WATER HAMMER

This chapter will introduce the concept of water hammer. The following topics will be covered:

Definition of water hammer or surge conditions

Causes and consequences of surge

Case study illustrating water hammer

Down-surges

Most pipeline engineers are quite familiar with the “water hammer” effect, also known as

‘surge’. This is the term given to the build up in pressure that occurs on the upstream side of a

valve as the valve is being closed to halt the flow of liquid. What is somewhat more obscure is

the sequence of events after the valve is closed and the flow has stopped at the valve. The course

of these events depends to a very large degree upon the piping and other equipment in the system

which can ‘see’ the disturbances caused by shutting the valve.

Causes and Consequences of Surge

Many different kinds of events can cause surge. Some of them include the following:

Opening, closing or chattering of valves in a pipeline.

Starting or stopping the pumps.

Starting up a hydraulic turbine, accepting or rejecting load.

Vibrations of the vanes of a runner or an impeller.

The following are a few examples of surges affect the system:

Pump tripping.

Check valve closure/flutter.

Maximum allowable operation pressure exceeded.

Column separation.

Pipeline rupture or collapse.

Figure 5.1 above, shows a simple example of what pressure transients look like under surge

conditions. These pressure transients are created by the closure of the valve. On the upstream

side, an upsurge condition is created, and on downstream side, a down surge condition is created.

The best way to illustrate the effects of water hammer or surge is to examine a case study which

is simple, but leads to somewhat striking consequences.

Case study Figure 5.2 schematically illustrates the system that we will study. A tall, water –filled stand pipe

on a hill maintains a constant pressure of 350 psi at the inlet of an inclined 48 inch, 80,000 foot

pipe, which drops 500 feet in elevation in the direction of flow. Initial flow is at 2,000 MB/D

with a 425 psi pressure drop across the valve at the bottom, which discharges into an open tank.

Valve Closure

Let’s start by closing the valve smoothly so as to reduce the flow to zero in about 40 seconds.

What happens?

Viewed at this time from a steady state point of view, we immediately bring up the mental

picture of a hydrostatic pressure gradient in the pipe which has a 565 psig pressure at the closed

valve, and no flow anywhere in the pipe. When viewed from a transient point of view, the

situation is a far different one. After a long time has elapsed, the steady state image is, of course,

a correct one. But the path by which we arrive at this condition is a very interesting, and

sometimes a very dangerous one.

A Look Between the Pipe Ends

In order to see what happens in the system, we must not confine our attention to the ends of the

pipe. To get the whole picture requires the recognition that the pipe is a unit in which

disturbances generated at one end move in very specific ways toward the other end. To see the

results of valve closure, we shall examine some time-lapse snapshots constructed by the

computer of the pressure and flow profiles of pressure transients due to water hammer effect,

caused by the closure of the valve. These are shown in Figure 5.3, and are numbered in order.

Graph (a) shows the flow rate at the tank and at the valve, and graph (b) shows the pressure at

the tank and at the valve.

Snapshot 1 is taken at the initial steady state conditions. Each subsequent snapshot corresponds

to a time lapse of 10 seconds, for a total elapsed time of 150 seconds or 2.5 minutes.

Snapshot 1 at time zero shows the original uniform flow at 2000 MB/D. The corresponding

pressure trace increases from 350 psig to 425 psig as we progress down the pipe, because the

gravity head more than overcomes the friction due to flow at this rate.

Snapshot 2 at 10 seconds shows the effect of beginning the decrease in flow rate as the valve

begins to close. Note the concurrent increase in pressure, to over 500 psig, due to a decrease of

only 20% in the flow rate.

Snapshot 3 at 20 seconds corresponds to an outflow rate of half the original and a pressure rise at

the valve to over 700 psig, or almost 150 psi above the maximum hydrostatic pressure possible

in the system.

Snapshot 4 at 30 seconds shows that between 20 and 30 seconds the flow rate entering the pipe

from the tank begins to decrease. If we calculate the sonic velocity, or the speed with which

disturbances move through the water pipe system, we find this velocity to be 3,370 ft/sec. Thus,

the time for a disturbance to travel the 80,000 feet length of the pipe is 24 seconds. We see the

effect of the disturbance generated by the start of valve closure is to ‘telegraph’ the fact that the

rate must go down; however, this fact arrives some 24 seconds after the closure begins. Another

way of saying this is that for 24 seconds after we start closing the valve at the outlet, water

continues to enter the pipe at the initial rate. Of course, the effect of this is to pack the pipe both

with water and with energy.

In Snapshot 5 at 40 seconds, we have just stopped flow at the valve. Also we have generated a

water hammer pressure rise to over 1,000 psig, or about 500 psi higher than the hydrostatic

pressure would have been. Note that the tank is forcing the inlet pressure to remain fixed at 350

psig , and the valve is forcing the outlet flow to remain at zero. The pressure; however, is free to

vary with time throughout the remainder of the pipe, and so it does. Similarly, the flow changes

rapidly and by large amounts, as we shall see.

The remarkable fact in Snapshot 6 at 50 seconds is that the flow throughout the pipe is now all in

the direction toward the tank. Also the pressure, which is anchored at 350 psig at the tank, is

sloping sharply upward all along the length of the pipe, with a gradient much higher than the

hydrostatic gradient. This unbalanced pressure gradient contains most of the energy that did

reside as kinetic energy in the moving water. This gradient must cause acceleration, which is

evident in Snapshot 7 at 60 seconds and Snapshot 8 at 70 seconds ,in which the flow is negative

(into the tank) with a magnitude which at 70 seconds is almost as large as the flow out of the

tank had been at the start.

This large negative flow unpacks the pipe, whose pressure begins to drop rapidly, and whose

compression energy is converted into the kinetic energy of flow. By, the time of 90 seconds in

Snapshot 10, the pressure at the valve is only slightly over 100 psig, while at a point about

halfway up the pipe the pressure falls to about 50 psig. Associated with this pressure profile is

again an almost zero flow and an unbalanced pressure gradient which when added to the gravity

head serves to accelerate the water toward the closed valve once more.

The effect of this acceleration can be seen in Snapshots, 11, 12, and 13 which show that flow

from the tank once more builds to over 1,800 MB/D. The similarity of Snapshots 13 and 4

strongly suggest that from this point the whole process is to be repeated many times over until it

finally dies out due to friction. The fact that the decay has been slight during the first cycle

suggests that many such cycles will occur in this natural period of resonance, before the

hydrostatic situation will be attained.

Why is all this important

The foregoing is more than an interesting exercise for the pipeline engineer. Generally, the

engineer does not have such a simple system and the disturbances generated at one point move

throughout the system causing other problems. The engineer must design for the water hammer

itself, to make certain that the closure of a valve will not cause pressures that rupture the pipe at

the valve where the surge is most severe.

But more far reaching is the fact that the surge is not a one time occurrence and the system

continues to resonate. If the system were branched, and other disturbances are generated in the

branch while the main pipe is resonating, the upsurges from the branch can add to those in the

main system. Therefore, while each branch taken separately may be protected against surges

developed by valve closings, pump failures, etc. within its immediate equipment, the system

taken as a whole might not be adequately protected.

Down-surges – The Paradox of Cavitation

Another interesting situation can be observed from Snapshot 10. Note that while the pressure at

the tank is 350 psig, the pressure in the middle of the pipe is at 50 psig. We might imagine

performing a similar study on a system with a pressure in the tank of 50 psig initially, with a

pressure at the valve of 125 psig. Then the process we studied would begin as before, the only

difference being that the pressure level everywhere would be 300 psig less. What then would be

the situation at 90 seconds corresponding to Snapshot 10? A pressure in the middle of the pipe

which is 300 psi less than that shown in Snapshot 10 would be -250 psig. This is of course not

physically realizable, because as the pressure falls below the vapor pressure of the fluid it will

boil, or form bubbles ,a situation known as cavitation, one of which the liquid simply pulls

apart. This situation is often very damaging, particularly if we had any kind of equipment

installed there that the cavitation might destroy. Even in the absence of equipment, cavitation is

undesirable because of the surge pressures which are caused later when the two separated

columns collapse together.

Our experiment at the lower pressure level with our conclusion that it leads to a very undesirable

result is disquieting. We begin with a tank full of water draining through a valve, with pressure

everywhere 50 psig or above. Then by simply turning off the valve we cause the paradoxical

result of cavitation, which one normally associates with the loss of supply. It is paradoxes such

as this that make the dynamic design of hydraulic systems of critical importance.

Summary

This chapter introduced water hammer, also known as surge. The discussion talked about the

causes and consequences of surge and discussed the dangers by examining a simple case study.