7 replies to this topic

[mark_84]

I have a very basic question regarding sizing overpressure relief for a pressure vessel resulting from a pressure regulator failure. ASME Sec. VIII, Div. 1 UG-133(a) tells me that I must size the relief for the "maximum quantity that can be generated or supplied to the attached equipment..."

It has been my practice to perform this sizing using the maximum upstream pressure. For example, if a steam header runs at 75 psig, and the header is relieved at 100 psig with a 110% relief valve protecting it, I would use 110 psig as my pressure regulator inlet pressure for calculating the "maximum quantity that can be... supplied".

I contend that there are cases where this is without question the necessary criteria. One could also subtract from the source pressure the piping losses from the source to the pressure regulator.

Question: Is there a case where it would be appropriate to size the maximum flow using either the header normal operating pressure (75 psig) or the relief valve set-point (100 psig) as the regulator inlet pressure?

This is not a hypathetical issue, as I have a current project where there are many valves that will be affected by this choice.

Thanks for your input.

[Art Montemayor]

Mark:

I’m going to try answering your basic question first and offer some comments.

I do not believe (employing what I perceive as the ASME intent) there is a logical basis for calculating the maximum relief quantity using the upstream pressure.

The worse, maximum, flow that can result from a failed open gas or vapor regulator is choked flow. This is the mass flow rate occurring when the regulator trim constraints produce a linear velocity that equals the sonic velocity within that fluid. That happens when the ratio of the absolute upstream pressure to the absolute downstream pressure is equal to or greater than [(k + 1) / 2] k / (k - 1), where k is the specific heat ratio of the discharged gas.

I have taken the liberty of quoting a learned and erudite Aggie buddy, Milton Beychok, in the previous statement. To date, I have not found a better and more succinct explanation of choked flow than Milton’s. I highly recommend you visit and read through his extraordinary website at mbeychok@air-dispersion.com. If we are lucky, we can receive a visit and post from Milton on this thread. He is a frequent visitor to this Forum and one of the main reasons engineers visit and read these electronic pages.

The mass flow rate during choked flow is independent of the pressure drop across the regulator trim and depends only upon the upstream pressure. In other words, for a given upstream pressure the mass flow rate during choked flow remains constant even if the downstream (pressure vessel & PSV) pressure is further decreased. Therefore, you need and want to identify this condition in order to size your downstream PSV.

Normally, in the case of a pressure regulator feeding a process vessel, I would expect to have the regulator’s maximum Cv as furnished by the manufacturer for the stated process fluid and conditions. This maximum Cv should identify with the maximum sonic flow you might expect across the regulator. Your PSV must have the maximum capacity expected at the worse case because if it falls short, it will cause pressure accumulation upstream and an over pressurization of your pressure vessel exceeding the 10% margin. The PSV, by ASME definition, is mounted on the pressure vessel and its upstream pressure encompasses the pressure vessel.

I don’t seem to understand your explanation because I perceive the sequence of events as follows:
1) The regulator feeding the pressure vessel fails in the open position;
2) The regulator strives to equate the pressure on both sides of its trim – the downstream side pressure rapidly rises, approaching the upstream pressure;
3) The pressure vessel reaches the set pressure on its PSV and the PSV opens to relieve;
4) The maximum rate at which the pressure is increasing on the downstream side (the PSV side) is related to the sonic (maximum) mass flowrate proceeding into the downstream side;
5) The maximum mass flowrate must be relieved at that same rate or else there will be an accumulation of pressure upstream of the PSV in excess of the 110% allowed.

The ultimate criterion to apply is one that does not allow the system upstream of the PSV to exceed the 110% allowed. Therefore, the maximum sonic mass flowrate must enter into consideration as the worse case scenario. Of course, there is a possibility that you may not have the capability of achieving choked flow. You should, in my opinion, establish if indeed you can avoid choked flow when dealing with compressible fluids. If that is the case, the situation is still one that must identify the maximum flow under the prevailing conditions and the manufacturer’s max. Cv can be applied to obtain that identification. Once identifying the maximum flowrate, you can proceed to quantify the respective pressure drops through the inlet nozzle and the discharge system.

Don't forget that if you avoid choked flow across the failed regulator, you still have the possibility of choked flow through the PSV. The latter is usually more likely.

Art Montemayor
Spring, TX

[rxnarang]

If we can draw a parallel of this situation to a failure of control valve from a high pressure vessel to a low pressure vessel, then API 521 para 3.10.3 is helpful. I have quoted the relevant text below.





In the event of loss of liquid level, the vapor flow into the
low-pressure system depends on what the interconnecting
system, which usually consists of wide-open valves and piping,
passes with a differential pressure based on the normal
operating pressure upstream and the relieving pressure on
equipment downstream. This pressure drop at initial conditions
frequently results in critical flow and may cause the rate
to be several times higher than the normal rate of vapor inflow
to the high-pressure system.

Does this help?

[mark_84]

rxnarang,

Thank you for your response.

I re-read API 521 section 3.10.3, and I think it is informative in my situation. However, in the case described, the loss of level in the high pressure vessel could be caused by many problems unrelated to the pressure in the high pressure vessel. To assume an overpressure in the high pressure vessel at the same time may superimpose two unrelated failures (not in all situations).

In the case I am concerned about, overpressure in the supply header could possibly cause the pressure regulator failure. This seems eapecially true if the up-stream pressure could rise dramatically or close to the limit of the regulator design. If this is the case, then it is when the upstream system is in relief that the pressure regulator is most likely to fail.

I think API 521 section 3.10.2 addresses it more specifically:

"In evaluating relieving requirements due to any cause, any automatic control valves that are not under consideration as causing a relieving requirement and which would tend to relieve the system should be assumed to remain in the position required for normal processing flow."

Any thoughts are welcome.

Thanks.

Mark

[rxnarang]

Mark,

I agree with you that IF the pressure regulator is expected to FAIL when upstream pressure rises to design, then it is not a case of double jeorpady.

However, I find it most unusual that a pressure regulator should fail. Normally the pressure regulator is designed for the highest upstream pressure ; interpret this as Design Pressure. I would double check the pressure regulator specs. In case it is an installed item, then you have very little choice. In case it is a new item, then it is in your control to buy the correct pressure regulator.

In the latter case both the upstream PSV relieving AND the pressure regulator failing would be considered as double jeopardy. You need not design for a double jeopardy.

The regulator failing is an interesting aspect. What do you mean by failing? I am not an expert on the internals of a regulator. It would be informative to get a photo and identify which parts fail.

Regards

[Art Montemayor]

Mark/ Rajiv:

I don’t know about the many other visitors to this thread, but I’ enjoying this very interesting and thought-provocative line of engineering logic. I have sat back and recalled the same scenario that Mark has described and challenged. Rajiv’s identification of the double jeopardy is well-reasoned and, in my opinion, correct.

The few things I can add to the line of thought are:

1) It is very possible for a process regulator to fail in the open position, allowing maximum flow through it in accordance with the pressure drop existing across it and the nature of the fluid. In Mark’s case the source (a header) is protected from achieving an accumulation pressure in excess of 110% of set pressure. The possibilities of the header achieving 110% set pressure level while at the same time the downstream regulator (being fed by the same header) fails are two independent and non-related actions; this constitutes double jeopardy – something I would not consider as statistically possible and not design for it happening.

2) However, note that I mention “not related”. If there is a pool fire in the area causing (by some mechanism) header over-pressure and, at the same time, damaging the regulator to the point where it fails mechanically in the open position, then there is no double jeopardy violation and both incidents are credible as being a possible scenario for over-pressurization downstream of the regulator with 100% inlet pressure into the regulator.

3) A gas or vapor regulator can fail mechanically in the open position in various ways – depending on the type of regulator, the design, and the process conditions. For example, a regulator can be a Fisher “Big Joe” or “Little Joe” type as well as a conventional control valve set up to function as a regulator. A Fisher 95H type (with SS diaphragm) can also be used. These are normally single-stage units. 2-stage regulators are also used when striving for better and more accurate control. The design of the action between the plug and the seat determines a lot with respect to the Cv and the “fail action” of the unit. The fluid (or the process) may cause a solid particle to lodge itself between the plug and the seat and render any closure of the unit as impossible. A lot of these features have to do with the manufacturer’s design and the purchaser’s specifications. There are regulators – and then there are “regulators”. One just can’t generalize. Every model and type should be analyzed for fail-safe action and shortcomings.

There may be a way to justifiably reduce the worse case flow for the regulator in question by replacing it with a 2-stage unit and/or installing a restriction orifice in the header, upstream of the regulator. This could, conceivably reduce the maximum vapor flow reaching the regulator at pressures in excess of the normal operating pressure. This is just a thought that I haven’t thoroughly studied, but am throwing it out to see if it has been thought of. This is would be similar to the concept of a safety interlock, except that it has no moving parts and is far more reliable.

I am accustomed to employing strategic restriction orifices in gaseous systems in order to ensure maximum flow is not exceeded in certain conditions. Perhaps some one else has similar experience.

Art Montemayor

[mark_84]

Gentlemen,

Thank you for your thoughts on this issue.

I agree that the failure of a properly applied regulator should not be related to the upstream pressure, unless there is a common event that causes both. The protection and configuration of the systems that I am working on makes this very unlikely.

That leaves the question of what upstream pressure to use in the calculations. My concern here is that I've seen far too many unstable plant-wide headers where the upstream relief valves open routinely. I have also seen situations where a single steam user can take up a large portion of a steam generator's usage. In this case, an upset caused by the failed regulator could cause the supply header to go out of control and into relief. (The low pressure section of a large paper machine is such a case.) If that were the case, I believe I would choose the higher header relief pressure as my upstream pressure.

In my case, I believe that the header pressure is very stable (This I will check out with plant operations and data logging system). The operator will also get quick notification of the event and shut off the supply, which is not the case in some situations.

However, I'm not ready to accept that the nominal header pressure will prevail at the time of the incident. I have not seen a header that operates at exactly the nominal pressure 100% of the time. I believe that it would be reasonable to gather data on the header pressure, and use the average pressure plus an allowance for variation. This can be determined statistically in my case.

Thanks again for your input on this issue. It's been a great help.

Mark