Process Integration and Design Optimization

Process Integration and Design Optimization

Results of Gradient-based Optimization

    • ciciola.2027298
      Subscriber

      Hi everyone,

      I completed an 2D Optimization of a wing profile, where I have varied, for every Design Iteration, only the angle of attack. My Input Parameter is the Angle of Attack and the output parameters are: drag, lift and efficiency (lift/drag). My Optimization had the objective of maximize the efficiency. So I had computed 5 Operating Condition and about 70 Design Iteration, amounting to about 350 Adjoint iteration (Right???). But, when I tried to plot results, I found only the Adjoint History of observable values (But only because was plotted for every design iteration, I couldn't replotted). I see, in the working directory, cas.h5 and dat.h5 files, for every design iteration. I want to plot efficiency, drag and lift for every design iteration and every operation condition, on Fluent. Also, I want to plot a sort of Pareto Front of the Optimization. 

      How can I do?

    • Murari Iyengar
      Ansys Employee

      To plot efficiency, drag, and lift for every design iteration and every operating condition in Fluent, you can utilize the post-processing capabilities of Fluent. The .cas.h5 and .dat.h5 files you mentioned are the case and data files for each design iteration. You can load these files into Fluent to visualize the results for each iteration.

      For plotting a Pareto front, you would typically need to extract the objective function values (in your case, efficiency) and the constraint values (drag and lift) for each design iteration. This data can then be plotted on a graph where the x-axis could represent drag, the y-axis lift, and the data points colored or sized by efficiency. Fluent itself may not have direct capabilities to plot a Pareto front, but you can export the necessary data to a tool that does, such as MATLAB or Python.

      Regarding the adjoint iterations, if you have computed 5 operating conditions and about 70 design iterations, and assuming each design iteration involves an adjoint calculation, then your total number of adjoint iterations would indeed be around 350, as you mentioned.

      If you need more detailed instructions on how to perform these tasks in Fluent, you may refer to the Ansys Fluent documentation or tutorials that cover optimization and post-processing.

    • ciciola.2027298
      Subscriber

      Thanks for your replying, 

      I found a solution without using other tools, but using only fluent. I applied a Execute command, in which I printed all the output parameters (I set alpha, drag, lift and efficiency lika output parameters, alpha is also a input parameter) in a .txt file and, after, I cleaned this file and transformed it in a .xy file. The execute command is the following:
      /define/parameters/output-parameters/write-all-to-file

      "name_file"

      [execution every 1 design iteration]

      I can't print the parameters for every operating condition, but that's not a problem, because I simulate these conditions separately.

Viewing 2 reply threads
  • You must be logged in to reply to this topic.