Artemis Kloess

Bayesian Reliability Analysis With Evolving, Insufficient, and Subjective Data Sets

A primary concern in product design is ensuring high system reliability amidst various uncertainties throughout a product life-cycle. To achieve high reliability, uncertainty data for complex product systems must be adequately collected, analyzed, and managed throughout the product life-cycle. However, despite years of research, system reliability assessment is still difficult, mainly due to the challenges of evolving, insufficient, and subjective data sets. Therefore, the objective of this research is to establish a new paradigm of reliability prediction that enables the use of evolving, insufficient, and subjective data sets (from expert knowledge, customer survey, system inspection & testing, and field data) over the entire product life-cycle. This research will integrate probability encoding methods to a Bayesian updating mechanism. It is referred to as Bayesian Information Toolkit (BIT). Likewise, Bayesian Reliability Toolkit (BRT) will be created by incorporating reliability analysis to the Bayesian updating mechanism. In this research, both BIT and BRT will be integrated to predict reliability even with evolving, insufficient, and subjective data sets. It is shown that the proposed Bayesian reliability analysis can predict the reliability of door closing performance in a vehicle body-door subsystem where the relevant data sets availability are limited, subjective, and evolving.

Usage of meshfree methods in reliability analysis

This paper describes the use of meshfree methods for response and design sensitivity calculations within structural reliability analysis when geometric shape is a random variable. Brief descriptions of meshfree methods and advanced probabilistic methods are provided. An existing interface between the probabilistic analysis and traditional finite element method is modified to allow the use of meshfree methods for response and design sensitivity calculations within the probabilistic analysis routine. Three examples that treat design shape and thickness as random variables are presented to assess the accuracy and use of meshfree methods for reliability analysis.

Bayesian Approach to Uncertainty Modeling and Reliability Analysis with Evolving Subjective Input Uncertainties

A primary concern in product design is ensuring high system reliability amidst various uncertainties throughout a product life-cycle. To achieve high reliability, uncertainty data for complex product systems must be adequately collected, analyzed, and managed throughout the product life-cycle. However, despite years of research, system reliability assessment is still difficult, mainly due to the challenges of evolving, insufficient, and subjective data sets. Therefore, the objective of this research is to establish a new paradigm of reliability prediction that enables the use of evolving, insufficient, and subjective data sets (from expert knowledge, customer survey, system inspection & testing, and field data) over the entire product life-cycle. This research will integrate probability encoding methods to a Bayesian updating mechanism. It is referred to as Bayesian Information Toolkit (BIT). Likewise, Bayesian Reliability Toolkit (BRT) will be created by incorporating reliability analysis to the Bayesian updating mechanism. In this research, both BIT and BRT will be integrated to predict reliability even with evolving, insufficient, and subjective data sets. It is shown that the proposed Bayesian reliability analysis can predict the reliability of door closing performance in a vehicle body-door subsystem where the relevant data sets availability are limited, subjective, and evolving.

A Design Optimization Method Using Possibility Theory

Early in the engineering design cycle, it is difficult to quantify product reliability or compliance to performance targets due to insufficient data or information for modeling the uncertainties. Design decisions are therefore, based on fuzzy information that is vague, imprecise qualitative, linguistic or incomplete. The uncertain information is usually available as intervals with lower and upper limits. In this work, the possibility theory is used to assess design reliability with incomplete information. The possibility theory can be viewed as a variant of fuzzy set theory. A possibility-based design optimization method is proposed where all design constraints are expressed possibilistically. It is shown that the method gives a conservative solution compared with all conventional reliability-based designs obtained with different probability distributions. A general possibility-based design optimization method is also presented which handles a combination of random and possibilistic design variables. Numerical examples demonstrate the application of possibility theory in design.

A Robust Procedure for Convergent Nonparametric Multivariate Metamodel Design

Fast-running metamodels (surrogates or response surfaces) that approximate multivariate input/output relationships of time-consuming CAE simulations facilitate effective design trade-offs and optimizations in the vehicle development process. While the cross-validated nonparametric metamodeling methods are capable of capturing the highly nonlinear input/output relationships, it is crucial to ensure the adequacy of the metamodel error estimates. Moreover, in order to circumvent the so-called curse-of-dimensionality in constructing any nonlinear multivariate metamodels from a realistic number of expensive simulations, it is necessary to reliably eliminate insignificant inputs and consequently reduce the metamodel prediction error by focusing on major contributors. This paper presents a robust data-adaptive nonparametric metamodeling procedure that combines a convergent variable screening process with a robust 2-level error assessment strategy to achieve better metamodel accuracy. A door seal gap example is presented to illustrate the effectiveness and efficiency of the procedure.