4 replies to this topic

[El Compostador]

Hello: I am in need of advice as I managed to ruin my first large pipe project. It is embarrassing to admit that I basically operated a pump with a valve in the effluent line in the closed position. What I don’t understand is why two well supported SCH 40 PVC fittings rated for 150 psi broke when the pump shut off pressure is only 45 psi.


I had a centrifugal sump pump pumping 7500 gpm water out of a 15 ft deep sump into two open tanks (sand filters). Please see attached piping layout in pdf form The 16 inch effluent from the pump is split into two 12” pvc lines via a 16x16x16 tee and two 16x 12” reducing couplings. Each of the 12” lines discharges into a sand filter via a 12 inch butterfly valve. I tested the system manually and managed to reach the desired set points of 4500 gpm into one filter and 3000 gpm into the other with no problems for about three hours. A pressure transmitter upstream of the butterfly valves normally read 0.9 psig but would sporadically read –0.1 psi!.

The problem arose when I tried to operate the valves automatically via an actuator controlled by the effluent signal from the filter effluent line flowmeters. The system ran ok for 15 minutes and as soon as I got to my office, operators called me with the news that both the elbow and tee were broken. Apparently, a problem with the control program caused both valves to close and the pump to shut down. To make things worst, my safety measure of automatically shutting down the pumps if the pressure exceeded 20 psi did NOT activate. I was trending the pressure every thirty seconds but I can’t see the pressure ever exceeding 1 psi. It is hard to believe that three months of work in putting this together ended in 30 seconds!

Since this was a temporary three-month project, I did not think it would be necessary to have a pressure relief valve since I though that the pipe could handle it. I also felt good about my pipe supports. The 40ft long pipe was anchored using two pipe support saddles and both ends and supported on wood blocks (8 inches above ground) every ten feet ; the tee was actually braced with a right/angle u-bolt setup. Obviously, I was wrong.

Could anyone also please explain why a good pressure transmitter would read vacuum (-0.1 psig) in a 12” pvc pipe that is flowing with 7,500 gpm of water? Does this mean that I had vapor bubbles?

It is actually easy to replace the elbow and tee but I don’t want to do it unless I understand why things happened.

[Art Montemayor]

compostador:

We can’t effectively give you constructive advice because you didn’t attach the .pdf file (which has a descriptive schematic sketch of the installation, I suppose).

However, I have some general comments/questions:

• I agree with your sound engineering decisions regarding the omission of a pressure relief valve. I suspect that you had a pressure surge (or water hammer) and a PSV would not have saved the Tee or the Elbow. It just cannot react fast enough – nor is it intended for that purpose.
• I would expect that you installed Schedule 40 PVC fittings, as well as pipe.
• I hope your pipe supports were stout and well-anchored and bolted down. Loose wooden blocks would not have been sufficient to absorb any hammer shocks and the displacement would have been transmitted to the weakest part or joint – which in this case, turns out to be the fittings. The fittings is where you have a change in direction and a potential for hydraulic forces concentrating any impact there. That is why I would install Schedule 40 fittings. I also presume the fitting were solvent-welded in a socket fashion.
• You are dealing with a considerable amount of flowrate. At 16” pipe, you are dealing just below a “big bore” size of pipe (24”) and you will develop very high hydraulic forces. You may not have much pressure, but remember that as the pipe cross-sectional area increases, the lateral forces increase (Pascal’s Law, Force = Pressure x Area).
• I hope you will supply details on the installation such as pipe supports, pipe & fittings details and descriptions, pipe and fittings joints, instrument types and reaction time, photos of the pipe/fittings failures, etc.

Await your attachment .pdf.

[El Compostador]

Mr. Montemayor:

Thank you so much for your reply. I hope this time I succeed in attaching the file. The devil is in the details and there are so many that I omitted in the interest of simplifying my inquire. For instance, as you can see from the drawing, I had a second pump as a backup connected to the 16 inch main via a check valve. However this second pump was always off. In regards to your comments:

- Fittings were socket type and solvent welded to flanged pipe nipples. The elbow and tee were fabricated schedule 40 pvc; long runs of pipe as well as flanges, and one 16x12 reducer were schedule 80. This mismatch was due to economical constraints as I was required to use as much material as I could find in our boneyard. The tee was recycled as you can see from the tarnishing of the pvc.

- the actual pipe supports have been removed (scavenged by others) but I am attaching manufacturer’s sketches. The pipe saddles and 90degree bracing bracket were hiltied with 3/8 concrete anchors rated for 3000 lbf shear/tension.

-Regarding the hydraulic forces, two operators claim that they did not hear the loud bang associated with a water hammer. Now that you mention this, the force could be calculated by the change in momentum due to the flow direction. In the case of the tee the velocity changes from about 12 ft/s to 0 ft/s; the force is then
Force = density x flowrate x velocity change = 64 lb/ft3 x 16.7 ft3/s x 12 ft/s x 1/32.2 lbft/s2 <400 lbf.
If this force is exerted over the area of one pipe diameter (1.6 ft2), then the pressure would be ~ 300 psi!

This alone could explain the fitting failure since it is only rated for 120 psi when new; the recycled fittings I used were probably not even that strong. I now see that you meant that you would use schedule 80 fittings instead of schedule 40.

I also suspect that the u-bolt support was a really bad idea since the U-bolt is only 5/8” diameter and its support area is very small and actually strains the fitting (acting like a pipe cutter?). Would you happen to know of a better support system for large diameter pipe?.

Thanks a lot for all your help so far. I now think that using Schedule 80 fittings, and better support system would ensure that this accident does not happen again.

Another missing detail is that the pump is not anchored to the floor. The basin is always full with water so it is impossible to anchor it. The first ninety degree elbow might still be in danger if the pump moves due to accidental shut down, but I prefer that this would be the piece to fail rather than underwater piping.

Would you also please comment on why the pressure transmitter would read negative pressure?

Attached Files

•  pvc_fitting_failure.pdf   280.29KB   28 downloads

[kkala]

I agree to measures mentioned. Besides following may be worthy of consideration:
1. Anchor pump, e.g. by making a concrete base in the sump. Connect pump to the pipe with an expansion joint (pump flanges should not be used as fixed points to equalize pipe displacements).
2. Distance is short but speed is rather high, so water hammer cannot be excluded. Check closure time of (butterfly) valves in automatic operation (we specify 2 min for such diameters, unless water hammer analysis has been performed).
3. Strength of PVC pipe is seriously reduced with temperature, as I sew when local PVC pipe had to operate at 60 oC ( reported strength=16 barg at ambient, rather referring to test pressure; Schedule No was not reported).
4. Assuming no reverse flow, -0.1 psig can mean 0.0 psig. The latter indicates no flow instantaneously, probably due to
a. insufficient NPSH available for the pump (max fluid temperature?), that is vapor bubbles at suction (as mentioned).
b. air pockets in the liquid of suction line (check air ingression at suction tank, if any).
c. instantaneous blockage of suction line, if possible (reported density does not indicate it, butterfly valves could, even wood pieces etc could cause it, a strainer would eliminate this possibility).

Edited by kkala, 24 April 2010 - 03:39 AM.

[El Compostador]

KKala:

Thank you for your much valued comments/suggestions.

I trust that using a flexible expansion joint will solve my final problem of the weak elbow. I can't anchor the pump since the basin is always full of water, but the expansion joint will allow for minor movements of the pump in case of accidental shutdown.

For completeness, I did check the hammer effect and estimated that the pressure rise was less than 150 Psi. (from pressure rise = density x velocity of sound in system x velocity change). This is much lower than expected because pvc is a flexible material (bulk modulus ~ 420,000 psi) and the pipe is of relatively large diameter (15 inches inside); the effective bulk modulus (water/pipe) used to calculate the sound velocity is lower than that of water alone. In view of this, I now think that a pump displacement plus inadequate supports is what did my fittings!

In regards to the negative pressure reading, I am actually able to see water come out continuously _even with negative gage pressures- from the two effluent pipes connected to the butterfly valves since these pipes discharge to an open tank with a water level five feet below grade. A zero gage reading would indicate that the pressure drop across the butterfly valves is negligible (valves fully open).

Since the pressure transmitter is only about two feet upstream of the butterfly valve and because the discharge point to atmosphere is only five feet downstream of the valve, the negative reading is most likely due to a unstable flow profile affecting the strain gauge.

Apologies for misleading you due to the lack of details.

Once again, thank you very much for your input. I now feel comfortable proceeding with this project and will do my best to contribute as best as I can to this forum in order to a part of this great resource.