24 replies to this topic

[Art Montemayor]

Kiran:

I've run into this argument/challenge several times. The logical answer is that it is illogical to expect that an operator could make the mistake of opening the bypass while the control valve has failed in the open position. And that is the only logical manner that one could have both open at the same time.

You are absolutely correct in stating that the reason the manual bypass valve is there in the first place is that it is to be used as a manual operation when doing maintenance on the CV or the CV has failed to open. In both of these events, an operator is expected to crack the bypass and block-in the CV in order to prepare to remove it or repair in situ. If an operator were to open the manual bypass while the CV failed in the open position it would constitute "Double Jeopardy" and would not be a credible design scenario. Both events are not related so the odds of both happening at the same time fall under the double jeopardy description. All design meetings and hazops where I've run into this dilemma have reached the same decision as you: The maximum Cv value of the control valve sets the maximum flow through it and is the basis for the relief capacity.

I wish you had posted this query in the Relief Valve Forum, so it would get the proper exposure.

[djack77494]

I had an interesting thought about this topic. Another situation, though rare, would occur when the control valve is removed for servicing and you are operating using the bypass valve. It might be in any position, including fully open. Should whoever is supposed to be paying attention to its operation become distracted, that situation would be very similar to a failed control valve. For that reason, I would propose that the relief scenario should be based on the LARGER of a failed open control valve OR a fully opened bypass valve. Any thoughts about this?

[Art Montemayor]

Doug:

You've not only made an excellent mental exercise, but you are absolutely correct as to your evaluation of the manual operation for control purposes. Although most by-passes are usually 1 size below the CV size, it is still possible to have the bypass contributing more flow during your described procedure - all depending on the type of valve and its Cv value.

Once again, your result has shown how a critical worse case can be camouflaged from the perception of those engineers trying to identify the controlling case. It is so very important to explore all possible operational and accidental occurances by carefully detailing a complete explanation of the scope of work.

[kdal]

I've experience of working for 2 international contractors in the O&G industry and interestingly they both take different views on the subject.

The first approach reflects Doug's previous post - it is feasible for the control valve to be in the normal operation and the bypass valve inadvertantly opened which produces an incremental increase in flow for relief. However it is double jeopardy for both the control valve and bypass valve to be wide open simultaneously.

The other contractor's approach is far more conservative and they make a point to say that the relief rate should be calculated with the control valve and its bypass fully open, to account for "start-up cases". I'm afraid that it doesn't go on to expand on what is meant by start-up cases!

Just thought that it was interesting how there is not a common approach and it very much comes down to the nature/ethos of the organisation for which you work. Does you're employer or perhaps client have a philosophy on this matter?

[djack77494]

My employers tended to "leave open" the development of relief case scenarios to the responsible engineer. I think it very difficult to do otherwise, since I can't imagine being able to enumerate all possible cases in the required detail. Of course, I've seen the lists of potential relief scenarios - this case would start with "CONTROL VALVE FAILURE". But that's it. There was no further guidance. Issues like discussed above were not discussed at all. End result is that it all depends on the experience and judgement of the engineer. Art, I don't want to appear contradictory, but I've often seen bypass valves the same size or even larger than the control valve. My current client, a major in the oil business, insists on the use of BALL valves for their bypass valves; they'll be one size smaller than full line size/certainly not smaller than the control valve. Wonder what their Cv's are?

[abhi_agrawa]

I'll just add something about the "start up" cases. Recently I was involved in the commissioning of a plant, where through a Control Valve, the flow of gas was more during the start up as compared to the normal operating case. This was because there was a reaction involved, and during the start up, the reaction was not effective, leading to a relatively large amunt of gas. However, for this scenarion, the control valve was chosen with a large Cv, so that during the start up, the bypass was not operated. The bypass was sized to allow the "normal" operating flowrate when fully opened. The relief valve was sized as per the maximum gas flow rate during the startup.
abhishek

[pleckner]

The purpose of the by-pass is to operate the process while taking the associated controll valve out of service (for whatever reason). These by-pass lines were never intended to be used for start-up (OK, a generalization but for the most part true). Those who use them for start-up are asking for trouble and it just isn't necessary. Just put the control valve in manual and open/close as needed through the control system.

If the by-pass is to be operated at the same time as the associated control valve then the total flow through both the by-pass and the control valve (fully opened) must be at least analyzed when determining the PSV sizing. Some years ago during a design of HFC facility I put in restriction orifices in the by-pass lines so that they would not be controlling in PSV sizing (yes, these lines were one-size smaller than the associated control valve). The flow was more than enough to allow the process to operate when the control valve was out of service but was restrictive enough during the relief scenario that it was not controlling over a fully opened control valve.

[gvdlans]

Phil, good to see you back on the forum.

One approach is to size the relief valve for largest flow through either fully open manual valve or fully open control valve (whichever results in the largest relief flow), and to check that downstream pressure remains below the hydrotest pressure when both control and manual valve are fully open.

Reasoning is that in the highly unlikely, but still credible case that both valves are fully open, there will still not be a loss of containment.

Note that the Cv values of the actually installed valves have to be used and that there should be some system in place to prevent installing of a valve with a higher Cv without checking the relief valves.

[pleckner]

I think the previous posts do talk about the scenarios of both a fully opened control valve and the opened by-pass. But if one is made significantly more restrictive than the other (as I had done in one design-see my post below), the relief flow will be governed other less restrictive path and will govern. To put the problem to rest, simply car-seal the by-pass valve closed.

Please note that API RP520 and RP521 are not codes but Recommended Practices. Saying this, they historically represent good engineering practice and thus should be followed as guidelines.

[pleckner]

If you want to run your system using your by-pass with no control whatsoever, feel free to do so. That is not what the by-pass is meant to be used for.

I guess we are just going to agree to disagree.

[Ritesh Suratwala]

I am totally agree that any unnecessory design margin in PSV capacity is not a good engineering practice and will lead to chattering.

And as far as line size is concern, you can not design line size for normal capacity and PSV for rated capccity. Your line size also must be based on rated capacity only. So, I will be concerned about point 1 only and will not bother about point 2.

If I come to main topic, my experince says eventhough it is "Double Jeopardy", selection depends on criticality of process. In LNG plans and Crude refineries, any cost of bigger size PSV is nothing when compared with downtime of plant. And in such places, client will definately ask to go even for "Double Jeopardy", just to minimize possibility of downtime of plant to zero extent. And I feed nothing wrong in this engineering practice.

[pleckner]

To clarify my stance on the by-pass line, the CV of a full-port manual block valve is usually significantly larger than the CV of a modulating valve. Therefore, one cannot put just a full line size by-pass around the modulating valve and hope not to get into trouble. That is basically why one needs at least a line size smaller by-pass and that is what is historically done in the chemical industry. As far as operating the system with only the by-pass, I wouldn't do it unless it was absolutely necessary for the very reason I stated above, no control.
However, it is often necessary and that is one reason why the by-pass is there, but only as a last resort to keep the system running (making product...making money) untill the control valve can be repaired/replaced. Operations must also be extremely dilligent in keeping the system under constant observation because there is no contol.

The by-pass is also there for start-up purposes when, let's say you may need to do an initial charge of a material or you may need to warm the downstream equiment up to prepare it to accept material. And this is why you can make the line more restrictive.

Saying all this, should one take into account the by-pass opened as well as the associated control valve in a PSV scenario? Let's look at this scenario. Our control valve failed during the midnight shift and the by-pass was opened to allow emergency but limited operation. Near lunch time, the control valve is fixed. The operators are somewhat occupied with lunch on their mind and forget about the opened by-pass. In their haste to get the system back up to full production, they stroke the newly fixed control valve to get it going. Double jeopardy? I think not.

So, yes, this scenario must be taken into account when sizing the relief system.

By the way, this whole discussion should be moved into the Relief Forum and should not be here.

[Ritesh Suratwala]

To remove control valve for maintainance, you start bypass, close control valve block valves and remove control valve. After maintainance, you fix it at position, open block valve, close bypass valve. This is a common practice has been followed and as a single action. There is no way operator can miss any of step in above process. If it happens, it will be extreem case of carelessness from operator.

Let us believe that it had happened. Now bypass valve and control valve both are in operation and plant is at full capacity mode. At this point, shift engineer must be monitoring control valve opening when control valve is taken in operation. Even if somebody missed to monitor it at that time, during short time it should come in notice. If no one monitors it, it is again case of carelessness from shift engineer.

After maintanance, control valve will not fail very soon. But you are again considering control valve failure at this time only. Now I will stop here sayimg all these under "double jeopardy".

Here, I will highlight one point which I forgot in my past reply. It is not control valve that must be responsible for relief load. It should be either control valve or bypass valve, whichever has higher CV. But again, not both. It is normal practice that Control valve CV is always higher than bypass valve. But final check required.


And if you want to design system for each and every event of carelessness during operation, what will you do for situation when one relief valve is taken out for maintanance without opening inlet/outlet block valves of spare relief valve? :-)