Relief Temperature --- Rediculous

[rocketscientist]

I have an unusual condition I am sizing a relief valve for:

low operating pressure but very high design pressure.

Here's my dilema: Pnormal = 50 psig (Pn=61.4 psia), barametric P = 11.4 psia; Pset pressure = 600 psig (for fire case, that's 737.4 psia); Tnormal = 127 F (586.7 R). Following equation 7b from page 17 of API RP 521, for the relieving pressure (constant pressure case):

Trelief = (737.4/61.4) X 586.7 R = 7,046 R (What?)

I seem to recall, and API hints at this, that the upper limit for carbon steel is 1,100 F (Of course, this is a 316SS tank).

Any consensus on what I should use for the real relief temperature? If 1,100 F, then does anyone have anything for stainless or other materials?

(Damn, I wish I hadn't stashed all my reference material in storage.)

Thanks in advance.

 

[dcasto]

You need to solve the two equations, the one you just did and the ASME stress equation (actually the flanges will fail first) to where the temperature stress are exceeded. First guess 712F. Vessel pressure 1172/586 * 61.4 = 123, ANSI 300 flange good for 520psig , second guess 950F 1410/586 *61.4 = 147 psia, flange rated at 140 psig. Maximum set point 130 psig. A quick look up says most carbon steels will be at 15% of their yield at 950F, thats about 110 psia with an original 600 psig.

The risk of failure under such conditions is so small, I would ignor this case.

 

[JLSeagull]

It is common for the fire sizing case to exceed the flange class. No problem exists if the size is based upon another case. However if you put the fire case on the specification data sheet the manufacturers may not sell the valve with a flange rated lower than the pressure and temperature for the metalurgy specified.

 

[MortenA]

In the specific case then i would say that the difference between sp and relief p is too great.

However as i have understood it in general - the fire case is only to work as a last resort against catastrophic failure. The vessel will/may be ruined afterward. Normal deluge should prevent the temperature/pressure ever getting so high that the valve will actually vent.

Best regards

Morten

 

[rocketscientist]

Here's my idea after some additional thought. I was incorrect in my assumption of constant pressure. (Must be my eyesight.) API uses an assumption of constant volume to represent the enclosed space of the area above the liquid (I am speculating here.). Constant volume is only an approximation. Because only a portion of the liquid is actually evaporated by the heat flux (I estimated 40% of the liquid.) that means that the volume is nearly constant for a large vessel. This would not be true for a medium or small vessel. The higher the H/D (horizontal) or L/D, the more inaccurate this assumption becomes, even for an ideal vapor, which this would not be.

Nor, would the assumption of constant pressure be correct.The pressure should gradually build over several minutes as the liquid level drops to produce the vapor to fill the space above it. No, I think perhaps, API is wrong. The evaporating liquid should cool the contents of the vessel. If the vessel is well-insulated, the heat loss by conduction and convection should be slight. From my experience, as long as the temperature is below 200 F, the radiation should be low; insulation will keep the skin temperature of the outside of the insulation jacket well below 200 F. So, what I am thinking is that adiabatic might be the best assumption to use to predict the relief temperature.

Here's the comparison:

Constant Volume (API-RP-521):

P1 = 737.4 psia (600 psi + 11.4 barometric + 21% overpressure); Pn = 50 psi +11.4 = 61.4 psia; Tn = 127 + 459.67 = 586.67 R.

T1 = (P1/Pn)Tn = (737.4/61.4)586.67 = 7,046 R (?)

Adiabatic (Q=0):

T1 = (P2/Pn)^m, where m = k-1/k and k=Cp/Cv (Don't use ideal gas for Cv=Cp-1 because Pr<1 and Tr<1; Z = 0.707 by corresponding states for water) k = 1.291 (Cp, Cv from NIST site online).

T1 = (737.4/61.4)^0.225 X 586.67 = 1,026 R (566.7 F)

Note, I am concerned because this approaches the 1100 F limit for carbon steel, although the allowable stress for 316SS is 13,000 psi at 1000 F compared to CS(A106) with an allowable stress of only 2500 psi at the same temperature.

Let me know what you think.

Dirk Willard

 

[JLSeagull]

About 15-years ago a PhD chemical engineer in our research organization took exception to some of the calculations in API RP 520. He proposed solutions with partial differentials etc.

I pointed out that API RP 520 is a recognized industry standards. Bring errors to their attention. Our contracts require compliance with API RP 520. We are not going to use another method of sizing and selecting relief valves based upon his clever technique. We would consider validating our relief valve calculations using recognized software. However the vendor's calculation would support the valves purchased. Also that the control systems engineers bought the relief valves. Most CSE's are electrical these days. Both chemical and mechanical engineers typically study more thermodynamics and fluidics than electrical engineers. If they want to assume the responsibility to specify and purchase the relief valves then they are welcome to take them over. The last that I knew PSV's were still handled by control systems engineers at that company - and my current home too.

 

[rocketscientist]

I ran comparison calculations and decided the best way to size the relief valve was to use the saturation temperature. The mass flow rate is calculated, for a fire case, using the latent heat of the liquid (in our case, water), at the relief pressure (design pressure X (1+.21) +barometric pressure). If we assume the evaporation cools the vessel, then it makes sense, since only a portion of the liquid is vaporized that the maximum temperature reached is the same as the temperature for the latent heat. (

FYI: this method of calculating the mass flow was proposed by A.K. Coker, "Size Relief Valves Sensibly," Chemical Engineering Progress, August 1992.)

W = Qload/DHlatent where Qload is calculated from one of the Q=F(C1)Aw^C2 equations. Be careful not to confuse storage tanks with pressure vessels. The equations for storage tanks from API 2000 should not be used for pressure vessels.