An Uncertainty Quantification and Management Methodology to Support Rework Decisions in Multifidelity Aeroelastic Load Cycles
Johnson, Brandon James
MetadataShow full item record
Cost overruns and schedule delays have plagued almost all major aerospace development programs and have resulted in billions of dollars lost. Design rework has attributed to these problems and one approach to mitigating this risk is reducing uncertainty. Failing to meet requirements during flight test results in one of the most significant and costly rework efforts. This type of rework is referred to as major rework and the main purpose of this thesis is to reduce this risk by improving the loads analysis process. Loads analysis is a crucial part of the design process for aerospace vehicles. Its main objective is to determine the worst-case loading conditions which will realistically be experienced in normal and abnormal flight operations. These conditions are called critical loads. With this information, a structure is designed and optimized to withstand such loads and certify the design. Observing the current approach to loads analysis has revealed some shortcomings related to uncertainty and the allocation of load and structural margins. The fields of uncertainty quantification and uncertainty management were chosen to address these limitations and a framework was proposed to support decisions for rework in loads analysis. Key aspects of the framework include utilizing a Bayesian network for modeling the loads process as well as propagating various uncertainty sources to the system response. Bayesian-based resource allocation optimization is another key aspect and used to reduce and manage uncertainty. Finally, the goal of the framework is to determine the optimal tradeoffs between aerodynamic fidelity and margin allocation to minimize the risk of major rework while considering their respective costs under a finite budget. Assigning costs related to fidelity and margins are intended to reflect the users' prioritization of uncertainty, computational cost and performance degradation through weight penalties. The demonstration model is the undeformed Common Research Model (uCRM) wing, which is representative of a transonic wide-body commercial transport. The modeling and simulation environment is multidisciplinary and anchored in three software programs to perform various analyses: NASCART-GT for computational fluid dynamics; NASTRAN for doublet-lattice method aerodynamics, structural analysis and aeroelastic analysis; and HyperSizer for failure analysis and structural optimization. Four experiments were conducted related to epistemic uncertainty quantification, uncertainty propagation and sensitivity analysis via Bayesian network, developing an uncertainty management system based on resource allocation for loads analysis and finally experiments to optimize and evaluate the overall framework against seven design scenarios to explore a potential decision makers' varying priorities and against a baseline model representing the current approach. Key findings reveal the structural required margins are the dominant factor in reducing the risk of rework but the aerodynamic fidelity and load margin are important for balancing performance and uncertainty when considering financial implications within a finite budget. The contributions of this thesis to the aerospace engineering community include; an integrated modeling and simulation environment for the load analysis process and structural design, uniquely applying and developing a Bayesian network for efficient uncertainty modeling and propagation and a viable cost-based uncertainty management system for loads analysis, among others.