5 replies to this topic

[csteen]

This relief valve forum is a fascinating read and I gained a better respect and understanding of relief valves from the forum, which I'm sure is one of the forum's goals. I work for a large company that started as a research company over 15 years ago. The facility probably has over 900 relief valves in service now with little to no documentation supporting their sizing basis. Obviously, this is a problem. We are in the process of obtaining as much nameplate information (legibility has been a problem) as possible before beginning the long and tedious task of reverse engineering the sizing basis of each and every valve. I have many questions, but let me start with just a few as follows:

1) When do you have to consider sizing a relief valve on a pressure vessel for a fire event? Is it always the case, or only when there is no other credible event that can lead to overpressure?

2) It seems to be clear from ASME Section VIII, Div. 1, UG-125, that any vessel that is designed, built and stamped in accordance with ASME Sec. VIII, Div. 1, must be fitting with a pressure relieving device to protect the shell even if there is no potential for overpressure. Please confirm this and what about the tubes of a heat exchanger? Is there a code requirement to have a relief device on the tube side or is it simply good engineering practice especially in a no-flow situation in the tubes full of water with steam applied to the shell?

3) Related to question 2 above, if you have a steam (shell) to water (tubes) heat exchanger that has a design pressure equal to the design pressure of the entire steam supply system right back to the boilers, do you still need to have a relief valve on the heat exchanger shell for the reason stated in UG-125? If no why?

4) Does anyone know of an off the shelf user-friendly and robust computer software program that can help my facility develop sizing calculations on our existing and future relief valves in an efficient and cost effective way?

Thank you for your time and I look forward to any responses !!

[pleckner]

WOW!

What you need is basically a book worth of explanations. Not only MUST you have the latest copies of the ASME regulations but API RP520 and API RP521 as well. And if you are outside the U.S., then you MUST have the regs that go with your country.

Now for that book..get a copy of "Guidelines for Pressure Relief and Effluent Handling Systems", CCPS/AIChE,(1998). A must have!

Seriously consider hiring a company experienced in doing process safety analysis, calcuations and documentation. If this appears to be too expensive, the consquences of not doing this exercise correctly will be even more expensive. It should not be undertaken by the inexperienced and it will take time. You should allocate about 6 hours per valve. Some will go quick, an hour, more or less. Some can take several days. Your average may vary considerably but 6 hours per valve should be a minimum average; that's how long it took me several years ago doing this exact same exercise for an Ethylbenzene/Styrene facility with about 125 valves to go through. Also, this time did not include data gathering for the most part. We got the data we needed from the plant.

Once the big job is completed, you should be able to handle any additions in-house based on the experienced gained while working with the consultant.

If you still indend on doing it in-house, don't hesitate to post any questions you may have on this message board. We're here to help where we can.

Now, let me see if I can start answering some of what you asked.

1. If you can have a sustainable fire, then you have a credible fire scenario, otherwise you don't. If it seems there are no credible scenarios then many plants use fire as the "excuse" but I don't. I put on a 3/4" x 1" relief valve and call it quits. Note that this is pretty much the size you would use for a simple thermal relief case and this is my justified relieving scenario in this case.

2. Yes, you need the PSV and see my response in #1 above. There is no clear direction from the codes and guides about protecting the tubes. It really is just good engineering practice. For this most part (but not in all cases), this would require only the use of that 3/4" x 1" PSV I talked about above.

3. If your heat exchanger shell cannot normally be isolated from the header PSV, then the shell can be protected by the header PSV and may not require its own PSV.

4. Sizing a PSV is very easy. The formulas for the most part are straight forward. The problem is in determining scenarios and the calculations of relieving rates. This may take process simulation to assist as some of the themodynamics can become hairy, especially if you get into two-phase relief. Many of the PSV vendors have pretty involved programs that you can get for free. One vendor is Farris at www.farrisengineering.com.

We use in our Company a software that allows us to calculate the relieving rate for the most frequent occurring upset scenario's (fire, exchanger tube rupture, control valve failure,...) define the fluid mixtures and get the relevant phyisical properties, calculate inlet and outlet lines sizes and populate a database where all the calculated scenario's data can be stored and used to generate process data sheets, relief load summary, etc... www.psvplus.com

[Art Montemayor]

CSTEEN:

Phil has given you some factual and actual advice. I led a team of 10 process engineers in 1997 on a similar project. Except we were charged with documenting and justifying over 1,500 PSVs in one very large Gulf Coast plant site. This was a task force assembled for a very large and prominent Chemical Processing company here in the Texas Gulf Coast. The outstanding fact that I remember is that we budgeted 10 hours per PSV and when all the work was done and the documents approved we found we averaged between 25 to 30 hours per PSV – and these were experienced, veteran ChE doing the analysis, the scenarios, and the calculations. I have since witnessed others doing the same thing at other plants and they have not improved on the manhours per PSV. I am convinced that it is not a matter of how experienced, fast, or smart the engineers are. It’s a matter of the local situation, the availability and field help in obtaining, confirming, and establishing basic data. The actual calculations of the size of the PSV and the orifice are nothing compared to the fact-finding and the basic data retrieval. I recall that we averaged between 1 to 2 hours on the actual calculations.

In my opinion you will find extensive problems in establishing the accuracy of existing records on the PSVs presently installed. I doubt if anyone kept a Management of Change procedure – and even if they did, the method was probably a “best effort” basis and probably not credible. The outstanding problems that really ate up the manhours were:

1. verifying the inlet nozzle pressure drop; just about every one had to be confirmed and this mean field as-built iso-sketches;
2. verifying that you don’t have a 2-phase possibility;
3. verifying that the P&IDs are literally “as-built”. I don’t remember coming across one P&ID that had not been changed and had to be as-built. This can be a killer on the man-hours.
4. Getting accurate information from the operations or engineering staff. Many times this just can’t happen because they are overloaded and can’t spare the time. This means more field trips.
5. Confirming that the actual PSV listed is the one that is installed; we found far too many PSVs had been changed and the records were wrong. And as if that was not enough, we couldn’t even read the tags on the installed PSVs because they had been painted over or were simply missing!
6. Obtaining credible physical properties on some of the specialty chemicals; this also was a surprise to find out that the producer didn’t have a complete properties database.
7. Running simulations on reactors and evaluating potential run-away scenarios. This can be a real headache.

Getting or buying a computer program is going to be the least of your problems. The computers and the programs can’t help you an iota with the troubles I’ve listed above. As Phil has stated and I reiterate: the actual calculations take very little time. The engineering analysis, the field information, and the evaluation of all the credible relief scenarios are the time-consuming factors. And these will vary with the type of process, the design and the type of equipment.

And don’t feel bad about your situation being an exceptionally bad one. The real sad news is that our biggest and strongest processing companies probably have worse situations than yours.

I don’t like being the bearer of sad news, but I feel it is better that you know the factual experience of others who have been down this road before. I wish you the best in this effort and that you succeed in establishing an accurate and maintained database on your PSVs. I am certain that the final results you publish will be eye-openers to all in your organization.

[gvdlans]

Again two exceptional good posts from Art and Phil. This thread is a must-read for everyone dealing with pressure safety devices!

[Adriaan]

Side note;

In some systems there are multiple safety (pressure relief) valves - that operate at different pressures - on a single line. Their purpose is to ensure that flow through the system is maintained (by having a lower pressure valve downstream) yet some maximum pressure is never exceeded (by having another, higher pressure, valve upstream; sometimes this setup is repeated several times up the line).

I mention this because it is a feature often found in heat exchanger setups (which you mentioned under point 3.).