Vent sizing for peroxide reactors 1

[mpercival]

I'm trying to size for a peroxide system but the more I read the more problematic I find them to be, i.e. vents being larger than the reactor.

Of the methods available, from Leung to Singh etc, which is the most appropriate for hand calculations? i.e. which does not oversize but remains safe

Thanks for your help.

 

[Montemayor]

mpercival:

I'm not trying to be critical, but aren't you trying to have your cake and eat it too?

Calculation methods don't make the correct and safe process. Your engineering logic and common sense do. It is obvious and logical that if you have a credible dangerous run-away reaction as a potential hazard, calculations aren't going to make it go away. You've got to take it into consideration in your PSV design. If reality tells you it could happen, then you've got to accept that fact and design from there. There may be a possible way of accepting probability risks; however, you're going to have to face those odds. There are also mitigation steps that may be available to you: isolation, block house design, explosion walls (or dikes), instrumentation, etc.

If it were my process, the #1 priority I would have would be to have all my operating personnel cognizant of all the facts - no matter how dire or dark they may appear. Their participation in the required Hazop would allow their contributions to a possible, workable and acceptable answer.
If you have information that tells you about a bad potential bad situation in your operations, it is your responsibility to let everyone know about this and proceed to resolve it expeditiously and safely.

I hope this helps you out in making your situation a safe and profitable one. Good Luck.

Art Montemayor
Spring, TX

 

[EGT01]

I have only little experience with assisting in the design of relief systems for runaway reaction. My suggestion would be to follow the experiences of other installations in your company unless you have found those experiences to be inadequate. If they have proven to be inadequate, then consider seeking an experienced consultant such as Fauske and Associates to mention just one.

http://www.fauske.com/NewOpener.htm

Of course it helps to learn as much as possible about the methods available but I don't think whether a method can be done by hand calculation is a criteria for choosing the best method (and I don't think that is what you really mean). If you have looked at any of the DIERS references, you need to consider the type of vessel flow model and vent flow model that might apply and have a good understanding of the characteristic behavior of the reaction itself such as from bench scale testing.

Not sure which peroxide you have in your system but one of the articles for download from the Fauske website has a short discussion about 3,5,5 trimethyl hexanoyl peroxide , dicumyl peroxide and t-butyl peroxy bensoate where it is listed as a Gassy reactive system. Get a copy of article C24 if you are interested.

http://www.fauske.com/flchemical.htm

I can't say I have enough experience to suggest a particular model to follow but perhaps others could advise if you offered more info about your reaction and vessel arrangement. Then maybe you could compare those methods to the ones you have found so far.

From the DIERS project manual (copyright 1992), there is a logic tree used as an approach for emergency relief system design assessment

Prevention ---> Moderation (relief) ---> Containment

If from your research so far, you determine that vent sizes larger than the reactor are required, then you probably need to consider designing the equipment adequate enough to contain a runaway or consider providing some type of highly reliable Safety Instrumented System to prevent the runaway from occurring.

 

[dcguffey]

Another way to minimize the hazard of a runaway reaction is to limit the amount of material in the reactor. This line of thinking is what led to "pipe reactors" or low volume tubular reactors.

I am ignorant of your particular case; however, I would think about changing the reactor size and thus the amount of reactor contents by perhaps a circulation reactor with peroxide injection and by installing several (more than one anyway) relief devices, then an apparent relief area greater than what results from a reactor diameter is possible.

 

[Rocco8]

The Diers Program calculates relief devices for Runaway Reactions. We have used it successfully and know that it works through field validation.

 

[CHD01]

In addition to limiting the batch size for the reactor, you may also be able to reduce the set pressure of the relief device to something less than the maximum allowable operating pressure of the reactor(ie - design pressure usually = MAWP). Why do this? Because this can reduce the total heat of reaction that occurs thus reducing the required relief rate at the MAWP of the vessel plus allowable overpressure. FOr example, if allowable overpressure for relief device is 10% and if reactor MAWP = 100 psig and normal reactor operating pressure = 10-15 psig. You might want to evaluate total heat of reaction developed using a relief set pressure of 40 psig or 50 psig instead of 100 psig. Just don't get so low that a relief becomes probable versus normal operating range.

Note: Doing this requires that you also determine what pressure above the relief set point you will allow; you need to specify to the relief valve vendor whether the relief valve set at 50 psig (for example) is to have an orifice sized to relieve the required rate at 10% above the set point OR 10% above the reactor MAWP (equal to 110/50 or 220% above the relief setpoint). If you want it 220% above the setpoint (10% above the rx MAWP) you need to verify with the vendor that the relief valve body flanges, etc can withstand the MAWP plus overpressure for the lower presssure spring.

This may sound strange to you but it is allowed, its just that not many designers or aware of this possibility.



The more you learn, the less you are certain of.

 

[pleckner]

To CHD01:

I've had the discussion of setting PSVs at lower than design or MAWP with many people and always seem to get into "arguments" over this. You are correct in that many are not aware of this possibility. But that brings me to your statement(s) concerning allowable overpressure. First, why bother setting the PSV at say 50 psig (for a 100 psig design pressure) if you are still going to allow the overpressure to reach 110 psig? The final temperature will still be coincident with the 110 psig and you will probably be back into your undersireable (run-away) status. You have gained nothing. The object of setting the PSV down is to avoid the run-away because this is nearly impossible to control and protect against with a PSV.

Second, code (ASME that is) requires the PSV flow capacity to be determined at 10% over the SET pressure, not 10% over design or MAWP. This is the basis for the stamped capacity and no vendor will be willing to guarantee the flow through their PSV at any overpressure other than this. This implies that even though ASME may allow a 220% overpressure, using your example, the stamped capacity of the PSV will not follow suite and thus, you will not have any guarantee of the PSV capacity at this 220% overpressure.

So begins the arguments....

 

[EGT01]

There are times when modest overpressures can be of benefit in reducing the size of vent required. A discussion of this can be found in one of the articles found at the Fauske web site above (see article C02, A Quick Approach to Reactor Vent Sizing). This works particularly well when reaction heat causes a component to vaporize which tends to cool the reaction as that material is vented but you have to be careful that depletion of an exotherm-controlling volatile doesn't result in a more severe event.

I think early venting also allows quicker removal of reacting material from the vessel which can also reduce the severity of the event.

In my recent experience with runaway relief design, an overpressure of 20% had a significant impact on reducing the required vent size but beyond that there was little benefit to higher overpressures. Of course different systems would have different considerations.

Not necessarily related to amount of overpressure but in case you haven't seen this reference, you might find this interesting, it has some worked examples
CRR 1998/136 - Workbook for chemical reactor relief system sizing
Main Report and Annexes

http://www.hse.gov.uk/research/crr_htm/1998/crr98136.htm

Anyway, I think you can say there are times when having the set pressure below MAWP provides some benefit. How much below MAWP, depends. But for spring loaded relief valves, seems like too much overpressure could result in a reclosing relief device becoming a non-reclosing device. Sure seems like I remember seeing a previous thread regarding the issue of high overpressure but couldn't find it.

As for vendor guaranteed flow, I believe that is what is stamped on the relief device. Usually that flow is given as either scfm air, lbs/hr steam or gpm water. But we use those same devices anyway in other fluid service. Of course ASME provides a method for capacity conversion of safety valves (Appendix 11) as well as a method for prorating the relieving capacity (UG-133) at any relieving pressure greater than 1.10p, as permitted under UG-125 (I'm not sure but this may permit prorating only up to 21% overpressure).

Then there are formulas specified by the National Board that can be used with certified valve coefficients in order to determine the certified relieving capacity and vendor formulas that use the valve coefficients to determine minimum required area. But the valve coefficients are either given for vapor/gases or liquids. So where does that leave us when it comes to vendor certified two-phase flow capacity? Probably lacking. My point is, in many cases, we are using relief valves in fluid services other than for what they have been certified. So I wonder how much guarantee we have of those relief valve capacities anyway.

 

[CHD01]

As to why bother lowering set pressure, I can only add that I have a real reactor that was studied in conjucntion with an honor guard of early DIERS founders seeking a solution for a reactor that could not safely relieve its fire case plus the heat of reaction. We never did reach the case of being able to handle the worst case but we were able to get close (what we fell short on was treated with numerous interlocks, safeguards, etc.); and one of the soutions was to decrease the set pressure to about 1/2 Design Pressure for the reactor. The actual simulation demonstrates very well a significant reduction in the PEAK release rate for which the relief system is designed. The system consists of a 12-in 3-way valve with a single T orifice 8x10 relief valve with a special 12-in rupture disk and holder preceeding it; plus a 12-in rupture disk on the other side of the 3-way valve to protect vessel should relief valve be out of service.

When I first got incolved with this problem they had a special 10x12 relief valve made and installed for protection. There was one problem with this valve, however, it was so big they could not inspect and refurbish it to meet required inspection intervals.

So in summary to first question, yes it does work - it was proven to work in an actual DIERS simulation; the PEAK relieving flow WAS REDUCED! It involves MORE than single component saturated temperature/pressure conditions at relief conditions.

Second - on certified relief capacity, the relief valve is certified at 10% based on PPH steam, SCFM Air, or GPM water regardless of the actual fluid being relieved (lets leave boilers and steam out of the discussion for simplicity). Further, if the allowable overpressure is 21% for the fire case, or if we have multiple valves, etc., the valve is still certified at 10% (there is one case that is special but I won't get into that). So the issue of what the valve is CERTIFIED at versus what the relief valve relieves does not require that they both be the same. What the design record for the valve should show besides the CERTIFIED CAPACITY (in air , water or steam) is the ACTUAL relief capcity required AND the MAXIMUM relief capacity for the fluid being relieved at the overpressure selected for the vessel being protected! What must be true is that the required flow is less than the capacity of the valve whether expressed in terms of pph steam or the actual fluid. Setting the relief valve below the MAWP is always an option - you only need to justify it. As previously stated, instead of applying the usual 10% overpressure above the SETPOINT - you should be aware of the potential to go to 10% above the MAWP even at the lower setpoint. Note my caution was to get with the relief vendor and verify spring, housing, flanges: will a) handle higher relieving pressure and safely relieve this with a lower set pressure spring.

Final Note: Note that with the lower set pressure and 220% over set pressure (10% over MAWP in our example) you wind up getting a smaller orifice for the required flow and get less pressure drop as a result versus going to valve with a setpoint equal to the MAWP.

Last Final Note: What we are talking about here in my case was a batch reactor where we were trying to prevent an explosion - trying to get the relief valve to reclose was not an issue!

Last Final Final Note: This was a complicated case, my point is that it can be done; should those not highly trained attempt it - perhaps it would be better not to. I think you all have taken excellent positions to either support this approach or to argue against it being used routinely - but it does have its place and that is the only point I want to make here.





The more you learn, the less you are certain of.