Hemanth K. Amarchinta

A Bayesian approach for quantification of model uncertainty

In most engineering problems, more than one model can be created to represent an engineering systems behavior. Uncertainty is inevitably involved in selecting the best model from among the models that are possible. Uncertainty in model selection cannot be ignored, especially when the differences between the predictions of competing models are significant. In this research, a methodology is proposed to quantify model uncertainty using measured differences between experimental data and model outcomes under a Bayesian statistical framework. The adjustment factor approach is used to propagate model uncertainty into prediction of a system response. A nonlinear vibration system is used to demonstrate the processes for implementing the adjustment factor approach. Finally, the methodology is applied on the engineering benefits of a laser peening process, and a confidence band for residual stresses is established to indicate the reliability of model prediction.

Material model validation for laser shock peening process simulation

Advanced mechanical surface enhancement techniques have been used successfully to increase the fatigue life of metallic components. These techniques impart deep compressive residual stresses into the component to counter potentially damage-inducing tensile stresses generated under service loading. Laser shock peening (LSP) is an advanced mechanical surface enhancement technique used predominantly in the aircraft industry. To reduce costs and make the technique available on a large-scale basis for industrial applications, simulation of the LSP process is required. Accurate simulation of the LSP process is a challenging task, because the process has many parameters such as laser spot size, pressure profile and material model that must be precisely determined. This work focuses on investigating the appropriate material model that could be used in simulation and design. In the LSP process material is subjected to strain rates of 106 s−1, which is very high compared with conventional strain rates. The importance of an accurate material model increases because the material behaves significantly different at such high strain rates. This work investigates the effect of multiple nonlinear material models for representing the elastic–plastic behavior of materials. Elastic perfectly plastic, Johnson–Cook and Zerilli–Armstrong models are used, and the performance of each model is compared with available experimental results.

Multi-Attribute Structural Optimization Based on Conjoint Analysis

Over the last 30 years, there have been tremendous advances in multidisciplinary design optimization techniques to find optimum solutions. Advances have come in areas such as reducing computational cost, developing algorithms for efficient sensitivity analysis, using function approximations, etc. Most of these efforts assumed a single objective function (attribute) and a multitude of constraints. The focus of this paper involves the designer preferences in the multidisciplinary design optimization. The concept of modeling preferences among multi-attribute alternatives is prevalent in consumer product marketing. In this paper, we adopt conjoint analysis, a popular technique used in marketing to assess consumer preferences. This technique is used to obtain part-worths of the attributes, which provide insightful knowledge of the product under study, and which are further used to create new products in the market. This paper presents integration of conjoint analysis and multidisciplinary design optimization applications. A cantilever beam, a fixed plate, and a composite lightweight torpedo are used as examples.

Regression Uncertainty Quantification Using Bootstrap Method for Residual Stress Field Predictions

[Abstract] Compressive residual stresses play an important role in increasing the fatigue life of metallic components. The nature of the residual stress field imparted by surface enhancement techniques such as shot peening, low plasticity burnishing, and laser peening depends on the process and the operating parameters. Variations in estimation of the residual stress field can cause a considerable uncertainty in fatigue life calculations because it is known that fatigue life estimations are sensitive to the residual stress field. This work develops a framework to quantify the variability in the residual stress field caused by uncertainty of predicting material behavior in the laser peening process simulation. The input uncertainty is considered to be the material model constant estimates obtained from a non-linear regression analysis to fit the experimental data of stress-strain curves. A statistical technique known as bootstrapping for regression is used to evaluate the multivariate normality assumption of the input uncertainty. The input uncertainty is propagated through finite element analysis to obtain the confidence bounds on the residual stress field. The upper and lower bounds are interpolated to obtain a confidence band for the entire residual stress field. A demonstration problem is shown to validate the methodology.

Reliable Material Models for High Strain rate Process Simulation

[Abstract] Advanced mechanical surface enhancement techniques have been used successfully to increase the fatigue life of metallic components. These techniques impart deep compressive residual stresses into the component to counter potentially damage-inducing tensile stresses generated under service loading. Laser shock peening is an advanced mechanical surface enhancement technique used predominantly in the aircraft industry. To reduce costs and make the technique available on a large-scale basis for industrial applications, simulation of the laser shock peening process is required. Accurate simulation of the laser shock peening process is a challenging task, because the process has many parameters such as laser spot size, pressure profile, and material model that must be precisely determined. This work focuses on investigating the appropriate material model that could be used in simulation and design. In the laser shock peening process material is

Combining Marketing and Engineering Tools for Multi-Attribute Optimization

Multidisciplinary design optimization has been an active topic of research in the past two decades in developing algorithms for reducing computational cost of re-analysis and also in developing efficient ways of calculating sensitivities. Most of the efforts were aimed at single objective function (attribute). Also very little work is done to include designer’s preferences inside the optimization. In this paper, conjoint analysis, a popular marketing technique to assess consumer preferences is used to involve the preferences of the designer. The optimization is driven by the designer’s preferences and a preferred design is obtained. Here, a novel way of combining tools from marketing and engineering is shown. A cantilever beam, and a composite lightweight torpedo are used as examples to demonstrate the method.Copyright