Link to Index Page

Example 11, Slide 3: Identification of some of the decision variables for optimization

Fig. 2 from the 2014 GENOPT paper.
Sample optimized cross section of half the width, WIDTH/2, of a corrugated panel with 8 major cylindrical segments over WIDTH/2 and with an alternating “convex surface up”/ “convex surface down” configuration. In this example there are no sub-segments. The panel cross section was optimized by GENOPT/span9/BIGBOSOR4 by means of the GENOPT processor called “SUPERDUPEROPT”. As shown in the next figure, the major segments are numbered starting with the left-most edge. PHIBIG is the overall arching angle in degrees. THICK(i) (the wall thickness of the ith major segment), SUBWID(i), and PHISEG(i), i = 1, 2, 3,…NSEG, are decision variable candidates. (“Candidate” means “eligible to be a decision variable”: a variable with designated Role No. 1 in Table 1.) The number of major segments, NSEG = 8, in this example. YPLATE(j), j = 1, 2, 3, …NSEG+1 are also decision variable candidates, and the overall arching angle, PHIBIG, is a decision variable candidate as well. In this example all of the decision variable candidates are decision variables except for YPLATE(1) and THICK(k), k = 2, 3, …, 8. THICK(k), 2 = 2, 3, …, 8, are all linked to THICK(1), so that the entire optimized panel cross section is of uniform thickness.

Page 144 / 190