9 replies to this topic

Hi all,

I am a very junior (just graduated) process engineer in the oil and gas industry.

Please see attached drawing. I have been asked to examine whether a relief valve on a 3-phase separator V-100, which has been sized for overpressure from blanket gas, needs to be sized for liquid relief as well.

Using API 520 Section 5.8 (Sizing for Liquid Relief: PRVs Requiring Capacity Certification), I have sized the RV and obtained a larger required area than for the gas relief case (~300 mm² vs. ~100 mm²). I did this using the approach whereby the Produced Water is entering the vessel at its maximum flow rate, and all the vessel outlets are closed.

My question is:

Is this scenario too conservative? Is it double jeopardy to assume that the vessel high high level trip has failed and all outlets are blocked? (I talked to someone who said it is feasible that during startup someone could have left the outlets closed and started to fill the vessel.)

Please advise if I have aproched the RV sizing correctly, or a more suitable approach to sizing it.

Just to reiterate, referring to the attached drawing, the RV-100A has currently been sized for gas relief, for a fail open of PCV-100A with PCV-100B remaining closed. What I have done is size it for a fail open of FCV-050A (maximum liquid pump-in rate), with all outlets blocked. I want to know the correct approach to sizing RV-100A.

Attached Files

•  RV schematic.pdf   51.67KB   140 downloads

Edited by Mark E. Smith, 21 December 2010 - 11:56 PM.

[fallah]

The scenario is double jeopardy because of simultaneous occurance of two unrelated causes of overpressure (operator error for outlets closed, and instrument failure for high high alarm trip).

[praveen.bajaj]

The PSV (RV-100A) should be sized to handle the following loads:

1. Fire
2. Partial power failure causing pump to shut down and LT-(high level shut down) failure (this high liquid level alarm failure can not be considered if we two out of three voting, in other words three level transmitters available to cut the incoming feed to the vessel). PCV-100B operating normally. High liquid level will cause increase in pressure due to reduction in vapor space and will cause PCV-100A to close and PCV-100B to open full. During calculation of PSV for this scenario, credit could be taken for the flow that will be relieved through PCV-100B.
3. Blanket gas valve failure with PCV-100B in closed condition

Most of the time, fire scenario becomes a dominant case but need to be looked into more carefully. Also since you stated hydrocarbons in produced water (which is generally in range of 300-5000 ppmv) latent heat of vaporization for liquid hydrocarbon need to be considered for fire scenario since there will be a pad of liquid hydrocarbons that will build inside the vessel. If latent heat of hydrocarbon is not available, GPSA recommends to use 50 BTU/lb as safety to come up with the desired relief flowrate.

Hope this will give you some guidance.

Thanks
Praveen

Edited by praveen.bajaj, 22 December 2010 - 10:03 AM.

Thank you for your help. The governing case ended up being HP gas breakthrough from upstream.

[paulhorth]

Mark,

The concept of double jeopardy is now generally obsolete in the offshore industry and so I disagree with Fallah's comment. For many years now, the offshore recommended practice API RP 14C has applied the principle that every cause of overpressure should have two independent means of protection. This is inherently accepting double jeopardy, that is, that two independent failures can occur: First, the malfunction itself (in your case, blocked outlet) and second, the failure of one protective instrument (in your case, the high level trip).
In this case, the two independent means of protection against overpressure due to blocked outlet would be (1) high level trip which closes the upstream inlet valve (2) a PSV sized for the maximum inflow rate. This solution would be recorded on a API RP14C SAT table.

You then go on to assess other causes of overpressure, such as fire, and gas blowby from upstream. I can well believe that this latter cause is the governing rate. Each of these causes should have two independent means of protection. In the case of fire, the first means of protection is usually the fire detection and protection system ( perhaps including deluge), and the second is the PSV.
In the case of gas blowby, the first means of protection should be a high pressure trip closing the inlet ESDV, and the second is the PSV. All of these solutions sholud be recorded on the SAT table.
The system should be HAZOPed to determine if all the causes of High Pressure have been identified.

Paul
Co-author, "Guidelines for the Safe and Optimum Design of Hydrocarbon Pressure Relief and Blowdown Systems", Institute of Petroleum, 2001. This is recommended for all junior process engineers.

[fallah]

The concept of double jeopardy is now generally obsolete in the offshore industry and so I disagree with Fallah's comment. For many years now, the offshore recommended practice API RP 14C has applied the principle that every cause of overpressure should have two independent means of protection. This is inherently accepting double jeopardy, that is, that two independent failures can occur: First, the malfunction itself (in your case, blocked outlet) and second, the failure of one protective instrument (in your case, the high level trip).
In this case, the two independent means of protection against overpressure due to blocked outlet would be (1) high level trip which closes the upstream inlet valve (2) a PSV sized for the maximum inflow rate. This solution would be recorded on a API RP14C SAT table.

1-Are you sure the raised issue by PO relates to offshore?

2-"two independent means of protection" differs from "two causes of overpressure"

3-Double jeopardy concept would be applied in HAZOP study and means of protection mostly in LOPA study.

Indeed,i disagree with you that API RP 14C by mentioning to necessity of two independent means of protection for each cause of overpressure inherently accepted the double jeopardy concept,but it just mandates having two independent means of protection as minimum layers of protections for offshore applications.

Finally,by looking at the attached page from API 521 one can find a general concept for double jeopardy without any exception within its applications.

Attached Files

•  Pages from API_STD_521Double Jeopardy.pdf   80.53KB   63 downloads

[paulhorth]

Fallah,

The causes of overpressure for this vessel are (1) blocked outlet, (2) fire(3) gas blowby. Each cause of overpressure can occur due to malfunctions. Malfunctions include operator error, instrument failure, power failure etc.

The key point of API RP 14C is to accept that two malfunctions occuring together , for any cause of overpressure, is not double jeopardy. Therefore, two independent means of protection should be provided for each cause of overpressure.
In your earlier post, you refer to two malfunctions, namely operator error and instrument failure, and you call this double jeopardy. I disagree with this, and so does API RP 14C. To repeat: Instrument failure is not a cause of overpressure in this vessel. Blocked outlet is the cause of overpressure.

Double jeopardy as referred to in API 521 means two independent causes of overpressure happening together, such as fire and exchanger tube rupture, or blocked outlet and power failure. This is made clear in the examples given in the document.
Consequently, I believe that Mark Smith was correct to consider overfilling the vessel with liquid as a valid relief case for this vessel. It was not the governing case, as he has now found.

Paul

[fallah]

Fallah,

The causes of overpressure for this vessel are (1) blocked outlet, (2) fire(3) gas blowby. Each cause of overpressure can occur due to malfunctions. Malfunctions include operator error, instrument failure, power failure etc.

The key point of API RP 14C is to accept that two malfunctions occuring together , for any cause of overpressure, is not double jeopardy. Therefore, two independent means of protection should be provided for each cause of overpressure.
In your earlier post, you refer to two malfunctions, namely operator error and instrument failure, and you call this double jeopardy. I disagree with this, and so does API RP 14C. To repeat: Instrument failure is not a cause of overpressure in this vessel. Blocked outlet is the cause of overpressure.

Double jeopardy as referred to in API 521 means two independent causes of overpressure happening together, such as fire and exchanger tube rupture, or blocked outlet and power failure. This is made clear in the examples given in the document.
Consequently, I believe that Mark Smith was correct to consider overfilling the vessel with liquid as a valid relief case for this vessel. It was not the governing case, as he has now found.

Paul

No one be able to put a red line between malfunction and cause of overpressure!

Looking above,for example,you mentioned power failure not only as a malfunction but also as a cause of overpressure.

Indeed,as an example,PIC malfunction (or failure) to close relevant PV full or more than required shall be considered as a cause of overpressure!

Also,looking at API 521 (sec. 4.2.3) contrary to your above mentioning,Operator error has been considered as a potential source (cause) of overpressure!

Edited by fallah, 26 January 2011 - 08:17 AM.

[paulhorth]

Obviously some causes of overpressure can also be counted as malfunctions. This simply reinforces the argument that the concept of "double jeopardy " should be applied very cautiously.

The point is that two malfunctions cannot be called double jeopardy. This is clear from API 14C. In the case of Mark Smith's separator, the failure of the high level trip could have happened months ago and remained undetected. This eliminates one level of protection. The overpressure would not occur until something else happened, namely blocked outlet. The two faults would not be simultaneous. So, at that time, a single malfunction would result in overpressure and, if the relief valve was not sized for this situation, the vessel could rupture.
You advised in your original post that it was not necessary to size the relief valve for liquid overfilling. I disagree, for the reasons I have given, which are also set out in API RP 14C.
You are free to make your own decisions. However, in the event of an accident, you should be ready to explain to a judge and to the victims' families why you decided not to apply international best practice and the API Recommended practice RP14C.
I can tell you that the approach in your original post would not have been accepted by any client that I have worked for, nor by my colleagues in HSE on the contractors side.

Paul

[fallah]

Obviously some causes of overpressure can also be counted as malfunctions. This simply reinforces the argument that the concept of "double jeopardy " should be applied very cautiously.

The point is that two malfunctions cannot be called double jeopardy. This is clear from API 14C.

You haven't still replied to my previous question as:"1-Are you sure the raised issue by PO relates to offshore?

In the case of Mark Smith's separator, the failure of the high level trip could have happened months ago and remained undetected.

Failure of high level trip for months and remained undetected!!!What about level of Integrity-Availability and Reliability for,let say as you insisted,a vessel under API RP 14C?Here instrument and control experts may submit a better judgement better than me.

This eliminates one level of protection. The overpressure would not occur until something else happened, namely blocked outlet. The two faults would not be simultaneous. So, at that time, a single malfunction would result in overpressure and, if the relief valve was not sized for this situation, the vessel could rupture.
You advised in your original post that it was not necessary to size the relief valve for liquid overfilling.

I hadn't advised about sizing the RV and just submitted my viewpoint about double jeopardy regarding mentioned scenario.

I disagree, for the reasons I have given, which are also set out in API RP 14C.