I have a customer who experianced the following which is something he does not want to repeat.  Does anyone have an idea what caused this?

 The Pump is a 3GT I-R Boiler Feed Pump, Steam Turbine Driven, 240-257 deg F water from DA tank
Pump was shut down and not turning but hot or at least warm
Discharge valve downstream of pump was closed first. This valve was some distance away from pump (20-30’?)
Pump has discharge check valve
Bypass Line to DA Tank from Pump discharge was open
Suction Isolation Valve (Gate valve) was manually closed – takes maybe 30 seconds
There is an expansion joint on suction side next to the pump
A loud explosion/noise was heard by the person closing the suction valve immediately when the suction line valve was closed and there was violent movement of the suction line to such an extent that the hold down anchors for the suction line pipe support (next to the pump) were ripped   from the floor. The expansion joint was extended to its maximum.

Operator was visibly shaken and doesn’t want to go through this again
 

If you have any ideas please let me know. I have asked the customer to check on a couple of theories I have but I want tell you what they are as I don’t want to color your thinking.

Sounds like surge, ie. water hammer, if as I am assuming they apparently had flow in progress from the suction through the pump to the deaerator when the suction valve was closed, since that branch to the DA is upstream of the discharge isolation valve.  Is it also upstream of the discharge check too, or is the discharge check immediately following the pump discharge nozzle?

Just trying to determine if it was (is) it possible for the DA or any other downstream pressure point to flow back towards the pump if the pump is shut down?

Are all pumps discharging into the same header?

Has a check valve been repaired since then?

It may not matter about the check, some are slow enough acting that fast transient surges are transmitted upstream right through them.  Saw that one happen on a nuclear plant feedwater sys.  They had a relief valve on the suction side to limit suction pressure buildups when the pump shutdown.  You might be experiencing something similar.

It does seem like there was a high pressure source from the pump discharge side and, since the pump was stopped it does indicate some vulnurability to reverse rotation and backflow from the discharge side to suction, although pump component failure could also be responsible.  The slow closing time of the suction line would point towards no generation of a suction line surge, as would a well placed working check valve in the discharge.  It should also be somewhat easier to tear up supports with the higher pressure source.  The description of an "explosion" would say that whatever happened occurred in a millisecond or so as well, so the possibility of reverse rotation being the cause lessens.  To me, it is shaping up like a pump component failure of some kind.  Is that how you see this so far?

I tend to favor the simplest answer in a problem like this.  I suspect that the pump pressured up to discharge pressure as the suction valve was closed and this was the source of the problem.  The source of high pressure could have been a leak backward through the discharge block valve and check valve.  Or, it could have been pressure from a common spill-back line to the DA if the valve on the end of that line at the DA was blocked in by mistake.  But simple leakage back through the block and check would be the most likely.  

You stated that the minimum flow spill-back was open to the DA but did not mention how this spill-back was controlled.  If there was an automatic spill-back valve (Yarway type), it would have defaulted open with no flow if it was working correctly.  But a control valve could have been closed depending on the logic programmed into the control system.  An orifice or pinched globe valve should have allowed the excess pressure to bleed off.  But then we are back to the possibility of operator error.  And once the failure occurred, the operator could have opened the valve that he had accidentally left closed.  Human nature would lead some to try and cover up their mistake.

I suspect the following scenario.  There was leakage back through the block and check on the discharge.  The spill-back line to the DA was blocked in or was otherwise restricted.  As the suction valve was closed, pressure built up.  The operator didn’t notice that the suction line expansion joint was moving.  It was only when the bolts snapped off (explosion) that he turned to see the suction line bellows extended and the pipe supports broken.  Depending on the pressure in the discharge and the design of the expansion joint, it may not be surprising that this joint extended but did not rupture.

More importantly than all of this, I need to point out that this system is very poorly designed.  If there must be an expansion joint in the suction line, it should be outside of the block valve.  The expansion joint should have tie rods in place and pinned with slight clearance so that it cannot extend as you describe.  If the expansion joint is in a location that could be exposed to discharge pressure by valve failure or operator error, there should be a PSV to relieve the pressure back to the DA.  Operators close the wrong valve all the time.  The system needs to be designed so that a simple error in the order of closing valves won’t result in serious injury or equipment damage.  

MORE NEWS