Flare Load Upon Instrument Air Failure 1

[Sawsan311]

Dear All,



Based on API 521 6th edition section 5.3.3, a common mode failure such as instrument air can have coincident opening of the emergency depressuring normally energized valves along with the pressure control valves installed across the plant which fail in open position upon loss of instrument air. I know that we can avoid the opening of the BDVs by providing sufficiently sized secured instrument air bottles. However, do we really have to consider or check the above case while designing the flare system? Will the basis of the fact that the starting pressure under the instrument air failure scenario would be relatively lower than the usual conservative basis of relief or emergency fire blowdown design calculations and hence the combined PCV + depressurization load would be hence covered? Do we also have to check the case of all PCVs opening simultaneously to flare across the entire plant?



Thanks

 

[don1980]

I don't mean to be cavalier, but are you asking us to guess the answer? Shall we flip a coin? One can't determine whether this global scenario (inst. air failure) is a significant or controlling scenario without doing the necessary work to determine that load. Certainly, instrument air failure is a credible scenario (unless there's a high-integrity back-up system that automatically activates to provide the necessary capacity). We can't justify ignoring this, or guessing at it, any more than we can justify ignoring or guessing at any of the other global scenario loads. You just gotta do the work.

 

[georgeverghese]

The basic concern you have here seems to be the additional load on the flare system resulting from PCVs' which are FO. Why do you have PCVs' which are connected to flare as FO ? To me, there is no loss in process systems integrity in making these FC. We must remember that these PCVs' are merely for operational pressure control, not safety overpressure devices.

The peak instantaneous flare load from plant wide IA failure would be the same as plant wide ESD with blowdown, provided these PCVs' to flare are all made FC.

 

[Sawsan311]

Mr. Don1980, I was not asking you to flip a coin or consider an imaginary scenario. Despite the fact that blowdown valves are usually provided with secured instrument air vessels and UPS supply, yet if you look at old designed, rarely Engineers considered these mitigation measures. The pressure control valves in some old designs are all fail open to flare considering them as mitigating measures. There is no guess in the answer, in fact API 521 even pointed out the consideration of having the EDV as well as the flare PCVs saying'' emergency depressuring valves and/or pressure control vent valves.

 

[Sawsan311]

Dear MR. georgeverghese, thank you for your kind response, we usually consider the flare PCVs to fail open on production separators and compressors suction manifolds although these PCVs play no role in the safety functions provided in the process systems yet they can contribute in reducing the demand rate on the SIF function as an independent layer of protection.

 

[georgeverghese]

With all these FO PCV's in your plant, think it has become more unsafe. You dont need these to be FO, especially when you have a good working emergency blowdown system in place. It adds un necessary load on the flare system.

 

[EmmanuelTop]

There usually is a 10-15 minutes buffer time in case of loss of supply of instrument air. Under such circumstances, a typical gas plant or refinery unit would likely be manually shutdown without depressurization, and hence the flare load would be little to none. This may not be true for chemical plants involving reactions, where there is a specific safety shutdown procedure involving relief to flare.

If the loss of instrument air would ultimately lead to uncontrolled opening of all those FCV's and PCV's, I'd suggest you investigate the solution to trip the plant at a preset value of instrument air pressure. Or, to evaluate the flare load and see if the flare system (sub-headers and headers as well) can take it.

There is no such thing as "common sense" - Apollo RCA

 

[don1980]

Sawsan, I'm not intending to be unkind. Instead, I'm trying to get you to recognize an obvious reasoning error that you've made. You have x number of FO control valves that will open to the flare header during an instrument air failure, and you're asking if the resulting load will be lower than the load from other global scenarios. The answer obviously depends on the sum of the loads from those X number of FO control valves, and the comparison of that sum against your other global scenario loads. Only you can answer this question, after summing the control valve loads.

If the FO control valve loads exceed the other global scenarios, then consider taking the actions mentioned by others, to safely solve this problem.

 

[Sawsan311]

Hi Don,

The actual question was not whether the sum of the FO PCV opening to flare would be higher than the blowdown load.. In fact the question was whether the design should evaluate the flare system capacity under COMBINED load of blowdown depressurization with the opening of fail open control valves.

By all means, even if the PCVs are retained to fail open to flare, the only lucky feature we have is that normally energized blowdown valves are provided with secured instrument air vessels and UPS power supply which tends to increase the survival of the plant for a duration of at least 30 minutes. During this duration the plant will be under pressurized shutdown state. The secured IA vessels on the other hand are not provided for the case of flare vent control valves.

In case an old system exists where blowdown valves are not provided with the features of secured vessels or UPS, then one can't deny the fact that upon instrument air failure, we have to evaluate the effect of PCV failing in open position in addition to whatever simultaneous blowdown load from the BDVs opening. The question wasn't inquiring which one is governing out of the two.. it was whether we should consider fail open PCVs to flare as additional load to the depressurization load.
In our design philosophy, we never put the flare PCVs as fail close, specially at compressor suction manifolds as they play a major role in protecting against stonewall conditions.

Regards,

 

[don1980]

Sawsan, yes you are absolutely correct, and thank you for setting me straight. I apologize for misreading your original question and for my responses which are unrelated to that question.

Whether there is overlapping flow from the PCVs and the depressurization valves requires a dynamic assessment of this instr air scenario. API 521 has a very high-level explanation in paragraph 5.3.4.2, but it can't provide details because each case is different. A dynamic evaluation results in a plot of load versus time. Each vessel isn't going to have the same rate of pressure rise, nor will each vessel vent for the same period of time. A dynamic evaluation is a laborious task, but it will almost always result is a much lower flare load than that which is calculated using the simplified assumptions which we ordinarily make (all sources within a global scenario relieving at the same time). The size (capacity) of the individual PCVs is another important factor, along with how long it will take before the pressure in the instr air header to bleed down.

Start by focusing on the vessels that will cause the highest vent loads. After plotting the results for a subset of these high-load vessels, run the flare header model to see if this sufficiently reduces the overall flare load/back-pressure. If so, then you're done. If not, then add the next highest loads to the dynamic analysis. Continue that iteration until the flare model reveals that the problem is solved.

 

[dean427]

To take credit for these secure air supplies you will need to have a very robust inspection & maintenance regime to ensure they are holding tight. The volume bottles may also need to be fitted with an alarm and dual class 1 check valves to ensure a robust secure design.

My plant's vent and flare system is not sized to take a coincident blowdown from all plant units (our depressurisation is sequenced to avoid overload and excessive back-pressure/radiation) we have such bottles but on an ageing plant it is difficult to keep on top of the inspection and maintenance program. As well as local bottles we also have dedicated common secure air receivers. From memory there are design rules in API 521 to consider a certain number of the largest depressuring valves opening simultaneously but I supposed you have checked that?