PART 1.10: Execute the PANDA2 processor called "DIPLOT" in order to generate Postscript files for plotting.
bush-> diplot
Enter the PANDA2 case name: cylstif
Print the plot file on the printer called: <lp> (y or n)? n
The PostScript files, cylstif.3.ps through cylstif.10.ps, contain the graphics for your plot. They can be printed on any PostScript
printer or viewed on the console with a PostScript previewing software program.
The files now existing in the working directory are as follows:
-rw-r--r-- 1 bush bush 76 Feb 21 18:29 cylstif.010 -rw-r--r-- 1 bush bush 30290 Feb 21 18:30 cylstif.5.ps -rw-r--r-- 1 bush bush 2046 Feb 21 17:56 cylstif.AL2 -rw-r--r-- 1 bush bush 2058 Feb 21 17:46 cylstif.AL3 -rw-r--r-- 1 bush bush 2058 Feb 21 17:56 cylstif.ALL -rw-r--r-- 1 bush bush 5604 Feb 21 11:18 cylstif.BEG -rw-r--r-- 1 bush bush 110324 Feb 21 17:57 cylstif.BL1 -rw-r--r-- 1 bush bush 110324 Feb 21 17:57 cylstif.BL2 -rw-r--r-- 1 bush bush 110324 Feb 21 17:56 cylstif.BL3 -rw-r--r-- 1 bush bush 110324 Feb 21 17:46 cylstif.BL4 -rw-r--r-- 1 bush bush 481 Feb 21 17:56 cylstif.BOS -rw-r--r-- 1 bush bush 182500 Feb 21 18:29 cylstif.CBL -rw-r--r-- 1 bush bush 360 Feb 21 18:29 cylstif.CPL -rw-r--r-- 1 bush bush 3126 Feb 21 17:45 cylstif.DEC -rw-r--r-- 1 bush bush 30 Feb 21 17:45 cylstif.NAM -rw-r--r-- 1 bush bush 2678 Feb 21 17:56 cylstif.OPA -rw-r--r-- 1 bush bush 12112 Feb 21 17:45 cylstif.OPB -rw-r--r-- 1 bush bush 7934 Feb 21 17:46 cylstif.OPD -rw-r--r-- 1 bush bush 12599 Feb 21 18:29 cylstif.OPL -rw-r--r-- 1 bush bush 2989030 Feb 21 17:57 cylstif.OPM -rw-r--r-- 1 bush bush 497082 Feb 21 17:57 cylstif.OPP -rw-r--r-- 1 bush bush 4547 Feb 21 12:44 cylstif.OPT -rw-r--r-- 1 bush bush 56760 Feb 21 17:57 cylstif.P11 -rw-r--r-- 1 bush bush 51084 Feb 21 17:57 cylstif.P21 -rw-r--r-- 1 bush bush 25752 Feb 21 17:57 cylstif.PL1 -rw-r--r-- 1 bush bush 0 Feb 21 18:26 cylstif.PL10 -rw-r--r-- 1 bush bush 34056 Feb 21 17:57 cylstif.PL2 -rw-r--r-- 1 bush bush 0 Feb 21 18:26 cylstif.PL3 -rw-r--r-- 1 bush bush 0 Feb 21 18:26 cylstif.PL4 -rw-r--r-- 1 bush bush 12074 Feb 21 18:29 cylstif.PL5 -rw-r--r-- 1 bush bush 0 Feb 21 18:26 cylstif.PL6 -rw-r--r-- 1 bush bush 0 Feb 21 18:26 cylstif.PL7 -rw-r--r-- 1 bush bush 0 Feb 21 18:26 cylstif.PL8 -rw-r--r-- 1 bush bush 0 Feb 21 18:26 cylstif.PL9 -rw-r--r-- 1 bush bush 34056 Feb 21 17:57 cylstif.PLD -rw-r--r-- 1 bush bush 854528 Feb 21 17:57 cylstif.RN1 -rw-r--r-- 1 bush bush 626688 Feb 21 17:56 cylstif.RN2 -rw-r--r-- 1 bush bush 574464 Feb 21 17:56 cylstif.RN3 -rw-r--r-- 1 bush bush 33792 Feb 21 17:46 cylstif.RN4 -rw-r--r-- 1 bush bush 79633 Feb 21 17:57 cylstif.TIT In this particular case there is only a single Postscript file generated by DIPLOT:
-rw-r--r-- 1 bush bush 30290 Feb 21 18:30 cylstif.5.ps This plot contains the objective versus design iterations during the SUPEROPT execution. In order to see the plot on your screen, type the following command:
gv cylstif.5.ps
("gv" means "ghost view", a utility for reading Postscript files and producing the plot image on your screen.)
A screen "snapshot" of the plot is taken and stored in the file:
1.cylstif.superopt1.objective.png (All the *.png files are
appended at the bottom of this file.) ---
PART 1.11: Execute the PANDA2 processors called "SUPEROPT"
and "DIPLOT"again in order to 1. search for a "global" optimum design that may weigh less than the optimum design determined so far, and 2. to plot the objective versus design iterations.
Next, run SUPEROPT again. Maybe there is a FEASIBLE or ALMOST FEASIBLE design with a smaller weight than that
found in PART 1.8 and listed at the end of PART 1.8: 1.221E+04 lb.
bush-> superopt
The purpose of SUPEROPT is to launch the batch run which performs multiple executions of the panda2 processors in the order:
autochange setup pandaopt pandaopt pandaopt . . . .
The processor autochange automatically changes the decision variables as follows:
y(i) = x(i)*(1. + dx(i)) (i = 1,2,3,...no. of dec. var) in which x(i) is the old value of the ith decision variable, y(i) is the new value, and dx(i) is a random number between -0.5 and +1.5 The purpose of the successive cycles of
autochange setup pandaopt pandaopt pandaopt . . . .
is to try to find a global optimum design by redesigning in each cycle from a different starting point. The user should use a small maximum number of design iterations (such as 5) in the file case.OPT, where case is the user-specified name of the case.
Enter case name: cylstif
Enter number of executions of pandaopt
for each execution of autochange (5 or 6 or 7 or 8 or 9 or 10):5 B (background), F (foreground), or Q (NQS - network queue system): b H (high) or L (low) priority: l
Diagnostics will be mailed to you upon program termination.
bush-> /usr/sbin/sendmail: No such file or directory
The abridged cylstif.OPP file includes the following
optimized design after completion of the second execution of SUPEROPT:
--- from the cylstif.OPP file --- VALUES OF DESIGN VARIABLES CORRESPONDING TO BEST FEASIBLE DESIGN
VAR. STR/ SEG. LAYER CURRENT
NO. RNG NO. NO. VALUE DEFINITION
1 0 0 1.011E+01 B(STR):stiffener spacing, b: STR seg=NA, layer=NA 2 STR 2 0 3.370E+00 B2(STR):width of stringer base, b2 (must be > 0, see Help): STR seg=2 , lay
3 STR 3 0 7.817E+00 H(STR):height of stiffener (type H for sketch), h:
STR seg=3 , layer=NA
4 STR 4 0 2.534E+00 W(STR):width of outstanding flange of stiffener, w:
STR seg=4 , layer=NA
5 SKN 1 1 7.690E-01 T(1 )(SKN):thickness for layer index no.(1 ): SKN seg=1 , layer=1
6 STR 3 1 2.038E-01 T(2 )(STR):thickness for layer index no.(2 ): STR seg=3 , layer=1
7 STR 4 1 8.450E-02 T(3 )(STR):thickness for layer index no.(3 ): STR seg=4 , layer=1
8 0 0 3.185E+01 B(RNG):stiffener spacing, b: RNG seg=NA, layer=NA 9 RNG 2 0 0.000E+00 B2(RNG):width of ring base, b2 (zero is allowed):
RNG seg=2 , layer=NA
10 RNG 3 0 9.783E+00 H(RNG):height of stiffener (type H for sketch), h:
RNG seg=3 , layer=NA
11 RNG 4 0 4.304E+00 W(RNG):width of outstanding flange of stiffener, w:
RNG seg=4 , layer=NA
12 RNG 3 1 5.673E-01 T(4 )(RNG):thickness for layer index no.(4 ): RNG seg=3 , layer=1
13 RNG 4 1 9.452E-01 T(5 )(RNG):thickness for layer index no.(5 ): RNG seg=4 , layer=1
***************************************************
******************** DESIGN OBJECTIVE *******************
************* ***********
CORRESPONDING VALUE OF THE OBJECTIVE FUNCTION:
VAR. STR/ SEG. LAYER CURRENT
NO. RNG NO. NO. VALUE DEFINITION
0 0 1.178E+04 WEIGHT OF THE ENTIRE PANEL ************* ************
******************** DESIGN OBJECTIVE *******************
*****************************************************
--- end of abridged cylstif.OPP file ---
It turns out there IS a slightly smaller optimized weight, 1.178E+04 lb versus the 1.221E+04 lb determined in PART 1.8.
Next, we want to obtain a plot of the objective versus
the design iterations during the second execution of SUPEROPT:
bush-> chooseplot
Please enter PANDA2 case name: cylstif (We use the same input file for CHOOSEPLOT that we used after the first execution of SUPREOPT. The interactive CHOOSEPLOT execution rolls by on the screen very fast.)
bush-> diplot
Enter the PANDA2 case name: cylstif
Print the plot file on the printer called: <lp> (y or n)? n
The PostScript files, cylstif.3.ps through cylstif.10.ps, contain the graphics for your plot. They can be printed on any PostScript
printer or viewed on the console with a PostScript previewing
software program.
The Postscript file for plotting is:
-rw-r--r-- 1 bush bush 30646 Feb 22 07:50 cylstif.5.ps This plot contains the objective versus design iterations during the SUPEROPT execution. In order to see the plot on your screen, type the following command:
gv cylstif.5.ps
("gv" means "ghost view", a utility for reading Postscript files and producing the plot image on your screen.)
A screen "snapshot" of the plot is taken and stored in the
2.cylstif.superopt2.objective.png (All the *.png files are appended at the end of this file.)
---
PART 1.12: Execute the PANDA2 processor called "MAINSETUP" again in order to set up a run for a fixed design (ITYPE=2):
the optimum design with the weight, 1.178E+04 lb.
Next, we wish to obtain results for the optimized design.
We edit the cylstif.OPT file (input for MAINSETUP) by changing ITYPE from 1 (optimization) to 2 (analysis of a fixed design):
2 $ Choose type of analysis (ITYPE = 1 or 2 or 3 or 4 or 5) We also change the output index, NPRINT, from 0 to -1, as
follows:
-1 $ NPRINT= output index (-1=min. 0=good, 1=ok, 2=more, 3=too much) This is done only for the purpose of this presentation.
Ordinarily, you would use NPRINT = 0 or NPRINT = 2
Next, we execute MAINSETUP with the new values for ITYPE (ITYPE=2) and NPRINT (NPRINT = -1):
bush-> mainsetup
Please enter PANDA2 case name: cylstif
***************** MAINSETUP ****************
The purpose of this processor is to permit you to choose loads and initial imperfections, Nx, Ny, Nxy, Mx, My, Nxo, Nyo, p, T(iseg), Wimp(global), Wimp(local), (up to 5 sets of them);
safety factors for general instability, panel instability, local instability (panel skin), local instability (stiffener parts), and stress; and strategy parameters for subsequent batch execution of an optimization analysis (analysis type 1);
or an analysis of a fixed design at fixed load levels (analysis type 2); or an analysis of a fixed design for a single load set for monotonically increasing load levels (test simulation: analysis type 3).
Results of the interactive session in MAINSETUP are saved on a file called cylstif.OPT, which will appear at the beginning of the cylstif.OPM file when the mainprocessor batch run launched by your command PANDAOPT has been completed.
NOTE: JUST HIT "RETURN" FOR DEFAULT VALUE OF INPUT DATUM. IF PANDA2 REQUIRES AN INPUT, IT WILL SAY "PLEASE SAY SOMETHING"
***************************************************
Are you correcting, adding to, or using an existing file?=y (The interactive "MAINSETUP" session rolls by on the screen very fast. We next give the command, PANDAOPT, and thereby obtain results for the fixed, optimized design.)
---
PART 1.13: Execute the PANDA2 processor called "PANDAOPT" for the analysis of the fixed, optimized design