DETERMINING SURGE POTENTIAL

IN WATER AND SEWAGE FORCEMAINS

AND PREVENTING OR MINIMIZING
THE CONSEQUENCES USING
AUTOMATIC CONTROL VALVES

by
Joseph E. Lescovich
Chief Engineer, GA Industries
March 1998

SINCE
1895

9025 MARSHALL ROAD

CRANBERRY TOWNSHIP, PA 16066 USA
PHONE: (724) 776-1020 ã FAX: (724) 776-1254
1
 DETERMINING SURGE POTENTIAL IN WATER AND SEWAGE
FORCEMAINS AND PREVENTING OR MINIMIZING THE
CONSEQUENCES USING AUTOMATIC CONTROL VALVES
By: J. E. Lescovich
A paper written by this author and presented at the Ductile Iron pipe “a” is nearly four times the speed of
Southwest Section of the AWWA in Austin, Texas in 1966 sound in air, or approximately 4000 fps. The surge wave
entitled “The Control of Water Hammer By Automatic velocity is a function of the pipe diameter, the pipe wall
Valves” has been given to interested people for over thirty thickness, and the modulus of the pipe material. The other
years. Now over thirty years later this writer has decided to factor affecting the velocity of the surge wave is the water
re-write the article to include some more useful informa- bulk modulus, which itself is affected by the water tem-
tion and incorporate some practical information gained after perature and its pressure. Changing water pressure and
more than forty-five years of minimizing pressure surge temperature, however, have little effect on the surge wave
problems by the use of automatic control valves. velocity. Dissolved air however, can have a significant
effect on the wave speed. It has been written that 1 part
Prevention of pressure transients (surges) should be of air in 10000 parts of water by volume at standard con-
one of the first considerations in an Engineer’s mind when ditions can reduce the wave speed by about 50 percent.
designing a pumping system or flow control facility. This The equation for determining the surge wave velocity is
article will emphasize some of the factors to be evaluated expressed by various equations. One such equation is
when designing a pumping system. While conceding that shown here. (A pipe restraint factor sometimes mentioned
the analysis of pressure transients in a pumping system being omitted.)
can be a very complex subject, and knowing there are many
fine technical publications and computer software programs a = 4720 / sq.. rt. (1 + k d / E t)
available, this writing will provide a shortened version for K represents water bulk modulus which is nearly
estimating the surge potential in a pumping system. Em- 300,000 psi. (NOTE: the modulus for steel is nearly
phasizing again, this paper should not replace a compre- 30,000,000 psi which means that water is 100 times
hensive surge analysis where one is required. The article more compressible than steel. All liquids have their
will suggest valves than can be employed to prevent the own bulk modulus, 230,000 PSI for crude oil, 130,000
surge problems or by the use of overpressure relief valves, psi for gasoline, etc. Illustrating that liquids are very
minimize the consequences. elastic substances.) “d” being the pipe diameter in
inches. “t” is the pipe wall thickness in inches. “E” is
REVIEW OF SURGE THEORY the pipe material modulus. A few examples are given.
Before attempting to deal with pressure surges, it would (All values are in psi.)
be appropriate to review some basic principles regarding Steel ............................... 30,000,000
the cause, magnitude and prevention of surge pressures. Ductile Iron ................... 24,000,000
Pressure Transients or Surge, sometimes called Cast Iron ........................ 15,000,000
waterhammer when a noise is associated with it, results Asbestos Cement ............. 3,400,000
when any attempt is made to alter the velocity of a column PVC .................................... 400,000
of water. It matters not if the column of water is at rest HDPE ................................. 110,000
and is being put into motion, or if the water is in motion
and is being brought to a stop or slowed down. The mag- A graphical chart is included which can solve for
nitude of the resulting pressure surge is directly propor- surge wave velocities for any liquid or pipe material up
tional to the rate of change of velocity produced on the to 96 inch dia. (See Figure 1)
water column. It is not the total velocity change produced For example, a 48 inch Ductile Iron pipe with a 0.51
over a period of time which is the prime concern, but the inch wall thickness would have a surge wave velocity of
rate of change. This factor will be elaborated on later in 3228 fps. The equation given earlier for hmax is derived
this writing. By applying Joukowsky’s equation for deter- from Newton’s force and impulse equations, or rate of
mining the pressure rise in a pipe, change of momentum. This should not be confused with
hmax = a dV / g Newton’s F=MA expression which applies to solid or rigid
bodies. Water is elastic, and there is no physical process
In English units, “hmax” equals the headrise in feet known where a force can be applied to all particles of an
of water, “dV” being the change in the water velocity in elastic substance simultaneously. The equation assumes
feet per second, and “g” being gravity or 32.14. The “a” the velocity change was produced instantly.
denoting the velocity at which the surge wave travels once
created. The surge wave travels at the speed of sound in It is commonly known that nothing can be done in-
that particular pipe. The velocity “a” is not the same for stantly, however, it is not commonly understood that a
every pipe. For example, in a typical small diameter flow need not be stopped instantly to obtain the same
result as an instantaneous flow stoppage. It so happens in Upon reaching the closed valve, the downsurge again
hydraulics, that any flow stoppage occurring within one becomes an upsurge and this cycle keeps repeating
surge period is equivalent to an instantaneous flow stop- resembling a square sine wave until pipe friction re-
page. duces the pressure in ever diminishing intensities to
static conditions. Should a downsurge intensity ex-
A surge period must now be defined. To simplify ceed the available head, water column separations may
this concept, visualize a long straight pipeline, for ex- occur, but such events are too involved to discuss in
ample, a 48 inch D.I. pipeline from a reservoir with a this writing.
valve at its discharge end. Assume also that water is dis-
charging from the pipe and valve at say ten fps neglecting A very important point to remember from the above
line friction. The potential surge or headrise would be description is the time required for the surge wave to make
hmax = 3228 x 10 / 32.2 = 1002 feet. You may have one round trip of the pipeline. That time is known as the
noticed the headrise in feet is 100 times the flowing ve- “Surge period” (T=2L/a) sometimes referred to as a “Criti-
locity in fps. This will be true for this pipe at any flowing cal period.” The reason this time period is so important
velocity because a constant can be made for 3228 / 32.2 is that any flow stoppage made within one surge period
which is nearly 100. At 5 fps, the resulting potential duration or less results in a surge equivalent to an instan-
headrise would be 500 feet. For each pipe diameter, taneous flow stoppage. In the previous example, the po-
wall thickness, and pipe material, a similar constant tential surge was 1000 feet plus the static head for the
can be determined. instantaneous stopping of a flowing velocity of 10 fps.
From the above statement, any total flow stoppage in one
When the discharge valve flowing at 10 fps is closed surge period (20 seconds) or less will result in an equiva-
suddenly, the flow at the valve inlet is suddenly stopped lent surge of 1000 feet. For a partial valve closure, the
with a pressure wave created of 1000 feet which begins initial surge will be calculated using the amount of ve-
its travel up the pipe at the speed of sound or 3228 fps. locity destroyed within one surge period.
The water at the valve’s inlet is of course stopped, but
water at the reservoir end continues to enter the pipe be- When closing a valve slowly and uniformly, the same
cause water there feels no effect of the discharge valve procedure is followed, except for the second increment
being closed, and it will not feel any effect until the surge of valve closure the initial upsurge is subtracted from
wave arrives at the reservoir. The time required for the second time period upsurge because, the first period up-
surge wave to arrive at the reservoir depends on the pipe- surge has traversed the system and is now beginning its
line length. If we assume, for simplicity, the pipe is 32280 second round trip as a downsurge while upsurges are still
feet long, then the surge wave will require 32280 / 3228 being generated. And so on, and so on, to where the re-
or 10 seconds to reach the reservoir. sults would appear as an arithmetic integration tabula-
tion. Computer programs perform these steps routinely.
At this time it should be remembered that the surge
resulting from closing the discharge valve is only the in- Pipelines frequently have branches which can com-
crease in pipeline pressure not the total pressure. The plicate the analysis. A surge wave reaching a pipe junc-
static pipeline head must be added to the surge pressure tion or lateral, causes the pressure wave to travel both
to obtain the total pressure. To re-emphasize, surge is branches. Should one branch be shorter than the other,
independent of the existing pressure. An upsurge is the the upsurge wave reflecting from the end of its branch,
increase in pressure, not the total pressure. Assuming the will begin a return to its point of origin. During its trav-
reservoir static head is 100 feet then, the total head in the els the upsurge may encounter downsurges or perhaps
pipe would be (Capital H) Hmax or 1000 plus 100 equal- upsurges traveling in the opposite direction. Pressure
ing 1100 feet. At time equals 10 seconds after valve clo- waves traveling in opposite direction will pass through
sure, the surge wave arrives at the reservoir, the pipeline each other without interfering with each other. The pres-
is now pressurized to 1100 feet of head. With a reservoir sure intensity at the point of interception is the algebraic
head of only 100 feet, the pipeline water in a compressed sum of each pressure intensity.
state (which is what pressure is), immediately begins to
relieve its excess pressure back into the reservoir creat-
ing a reversing pressure wave traveling at back towards VALVE CHARACTERISTICS
the closed valve at the same speed of sound. This return- In the above example, an unspecified valve at the
ing surge wave will require 10 seconds to reach the closed pipe’s end was closed suddenly. It was shown that clos-
valve. At that time the head in the pipe for an instant is ing that valve instantly or in any time up to 20 seconds
equal to the static head, or 100 feet. The required time would produce the same results. We will now review the
for the round trip of the initial upsurge is therefore, 20 effects of various real valve types should they have been
seconds. employed.
A second surge wave is immediately created which Every valve type during its opening and closing stroke
is a downsurge and a mirror image of the upsurge but will effect flow differentially, which is called the valve’s
of the opposite intensity. This downsurge wave like- characteristic. For example, assume the valve at the end
wise travels to the reservoir and returns as a downsurge. of the pipe is a common gate valve, and assume this valve

3
was closed uniformly in say 60 seconds. Also assume minimize surges. An example of estimating a required
the valve’s flow and stroke were recorded during its clo- valve closing time may be seen in the following example.
sure. It would be found that the gate valve could be closed
about 50 percent with little noticeable change in the flow. Using our previous example of a pipeline 32280 feet
Closing the gate valve another 10 percent may reduce long, flowing at 10 fps, select a gate valve closing time
the flow only about 4 percent. Notice, at 60 percent which will limit the surge to 100 feet, and also select a
closed, this gate valve has 96 percent of the flow still closing time for a GA Vee-Ported valve. (For headrise
passing through it. It is during the final 20 percent of the versus valve closure -See Figure 4)
valve stroke where about 75 percent of the flow is cut- The potential surge established previously was 1000
off. The cut-off characteristic resembles parabolic curve feet for a total flow stoppage in 20 seconds or less. Lets
from which it can be concluded that an effective closing assume the allowable surge is 100 feet or 10 percent of
time of a gate valve is roughly about 30 to 35 percent of the potential surge. Referring to the graph fig. 4 which
its total time. (See Figure 2) shows headrise versus valve closure time. The vertical
scale represents percent of surge for an instantaneous flow
COMPARISON OF VALVE EFFECTIVE stoppage, and the horizontal scale showing valve closure
CLOSING TIMES time in units of Surge Periods (2L/a). In our example,
An interesting observation can be made, that closing each unit on the “Perfect” valve scale represents 20 sec-
this gate valve in 60 seconds stopped the flow in an onds. Calculate the “K” or pipeline constant. The equa-
equivalent time of about 20 seconds. Remembering that tion is given where K = a v / (2 g H). “a” = the surge
one surge period is 20 seconds, the resulting upsurge wave velocity in fps as given previously. “v” = the pipe-
would still be 1000 feet, the same as for an instantaneous line water velocity in fps. “g” = 32.14. “H” = the normal
flow stoppage, even though the valve was closed in one system head, in this case reservoir head. For our example,
minute. It should now be apparent that the valve closing K = 3228 x 10 / (64.34 x 100) = 5 On fig. 4 graph, read
time does not equal flow stoppage time. For the gate horizontally to the “5” curve, then down to the result of
valve it was the final 30 percent of the valve stroke which about 7.5 on the “Perfect” scale. This scale represents a
produced a very high rate of change in the water column perfect valve with a perfectly linear closing characteris-
velocity. The effective closing time for a Butterfly valve tic. Such a valve would need to be closed uniformly in
may be assumed as about 20 percent of its total time. A 7.5 surge periods or 7.5 times 20 seconds or 150 sec-
Rotary Ball valve or Cone valve will have an effective onds. The valve being a gate valve, requires reading down
closing time of about 50 percent of its total time. to the gate valve scale, to 19 surge periods which is 19
times 20 or 380 seconds (6.33 minutes) of valve closure
The Vee-Ported Differential Piston valve as manu- time. A GA Vee-Ported valve would need to be closed in
factured by GA Industries, is comprised of a piston which 14.5 surge periods, or 14.5 times 20 equals 290 seconds
moves up and down within a Vee-ported cage. As the (4.8 minutes). These results are fairly good approxima-
piston rises six or eight small triangles initiate flow. As tions, but to be more sure using the actual valves installed
the valve opens further, the triangles become larger in- characteristic curve with any of the Transient analysis
creasing the flow in a programmed manner. When the computer programs will provide more comprehensive
valve is fully open, the total areas of the Vee-Ports equals results. Pipeline constants are typically about 1.0.
the full area of the nominal pipe diameter. During its When pipeline constant K’s become larger this implies
closing stroke, the Vee-Ports become increasingly smaller a larger momentum exists for the water column. Wa-
similar to a vee notch weir pinching off the flow in an ter column separation potential should then be investi-
ideal manner. This provides the Vee-Ported valve with gated.
an extended effected closing time of over 60 percent,
meaning the Vee-Ported valve can be closed more quickly
VALVE CHARACTERISTICS
than the above valves with less resulting pressure rise.
(For comparison of valves 50% open - See Figure 3) The fact that all valves do not perform equally was
mentioned before. Every valve type can be represented
To be most effective, the Vee-Ports should be down-
by its Inherent Characteristic curve. Such a curve plots
stream of valve’s seating surface. Throttling is then done
the valve stroke versus its Cv. The Cv as you may know
by the Vee-Ports and not by the valve seat where mainte-
is a flow capacity coefficient, and for water is found by
nance would be a problem. It is possible to close valves
dividing the GPM flow through the valve by the square
that have unsatisfactory closing characteristics in two
root of the differential pressure across the valve in psi.
stages. The initial closing stroke made at one speed and
Plotting a curve of Cv versus stroke for a valve for each
the final closing at a slower speed. This requires some
increment of its stroke produces the Inherent Character-
additional control piping.
istic curve that can be used to compare valves on an equal
SURGE PREVENTION basis. In simplified terms, the curve shows how many
gallons per minute of flow is required to produce a one
From the preceding discussion, the reader should have psi pressure drop at that particular stroke position. (In
become aware that long pipelines equals long surge peri- European units the Cv becomes “K” with flow is in
ods, and slower the valve’s closing times are required to M3/hr, and pressure in Bars.)
 There is another valve characteristic curve called which is a standard feature, permits the check valve to
the Installed characteristic curve which depicts the flow close at a much faster but adjustable rate to minimize
through a particular valve in an actual installation. pump reversal.
Curves are drawn by stroking the valve fully and plot-
ting stroke versus actual flow. The valve’s differen- The pump Control Check valve can be provided with
tial pressure is automatically accounted for. This curve additional controls to perform secondary functions. This
will change for the same valve in every installation. could be a back pressure sustaining or discharge pres-
sure control. There are many options that the valve manu-
facturer can inform you about. An article entitled “Check
SURGE PREVENTION (CHECK VALVE Valves With Brains” by this author offers a description
SELECTION) of several of these features.
The previous discussions described how surges are
created. The obvious conclusion to minimizing surges, SURGE DISSIPATION
is to produce gradual changes in the fluid velocity. The The Electric Check valve just described is a valve
importance of valve closing time is not specifically in which prevents surges normally associated with the start-
seconds, but in units of surge periods. A pipeline that is ing and stopping of pumps. However, surges cannot al-
only 400 feet long, for example, with a surge wave ve- ways be prevented such as those resulting from a pump
locity of say 4000 fps, will have a surge period of only 2 power failure. For such occasions, devices are employed
x 400 / 4000 = 0.2 seconds. Each second of valve clo- to minimize the consequences. There are Surge Tanks,
sure time or each second of pump coastdown time rep- Surge Chambers, Surge Relief Valves, Surge Anticipa-
resents 5 surge periods. Pumps often coast for several tors, and Accumulators to mention a few. Tanks and
seconds on a power failure, except for very high head chambers are large unsightly expensive and require fre-
installations. There are pumps that coast for long pe- quent inspections. This article will elaborate mainly on
riods producing in effect a gradual flow stoppage. (See the Surge Relief valve and Surge Anticipators to mini-
Figure 5, 6, 7, & 8) mize surge intensities.
Any simple mechanical check valve, such as a swing
check, tilting disc, wafer check, etc. are at the mercy of SURGE RELIEF VALVE
the flow. Such valves can only respond to the flow, and
therefore cannot affect a gradual rate of change in the A Surge Relief valve is normally installed on a tee
fluid velocity. Every pump shutdown with simple check directly downstream of the pump check valve. Surges
valves is therefore equivalent to a pump power failure. are always greatest at their point of creation, so it is logi-
The resulting surge can be determined as described ear- cal that the over pressure control be located near there.
lier. Shutting down a pump employing a simple mechani- Surge Relief valves are usually arranged to discharge to
cal check valve may result in a check valve bang. Do not atmosphere, sometimes to a wet well or clear well. Surge
equate the noise with a surge. The check valve slam is a valves can sometimes be arranged to discharge back to
mechanical problem - there may actually be little surge. the Pump suction if conditions there are considered. The
Also, do not assume that a quietly closing check valve Surge valve opening setpoint is typically a pressure about
means no surge occurred. Surge is an hydraulic phe- 10% to 15% above the pump maximum discharge pres-
nomenon. Pump control check valves such as power sure, and when responding to overpressure, the valve
operated Cone valves, Rotary Ball valves, GA Elec- opens only as much as required to prevent further pres-
tric Check valves and the like, change the flow gradu- sure rise. (See Figure 11 thru 16)
ally on startup or shutdown of the pump. Surges there- There are Engineers that view the Surge Relief valve
fore can be limited to any desired intensity. (See Fig- as a safety device and therefore prefer to have two valves
ure 9 & 10) (sometimes more) in parallel. In such instances, each
valve has a slightly higher set point. The valves then
A pump with a GA Electric Check valve installed
open in sequence, but only as much as necessary. The
would function as follows. When the pump is first started
Relief valve is a pilot operated valve that does not have
the pump check valve is closed. As the pump pressure
the accumulation problem inherent in a direct spring
exceeds the system static head, the Electric Valve is en-
loaded valve. (See Figure 17)
ergized to open at a controlled rate producing a gradual
acceleration of the water column. The surge can be held A question often asked about Relief valves, is the
to any desired minimum. When a pump shutdown com- valve’s ability to respond to surges traveling at the speed
mand is given, the pump continues to run while the check of a bullet. The answer is that surge waves do travel at
valve begins a controlled closing sequence resulting in a great speeds, but the rise in pressure is at the same rate
gradual deceleration of the water column. Just prior to the surge was created. A pump coastdown of just a few
the check valve seating, the check valve actuates a limit seconds, produces a surge wave with a front resembling
switch to shut down the pump. Valve operating times an inclined plane traveling very fast. The horizontal
can be adjusted for several minutes if line lengths require movement is rapid, but the vertical pressure rise is gradual
it. Should a power failure occur, an emergency solenoid compared to surge valve response time.

5
SURGE ANTICIPATOR VALVE SIZING THE SURGE ANTICIPATOR VALVE
In pumping installations where the surge pressure rise We stated earlier that over pressure relief valves can-
is very abrupt, or in installations where three way surge not be oversized. However, the Anticipator valve should
protection is desired, a Surge Anticipator valve can be not be oversized. The Anticipator valve opening on power
employed. The Surge Anticipator valve is programmed failure or pump failure will open fully. The valve could
to respond to the cause of a surge prior to the arrival of then establish an undesirable and unnecessary excess re-
the surge itself. (See Figure 18) verse flow.
A power failure will result in a surge, so will a pump It is difficult to present in this writing a sizing method
failure which is not a power failure. The Surge Anticipa- which is correct for every case, or knowing if the reader
tor, referred to as the Surge Sentinel, opens immediately will apply the input data correctly.
upon pump power failure by way of a solenoid pilot valve,
or a pump failure such a broken pump shaft, motor wind- A sizing procedure will be presented here mainly to
ing failure, etc., by way of a pressure switch. The Surge be used as a guide. Refer to fig. 19 chart. Read across
Anticipator will remain fully open for an exact number the bottom scale to the pipeline velocity. Next, read up
of seconds, awaiting the completion of the downsurge to the diagonal line representing the head in feet, and
cycle and creation of the upsurge, then begin a controlled then left to the vertical scale. Read the required anticipa-
closing sequence limiting the upsurge to an acceptable tor valve discharge capacity as a percent of the line flow.
pressure. Any subsequent over pressures that occur above To determine the proper angle body valve or globe body
a preset pilot setpoint will cause the valve to reopen as valve refer to charts fig. 20-23. On the USGPM scale
much as necessary as an over pressure relief valve. locate the required discharge rate obtained above and read
vertically to the horizontal line representing the valve
relief setpoint in feet. Chances are you will arrive at a
SURGE VALVE SIZING
point between two valves sizes. The conservative deci-
Sizing an overpressure relief for water service either sion is to select the larger of the two sizes. If more than
raw water or filtered water, equation labeled “Relief Valve one relief valve is preferred, then divide the flow by the
Sizing” may be used. Relief valves such as the fig. no. number of valves, and refer to the charts again. The dis-
6600-D (Angle pattern), or the fig. no. 6700-D (Globe charge charts apply to the valves alone, if a long or com-
pattern) cannot be oversized. If any back pressure ex- plex relief valve discharge piping exists that could affect
ists, or any extraordinary relief valve discharge piping the discharge capacity of the relief valve, then the relief
exists, then the valve size may need to be increased. valve size may need to be increased.
When applying a surge anticipator to a pumping sys-
RELIEF VALVE SIZING EQUATION tem with a widely varying flow rate, it is suggested that
A common equation for sizing pilot operated relief there be two valves selected. One valve being a surge
valves for water is, Relief Valve Size for water = C times anticipator valve and the other being a standard over pres-
Sqrt (gpm / sqRt (PSI)). C = 0.284 for Angle Valves, or sure relief valve. Being both the same size, their parts
“C” = 0.326 for Globe Valves. are interchangeable. If desired, the relief valve can be
converted to an anticipator. This will prevent oversizing
The sewage surge valve size may be determined by the anticipator at lower flow rates.
referring to chart fig. 24 with similar comments as above.
The valve fig. no. would be 625-D for an LR Elbow pat- Should a valve application problem arise, contacting
tern valve in sizes to 8 inch. Valve sizes 10 inch to 12 a valve manufacturer experienced in this field will afford
inch in the LR Elbow pattern valves have a fig. no. of you a satisfactory solution.
624-D. The fig. 626-D being a Wye pattern valve may
be specified when an inline flange valve is preferred.
7
9
11
13
15
17
19
21
23
25
27
29