Introduction To Pressure Surge In Liquid Systems

Introduction to Pressure Surge in Liquid Systems

0 INTRODUCTION/PURPOSE
1 SCOPE
2 FIELD OF APPLICATION
3 DEFINITIONS
4 CAUSES OF PRESSURE SURGE
4.1 Start-up

5 CONSEQUENCES OF PRESSURE SURGES

6 PRELIMINARY CALCULATIONS
6.1 Estimation of the Sonic Velocity
6.2 Pipeline Period

7 CALCULATION OF PEAK PRESSURES
7.1 Rigid Liquid Column Theory
7.2 Sudden Changes in Flowrate
7.3 Moderately Rapid Changes in Flowrate
7.4 Reflections and Attenuations
7.5 Vapor Cavity Formation
7.6 Complex Piping Systems

8 FORCES ON PIPE SUPPORTS.

9 METHODS OF REDUCING THE EFFECTS OF
PRESSURE SURGE
9.1 Flowrate
9.2 Pipe Diameter
9.3 Valve Selection and Operation
9.4 Pump Start-up/Shut-down
9.5 Surge Tanks and Accumulators
9.6 Vacuum Breakers
9.7 Changes to Equipment

10 DETAILED ANALYSIS
10.1 Data Requirements
10.2 Interpretation of Results

11 GUIDELINES FOR CALCULATIONS

12 EXAMPLES OF PRESSURE SURGE INCIDENTS
12.1 Caustic Soda Pipeline Movement
12.2 Ammonia Pipe Movement
12.3 Propylene Reactor Start-up
12.4 Cooling Water Failure
12.5 Dry Riser Fire Sprinkler Systems
12.6 Cast Iron Fire Main Pressurization

1. A,
CPT
CATALYST, PROCESS TECHNOLOGY
CONSULTANCY

Process Engineering Guide:

GBHE-PEG-FLO-305

Introduction to Pressure Surge in
Liquid Systems

2. DEFINITIONS
CAUSES OF PRESSURE SURGE

Start-up

CONSEQUENCES OF PRESSURE SURGES
PRELIMINARY CALCULATIONS

Estimation of the Sonic Velocity

Pipeline Period

CALCULATION OF PEAK PRE:

Rigid Liquid Column Theory
‘Sudden Changes in Flowrate
Moderately Rapid Changes in Flowrate

3. DETAILED ANALYSIS
Data Requirer

Interpretation of Resu

GUIDELINES FOR CALCULATIONS

EXAMPLES OF PRESSURE SURGE INCIDENTS

Caustic Soda Pipeline M

‘Ammonia Pipe
Propylene Re
jooling Water Failure
Dry Riser Fire
Gast Iron Fire Main Pressurization

4. PIPE MATERIALS

GUIDELINES FOR PRESSURE SURGE ANALYSIS
OF PIPING SYSTEI

FIGURES

1

WAVESPEED IN PIPES FILLED WITH WATER

FORCES ON A PIPING SYSTEM IN STEADY
STATE FLOW

FORCES ON A PIPING SYSTEM IN
TRANSIENT Fl

EFFECT OF VALVE CLOSURE TIME ON OUT-OF
BALANCE FORCES

EFFECT OF VALVE TYPE ON FLOWRATE DURING
TRANSIENT FLOW

5. fe phenomenon
ure surge

jow to make preliminary estimations of the likely magnitude of
effects. Methods of reducing the magnitude ofthe effects are discussed.
Guidelines are given to help in assessing which piping system should be
analyzed in detail and the data requirements to perform a det mputer

analysis of the system are given. Some examples are given of problems
encountered by GBH Enterprises due to pressure surge, Tables 1, 2 and 3
contain some useful data to assist in surge calculations.

This Guide is not a comprehensive treatise on pressure transients. More detailed
information is available in standard texts on the subject, for example references
4, 2and3.

This Guide does not give advice on the mechanical design of piping systems
subjected to pressure surge, although it does indicate how to calculate the loads

6. ‘Any operation which can result in a rapid change in velocity is a potential cause
of serious pressure surge. Typical operations include:

(a) Rapid change in position, either closing or opening, of a control or
olation valve.

Opening ofa safety valve or rupture of a busting dis
Starting or stopping of a pump.

Priming of an empty pipeline.

Unsteady low generated by a reciprocating pump.

apid phase change, due to thermal effects, rapid chemical react
the collapse of a

Traditionally, proces ciated pressure surge problem
almost e» ollowing the RAPID closure of
be shown later that

7. tart-up itis quite possible that vapor cavities will be present at Ihe high points.
yystem, which will collapse as the system is pressurized.

4.1.2. Gas pockets

Gas pockets may reduce fluid friction and allow much higher velocities than
‘occur in normal full liquid operation. The deceleration of the liquid by the gas
compression can lead under certain circumstances to very hi

and temperatures. Explosive ignition of flammable mixtures has been known,

413 Venting

The volumetric flow of a gas through an orifice is considerably greater than that

of a liquid for the same pressure drop. When venting gas from a system, high
rales can occur, which cannot be maintained when the

expelled; a sudden flow redu curs, resulting in pressure surge. Speci

examples include priming dis ‘blowing in systems

8. pressure surge. Experiences indicate that failure of pipework supports a
of pressure surge is more likely than pipeline rupture due to over-pressur

e examples of pr incidents are given in Clause 12.

6 PRELIMINARY CALCULATIONS.

Detailed pressure surge analysis of most piping syster mplex and
ngthy operation, usually involving the use of a computer program, and is
beyond the scope of this Guide. However, some simple preliminary calculations
are possible in order to estimate the likely magnitude of any effects. These m
jow that no significant problem exists, or may indicate the need for a mo
detailed study. Guidelines for determining whether piping systems require
detailed analysis are given in Table 3.

6.1 Estimation of the Sonic Velocity

Pressure surges are propagated through a piping system at the local sonic
velocity. An estimation ofthis is basic to all calculations.

In a rigid isturbance:

9. the sonic velocity (m/s)

the bulk modulus of ela liquid (Nm

tial surge damage.

The bulk modulus of elasticity is not readily obtainable for many liquids. I is not
an item which is stored in the GBH Enterprises; “The VAULT" physical
properties data bank.

‘Some commercially available programs purport to calculate the speed of sound
in liquids, but the method used is of doubtful validity. It should not be used.

Table 1 gives typical values of bulk m
me liquids. Note that dife
for these
where the sonic velocity cal
urce difers by a factor of 3 from another ‘measurement
of sonic velocity. In general, a higher assumed value for the sonic velocity
can be expected to lead to predictions of higher surge pressure:

The velocity of propagation of a pressure wave in a thin walled elastic pipe is
lower than the sonic velocity in a rigid pipe, and can be calculated from

10. where: B =a constant- see below
= velocity of propagation of a pressure wave in a p
vo en by equation (1)
eter (m)
dus ofthe pipe material (Nim
mess (m)
the bulk modulus of elasicity of the quid (N/m2)

‘The value of B depends on the method of pipe support and the Poisson ratio of

the pipe material. When the Young's Modulus, E, is large, as for metal pipes, the
numerical value of B may be taken as unity without significant error. In other

ained pip

constrained pipes

the Poisson's ratio for the pipe wall material

's Modulus

12. where Lis the length of the pipeline from the source of the disturbance to the end
of the pipe (m) and locity of propagation of Ihe pressure wave (mi
associated with a single pipe having a val
ed on the upstream length and one on th
downstream length,

7 CALCULATION OF PEAK PRESSURES

7.1 Rigid Liquid Column Theory

If the velocity of a liquid in a pipeline is changed gradually and steadily, such that
the change takes place over more than about 10 pipeline periods, the simplifying
assumptions can be made that the rate of change of velocity isthe same at all
points along the pipeline. Compressibiity effects can be ignored and the liquid
treated as a rigid column. Pressure changes can then be calculated by
considering the momentum effects. For the upstream side of the disturbance:

13. increase in stat pressure on the upsieam side of a disturbance
Nim?)
pressure 19 fiction fon o to the point

der consideration under

pressure loss due to Inca from the slat of a pipeline tothe point
Under consideration under nal condone (Nim?)

tar of the pi

= ui denciy ag

ream side of the disturbancs

rm. Most types
such that most ofthe effect

osure times of these val nsiderably less than the nominal times
This has to be remembered when determining if the closure can be considered
‘slow. Note also that the effects are very much dependent on valve size, with
much better performance being obtained when small valves in large
lines. See sub claus

72 Sudden Changes in Flowrate.

‘Any change in flowrate which is completed in under one pipeline period can be
considered instantaneous as far as the initial pressure transient is concerned.

14. ah =

inal change in stave head (m)

inital change in pressure (N

ange in velocity (m/s)

quid density (Kam)
This pressure change is usually known as the Joukowski Pressure, or, i
expressed in terms of a head of the liquid, the Joukowski Head. Note that th
equations do not include the length of the pipeline. The same value will be

obtained in a short pipe as in a long one, provided the closure is sufficiently rapid
to occur within one pipeline period.

If the disturbance is caused by the sudden closur
sure will initially incre

15. 7.4 Reflections and Attenuations

When a pressure surge reaches the end of a pipeline, itis reflected back down
is open, the magnitude ofthe reflection is the same
sed. Thus a p
When a surge r
double the incoming magnitude.

and down the line, being reflected from each end. Due to friction effects, the
magnitude of the surges gradually dies away,

75 Vapor Cavity Formation

pressure remains low. Subsequent positive pressures will

collapse. This will lad to a large abrupt pressure surge

condensed. Experience shows that this is one of th

deal with. If preliminary calculations indicate that cavity formation is likely, a
ore detailed analysis is recommended.

16. Pipe inside diameter (m
potential force inthe di
pressure inthe pipe (Nim?)
pressure outside the pipe (Nm
velocity of flowing Mu (avs)

‘uid density (kg

When a fluid is flowing in steady flow through a pipeline, as it approaches a
bend, it will exert a force on that bend, equal to the potential force in the direction
hich it approaches the bend. After flowing round the bend, it will exert a
backward force of the same magnitude in the new direction of flow. The resultant
force acts outwards on the bend in a direction bisecting the angle of the bend.

iting force (F) is given by

17. However, if a pressure transient is passing through the piping system, the forces
on the pipework are no longer in balance. Consider the section of pipe shown in

FIGURE 3. FORCES ON A PIPING SYSTEM IN TRANSIENT FLOW

Shock front moving ups! locity €

mt

Fy

Fluid stationary
sure = P+

Fluid fo

18. a (arte

oe along the pipe tom the downstream bend is

In the simple system considered here, where the pressure remains at the higher
value for a significant time after the transient has passed, this force will act for a
time Ar, the time taken for the pressure transient to pass along the length of pipe
between the two bends:

L

19. kTor pressure

closed gradually rather than suddenly, not only will the magnitude
re be reduced, but the pressure wil rise gradually rather than
ise fashion. The effect of this is to reduce significantly the magnitude of

the out of balance forces. This is ilustrated in Figure 4,

FIGURE 4 EFFECT OF VALVE CLOSURE TIME ON OUT-OF-BALANCE
FORCES

20. For most pipelines, the required flowrate is determined by the process. He
ne operations, such as the batchwise transfer of liquid from one ves
ible to increase the transfer time, thus reducing the flo

ty, and hence a reduced surg

Pipe Diameter

1 the diameter of a pipeline for a given flowrate will reduce the velocity,

and hence the peak surge pre: ach is likely to be
expensive compared with the alternat loreover, although it will reduce the

‘magnitude of the peak pressure, the out-of balance forces will remain
‚bstantialy the same, as the reduction in pressure is balanced by an increase in
ional area, This method will not be considered further here.

Valve Selection and Operation
9.3.1. Isolation valves

The unthinking operation of an isolation valve when the liquid is flowing in
the pipeline can have serious consequences, es as many types of

21. ¡ve may be reduced by selection of a suitable actuator. Certain y
valve, particularly butte es, have unfavorable
ost of their effect
es, a split
fave a rapid rate of closure over
the part of the char o rate does not vary much with v
position, and a slower response during the last stages of closure.

9322 Valvetype
ristis

cations permit, a change in valve type may have significant eft
gnitude and rapidity of pressure changes in the system.

In order toilustrate this, a model of a simple piping system has been set
up using the “The Vaulf (see Clause 10). The system modeled consists
of a feed tank, a pump, a length of pipeline with a pipeline period of 2

a control valve and a receiving tank.

eristics (loss coefficient as a

22. FIGURE 5 EFFECT OF VALVE TYPE ON FLOWRATE DURING TRANSIENT
FLOW

Resvees
Eau percent

23. It should be emphasized that these two Figures are illustrat
clual magnitude ofthe differences between different valve types
depend on the system. The effects of valve size should also be
considered; a large valve will have worse char than a small one
of the same type, as the intial stages of closure of a large valve result in
rate. It cannot be
the best results.

24. el defined
flowrate is unsafe.

Pump shut-downs should also be done in a controlled manner. Unfortunately,
this is not always possible, as power failures ‘ample, can result in a pump
trip.

it may be possible to increase the run-up time of a pump by choice of a suitable
now control systems available which allow the start-up to take

place over an extended period, say 2 minutes. These have been used
fully in many locations. Alternatively, addition of a flywheel to the pump

will increase both run-up and run-down times. The motor will obviously
10 be suitable to deal with the higher inertia. Unfortunately, this approach is
ble with a canned pump.

The use of a non-retum valve on a pump delivery may itself be a cause ol
pressure surge:

95 Surge Tanks and Accumulators

25. 95.2 Accumulators

An accumulator is

nected to the pipeline at a 9

ch absorbs some of the pressure surges passing along the pipe by
allowing some of the flow to be diverted into or out of the accumulator

umulators are common on reciprocating pumps, where they reduce
the pressure fluctuations. They may be located on the suction side,
delivery side or both, depending on the perceived problems. Like surge
tanks, accumulators have to be sized to allow for the magnitude of the
flows. This may make them large and prohibitively costly for high flow
lines.

The main disadvantage of accumulators apart from cost is the problem of
ensuring that they are in a suitable condition to operate when required. I
the gas isin direct contact with the liquid it may dissolve over a period of
time, and some means of checking for this is necessary. Some designs of
accumulator get round this by the use of a bladder or diaphragm.

However, there is always a danger that the diaphragm may rupture. It may
not be easy to determine if this has happened. It may also be difficult to
find a material for the diaphragm which is compatible with the chemicals
being handled.

26. neglect
fety device where the consequences of failure to operate are
’eptabie. If it does prove unavoidable to use them in such
table maintenance procedures are
implemented and can be demonstrated. Inspection of accumulators
should be included in safety inspection routines,

9.6 Vacuum Breakers

Ita pipeline passes through a high point, during shut-down conditions the
pressure at that point may fall below the vapor pressure and a vapor cavity may

form, When flow is re-started in the line, this collapse and initiate a
pressure surge. Ideally, pipelines should be routed to avoid such high points, but
this is rarely possible in practice.

ita vacuum breaking valve is installed atthe high point, the vapor cavity will be

replaced by an air pocket. When the flow is re-started in the line, this air pocket
umn ahead of it and reducing
sumulator

27. 10 DETAILED ANALYSIS

The preliminary hand calculations described in Clauses 6-8 can give an estimate
ofthe likely magnitude of a pressure surge, but if these indicate a possible

problem, (for example, predicted Joukowski Pressure close to the design
pressure, the potential to form vapor cavities, or large predicted out of balar
forces), expert advic sgh.

The analysis of all but the simplest systems requires the use of a computer
program.

Experience is necessary to obtain the best results from these programs, and to
interpret the ans

h modeling press!

28. needed to determine the conditions under
Ice,

Bulk modulus of elasticity of the liquid. This is the in

compressibility. Combined with information on the pipe wall thi

used to determine the speed of propagation of a pressure wave in the

piping system, using equation

for some liquids are given in Table 1, but for other liquids it will probably

be necessary to consult a physical properties specialist for assistance. In
any reliable data, an estimate of the wave speed will have

to be made by analogy with similar liquid.

An isometric drawing of the pipeline installation, including all valves and
other fítings, and giving pipe lengths and elevations. It is not necessary
for lengths to be highly sis, the pipeline
be divided up into sections corresponding to the distance travelled by a
pressure wave in one time-step, and some rounding of the data will

be necessary to ensure an integer number of these sections. Section
lengths in a typical analysis might be 10 to 50 meters. The re

nable accura

29. 0

when fluid is flowing, a res ressed as a K
include the diameter on which the K value is based). Note that the experts
running the analyses can usually provide these data for most common
fitings, provided the type is know

Pump data. During the course of a pressure transient, the differential
pressure across a pump, and hence the flowrate through it, will change. I
the effects of pump start-up and shut-down are not to be modeled, only
the pump characteristic curves (flowrate against differential head) are
necessary. If start-up and shut-down are to be modeled, several other
items of data are required, which are less readily available. These are: the
moment of inertia of the pump and d ction torque; the
speeditorque chara ions should be
‘approached for these data.

The boundaries for an analysis are normally vesss
liquid from it, and which can be regarded as regions of
ble to analyze a section of pip

30. generally results in a
other surge events may give a more gradual, but still rapid, change. Bec
the calculation methods used by “The VAULT, any pressure chant
takes place over a single time step is considered as instantaneous.

For an instantaneous pre range, the magnitude of the out-of-balance

forces at each pipe bend can be estimated as the product of the pressure ct
and the pipe cross sectional area.

the out-o balance force on the pipe section in Ne

31. the rate of pressure change in Nim

the length oft

ure wave in a pipe in mis

gi it is subject to a maximum
value of the total pressure change x the cross sectional area.

In many cases, the frictional resistance between the pipe and the supports will be
nificant compared with the above force. The frictional resistance between à
pipe and skid supports is proportional to the weight of the pipe plus contents.

the length of the (straight) pipe section in m

32. For a pipe supported on hangers rather than skids, there is no significant
frictional force, and any out ce will tend to result in some
movement. However, as the pipe swings on the hangers they will mo
from the vertical and thus exert a restraining force on the pipe. For a di
on the behavior of pipes supported on hangers when subjected to dynami
forces is given contact GBH Enterprises.

11 GUIDELINES FOR CALCULATIONS.

Itis obviously impractical to perform a detailed surge analysis of every pipeline
on a plant. However, itis not al ious which lines should be analyzed.
The guidelines in Table 3 were developed during discussions between the author
and other members of GBH Enterprises, and are proposed for use throughout
the company. In developing these guideline ideration was also given to the
practices within other major nationalmultinational organizations. Table 3 should
be read in conjunction with the more detailed explanations in the main body of
this Guid

The ultimate responsibility for determining whether a s id be
carried out lies with the responsible engineer. If in doubt, a specialist should be
consulted for a

33. ¡pporis, through a 90
‘supported on hangers and

anchored near to the
discharge, but only supported betw
restraint

Although the exact cause of the incident is not
xplanation is that at some unknown time the
while the line was discharging to the tank. This
formation of a vapor cz

lapse of this cavity sent a positive pressure surge back down the line to
the tank. The resulting out of balance force on the bend displaced the line
by about 0.15 m, jpports and hangers.

ry to do the observed damage
‘agreement with that calculated
ge analysis. Points to note from this incident are:

The peak pressures calcula I within the design pressure for
the piping system.

35. Propylene Reactor Start-up

Liquid propylene through restriction ori reactor. In order
ist start-up, propylene vapor rather than liquid was fed through a
side branch, the liquid line being closed. Inadvertent opening of the liquid
line caused the vapor packet to collapse, allowing a very high liquid
city in the feed line. This flowrate c+ stained through the
liquid hammer caused the non-return
to slam shut. The bolts on the valve cover stretched, allowing
propylene to jet from the broken joint

Cooling Water Failure

The cooling water pump tripped on a large cooling water circuit on an
ammonia plant. During the short delay between the trip and the automatic
start-up of the stand-by pump a vapor pocket formed at an elevated he:
exchanger. The present y allowed high flow from the stand-by
jammer which arose from.

37. The VAULT” Technical Guide. GBH Enterprises.

The VAULT” User Manual. GBH Enterprises.

HTFS Handbook Sheet FM13. "Pressure surges in a pipeline with liquid
flow due to valve closure." Smith R A, 1990.

iler D S. "Internal Flow Systems." 2nd edition (1990). BHRA. ISBN 0.
947711-77-5

38. NOMENCLATURE

A

velcty of propagation of a pressure wave in à pipe
pipe ins diameter
frictonal resistance between pipe a
force ane upsteam bend
intial change in state head

(a) pipe enath
(6) Sth pe length between bend

pressure loss duet ction fom the pom under consideration 1 the end of

pressure lose duet cion rom the point under consideration tothe end of
eine under fal conatons
pressure lass duet fiction from the sat oa peke othe pont unde

pressure loss duet fiction from the sat ofa ppele othe pain under
Eonsieration under nal constone

rate of pressure change

change in velocty
24 distance fom point under censideraten tothe end of the pipeine

uk modulus of elastty of igus
M (6 Poon rato

43. imal of Chemical Engineering Data. Volume 23 (3) 1978, page
“The density and
dium hydroxide solutions at high
s in quid
from. Journal of th
‘The Handbook of Chem
GW. Kaye & T.H. Laby. Tabl
BHRA Report TN 411. Plinton

1.8. Pearsall, The velocity of water hammer waves. Symposium on
surges in pipelines.

FW Bridgeman. The physics of high pressure.

The author would be grateful for any additional data to supplement the

44. Young's Modules Source
a)

GRP properties will vary according to the proportion of fibre, bonding material
and method of manufacture. Information from external consultants, suggests that
the very low values quoted in reference 3 may be most appropriate for certain
types of piping,

45. Effective flow

arrest tme(t] Action

‘Modify ne ping system

‘Cary out fd How modelin
mod the piping system
adate denied pr

acton requred.
Critical piping systems -
ry out fu Now modeling, and
y te piping system to
accommodate identified p

Rationale

Fal Joukowsk Pressure and steep
These may result in

Rupture ofthe pre

High pipe sup

Large ping system delectons

As above, but with reduced severty

Fal Jookowsk Pressure and steep
‘but cannot be ignored for cical piping

cab

46. GBH ENTERPRISES, LTD

VULCAN Catalyst Process
Technology Consultancy
Sales and Service

Gerard B. Hawkins
Managing Director, C.E.0.

Skype: GBHEnterprises.
Office: [phone]
Cell: [phone]

[email protected]

[email protected]

Inked comin gearconawains
Enterprises