Optimizing Based on Frequency Data


The speed of the optimization algorithms makes running several optimizations with slightly different parameters easy to run and review.


Using the criteria defined in the last section for Gas Detector Layout 02 - 10%LEL Point Alarm (Lo) and 1 LEL.m Open Path Alarm (Lo), etc. Except only change the Target Basis to Frequency and click Run.


Setting the target basis to frequency tells the Greedy algorithm to find a detector in a region with the highest frequency first, once added to the list the detected simulations are removed, and the process is repeated to place a detector in the next region with the highest frequency.


Tutorial 12 - Figure 16 - Resulting coverage for a target basis using frequency


Gas Detector Layout 03 shows a slightly steeper curve of the 1ooN coverage than when using the scenario basis. Meaning the first few detectors are being more effective when looking at the frequency. This can be because of several reasons: proximity of the devices to the leaks with the highest frequency, location of the devices downwind of the leaks are just two of the possible reasons.


It is typical for optimizations to have the first few detectors provide a great deal of coverage and the latter devices only contribute a small amount. This is most apparent in Figure 16 where 8 detectors achieve almost the same coverage as the entire layout with 10 devices. A cost/benefit analysis could then be used to decide if it is beneficial to have the last 2 detectors in the layout based on the project performance standards.


The coverage factor here of 80% can be thought of as the reduction in frequency of undetected leaks.


The next section looks at optimizing based on risk of the simulations run. Recall, the risk for each case is calculated by multiplying the frequency by the consequence. If an equal consequence weighting was applied to each simulation, then this would mean that the optimization would try to only detect very small gas clouds because those have the highest frequency (what was done in this section). This is an important concept as running an optimization by frequency alone would generally lead to larger detector counts and may not detect larger gas clouds (which is accounted for when performing optimizations with risk).