Metamodeling uncertainty quantification in multi-level engineering system design

Design of complex system often involves a multitude of decision variables and parameters. In order to relieve the computational burden, manage the complexity in design processes, and achieve concurrent analysis and design, a complex system can oftentimes be decomposed into several subsystems with hierarchical relation according to their functional attributes, physical structures, or scale magnitudes etc., and each subsystem can be analyzed and designed independently. In addition, metamodeling techniques are widely used in hierarchical system design and analysis to replace the original time-consuming simulation models, like finite element simulation, molecular dynamics simulation etc., so as to future reduce computational burden. However, due to the limited sample points, metamodeling may contain predictive uncertainty at sites with no sample point. In order to investigate the impact of the metamodeling uncertainty on system performance in multi-level engineering system design, this paper proposes a metamodeling uncertainty quantification method for multi-level engineering system, and derives uncertainty of system performance as a function of metamodel uncertainties of subsystem models. Two numerical examples are used to examine the contribution of the uncertainty of subsystem metamodels to the uncertainty of top-level performance of studied systems.