Reverse Flow

[Bill3752]

We are looking at a reverse flow case. Normally liquid CO2 flows through a reciprocating pump, then is combined with vaporized CO2, the combination is a vapor. A PHA team found a potential source of overpressure on the suction side of the pump should the pump be shut down. We are modeling the reverse flow case by assuming an orifice at the PD pump.

Referring to the attached sketch, I am looking at three cases. In case 1, at the beginning of the scenario, liquid CO2 will flow back from C to B through the "orifice" and relieved at A. Skipping to case 3, in the later portion of the relief after the liquid had been removed the system will be filled by vapor, hence D to B. The relief from both are OK.

It is case 2 that I am interested in gaining input. Assuming for a minute that we have "plug" flow through the system, consider the situation wherein there is vapor at point B, and liquid from B to A. I am thinking that as the vapor pushes the liquid ahead of it, I must handle the volumetric rate of liquid that would be displaced by the vapor (not the mass rate). I.e., I am assuming that if the volumetric rate of vapor through the orifice is 250 cfh, then I must relief 250 cfh of liquid.

How have you handled this "vapor pushing liquid" in the past?
On a separate note, how do you deal with reverse flow through a reciprocating piston pump - is this considered a viable scenario?

Thanks

 

[georgeverghese]

To give you some useful suggestions, pls update this sketch to include location of check valves, design pressure break locations, relief valve locations and setpoints, and location of recip pump isolation block valves on both suction and discharge. Also tell us if this pump has an online spare.

 

[Bill3752]

George,
I have attached a drawing with more info. We are assuming the 3" check downstream of the pump "doesn't exist". We are using the check valves in the PD pump as our basis, and applying API 521 - which allows for calculation using 10% of the larger of the two diameters.

Not shown are the pump isolation valves, which are just ahead of the suction side accumulator, and just ahead of the 3" check valve on the discharge.

I originally provided the simplified drawing to help drill down to try to answer the two questions: (1) does one normally consider back flow through a reciprocating (triplex) pump, and (2) assuming plug flow back through the system, after the vapor is at point B, would one normally base the relief rate at A on the volumetric rate of the vapor at B, or the mass rate?

Thanks for your interest.

 

[georgeverghese]

Your focus seems to be on overpressure protection for the suction vessel through PSV A, but another overpressure concern would be the pump suction piping downstream of the pump isolation valve, where you have the limiting design pressure of 550psig at the accumulator. We'll come back to this later, and for now here are responses to your questions:

Agreed, reverse flow through a recip pump would result in a small, but non negligible relief rate upstream of the pump, since you've got 3 check valves in series (one which you wish to consider as nonexistent, and another 2 within the pump ie discharge and suction side unidirectional plate valves). Do you not have another check valve just downstream of the centrifugal pump FCV tee off from the the main suction line? - that will make it 4 check valves in series. Given the number of check valves in series here, I would say the relief quantity appearing at PSV A due to this reverse flow would be, by inspection alone, well within the current capacity of PSV A, given my assumption that this would be a good sized PSV sized for external pool fire at vessel A.

The other overpressure scenario at the 550psig suction side accumulator appears to be more of a concern, since the PSV B here may be a small capacity RV. I would look at the reverse flow resulting from a restriction equivalent to 10% of the dia of this 0.11inch internal plate valve at the pump for each of both the scenarios you've identified (a) liquid relief resulting from the corresponding volumetric vapor displacement rate (b) vapor mass flow rate. Check to see if this existing 550psig PSV B can handle each of these rates. There may be some wiggle room to allow for accumulation higher than 10% at this PSV if this accumulator is a pipe and not an ASME coded vessel.

It is these 2 relief rates calculated at (a) and (b) which I assume are well within the capacity of PSV A. If this 4th check valve does exist ( downstream of booster pump min flow recycle valve tee off), would say you could take credit for this also, and trim down these relief rates to 10% of that calculated, for a rating check at PSV A if you want to quantify adequacy.

 

[Bill3752]

George,
Thanks for the input. There is no check valve on the suction side, and the third (omitted valve) is very large. You are correct that we need to also consider the 550 psig valve, but it is coincidently the same size as the lower set "A" valve, so the A valve governs for this discussion.

Your most important comment is that I need to consider the liquid case (based on vapor volumetric rate)- for this case the RV is undersized. For the liquid case based on a liquid orifice calc, and for the vapor only case, the relief valves are ok.

Thanks again.

Bill