API PSV Sizing on Tube-side of Shell & Tube Heat Exchanger

[Ally0138]

Hi guys.

Can anyone help me out with the following?:

I have been asked to size a relief valve for the tube-side (process side) of a shell and tube heat exchanger.
This particular exchanger has hydrocarbon gas on the tube-side, and a heating medium on the shell-side.
The relevant sizing cases appear to be a shut-in fire case and/or a process upset case whereby the process side is shut-in, but the heating medium continues to flow.
The tube failure case is not a concern in this instance - it is protected against via a locked open route from the shell-side of the exchanger to another vessel which is open to an atmospheric vent system.

My question is: How do I determine the required relief rate (for either the fire case, or process blocked-in, but heating medium still running case)?

I'm currently looking at the fire case, and the standard method for a vessel is to calculate the total wetted area and the heat absorption then calculate the required relief rate as the total heat absorbed divided by the latent heat of liquid per unit mass.

Complications in this case include the fact that the temperature of the hydrocarbon gas at relieving pressure, is below the normal heating medium supply temperature. To my mind, it doesn't make sense to take credit for the latent heat of the heating medium in reducing the heat input into the process stream when it's temperature will already be well above that of the relieving stream. Also, intuitively, it seems like the heat input from a fire ought to be greater than that from the heating medium, so this ought to be a worse case, but I've managed to confuse myself so much on this already that I'm not sure how to proceed from here.

Apologies if this has been asked many times in the past. I have run a couple of searches and can't quite find the information I'm looking for.

Appreciate any clarity / guidance anyone can provide on this. Thanking you in advance for your answers.
-Ally-

 

[Latexman]

First, please don't withhold information that is not proprietary. A "heating media"? What is it, exactly?

Second, a drawing/picture is worth a thousand words. A P&ID and equipment drawing is the ultimate!

Are you using API 520 Part I and II and API 521? They should give you some good guidance.

You have to break up the heat input from a fire on a shut-in S&T HX. The heat to the heads/bonnets go into the tube side. The heat to the shell goes into the shell side, but there can be heat transfer to the tubes. It's a little complicated, right? You can make some conservative assumptions to simplify matters and see if you end up with a reasonable sized PSV's. Like, all the heat goes to tubes and all the heat goes to shell. Or, the shell wetted area heat goes to shell, and the head/bonnet wetted area heat goes to tubes (you may have to add some heat to the tube side if the relieving T on shell side is higher than the relieving T on the tube side).

I could continue, but I'd like to hear your response first. I don't care to write a tome today.



Good luck,
Latexman

To a ChE, the glass is always full - 1/2 air and 1/2 water.

 

[Ally0138]

Thanks for the response Latexman.

First, what makes you assume that the details of the heating medium is not proprietary information? In this case it is, which is exactly why I withheld that information. Suffice it to say in this context it is a synthetic heat transfer fluid used as an alternative to hot oil.

Second, here is a P&ID, with proprietary/identifying information redacted:

Third, yes, I am using API 520/521/526. I have also been referring to ABB's 'Pressure Relief - A Proven Approach' course notes - although they do not appear to give any guidance for fire case for a heat exchanger, only for more simple vessels.

Your suggested approach makes sense to me. If I understand you correctly, I'm simplify the problem by essentially considering the heat exchanger as two separate vessels, the vessel body (shell-side) and heads (tube-side) and doing two separate calculations to estimate the heat input into each.

I'm with you up to this point, but I'm not sure where I'd go after this? i.e. how do I then turn this heat input into a relieving rate? As I mentioned in my OP, the standard method for a vessel is to simply divide the heat input by the latent heat of the liquid, but in this case the normal inlet temperature of the heating medium is already higher than the reliving temperature of the process gas, so it's this stage of the calculation that I'm not sure how to approach.

API 521 suggests that a minimum "latent heat" value of 115 kJ/kg is "sometimes acceptable as an approximation" in cases where there is no liquid, or where the fluid approaches it's critical point and therefore latent heat approaches zero. It also states "If pressure relieving conditions are above the critical point, the rate of vapor discharge typically depends on the rate at which the fluid expands as a result of the heat input because a phase change does not occur."

After sleeping on this, I reckon my approach is therefore to estimate the total heat input as discussed above, calculate how much the gas temperature rises as a result of this heat input and therefore how much the gas expands, which gives me how much gas would need to be relieved. I'm still not quite sure how to turn this into a relief rate though, I somehow need to estimate how long this process takes to turn my gas volume into a flow rate.

Any thoughts?

Thanks,
-Ally-

 

[Latexman]

Because in 38 years experience, I've never seen a proprietary heating media, but if it is to your process, that's cool.

You are on the right path. There are a few good magazine articles on sizing reliefs for supercritical fluids out on the web. If your Google-fu doesn't come up with them, let us know.

It may be easier to assume the tube side is shut-in and the heating media is flowing. Assume your HC gas is at the sizing pressure of it's PSV. I suspect you can figure out what the temperature might be at this point. Then, estimate the heat transfer to the HC gas using a clean htc (no fouling). This heat will raise the pressure. Figure out the volumetric rate that keeps the pressure at the sizing pressure. That is your design volumetric flow rate.

Good luck,
Latexman

To a ChE, the glass is always full - 1/2 air and 1/2 water.