Pipe line failure statistics

[ssk]

Dear all,
In my industry, we have 1.5" to 40" size gas pipe lines containing toxic gas. What size of pipe line failure is to be considered for risk assessment. Do we have to consider full bore rupture of 40" size or full bore rupture of smallest pipe line? Are there any standards for this? Thanks in advance for your positive reply
sskumar

[ARMOFI]

Consider the worst scenario that afflicts maximum damages. Here it can be full bore rupture of the size you consider. Also calculate all possible consequences, frequencies & severity of these scenario to summarize highest risk available at the site.

[safetyuser]

Dear all,
In my industry, we have 1.5" to 40" size gas pipe lines containing toxic gas. What size of pipe line failure is to be considered for risk assessment. Do we have to consider full bore rupture of 40" size or full bore rupture of smallest pipe line? Are there any standards for this? Thanks in advance for your positive reply
sskumar

Dear ssk, I assume it is onshore pipeline.

fullbore rupture alone of a pipeline is the worst case scenario and should be considered only for benchmark purposes when performing emergency planning or conceptual study.
Indeed I would use a quantitative approach and give a look to the report by European Gas Pipeline Incindent Data Group (EGIG) comprising 3 hole sizes with their frequency class: pinhole, hole, rupture.

CONCAWE statistics define the ratio HoleArea/Pipeline cross section as "pinhole" up to 0.003%, "fissure" up to 0.08%, "hole" up to 0.9%, "split" up to 12% and rupture from 48% and above.

Approximately, a review of the most reputable reports on pipelines leaks show that the main differentiation in the data set is between a leak and a rupture. A leak is a loss of product that does not result in the impairment of the operation of the line, and a rupture is a larger release that does. The proportion of these is about 87% leaks to 13% ruptures.

You could say pinhole release is hardly detectable by process measurement whilst holes and ruptures are.

hope this helps

[ARMOFI]

The failure frequency rate are not uniform. It depends more on data available on hand. What is observed in Europe may not be true in Asia or in US. These rate can be taken from your company or group of company in your region. Or select a rate from a region that is close to you so that the predictions are more realistic.

As earlier, one sr. member brought out, there are various size of leaks (one such scenario), fire, rupture & explosions are also possible. You need to predict exactly from the statistics available in hand "the Failure rates". Moreover there are chances for you to calculate the failure rates of various accessories such as pumps, flanges, com pressures, valves etc.. In total, you need to analysis the risk rating.

Deutch Risk models as explained in blue, purple,yellow book may be consulted. or PHAST / CIRRUS can be used if you have knowledge on these.









'.

[ssk]

Thanks safetyuser and ARMOFI. The Health and Safety Executive (HSE) of UK says that the failure scenario is to be considered if the frequency is more than one in a million (1x10^-6. Where as , the purple book says the LOC(loss of containment) event is to be considered only when the frequency is 1x10^-8 and the lethal damage(1%probability) occurs outside the establishment's boundary. My question is how one can decide whether scenario considered is credible or not. Thanks.

[safetyuser]

Thanks safetyuser and ARMOFI. The Health and Safety Executive (HSE) of UK says that the failure scenario is to be considered if the frequency is more than one in a million (1x10^-6. Where as , the purple book says the LOC(loss of containment) event is to be considered only when the frequency is 1x10^-8 and the lethal damage(1%probability) occurs outside the establishment's boundary. My question is how one can decide whether scenario considered is credible or not. Thanks.

Hi
first I would say to search an advise about the Local regulations in force in the country for your Project to be based. Maybe you'll find the true values for risk acceptance criteria.

2nd: I feel to correct your approach since in UK regulations the limit 1E-6 per year is not the definition of what you call "a credible scenario" but it is the Broadly Acceptable limit for the Individual Risk for members of the public. This is only one of the possibile risk measures to be examinated (e.g. other measure maybe the societal risk). Seek advise by your QRA specialist on this matter if you don't feel confident on the matter.
The same apply for the purple book which follow the Dutch regulations where 1E-8 is used instead.

3rd. Now you are introducing a new issue with respect your first post: the former was about What size of pipe line failure is to be considered for risk assessment, but now you are talking about tolerability criteria of the risk i.e. how much tolerable is the resulting risk AFTER you have completed the full calculation process. Again on this matter you should seek advise to your team QRA specialist, and the purple book you have mentioned above is a very good start on this technical academics.

4th. "how one can decide whether scenario considered is credible or not?". Once you have the statistics by hole size in your pipeline ALL the scenario are credible since each of them contributes to the overall risk for people: maybe the least frequent scenario by itself contribute in negligible way but be sure that tens of 1E-7 scenario, when superimposed in one location, become >1E-6 outcomes and very risky for people.

regards

[ARMOFI]

As explained here by the Sr. member, it is merely conforming to particular standard, regulations. In UK, the HSE classifies acceptance criteria ie 1E-3 & 1E-4 are unacceptable & 1E-6 is generally acceptable. The region between is acceptable but with ALARP (As Low As Reasonably Practical) measure. So, the tolerance level is dictated by HSE when we measure a risk for an Individual or Society.Purple book is Dutch Model obviously portraying their tolerance level .There is no harm in it. If you see closer 1E-8 even stricter than 1E-6. Just see what is 1 % of 1E-8. Thus it is only a compliance to regulation of a particular region. Nothing to worry about credibility.

[ssk]

Thanks SAFETYUSER and ARMOFI. Sorry for the delay in thanking you for useful inputs. Still I am little bit confused. Why one should consider rupture scenarios ( either vessel or pipe line) when there is no past history of ruptures in that particular type of Industry. Failure frequencies given in many data bases are based on the data published in 1970 and mainly based on steam generators or chlorine vessels or Ammonia vessels and that too on older technologies. Suppose I have got a vessel which handles Hydrogen Sulfide. Now, the data given in the failure data bases can not be applied for my vessel as services are different. But I have got 100 years of data on Hydrogen sulfide vessels and pipe lines, which says that there is not even a single rupture of either pipeline or vessel, only pinhole/----- failure were reported. Now for my QRA, whether should I consider the rupture scenario? Going by World Bank guidelines, IAEA guidelines, AIChE publication (QRA with data tables), HSE guidelines, one should give preference to plant specific data. NOW, please let me know whether should I consider rupture scenarios? If it is to be considered, which did not happen for 100 years of experience, then plant can not run as H2S, you know, highly toxic.

[ARMOFI]

Thanks SAFETYUSER and ARMOFI. Sorry for the delay in thanking you for useful inputs. Still I am little bit confused. Why one should consider rupture scenarios ( either vessel or pipe line) when there is no past history of ruptures in that particular type of Industry. Failure frequencies given in many data bases are based on the data published in 1970 and mainly based on steam generators or chlorine vessels or Ammonia vessels and that too on older technologies. Suppose I have got a vessel which handles Hydrogen Sulfide. Now, the data given in the failure data bases can not be applied for my vessel as services are different. But I have got 100 years of data on Hydrogen sulfide vessels and pipe lines, which says that there is not even a single rupture of either pipeline or vessel, only pinhole/----- failure were reported. Now for my QRA, whether should I consider the rupture scenario? Going by World Bank guidelines, IAEA guidelines, AIChE publication (QRA with data tables), HSE guidelines, one should give preference to plant specific data. NOW, please let me know whether should I consider rupture scenarios? If it is to be considered, which did not happen for 100 years of experience, then plant can not run as H2S, you know, highly toxic.

As far as my knowledge goes, consider the pipeline/vessel failure frequency irrespective of its' content. Content properties are analyzed in consequence analysis such as flammability, Radiation, toxic effect, over pressure effects. etc. QRA considers all potential release models. Subsequently it's consequence, frequency, severity ,risk rating predicted.

[safetyuser]

Now, the data given in the failure data bases can not be applied for my vessel as services are different. But I have got 100 years of data on Hydrogen sulfide vessels and pipe lines, which says that there is not even a single rupture of either pipeline or vessel, only pinhole/----- failure were reported. Now for my QRA, whether should I consider the rupture scenario?

One could reply that you must consider pipeline rupture because so is requested by the regulation in force and it is the" good engineering practice" or the state of the art in the QRA field.

Another one (included me) could reply that my failure database do consider at least 1 rupture of pipelines and that is in principle a theoretical chance. Theoretical, of course, but not sufficient to be neglegted on "a priori" basis.

If it is to be considered, which did not happen for 100 years of experience ...

Also Piper Alpha destruction accident did not occur before of that time, and I guess some people made the same consideration: "It never happened for 100 years of experience". Think about Deepwater Horizon ...... again in that case the specific problem "never happened for 100 years of experience" but ... shit it happened!

Please bear in mind that we're talking of events with a failure period in the order of thousands of years (1E-3 for acceptance criteria): neither me nor (possibly) the majority of this forum members will survive so long to witness such a catastrophic failure event........but it is still a credible event to account for.

NOW, please let me know whether should I consider rupture scenarios? If it is to be considered, which did not happen for 100 years of experience, then plant can not run as H2S, you know, highly toxic....

Again, you are here misusing the concept of "credible events" with "tolerable or acceptable risk" .....