Leak size and frequencies

Crawley and Grant (1997) have developed a risk assessment tool for new offshore installations. They have examined typical leak frequencies of equipment items and the ignition probability of these leaks in four pressure bands. With this information it was possible to define leak size and frequency for any piece of equipment and the ignited leak frequency. In off-shore installations gas separation vessels were found to have ten times higher ignited event frequency than oil pumps. 

The size and frequency of leaks can be reduced by using spiral-wound gaskets in place of compressed asbestos fiber ones. Screwed joints should not be used. 

Given the difficulty of determining leak sizes and the frequencies of their occurrence these are usually fixed in safety reports (deterministic approach). An important role is played by the leak before break criterion, which implies that a leak of stable size formed before large area leaks in vessels or full cross section ruptures of pipes occur. However, the applicability of this criterion is subject to numerous prerequisites being fulfilled. Details can be found in [7]. 

In Eqs. (10.1)-(10.3) Dl denotes the diameter of the leak in mm (a circular leak geometry is assumed) and DN is the nominal diameter (occasionally called nominal bore) of the pipe (approximately equal to the internal diameter in mm). Equation (10.3) is the only one to estabhsh a relationship between leak size and its expected annual frequency of occurrence h. The latter refers to a length of 1 m and must therefore be multiplied by the length of the pipe under consideration. Equation (10.3) is based on evaluations for steel pipes in the process and petrochemical industries. 

The methods of calculation employed are those which are used as well for deterministic analyses. The difference is that stochastic boundary conditions, which are closer to reality, are used for the calculations. For example, instead of a fixed leak size a whole spectrum of leak sizes is treated with pertinent expected frequencies of occurrence being assigned to the different leak sizes. Instead of calculating the dispersion of a toxic substance based on a specific weather situation, different possible weather situations with their corresponding probabilities of occurrence are accounted for. This is reasonable, since the instant in time of the accident and the weather condition, which then prevails, are not known beforehand. 

It is extremely difficult to determine frequencies of occurrence for the loss of containment (leaks and ruptures). These frequencies depend on the size and length of pipes, the number of valves, the design of vessels and other medium containing components (e.g. pump casings). An important part is also played by the number of elbows, flanges, pipe branches, instrumentation ports etc. The properties of the medium involved, the pressures and temperatures just as their variations with time are of relevance, too. The frequency and quality of maintenance should also not be forgotten. 

Table 10.1 gives expected frequencies for some classes of ruptures and leaks. Both the frequency and the indication of the size are affected by large uncertainties. If uncertainties are stated the frequencies are represented by parameters of log-normal distributions (vid. Sect. 9.3.4), as often is the case with probabilistic safety analyses. 

Distribution of fire exceedance plots on the x-axis came from selection of different leak sizes in each set of representative scenarios. On the other hand, distribution of fire exceedance plots on the y-axis included information of uncertainty in fire frequency analysis. As a result, different fire exceedance plots involved information about uncertainty of the probabilistic fire load procedure. However, to cover the full extent of parameters, much more scenarios should be generated and simulated. In this situation, the non-CFD model was only qualified for the consequence analysis due to simple input and output. 

Bubble Dynamics. To adequately describe the jet, the bubble size generated by the jet needs to be studied. A substantial amount of gas leaks from the bubble, to the emulsion phase during bubble formation stage, particularly when the bed is less than minimally fluidized. A model developed on the basis of this mechanism predicted the experimental bubble diameter well when the experimental bubble frequency was used as an input. The experimentally observed bubble frequency is smaller by a factor of 3 to 5 than that calculated from the Davidson and Harrison model (1963), which assumed no net gas interchange between the bubble and the emulsion phase. This discrepancy is due primarily to the extensive bubble coalescence above the jet nozzle and the assumption that no gas leaks from the bubble phase. 

Unfortunately, fires, explosions, chemical leaks, and other incidents happen in the process industries. Depending upon the type of plant, its size, work force training, experience, and safety culture, a plant site may experience one or two incidents a year. An incident does not necessarily mean a serious fire or explosion. The most common types of incidents ranked by frequency are ... 

The assumptions made as to the size of leaks from flanges, fittings, piping, instruments, and vessels need to be documented. Similarly, an estimate as to the frequency with which leaks can occur is required. The frequency value will generally vary inversely with hole size. 

Platypus models the left hand or fault-tree side of a bowtie model for process safety by using a Bayesian Belief Net (BBN). That is to say, it calculates the frequency of occurrence of leaks of arbitrary size in process equipment. A leak or Loss of Containment (LoC) event does not necessarily lead to a catastrophe but in some cases leads a loss of containment can have severe consequences ranging from accidents or catastrophes. Most of the BBN-based method was develop in 2011 and 2012. The background information for the model can be found... 

The results from the calculations yield a distribution function for the LoC frequency. Ligure 1 shows the results for the base case. The figure should be read as follows. On the x-axis is the frequency of an LoC event of arbitrary size in [year ]. On the y-axis is the percentage of the frequency distribution histogram where the histogram has 100 equidistant size bins between the minimum LoC frequency (5.15 year ) and the maximum LoC frequency (11.2 year ). This means that it is estimated that the LoC frequency is between 5.15 and 11.2. Any leak frequency rate between those Umits may be expected in the observation of the unit. There is a 90% probabiUty interval between 5.91 year and 8.21 year, which narrows the likely occurrence of an observable LoC rate. The mean is 6.68 year the standard deviation is 0.674 year. The median is 6.55 year. ... 

Frequency information is required to match the event specification and consequence calculation. In general, leak frequencies of hole sizes (small, medium, large, rupture), see Table 1, in risk assessments are estimated from generic historical leak statistics, which works well for common items (pipes, valves, pumps, etc.) that have been collected and provided by COVO (Rijnmond Public Authority 1982 Crossthwaite P.J., Fitzpatrick, R.D. and Hurst N.W. 1988), DNV (DNV consulting 2005) and API (API 2008). 

---