Smart Shape Optimization with the ANSYS Adjoint Solver

With ANSYS 16.0, the adjoint solver is even more powerful and usable.

  • The adjoint solver can now be used for simulations using up to 50 million or more mesh elements.
  • The robustness of the adjoint solution has been significantly improved thanks to automatic solution advancement, which essentially eliminates the need for users to tweak parameters during the course of the solution.
  • The new Design Tool allows users to better manage design changes for improved and more efficient multi-objective design, better control of mesh morphing, and adherence to design constraints — all of which are possible in combination, too.
  • The adjoint solver can now also be applied to applications including porous media and user-defined sources, and for volume observables (e.g., volumetric integrals, averages and variances).

Imagine that you have to design a car and minimize its drag, or engineer a piping system and minimize pressure drop. In both cases, the actual shape of the design is the most important factor. When setting parameters for simulation, usually you define the shape and run parametric variations, often with the help of optimization tools. While this is a good approach, it has many limitations:

  • Design shapes can be extremely complex, governed by hundreds of parameters (or more). It is impossible to consider all of them. How do you make sure that you select the relevant parameters?
  • Even if you select one set of key design shape parameters, you still have a very large number of designs to evaluate. Simulating all these can be extremely time consuming.

Watch the video to see how smart shape optimization is possible with ANSYS adjoint solver

For these reasons, you need a smart shape optimization tool — one that:

  • Automatically identifies the section of the design (shape) that needs to be modified
  • Automatically guides shape optimization by determining how to modify the shape directly from simulation results, without the need for trial-and-error simulation run after run
  • Quickly performs design shape optimization with the minimum amount of simulations, and performing those simulations as fast as possible
In this race car design, a smart shape optimization tool identified the section of the design that needed to be modified, along with best method to increase the downforce. Arrows indicate increased chamber in the rear wing, which can generate greater downforce.

What is the smart shape optimization tool?

The ANSYS smart shape optimization tool is called adjoint technology. The tool is actually a solver that uses CFD simulation results to find an optimal solution based on stated goals (reduced drag, maximized lift-over-draft ratio, reduced pressure drop, etc.). But it doesn’t stop there: It also computes how to specifically modify the design. Because it is a solver, it has many advantages:

  • It directly computes which section of the design needs to be modified and how. You do not need to define any parameters.
  • It directly determines a better-performing shape as well as the associated performance improvement, all without needing another CFD simulation.
  • It can, in a minimal number of simulations, determine the optimal shape. At each iteration, design performance increases until the optimal design is reached.

Why is this smart shape optimization tool so fast?

Because the adjoint solver directly determines which section of the shape to modify and how to do it, it reaches the optimal geometry faster. Because the adjoint solver works hand-in-hand with mesh morphing technologies, you do not need to redefine the geometry nor recreate the computational mesh; rather you simply morph the mesh to the new shape. In summary, this solution is fast because:

  • The adjoint solver determines directly how to improve performance, so there is no time wasted on trial-and-error processes.
  • A mesh morpher automatically adjusts the design shape and computational mesh following the adjoint solver recommendations, saving even more time.