locations, Kwong and Edge [ 125] chose to minimise a
weighted sum of the transfer functions between the
force at the point of interest and the fluid motion at the
entry to the piping system.
Twenty-seven discrete
frequencies were chosen which targeted harmonics of
pumping frequency and their immediate neighbours.
Neighbouring frequencies were included in order to
desensitise the solution to possible changes in the
resonant characteristics of the system which might arise
in practice (as a result of changes in temperature for
example). The next stage will be the extension of the
methodology to handle more complex and realistic
circuits. This is likely to necessitate finding methods of
sub-dividing complex systems into more manageable
sub-circuits.

3.5 Design issues


Computational tools are beginning to be developed to
assist both component and system designers in the
analysis of SBN. The application of FEM techniques
for pump (and potentially motor) analysis are now
showing real promise and can be used to undertake
sensitivity studies for example [ 128]. The coupling of
pumping dynamics models with FEM models has begun
[115] although only appears to have been attempted for
the case of axial piston pumps at the present time.
Keiper [128] points out that an integrated pumpingdynamics/FEM model has yet to be developed.

19

But again, as with FBN, the development of design


tools still appears to be in its infancy: the design
process tends to be a trial-and-error process largely
guided by experience. The development of design tools
to reduce circuit SBN has made somewhat further
progress, largely as a result of the use of optimisation
algorithms. Nonetheless, there is still much more to be
done.

With the development of sound intensity measurement


techniques, special test environments are no longer
necessary
[132].
Nonetheless,
for comparative
purposes, it is important to establish well-defined test
conditions when using sound intensity instrumentation.
At the present time, work is being undertaken to
establish a Standard Test Code which will allow
comparative measurements to he undertaken on pumps
[133].
For component design investigations, it is desirable to
identify regions
on the component
casing
and
supporting structure which are the principal sources of
ABN generation.
Early work by McCandlish
&
Petrusewicz [134] and McCandlish and Linney [108]
demonstrated
experimentally
a close
correlation
between pump casing velocity and ABN levels in the
near field. Casing velocity is readily established by
integration of an accelerometer signal and hence this

4. AIR BORNE NOISE


The sole objective in the design of quieter systems is the
reduction of ABN through the reduction of FBN and
SBN. There may of course be related benefits in terms
of improved overall reliability following on from
extended component fatigue life and reduced leakage
from pipe fittings. However, the success of any noise
abatement exercise must ultimately be judged by
meeting or exceeding target air borne noise levels and
achieving customer satisfaction. The latter introduces
the interesting and potentially problematic issue of
pschoacoustics: it is not just the noise level of a
machine which needs to be considered but also the
quality of the noise. For example, a machine may be
deemed "quiet" but nonetheless "irritating" as a
consequence of the tonal quality. In addition, tones
generated in the speech range (which is commonly the
case with hydraulic machinery) are known to interfere
with speech recognition [129]. This has a clear impact
on the quality of the working environment and may also
result in hazards arising from misunderstood verbal
instructions.
More work on the psychoacoustical
aspects of hydraulic system noise generation is
undoubtedly required.

provided a useful means of developing prototypes.


However, the degree of correlation between casing
velocity and ABN was dependent on accelerometer
location and hence the technique is somewhat empirical
in its application.
More sophisticated evaluation of
ABN characteristics is possible by mapping the sound
intensity vectors over the surface of a machine [135,
136]. This technique is very effective for establishing
areas of high acoustic energy generation and hence is a
valuable design tool.
4.2 Design Issues
Considerable emphasis has been placed, and continues
to be placed, on reducing the ABN noise levels of
pumps. This is entirely appropriate.
However, it is
often poor installation of the pump which is responsible
for unnecessarily high noise levels.
Although "best
practice" advice continues to be published [1 17, 137,
138, for example], it is not always heeded.
The prediction of ABN at the design stage is still
embrionic. At the component level, some progress has
been made by Pettersson et al [107]. They have shown
that it is possible to establish for a given pump structure
the relationship between sound pressure at certain
locations around the pump and the internal forces and
moments.
However,
this requires experimental
measures of the basic design to permit predictions to be
made. Wataridachi et al [135] predicted the sound field
by sub-dividing a pump housing into a large number of
individual elements each acting as single sound sources.
The velocity of each element was predicted from
knowledge of the modal characteristics of the pump
casing (obtained experimentally)
and the internal
excitation. Good agreement was achieved between the

4.1 Measurement of ABN


System designers usually require sufficient information
to allow a comparison to be drawn between rival
components; component designers generally require
more detailed information to aid the design process.
Until fairly recently, the most reliable method of
establishing component ABN levels for comparison
purposes was to perform tests in a special test
environment, most commonly an anechoic chamber. In
performing tests on pumps and motors, it is necessary to
locate the prime mover or loading system, as
appropriate, outside of the chamber. This results in a
complex and expensive test procedure. Nonetheless,
such requirements are embodied in Standard test
procedures [130, 131] which enable machines to be
compared in terms of one-third-octave frequency
spectra or in terms of an overall A-weighted figure of
merit. As with FBN and SBN testing, the results of
such tests relate to a special environment and are not
necessarily representative of behaviour in real
installations.

predicted and measured sound pressure at one particular


frequency.
Agreement at other frequencies is not
discussed in the paper and no information is given on
the geometry of the pump case.
More detailed

20

appropriate algorithms to minimise an object function in


order to optimise a design. This approach has been
adopted successfully in selecting hydraulic components
to meet a dynamic performance specification [143, for
example]. However, little effort appears to have been
devoted to using this technique in the context of
hydraulic system noise reduction. Notable exceptions
include work by the following researchers: Longmore
et al [73], who successfully used a genetic algorithm to
establish the properties of a hose by achieving a "best
fit" (in the least squares sense) between modelled and
measured behaviour of over a selected frequency range;
Kojima and Ichiyanagi [67] who used a constrained
optimisation technique to design the dimensions of a
silencer; finally, Kwong and Edge [125, for example]
and Bihhadi and Edge [144] who used a genetic
algorithm for optimising, respectively, pipe clamp and
component locations in order to minimise energy
transfer to a supporting structure.
Optimisation
techniques have clear potential for many aspects of
FBN, SBN and ABN reduction at both the component
and system level. As with other applications the biggest
difficulty is likely to be in the selection of the object
function to be minimised. Recent work by Andersson
[145] in the general field of engineering system design
describes a possible way forward.
Despite the specialist nature of hydraulic system noise
and the many issues involved (Fig. 19) it is essential
that both designers and commissioning engineers are
appropriately educated. Knowledge at the level of
"guidelines" and "rules of thumb" has its place , but a
familiarity of what is now achievable through computer
simulation deserves to be widely known.

information on an investigation into the use of FEM


analysis for the prediction of pump ABN is given by
Shimizu and Kojima [139].
They achieve good
agreement between predicted and measured behaviour
(obtained from sound intensity measurements) over a
wide frequency range.
They also demonstrate how
modifications to the portplate can have a significant
impact on pump ABN.
A systems approach to the design of the pump and prime
mover combination
has been described by Balmer
[140]. By integrating an oil-cooled induction motor and
pump in a specially-designed
housing, appropriate
measures can be taken to achieve a significantly quieter
sub-assembly.
Kwong and Edge [122] have proposed that it should be
possible to estimate the sound field around pipework
from predictions
of structural vibration using the
Boundary Element Method. The merit of pursuing this
approach really depends on the contribution to ABN
nlade from the individual elements of the system. It has
been shown [141] that the sound power generated by
1m of steel pipe can be greater than that generated by a
pump. Similar levels of ABN from pipework were
reported by Petrusewicz and Stammers [142]. However,
the supporting structure is usually a much more
effective radiator of acoustic energy than the pipework
itself and hence this is the area which is worthy of more
attention. Little work seems to have been devoted to
this aspect of system design.
5. CONCLUDING

REMARKS

Tools for the analysis of FBN, SBN and, to a lesser


extent, ABN are now becoming available. These tools
are assisting in the design process and are leading to
quieter machines and systems. However, there is still
much to be done. Three areas are worthy of special
mention:
(i)
(ii)
(iii)

software integration
optimisation
education

The coupling between FBN, SBN and ABN is


inherently complex and as a consequence design and
analysis software tends to focus on one area in isolation
from the others. To achieve the ultimate objective of
the lowest levels of ABN, it is important that efforts are
made to integrate software tools so that component and
system designers are able to take full advantage of the
techniques developed in the research laboratory. Some
progress has already been made in this area, most
notably in the area of pump design, but less effort has
been devoted to the design of other components and
systems.
With the continuing
improvements
in computer
processor speed, it is now a practical proposition to use

Figure

19 The

domain

generation

21

of hydraulic
and transmission

system

noise

ACKNOWLEDGEMENTS

13.

The author gratefully acknowledges the Engineering


and Physical Sciences Research Council for supporting
some of the work reported in this paper (grant refs.
GR/H78054 and GR/L26858).

14.
REFERENCES

1.

2.

3.
4.
5.

6.

7.

8.

9.

10.

11.

12.

Council Directive 89/392/EEC of 14 June 1989


on the approximation of the laws of the member
States relating
to machinery.
O.J. of the
European Communities, L183, 29 June 1989
Council Directive 86/188/EEC of 14 May 1986
on the protection of workers from the risks
related to exposure to noise at work. O.J. of the
European Communities, L137, 24 may 1986
Seminar on Quiet Oil Hydraulic
Systems,
IMechE, London, 2-3 November, 1977
Seminar on Quieter Oil Hydraulics, IMechE,
London, 29-30 October, 1980
Edge KA & Liu Y. Reduction of piston pump
pressure ripple. Proc 2nd International Conf. On
Fluid
Power
Transmission
and
Control.
Hangzhou, P.R.China, March 20-22, 1989, Paper
9.1, pp 779-784
Palmberg J-O. Modelling of flow ripple from
uid power piston pumps. Proc of Second flBath
International Fluid Power Workshop, September
1989, pp 207-227
Foster
K & Hannan
DM.
Fundamental
uidborne and airborne noise generation
fl
of axial
piston pumps. Seminar on Quiet Oil Hydraulic
Systems, IMechE, London, 2-3 November, 1977,
C257/77, pp 29-40
Kelsey JS, Taylor R & Foster K.
Fluid
properties: the effect of the fluid being pumped
on the noise emitted by an axial piston pump.
Seminar on Quieter Oil Hydraulics, IMechE,
London, 29-30 October, 1980, C384/80, pp 7175
Edge KA & Darling J. Cylinder pressure
transients in oil hydraulic pumps with sliding plate
valves. Proc Instn Mech Engrs, Vol. 200, No. B 1,
1986, pp 45-54
Harris RM. The CAPPA Suite: Bathfp model
reference
guide for computer aided pump

15.

16.

17.

18.

19.

20.

21.

22.

23.

performance analysis, parts 1 and 2. University of


Bath, School of Mechanical Engineering, Aug
1991 and Aug 1992
Harrison AM. Reduction of axial piston pump

24.

pressure ripple. PhD Thesis, University of Bath,


Bath, UK, 1997
Pettersson M, Weddfelt K & Palmberg J-O.
Reduction of flow ripple from fluid power piston
machines by means of a precompression filter
volume. 10th Aachen Colloquium on Fluid Power

25.

22

Technology, Aachen, Germany, 17-19 March


1992
Helgestad BO, Foster K & Banninster FK.
Pressure transients in an axial piston hydraulic
pump. Proc Instn Mech Engrs, Vol 188 17/74,
1974, pp 189-199
Grahl,
T."Gerauschminderung
an
Axialkolbenpumpen
durch
variable
Umsteuersysteme"
(In German) Olhydraulik
and Pneumatik, Vol 33, 1989, no.5, pp 437-443
Weddfelt K, Pettersson M & Palmberg J-O. An
investigation of the possibilities of using fluidic
components
in
the
fluid
power
area.
FLUCMOME '94, Toulouse, France, 29 Aug-1
Sept 1994, pp 137-142
Harrison AM & Edge KA. Reduction of axial
piston pump pressure ripple. Submitted for
consideration for Proc Instn Mech Engrs, Part I,
1999
Johnston DN, Edge KA, Vaughan DN & Raggett
M. Experimental and computational study of
pumping dynamics and valve operation in a
reciprocating
pump.
ASME WAM, New
Orleans, USA, November, 1993, 93-WA/FPST-5
Nguyen-Schaefer H. Flow-induced noise in a
hydraulic plunger pump. Proc of Bath Workshop
on Power Transmission
and Motion Control
(PTMC '98), September ,1998, pp 219-231
Keen GV.
Performance of an oil hydraulic
wobble-late pump. Project report 681, School of
Engineering, University of Bath, UK, 1984
Yudin EM. Gear pumps, principal parameters
and their calculation (translation by E Harris), of
the Russian book 1967 (National Lending
Library for Science and Technology)
Molton GR. Techniques for reducing fluid borne
noise from gear pumps and their circuits. Proc 7th
International Fluid Power Symposium, BHRA,
Bath, 16-18 September 1986, pp 261-267
Schwuchow
D.
Pulsationssenkung
bei
Zahnradpumpen.
Proc of 1. Internationales
Fluidtechnisches
Kolloquium,
Aachen, 17-18
March, 1998, Band 1, pp 79-92
Li, J. Research on optimized gear parameters
and the nomograms of noise-free external gear
pumps.
1985 ICFP, Zheijiang
University,
Hangzhou, P.R. China, 10-14 September, 1985,
III.13, 859-878
Mitome K & Seki K. A new continuous contact
low-noise gear pump. Trans of the ASME Jnl of
Mechanisms, Transmissions, and Automation in
Design, December, 1983, Vol 105, pp736-741
Ishibashi A, Muta S & Hoyashita S. Gear pumps
with new release ports suitable for various
running conditions. JSME International Journal,
Series III, Vol. 32, No. 3, pp 442-447, 1989

26.

27.
28.

29.

30.

31.
32

33.

34.

35.

36.

37.

38.

39.

Bonacini C. On flow in gear pumps. The


Engineer, Journal of the Italian National
Association of Engineers and Architects (In
Italian), No. 9, 1961
Read L. Quieter pump has the teeth to succeed.
In: Eureka on Campus, Autumn 1998, p 14
Hobbs J & Fricke H-J. New method of reducing
noise in external gear pumps. International OffHighway Meeting and Exposition, MECCA,
Milwaukee, 8-11 September, 801005, 1980
Headrick DC & Maroney GE. A quiet split-gear
pump for reduced fluidborne noise. The BFPR
Journal, 14, 1, Fluid Power Research Center,
Oklahoma State University, USA, 1980, pp 83-86
Duke KT & Dransfield P. Improving gear pump
relief groove design. 32nd National Conference
on Fluid Power, Oct 26-28, 1976, pp 449-461
Edge KA & Lipscombe BR. The reduction of gear
pump pressure ripple. Proc Instn Mech Engrs,
Vol. 201, No. B2, 1987, pp 99-106.
McCandlish D, Edge KA & Tilley DG. Fluid
borne noise generated by positive displacement
pumps.
Seminar on Quiet Oil Hydraulic
Systems, IMechE, London, 2-3 November 1977,
C65/77, pp 103-114
Weddfelt K. On modelling, simulation and
measurements of fluid
power pumps and
pipelines with special reference to flow
pulsations. PhD Thesis, Linkoping University,
Linkoping, Sweden, 1992
Dickinson AL, Edge KA & Johnston DN.
Measurement and prediction of power steering
vane pump fluidborne noise. SAE Trans. Jnl. of
Passenger Cars, paper no. 931294, 1993
Nishiumi N & Maeda T. The relationship
between vane motion and chamber pressure in a
vane pump. Proc of the 2nd JHPS International
Symposium on Fluid Power, Tokyo, 6-9
September, 1993, paper 22, pp 209-214
Matten N. & Lang CM. Flugelabheben bei
einpoligen,
verstellund
regelbaren
gelzellenpumpen.
Olhydraulik
flu
und
Pneumatik, Vol 33, No 5, 1989, pp 427-434
Ueno H, Shintani R & Okajima A. Pressure and
flow ripples of a variable-displacement vanepump. Proc of the 1st JHPS International
Symposium on Fluid Power, Tokyo, 1989, A413, pp 139-145
Nagata K, Takahashi K & Saitoh K.
A
simulation technique for pressure fluctuations in
a vane pump. Proc of Eighth Bath International
Fluid Power Workshop, September, 1995, pp
169-183
Kojima E, Shinada M & Yoshino T.
Characteristics of fluidborne noise generated by
fluid power pump (2ndreport, pressure pulsation

40.
41.

42.

43.

44.

45.
46.

47.

48.

49.

50.

51.
52.

23

in balanced vane pump). Bull JMSE, Vol 27, No


225, March, 1984, pp 475-482
Skaistis S.
Noise control of hydraulic
machinery. Marcel Dekker, 1988
Edge KA & Freitas FJT. A study of pressure
uctuations in the suction lines of flpositive
displacement pumps. Proc Instn Mech Engrs, Vol.
199, No. B4, 1985, pp 211-217
Jones BL, Johnston DN & Longmore DK.
Simulation of suction flow ripple in power
steering pumps.
SAE Int. Off-Highway
Congress & Exposition, USA, September, 1998
Harris RM, Edge KA & Tilley DG. Reduction of
piston pump cavitation by means of a preexpansion volume.
Proc of Fifth Bath
International Fluid Power Workshop, September,
1992, pp 167-183
Conrad F & Adelstorp A. Pure water hydraulic
systems and applications. Proc of Ninth Bath
International Fluid Power Workshop, September
1996, pp 158-180
Trostmann E. Water hydraulics
control
technology. Marcel Dekker, 1999
Tilley DG & Butler MD. The generation and
transmission of fluid borne pressure ripple in
hydraulic systems. Seminar on Quieter Oil
Hydraulics, IMechE, London, 29-30 October,
C374/80, 1980, pp 7-13
Kojima E & Shinada M. Characteristics of
uidborne noise generated by a fluid power pump
fl
(4th report, pressure ripple in hydrostatic power
transmission). Bull of JSME, Vol 29, No 258,
December, 1986, pp 4147-4155
Ichiyanagi,T, Kojima, E & Edge KA. The fluid
borne noise characteristics of a bent axis motor
established using the 'secondary source' method.
Fifth Scandinavian International Conference on
Fluid Power, Linkoping, Sweden, 28-30 May,
1997
Shorin V & Sverbilov V. On possibility of
suppressing self-excited oscillations of relief and
check valves by actions on pipeline performance.
Proc of the 2nd JHPS International Symposium
on Fluid Power, Tokyo, 6-9 September, 1993,
paper 63, pp 465-470
Maeda T. Generation mechanism of auditory
noise from brake valve for vehicle. Fourth
Scandinavian
International
Conference,
Tampere, Finland, 26-29 September, 1995, pp
1182-1194
Heron RA. The control of cavitation in valves.
Proc 7th International Fluid Power Symposium,
BHRA, Bath, 16-18 September 1986, pp 275-283
Stecki JS & Davis DC. Fluid transmission linesdistributed parameter models. Part 1: a review of
the state of the art. Proc Instn Mech Engrs, Vol
200 No. A4, 1986, pp 215-228

53.

54.

55.

56.

67.

Edge KA & Tilley DG. The use of plane wave


theory in the modelling of pressure ripple in
hydraulic systems. Trans. Inst. M. C., Vol5, No.4,
Oct-Dec, 1983, pp 171-178.
Edge KA. The theoretical prediction of the
impedance
of positive displacement
pumps.
Seminar on Quieter Oil Hydraulics, IMechE,
London, 29-30 October 1980, Paper C387/80, pp
95-103.
Johnston DN & Drew JE.
Measurement of
positive displacement flow ripple and impedance.
Proc Instn Mech Engrs, Vol 210, 1996, pp 65-74
Johnston DN & Edge KA. Simulation of the

68.

69.

pressure
ripple characteristics
of hydraulic
circuits. Proc Instn Mech Engrs, Vol 203, 1989,
57.

58.

59.

60.

61.

62.

63.

64.

65.

66.

70.

pp 275-282
Lecerf J-P. A methodology to describe and predict
hydraulic noise in fluid power systems. Proc of
Second Bath International Fluid Power Workshop,
September 1989, pp 177-206
Kojima E, Sinada M & Matsui M. Development
of a general-purpose
simulation package for
pulsation in hydraulic pipeline generated by
pumps. Tans. JSME (in Japanese). Ser. B, Vol
58, No 546 (1992-2), pp 356-363
Wacker K. Flussigkeitsschall in Rohrleitungssystemen mit kleinem durchmesser. Olhydraulik
und pneumatik, Vol 29, Nr 11, 1985
Weddfelt K. A general simulation program for

71.

72.

73.

pressure ripple of axial piston machines.


An
application program to HOPSAN. LiTH-IKP-R565, Linkoping University, 1990
Edge KA & Johnston DN. The secondary source
method for the measurement of pump pressure
ripple characteristics.
Part 2 - Experimental
results. Proc Instn Mech Engrs, Vol 204, 1990,

74.
75.

pp 41-46
Lallement J.
Hydropneumatic
accumulators.
Proc of Second Bath International Fluid Power
Workshop, September 1989, pp 43-57
Edge KA & Johnston DN.
The impedance
characteristics of fluid power components: Relief
valves and accumulators. Proc Instn Mech Engrs,
Part I, Vol 205, 1991, pp 11-22
Johnston DN & Edge KA. The impedance
characteristics
of fluid
power components:
Restrictor and flow control valves. Proc Instn
Mech Engrs, Part I, Vol 205, 1991, pp 3-10
Lau KK. Impedance characteristics of hydraulic
components -progress
report 022/95/a. School
of Mechanical Engineering, University of Bath,
March, 1996
Whitson RJ. The measured transmission loss
characteristics
of some hydraulic attenuators.
Seminar
on
Quieter
Oil
Hydraulics
(Supplement), IMechE, London, 29-30 October,
1980, C383/80, pp 105-115

76.

77.

78.

24

Kojima E, & Ichiyanagi T.


Development
research of new types of multiple volume
resonators. Proc of Bath Workshop on Power
Transmission and Motion Control (PTMC '98),
September, 1998, pp 193-206
Muto T, Yamada H, Fukumori J & Suematsu Y.
Simulation of pressure pulsation induced in fluid
transmission lines including a viscoelastic house.
Proc of 1. Internationales Fluidtechnishes
Kolloquium, Aachen, 17-18 March 1998, pp
511-520
Longmore DK & Schlesinger A. Transmission
of vibration and pressure fluctuations through
hydraulic hoses. Proc Instn Mech Engrs, Part I,
Vol 205, 1991, pp 97-104
Longmore DK & Schlesinger A. Relative
importance of the various vibration transmitting
mechanisms in hoses. Proc Instn Mech Engrs,
Part I, Vol 205, 1991, pp 105-111
Yu J & Kojima E. Dynamic response of
unhomogeneously viscoelastic fluid
lines.
American Control Conference, Seattle, USA,
1995, pp 1821-1825
Yu J & Kojima E.
Wave propagation
characteristics of fluid contained in a finitelength hydraulic high-pressure flexible hose (1"
and 2ndreports). Trans JSME, Vol 62, No 598,
Series C, 1996-6, pp 2207-2222
Longmore DK, Johnston DN & Drew JE.
Measurement of the dynamic properties of hose
walls required for modelling fluid-borne. ASME
IMECE, FPST Division, Dallas, USA,
November, 1997
Drew JE. The use of flexible hose to reduce
pressure ripple in power steering. PhD Thesis,
University of Bath, 1997
Kojima E, Ichianagi T & Edge KA. Performance
of flexible hoses with spiral wound tubular
inserts for reduction of fluid-borne pressure
pulsations in automotive hydraulic power
steering systems (1streport,
experimental
considerations). Proc of Ninth Bath International
Fluid Power Workshop, September, 1996, pp
195-221
Bowns DE, Edge KA & Tilley DG. The
assessment of pump fluid borne noise. Seminar
on Quiet Oil Hydraulic Systems, IMechE,
London, 2-3 November, 1977, C266/77, pp 115125
Theissen H.
Volumenstrompulsation von
kolbenpumpen. Olhydraulik und pneumatik, Vol
24, No.8, 1980, pp 588-591
Larsson P, Palmberg J-O & Weddfelt K.
Analysis and measurement of pressure ripple of
uid
power
pumps.
Eighth
IASTED
fl
International Symposium MECO 86, Taormina,
Italy, 3-15 September, 1986

79.

80.

81.

82.

83.

84.

85.

86.

87.

88.

89.

90.

91.

ISO/FDIS 10767-2 Hydraulic fluid


powerDetermination
of pressure ripple levels in
systems and components. Part 2 - Simplified
method for pumps.
Unruh DR. Outlet pressure ripple measurement
of positive
displacement
hydraulic
pumps.
National Conference on Fluid Power, Chicago,
21-23 October, 1975
Esser J & Backe W. Development of a new flow
ripple sensor for the measurement of pump
ripple. Seventh Bath International Fluid Power
Workshop, September 1994, pp 262-278
Murrenhoff H & Ristic M.
Recent results
concerning direct measuring of flow ripple by
two
new sensors.
Proc
of Ninth
Bath
International Fluid Power Workshop, September
1996, pp 212-228
Edge KA & Johnston DN. The secondary source
method for the measurement of pump pressure
ripple characteristics. Part 1 - Description of
Method. Proc Instn Mech Engrs, Vol 204, 1990,

92.

93.

94.

95.

pp 33-40
Kojima E. A new method for the experimental
determination
of
pump fluid-borne
noise
characteristics. Proc of Fifth Bath International
Fluid Power Workshop, September 1992, pp
111-137
Kim D-T, Yokota S & Nakano K. An approach
for remote measurement of instantaneous flow
rate by making use of hydraulic pipeline
dynamics (2nd report: A study on the quasiremote instantaneous flow rate measurement in
real time). Jnl of JHPS, Vol 22, No 2, Mar.
1991, pp 77-83
Johnston & Edge. In-situ measurement of the
wavespead and bulk modulus in hydraulic lines.
Proc Instn Mech Engrs, Part I, Vol 205, 1991, pp
191-197
Weddfelt K. An accurate method to derive the
Fourier series coefficients from a FFT-spectrum
of harmonic signals.
Proc of Eighth Bath
International Fluid Power Workshop, September
1995, pp 204-220
ISO 10761-1: 1996 Hydraulic fluid
powerDetermination of pressure ripple levels generated
in systems and components.
Part 1 precision
method for pumps
Johnston DN & Longmore DK.
Rating of
pumpfluid-borne noise. ASME IMECE, FPST
Vol 1, Chicago, USA, November, 1994
ISO/DIS
10767-3 Hydraulic fluid
powerDetermination of pressure ripple levels generated
in systems and components.
Part 3 Precision
method for motors
Lau KK. Impedance characteristics of hydraulic
components-Final
progress report 35/96,

96.

97.
98.

99.

100.

101.
102.

103.

104.

25

School of Mechanical Engineering, University of


Bath, February, 1996
Bais R-J & Toet G. How to define the properties
of a hydraulic silencer in a hydraulic system.
Proc of Fifth Bath International Fluid Power
Workshop, September,1992, pp 138-154
Johnston DN, Longmore DK & Drew JE. A
technique for the measurement of the transfer
matrix characteristics of two-port hydraulic
components.
ASME IMECE, FPST Vol 1,
Chicago, USA, November, 1994
Kojima E & Edge KA.
Experimental
determination of hydraulic silencer transfer
matrices and assessment of the method for use as
a standard test procedure.
Seventh Bath
International Fluid Power Workshop, September,
1994, pp 221-241
Lau KK, Johnston DN & Edge KA. Fluid borne
noise characteristics of hydraulic filters and
silencers. Proc of Seventh Bath International
Fluid Power Workshop, September, 1994, pp
242-261
Drew JE, Longmore DK & Johnston DN. The
noise characteristics of power steering systems.
IMechE International Conference on vehicle
NVH and refinement, Birmingham, May, 1994,
C487/001/94, pp 167-175
Klees GT. Attenuation device. United States
patent No 3,323,305, 1967
Yokota S, Sanada H & Yamaguchi H. A study
of active accumulator. JHPS 90th Hydraulics &
Pneumatics Symposium, November, 1991, pp
133-136
Yamaoka T, Kojima E, Shinada M & Izawa N.
Development of an active attenuator for pressure
pulsation in liquid piping systems (5th report,
construction of the adaptive control system and
fundamental experiments). Trans JSME, part B,
Vol 61, No 588, August, 1995, pp 2987-2995
Brennan MJ, Elliot SJ & Pinnington RJ. Active
control of fluid waves in a pipe. Active 95,
California, USA, 608 July, 1995, pp 383-394
Anon. Noise contribution of electric motors in
industrial hydraulic power units. Vickers Inc,
USA, 1994
Wicke V, Edge KA & Vaughan ND.
Investigation of the effects of swashplate angle
and suction timing on the noise generation
potential of an axial piston pump. ASME
IMECE, FPST Division, Anaheim, CA, USA,
November, 1998
Helgestad BO, Foster K, Banninster FK. Noise
in an axial piston pump. IMechE Seminar on
noise emitted by fluid power equipment-its
causes and control, 1973, C86/73
Haarhaus M.
Gerauschentstehung und
Gerauschminderung bei axialkolbenpumpen in

105.

106.

107.

108.

109.

110.

111.

112.

113.

114.

115.

116.

Schragscheibenbauweise. Disertation, Aachen,


September, 1984
Bergemann M. Noise problems of hydraulic
piston pumps with odd and even number of
cylinders. 9th International Conference on Fluid
Power, Cambridge, UK, 25-27 April, 1990
Masuda K & Ohuchi H. Noise reduction of a
variable piston pump with even number of
cylinders.. Proc of the 3rd JHPS International
Symposium on Fluid Power, Yokohama, 4-6
November, 1996, paper 114, pp 91-96
Pettersson M, Weddfelt K & Palmberg J-O.
Prediction of structural and audible noise from
axial piston pumps using transfer functions. Proc
of Eighth Bath International Fluid Power
Workshop, September, 1995, pp 184-218
McCandlish D & Linney G.
Noise
characteristics of gear pumps. 4th International
Fluid Power Symposium, BHRA, 16-18 April,
1975, paper G1
Kojima E & Shinada M. Characteristics of the
uidborne noise generated by fluid power pump
fl
(3rd report, discharge pressure pulsation of
external gear pump). Bull of JSME, Vol 27, No
232, October, 1984, pp 2188-2195
Manco S & Nervegna N. Pressure transients in
an external gear hydraulic pump. Proc of the 2nd
JHPS International Symposium on Fluid Power,
Tokyo, 6-9 September, 1993, 221-228
Manco S & Nervegna N. Simulation of an
external
gear pump
and
experimental
verification. Proc of the 1st JHPS International
Symposium on Fluid Power, Tokyo, 1989, A414, PP 147-160
Dunlop RW.
Determination of the natural
frequencies, mode shapes and modal damping
coefficients of a positive displacement pump.
Seminar
on
Quiet
Oil
Hydraulic
Systems,IMechE, London, 2-3 November, 1977,
C268/77, 139-154
Kojima E, Iwata H, Hyodo N, Matsushima S &
Watanabe K. Modal analysis of the structural
vibration of an oil-hydraulic pump (in Japanese).
Trans JHPS, Vol 27, No 2, 1996, pp 315-327
Breuer-Stercken A. & Jochum S.
Finiteelemente-berechnungen
zur
Unterstutzung
larmarmer konstruktionen am beispiel von
kolbenpumpengehausen.
11.Aachener Fluidtechnisches Kolloquium, 8-10 March, 1994,
Band 1, pp 107-125
Kojima E, Matsushima S, Watanabe K &
Shimizu S. FEM analysis of the structural
vibration of an oil-hydraulic pump under real
operating conditions (in Japanese). Trans JHPS,
Vol 29, No 3, 1998, pp 79-86
Fiebig W. & Heisel U. Schwingungs- und
gerauschoptimierung von hydropumpen durch

117.

118.

119.

120.

121.

122.

123.

124.

125.

126.

127.

128.
129.

26

analyse des schwingungsverhaltens


von ihren
gehausen.
11 Aachener
Fluidtechnisches
Kolloquium, 8-10 March, 1994, Band 1, pp 171188
Heron RA & May J. The use of isolated pump
mounts for fluid power system noise control.
Seminar on Quieter Oil Hydraulics, IMechE,
London, 29-30 October, 1980, C379/80, pp 4753
Donaldson CJ & Dunlop RW. Design of low
noise valves. IMechE Seminar on Quieter Oil
Hydraulics, 29-30 October, 1980, paper C383/80,
pp 117-122
Weber
ST, Porteiro
JLF,
Rahman
MM,
Pennington II RE & Musallam AE. Proc of Bath
Workshop on Power Transmission and Motion
Control (PTMC '98), September 1998, pp 207217
Wiggert DC, Hatfield FJ & Stuckenbruck S.
Analysis of liquid and structural transients in
piping by the method of characteristics.
ASME
Jnl Fluids Engineering, 1987, Vol 109, pp 161165
Brown FT & Tentarelli SC. Analysis of noise
and vibration in complex tubing systems with
uid-wall interaction. Proc 43(1
Conf Fluid
fl
Power, 1988, pp 139-149
Kwong AHM & Edge KA. Towards the design
of quiet hydraulic circuits. ASME Engineering
Systems
Design
& Analysis
Conference,
ondon, June 1994, Part B, Vol 8, pp 421-440
Kwong AHM & Edge KA. Wave propagation in
uid-filled pipe systems.
Euro-Noise
'95
fl
Conference on 'Software for Noise Control',
Lyon, France, 21-23 March, 1995
de Jong CAF.
Analysis of pulsations and
vibrations in fluid
filled pipe systems.
PhD
Thesis, TNO Institute of Applied Physics, Delft,
The Netherlands, 1994
Kwong AHM & Edge KA. A method to reduce
noise in hydraulic systems by optimising pipe
clamp locations.
Proc Instn Mech Engineers,
Part I, Vol 212, 1998, pp 267-280
Hubner G. Methods characterizing the structure
borne
sound
emission
of machines
and
equipment. Report of the work of ISO/TC43/SC 1
Working Group 22. Proc of Internoise 88
Janssens, MHA & Verheij JW. Two applications
of the pseudo forces method for characterizing
the source strength for structure-borne
sound.
Proc of Internoise 96, Liverpool, UK, 1996, pp
1335-1338
Keiper W. Noise and vibration in hydraulic
drives. European Jnl of Fluid Power, May 1992
Bies DA & Hansen CH.
Engineering Noise
Control Theory and Practice. 2nd Ed, publ Sporn
1996 (page 115)

130.

131.
132.

133.

134.

135.

136.

137.
138.
139.

140.
141.

142.

143.

144.

ISO 4412-1: 1991 Hydraulic fluid power-test


code for the determination of airborne noise
levels-Part 1: pumps.
ISO 4412-2: 1991 Hydraulic fluid power-test
code for the determination of airborne noise
levels-Part 2: motors
Ngah Z & Edge KA. The measurement of
positive displacement pump and motor noise
using sound intensity techniques. Trans. SAE,
Int. Off-Highway & Powerplant Congress &
Exposition, Milwaukee, USA, 9-12 September,
1991, paper 911760.
ISO/CD 16902: Hydraulic fluid power-test
code for the determination of sound power levels
of pumps and motors using sound intensity
techniques, 1999
McCandlish D & Petrusewicz SA. Assessment
of noise generated by hydraulic pumps using
accelerometers. IMechE Conf on Noise Emitted
by fluid Power Equipment, 15 March, 1973,
C83/73, pp 43-49
Wataridachi K, Miki K, Inakuka T & Fujmoto S.
Analysis of the sound field for oil hydraulic
piston pump. Proc of Internoise 90, 1990, pp
1065-1068
Carletti E & Vecchi I.
Sound intensity
technique: a suitable aid in the design of silent
hydraulic pumps. Proc of Internoise 95, Newport
Beach, CA, USA, 10-12 July, 1995, 109-112
Anon. Reducing noise in hydraulic systems.
uid Power. Oct/Nov, 1992, pp 32-34 fl
Anon. Do hydraulics have to be noisy? fluid
and Air Technology, No 22, 1996, 28-30
Shimizu S & Kojima E. Estimation of sound
radiation power of hydraulic fluid power pump
by FEM analysis. (in Japanese). Accepted for
publication in Trans JHPS, 1999
Balmer C. Ways to reduce hydraulic component
noise. fluid
and Air International, No 36,
December, 1998, pp 26-28
Anon. Air-borne noise from pipework. In:
Guidelines to the design of quieter hydraulic
uid power systems, The British fluidfl Power
Association, P4, 1986, pp 11-15
Petrusewicz SA & Stammers CW. The prediction
of the noise radiated by hydraulic piping. fifth
International fluid Power Symposium, Durham,
UK, September, 1978, paper F3
Connor AM & Tilley DG. Optimization of
power Transmission systems using a discrete
Tabu Search method. Proc of Bath Workshop on
Power Transmission and Motion Control (PTMC
'98)
, September ,1998, pp153-163
Bihhadi A & Edge KA. Minimisation of noise
and vibration in a hydraulic circuit using genetic
algorithms. Proc of Internoise 98, Christchurch,

145.

27

New Zealand,
16-18 November
1998, pp 119122
Andersson J. On engineering
systems design-a
simulation and optimisation
approach.
Thesis No
765, Linkoping University, Sweden, 1999

Anda mungkin juga menyukai