Page 356 - Practical Design Ships and Floating Structures

F'ractical Design of Ships and Other Floating Structures 33 1
You-Sheng Wu, Wei-Cheng Cui and Guo-Jun Zhou (E&)
0 2001 Elsevier Science Ltd. All rights reserved

OPTIMIZATION OF THE DESIGN OF SHIP STRUCTURES
USING RESPONSE SURFACE METHODOLOGY

Makoto Arai' and Taito Shimizu'
'Department of Naval Architecture & Ocean Engineering, Yokohama National University,
79-5 Tokiwadai, Hodogaya-ku, Yokohama 240-850 1, Japan

ABSTRACT

In this paper, the applicability of response surface methodology (RSM) to the design of ship structures
is examined. This methodology involves the following mathematical techniques: a method of
designing experiments that allows an efficient and accurate grasp of response features using a minimal
number of analyses, the least squares method to obtain an approximate mathematical expression of the
response, and the nonlinear optimization method to determine the optimal design, i.e., the minimum or
maximum of the response. RSM allows easy estimation of changes on the response surface as changes
are imposed on the design parameters. This is particularly effective in ship structural designs wherein
trade-offs between design variables are inevitable. However, before RSM is applied to the actual
design, the order of the approximate polynomial equation that represents the response surface, and the
necessary number of response evaluations, etc., must be examined. Thus, the optimization of the
transverse bulkhead structure of a crude oil tanker is performed as an example of the application of
RSM, wherein the method's effectiveness together with the results of above-mentioned basic
examinations are shown.

KEYWORDS

Optimization, Structural Design, Response Surface Methodology, Bulkhead Structure

1 INTRODUCTION
When a ship designer creates ship structural design, he first considers several combinations of design
variables related to structural form, structural size, and so on. Next, he performs response analysis
using FEM or some other method, and evaluates the performance of each design. Finally, he selects the
best design from among the candidates.
However, this design-selection procedure carries with it the possibility that the selected combination of
design variables may not always be the optimal one, due to the limited number of candidate designs
analyzed.
Thus, response surface methodology (RSM) is applied in this study to optimize ship structural design.
This methodology involves a few sets of mathematical techniques: the method of designing
experiments to efficiently grasp the accurate features of a structural response using a minimal number