|
[Sponsors] |
Objective function increases after decreasing |
|
LinkBack | Thread Tools | Search this Thread | Display Modes |
March 15, 2017, 05:47 |
Objective function increases after decreasing
|
#1 |
New Member
Floris
Join Date: Sep 2016
Location: Enschede, The Netherlands
Posts: 21
Rep Power: 9 |
Hello guys,
I run a simulation whereby the drag has to be optimized. Which is done for the first +- 60 designs ( 'NIT' = 8 ). However after this point the objective function (drag) increases again. I checked the residuals of the direct- and adjoint calculations, both were nicely converged. At this point, I don't know what is happening inside 'the black box' and any help is welcome. I would like to know why the objective function is increasing after a couple of 'NIT'. Also, some tips to avoid this behavior are welcome . Kind regards, Floris (config file: turb_SA_RAE2822.txt ) |
|
April 6, 2017, 04:43 |
|
#2 |
New Member
Floris
Join Date: Sep 2016
Location: Enschede, The Netherlands
Posts: 21
Rep Power: 9 |
Hello all,
This time I ran another case, which contains the same behavior as the test case before. The objective of this case is to minimize the drag, just for testing purposes. During the line search process the optimizer does not choose the design with the lowest drag, but it chooses the design with the highest efficiency. Can anybody tell me, why the optimizer is choosing for highest efficiency instead of choosing the design with lowest drag? Any help is appreciated Cheers! Floris config file: turb_SA_RAE2822.txt history_project.dat:history_project.txt log.txt: New Project: ./ Sequential Least SQuares Programming (SLSQP) parameters: Number of design variables: 22 ( 22 ) Objective function scaling factor: [0.1] Maximum number of iterations: 100 Requested accuracy: 1e-11 Initial guess for the independent variable(s): [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0] Lower and upper bound for each independent variable: [(-0.01, 0.01), (-0.01, 0.01), (-0.01, 0.01), (-0.01, 0.01), (-0.01, 0.01), (-0.01, 0.01), (-0.01, 0.01), (-0.01, 0.01), (-0.01, 0.01), (-0.01, 0.01), (-0.01, 0.01), (-0.01, 0.01), (-0.01, 0.01), (-0.01, 0.01), (-0.01, 0.01), (-0.01, 0.01), (-0.01, 0.01), (-0.01, 0.01), (-0.01, 0.01), (-0.01, 0.01), (-0.01, 0.01), (-0.01, 0.01)] NIT FC OBJFUN GNORM 1 1 1.412388E-03 1.995497E-01 2 12 1.426525E-03 2.079778E-01 3 23 1.426297E-03 2.077896E-01 4 34 1.425957E-03 2.075662E-01 5 40 1.406145E-03 2.047457E-01 6 51 1.405817E-03 2.046707E-01 7 62 1.405490E-03 2.045980E-01 8 73 1.405164E-03 2.045267E-01 9 84 1.404844E-03 2.044569E-01 11 95 1.404627E-03 2.042543E-01 Last edited by WillemFloris; April 24, 2017 at 03:46. |
|
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
[Other] Contribution a new utility: refine wall layer mesh based on yPlus field | lakeat | OpenFOAM Community Contributions | 58 | December 23, 2021 02:36 |
channelFoam for a 3D pipe | AlmostSurelyRob | OpenFOAM | 3 | June 24, 2011 13:06 |
latest OpenFOAM-1.6.x from git failed to compile | phsieh2005 | OpenFOAM Bugs | 25 | February 9, 2010 04:37 |
Compilation errors in ThirdPartymallochoard | feng_w | OpenFOAM Installation | 1 | January 25, 2009 06:59 |
Problem with compile the setParabolicInlet | ivanyao | OpenFOAM Running, Solving & CFD | 6 | September 5, 2008 20:50 |