Relance on operator responce API521 to discount a case

[jamesbanda]

Dear all,

I've noted in API 521 page 24: 5.4 Effect of operator comments are made regarding allowance for operator responce. It stated along the lines that you dont need to size for a case if an operator responce time is 10-30 minutes and they can reasonably be expected to take the right respoce. It has room for interpretation - which makes sense.

Has anyone actually used this argument? not to size for a process case or even calculate the relief rate

Note: I dont want to prejudge or bias peoples responces by explaining my perpsective.

I would appeciate candid comments on this:

 

[ggordil]

I have on occasion used this to take a scenario from 10% accumulation to a remote contingency and get 50% accumulation. Mostly overfilling scenarios on towers.

 

[sheiko]

Hello,

I have used this criteria for the following 2 scenarii:

- Fire case: if the amount of wetted surface area in a pressure vessel is totally vaporized by fire before operator response, one should treat the vessel as a gas-filled vessel.

- Liquid blocked outlet: when the liquid filling time is less than the time for operator response, then the scenario is credible.

In any case, be conservative and use 30 min. It highly depends on the location of the equipement however...

"We don't believe things because they are true, things are true because we believe them."

 

[skearse]

I would be very careful in using this to discredit any scenarios. API themselves (in a training course I attended), in lieu of the Texas City incident, is shying away from the 10-30 minutes recommendation (at least for taking credit for an overfill scenario) and is moving toward simply requiring a risk analysis, without the time recommendations. Section 5.23 was re-written in response to the CSBs recommendations in the Texas City final report.

 

[Chance17]

My client argues that operator response is the prevention of overpressure events when the plant is shutdown. Outages are infrequent, well planned and operators well trained. Any contingency during plant outage is viewed not applicable.

 

[sheiko]

Sorry Yitbos but english is not my first language. What do you mean by outage?

"We don't believe things because they are true, things are true because we believe them."

 

[jamesbanda]

Thankyou all for your responses.

I felt strongly that it is inapprorpate to take credit for operator responces without a proper consideration of the risks. I believe this can be done effectively if you document your basis in your risk assessment or hazard study.

Otherwise i would not take credit for it in rv sizing,

 

[Latexman]

I am not personally familiar with API 521. I work for a large company and our "experts" write company policy, design standards, procedures, etc. such that Code is met, so please bear with me.

We have standardized on LOPA for simple risk analysis of most scenarios. We have more complex tools for complicated scenarios.

In my training and experience, an operator response alone does not justify making a scenario non-credible. It also does not justify reducing the severity of a scenario or extending the protection level of another layer of protection; you add enough independent protection layers (IPLs) to balance the full scenario, which is evaluated assuming NO controls (i.e. no operator). Severe consequences need many IPLs, and an operator response that requires a 10-30 minute response time qualifies for only 1 IPL credit. Under the duress of an emergency situation, well-trained operators have a fairly high failure rate; thus only the 1 credit.

Good luck,
Latexman