Water hammer caused by isolation of pressure relief valve

[dsg1985]

Hello,

In the design of a recent pipeline a fellow engineer and I discovered that the largest transient pressures in our pipeline would be caused by a pressure relief valve activating, and an operator shutting off the pipeline while it was activating. We were wondering if this is normally considered in design (neither of us are very experienced).

The pipeline is a raw water line feeding a mine. The line descends down a drift about 3km long and 250m deep. To limit static pressures a pressure reducing valve is located about midway down the drift (125m elevation point). Immediately downstream of the pressure reducing valve is a pressure relief valve set close to the downstream pressure setting of the reducing valve. If the reducing valve fails, the relief valve operates to ensure the static pressures at the bottom of the drift are not excessive.

To illustrate:

/////////z/////////Xo----------------

////////////////// : First 1.5km of pipeline
------------------ : Second 1.5km of pipeline
X: pressure reducing valve
o: pressure relief valve
z: isolation valve

When the reducing valve fails the PRV operates and needs to discharge 100L/second to cause a loss in the '///' section necessary to bring the pressure down at point X to the pressure of the reducing valve's downstream setting. This is a very high flow for the pipeline (normal flow is about 50L/s). The issue is the maximum pressures we had designed for up until this point were for the transients associated with suddenly stopping a flow of 50L/s. Now we had effectively doubled our transient pressure increase!

We did some calculations and worked out that if an operator suddenly stopped the flow at z, they would cause a transient pressure increase that would be beyond the burst pressure of the pipeline.

We solved the issue by putting the relief valve setting up slightly so that it only needed to discharge about 70L/second.

Has anyone else encountered this issue before?

Disclaimer - I used figures of 100L/s and 50L/s to simplify things, the flows were quite different, but the same principle applied.

 

[BigInch]

It is relatively easy to get very high pressures by combining wrong actions at precisely the right moment. It helps if the design isn't so good too.

In the event of an emergency, it must be possible shut down in a controlled and safe manner. Pressures and flowrates must be maintained within design conditions. That is a stable shut down. A stable shutdown procedure is required for safe operations. Seems to be that you have an unstable shutdown procedure and you should fix that.

Not all pipelines will operate at their design maximum pressures, mostly because transient pressures are not part of the standrard wall thickness sizing equation and transient pressures are customarily investigated after the preliminary wall thickness has been decided. Good thing you found it now.

Only put off until tomorrow what you are willing to die having left undone. - Pablo Picasso

 

[davefitz]

One measure used to reduce the magnitude of the water hammer is to use a very low speed in the final 5% of the closure stroke of the isolation valve. I have also heard of some pipelines using a mechanical stop to prevent remote auto closure to below 5% open- the "isolation valve" will always leak, and true system isolation would depend on a separate manual handwheel valve.

Supposedly the failure of the 100 mile long UG gasoline pipeline in Bellingham Washington ( 1999) was caused by waterhammer, when the DCS was reset following a DCS power supply failure ( a not-so -bumpless transfer)- the fail closed signal to a remote auto isolation valve fast closed it to 0% open- they had forgotten to install the mechanical stop at 5% open. Failure occurred at a section of pipe that was damaged by a contractor's backhoe a few years earlier.

 

[BigInch]

What kind of isolation valve design is to be left open 5%?
That was a failure to design the pipeline for full valve closure as it should have been designed in the first place. A ball valve at 90% travel to closure could still be 25% open. There can be a big difference in %travel and % closure, especially with ball valves.

Only put off until tomorrow what you are willing to die having left undone. - Pablo Picasso

 

[dsg1985]

Funnily enough a pressure reducing valve failure followed by a downstream relief valve activation is just the sort of situation that can cause an operator to go 'oh shit, leak' (and at 100L/sec thats quite a leak) and immediately run to the nearest isolation valve and close it *as quickly as possible*.

We did not have the option of offering a slow close solution in this case due to the length of the pipeline - it would need to be incredibly slow closing - partially because, as big inch pointed out, the majority of throttling can occur over the final 25% closure of the valve. Another reason was the fact one of the pipelines where we had this issue was poly (slow wave speed etc.)

We could not automate all the isolation valves for cost reasons (the fact this was a coal mine, which requires intrinsically safe electronics, drives this up even more).

We had one relief valve manufacturer offering a slow close mechanism on the relief valve itself, but due to the length of one of the pipelines (we had this issue on a few), and the fact it was poly, ruled out the possibility of using it (even at a 30s closure time, from memory, it was too fast).

In the end, as I said, we just made sure the transient pressures generated by the operator closing the valve ASAP were below the burst pressure of the piping, and had to adjust the relief valve settings to accomodate this. It became a bit of a juggling act calculations wise, but we got there.

 

[BigInch]

"run to the nearest isolation valve and close it *as quickly as possible*" and that's exactly what they should do in most, or certainly, similar situations, which is why the pipeline must shutdown within a stable state when they push that button.

The timing of startup and shutdown sequences can be critical, as you have found out. It is common for relief valves to reduce pressure in localized areas, which tends to accelerate inflow from other parts of the network, especially if pressures normally run relatively high and the pumps have not started to, or haven't finished shutting down. This can cause high pressures to be reached again, accompanied by another relief valve open/close cycle. The cycling can continue until the pumps finally shut down and the pipework is drained. I have had it happen on startups where shutting down the pumps was not part of the desired result. Sometimes you just have to resort to pulling a rabbit out of the hat.

Glad you sorted out the problem. Well done. I always feel better when eliminating the possibility of overpressures, rather than just making the symptoms tolerable.

Only put off until tomorrow what you are willing to die having left undone. - Pablo Picasso