Research Papers: Design Automation

Optimal Design of Nonlinear Multimaterial Structures for Crashworthiness Using Cluster Analysis

[+] Author and Article Information
Kai Liu

School of Mechanical Engineering,
Purdue University,
West Lafayette, IN 47907
e-mail: liu915@purdue.edu

Duane Detwiler

Honda R&D Americas, Inc.,
Raymond, OH 43067
e-mail: ddetwiler@oh.hra.com

Andres Tovar

Associate Professor
Department of Mechanical Engineering,
Indiana University-Purdue
University Indianapolis,
Indianapolis, IN 46202
e-mail: tovara@iupui.edu

1Corresponding author.

Contributed by the Design Automation Committee of ASME for publication in the JOURNAL OF MECHANICAL DESIGN. Manuscript received September 12, 2016; final manuscript received August 2, 2017; published online August 30, 2017. Assoc. Editor: Nam H. Kim.

J. Mech. Des 139(10), 101401 (Aug 30, 2017) (11 pages) Paper No: MD-16-1637; doi: 10.1115/1.4037620 History: Received September 12, 2016; Revised August 02, 2017

This study presents an efficient multimaterial design optimization algorithm that is suitable for nonlinear structures. The proposed algorithm consists of three steps: conceptual design generation, clustering, and metamodel-based global optimization. The conceptual design is generated using a structural optimization algorithm for linear models or a heuristic design algorithm for nonlinear models. Then, the conceptual design is clustered into a predefined number of clusters (materials) using a machine learning algorithm. Finally, the global optimization problem aims to find the optimal material parameters of the clustered design using metamodels. The metamodels are built using sampling and cross-validation and sequentially updated using an expected improvement function until convergence. The proposed methodology is demonstrated using examples from multiple physics and compared with traditional multimaterial topology optimization (MTOP) method. The proposed approach is applied to a nonlinear, multi-objective design problems for crashworthiness.

Copyright © 2017 by ASME
Your Session has timed out. Please sign back in to continue.



Grahic Jump Location
Fig. 1

LHSs with P = 5 and (a) high correlation and (b) low correlation achieved using the maximization of the minimum inter-site distances

Grahic Jump Location
Fig. 2

Leave-one-out cross-validation at the second sampled point exemplified by fitting a function with a Kriging metamodel

Grahic Jump Location
Fig. 3

Thermal compliance problem—design domain and Dirichlet boundary condition: constant surface temperature u = 0

Grahic Jump Location
Fig. 4

Thermal compliance problem—conceptual design with f = 1.94 × 106

Grahic Jump Location
Fig. 5

Thermal compliance problem—clustered design with f = 1.97 × 106

Grahic Jump Location
Fig. 6

Thermal compliance problem—comparison of the predicted number of function evaluations

Grahic Jump Location
Fig. 7

Mechanical compliance problem—design domain and boundary conditions for the three-dimensional beam in cantilever

Grahic Jump Location
Fig. 8

Mechanical compliance problem—conceptual design with 8320 distinct density values and f = 2217

Grahic Jump Location
Fig. 9

Mechanical compliance problem—clustered design with three clusters and f = 2432

Grahic Jump Location
Fig. 10

Armor plate problem—finite element model

Grahic Jump Location
Fig. 11

Armor plate problem—initial design (left) and impact simulation (right) with maximum penetration f1 = 12.05 mm and mass fraction f2 = 0.50

Grahic Jump Location
Fig. 12

Armor plate problem—conceptual design (left) and impact simulation (right) with maximum penetration f1 = 9.33 mm and mass fraction f2 = 0.50

Grahic Jump Location
Fig. 13

Armor plate problem—clustered designs with K=1,…,18

Grahic Jump Location
Fig. 14

Armor plate problem—maximum penetration and mass fraction as functions of the number of clusters

Grahic Jump Location
Fig. 15

Armor plate problem—Pareto front of the design optimization problem and the optimized clustered design (K = 4) with maximum penetration f1 = 9.46 mm and mass fraction f2 = 0.50

Grahic Jump Location
Fig. 16

S-rail problem—geometry of the thin-walled S-rail (side view). The cross section is squared of dimensions H × H and thickness xe.

Grahic Jump Location
Fig. 17

S-rail problem—locations of input and output ports for a thin-walled S-tube following the wavelength λ corresponding to the progressive buckling after an ideal axial crushing condition

Grahic Jump Location
Fig. 18

S-rail problem—initial design represented by a uniform thickness distribution in the “unfolded” thin-walled structure (left). The initial design depicts Euler-type buckling (right). The corresponding crashworthiness indicators are SEA = 3.39 kJ/kg and PCF = 267 kN.

Grahic Jump Location
Fig. 19

S-rail problem—conceptual design represented the thickness distribution in the “unfolded” thin-walled structure (left). The conceptual design depicts progressive folding (right).The corresponding crashworthiness indicators are SEA = 5.05 kJ/kg and PCF = 359 kN.

Grahic Jump Location
Fig. 20

S-rail problem—clustered designs with K=1,…,12

Grahic Jump Location
Fig. 21

S-rail problem—SEA and PCF values as a function of the number of clusters K

Grahic Jump Location
Fig. 22

S-rail problem—Pareto fronts for K=1,…,4. Initial, conceptual, and clustered designs are dominated. Clustered designs have the following (−SEA, PCF) coordinates: for K = 1: (−3.82, 274), for K = 2: (−4.81, 349), for K = 3: (−3.98, 423), and for K = 4: (−3.90, 381).




Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In