Cluster-Based Structural Optimization and Applications to Crashworthiness
Abstract
In the event of a crash, a vehicle structure should be properly designed to manage the impact energy to protect all interior spaces. This requirement, known as crashworthiness, is a primary consideration in the design of vehicle frames and bodies. Currently, automakers are considering new and multiple materials, innovative designs, and advanced manufacturing methods to reduce vehicle glider weight and to meet market demands for fuel-efficient vehicles, and global environmental and fuel-economy regulations. The integration of design optimization methods for crashworthiness potentially reduce the design cycle time and increase the effectiveness of the structural component. However, most of the (multimaterial) design optimization methods are limited to solve linear problems on continuous design space, and the solutions are usually local optimum. Existing material selection methods, which search over a categorical/discrete design space, require ranking the candidate materials. The objective of this dissertation is to develop a systematic approach for multimaterial structural optimization. The approach developed in this work is a cluster-based structural optimization (CBSO) algorithm, which can efficiently solve global optimization problems involving nonlinear analysis such as crash simulations on a continuous, non-ordinal categorical, or mixed design spaces. The proposed CBSO algorithm consists of three steps: (i) conceptual design generation, (ii) design clustering, and (iii) metamodel-based global optimization. The CBSO conceptual design is generated by using structural optimization for linear problems or using heuristic methods, such as hybrid cellular automata, for nonlinear problems. Design clustering algorithm, such as a proposed threshold clustering algorithm, is applied to downsizing the dimension of the conceptual design space to enable metamodel-based global optimization. The global optimization algorithm can search over (i) a continuous design space for sizing optimization, (ii) a categorical design space for material selection, or (iii) a mixed design space for concurrent sizing optimization and material selection. With the proposed approach, materials are optimally selected and distributed based on multiple attributes and multiple objectives without the need for material ranking. The proposed CBSO algorithm is also applied to the design of cellular materials and structures as well as the design under uncertainty.
Degree
Ph.D.
Advisors
Panchal, Purdue University.
Subject Area
Design
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.