Risk assessment of fuel property variability using combined Monte-Carlo/design of experiments methodologies 2019-01-1387
Increases in on-board heat generation in modern military aircraft have led to a more prominent reliance on thermal management techniques using fuel as a primary heat sink. Complicating matters, recent studies have found that fuel properties such as specific heat can vary from batch to batch by non-trivial amounts, impacting the amount of heat that can be delivered to the fuel. With modern systems utilizing the majority of available aircraft heat sink, improved understanding of the effects of fuel property variability on the overall system response is important. One way to determine whether the property variability inside a thermal system will cause failure is to perform an uncertainty analysis and compare the results to a risk assessment metric. In addition to highlighting areas where a thermal system failure occurs, a sensitivity analysis can be performed on the properties which cause the inherent system variability to determine which contributes the most significant impact. Several methods exist to perform uncertainty analysis, which vary based on complexity and intrusiveness. For the current study, an analysis was performed by combining two of the most popular uncertainty methods: sampling based using an orthogonal array approach and a design of experiments approach. The objective of combining the two uncertainty analysis methodologies was to obtain the statistical information provided by the sampling based approach and the sensitivity information provided by the design of experiments approach all in one test series. Performing two methods simultaneously allowed for the advantages of both methods to be obtained while still reducing the number of trials required to obtain a statistical sample. A series of tests was performed using varying sample sizes to test the versatility of a combined uncertainty analysis approach. The sample sizes were compared using computation time and sample size effect on statistical variation to determine the optimal setting for risk analysis.