Engineers Harness the Power of the Adjoint Solver

In a previous blog, I shared with you my excitement about the power of the adjoint solver technology for shape optimization from ANSYS. Since then I have been working tirelessly to make this remarkable technology even more capable. CFD engineers can now understand their designs better and can perform smart shape optimization, all for larger problems with richer physics thanks to the adjoint solver technology.

My numerous interactions with people from all around the world confirmed what I knew: the adjoint solver technology is powerful and has the capability to enable a sea-change in the fluid design process. The technology is already having a positive disruptive impact on design, especially among the early adopters. Products are being improved. Established concepts about some types of fluid systems and how they function have been overturned. New manufacturing procedures are being attempted in order to produce the shapes indicated by the adjoint.

But I also discovered something else: we still needed to make it more accessible and easier to use!

Aerodynamic simulation (CFD) of a Formula 1. The adjoint solver indicates how to modify this Formula 1 wing geometry for optimized downforce.

Aerodynamic simulation (CFD) of a Formula 1. The adjoint solver indicates how to modify this Formula 1 wing geometry for optimized downforce.

Making sure that every CFD engineer can harness the power of the adjoint solver technology is a priority for me and my team. Thanks to thoughtful feedback and suggestions from ANSYS clients, we worked on 3 key aspects of the technology:

  1. The adjoint solver is now available to perform smart shape optimization on many more applications: reduce drag or pressure drop, minimize (or maximize!) heat transfer, etc. Also, you can combine those different objectives (e.g. optimize a design for best lift over drag ratio).
  2. To easily set up the optimization process, we added a “design tool” so that users can easily set up the optimization goals, define the region to optimize, and add geometrical constraints. Multi-objective optimization can be performed and the effect of making prescribed deformations of a particular form can also be seen.
  3. We made the solver more robust. Fire and forget! Launch the adjoint solver and it will converge robustly on simulations of up-to 100 million cells

But enough talking, lets do some showing. Please watch the video below where the power of the adjoint solver is demonstrated. Looking for more information? Please visit the adjoint Tech Tip page and listen to the webinar available on the page!

 

 

2 thoughts on “Engineers Harness the Power of the Adjoint Solver

  1. Great!
    I agree that adjoint has yet a great undiscovered potential. At the University of Rome Tor Vergata in cooperation with the software partner RBF Morph we did interesting studies.
    The mesh morphing tool RBF Morph allows to quickly compute sensitivities with respect to shape parameters defined by the user. As many shape parameters as desired (even 100) without the need to recompute CFD and adjoint solution.

    More details on the presentation used in 2014 events:
    http://www.rbf-morph.com/images/download/2014-05-ugm2014_italy_rbfmorph-fluent-adjoint.pdf

Comments are closed.