When evaluating an observable using the adjoint solver (goal-based adjoint calculation), you can use the Goal-Based Error Indicator, a predefined Error-based criteria within the Automatic Mesh Adaption dialog box, to identify error in an observable calculation due to mesh refinement. This indicator can be used to adapt the mesh, in order to reduce mesh refinement affects on observables that are dependent on flow features. Prior to setting up mesh adaption for an observable evaluation, you must first compute the conventional flow solution, then calculate the adjoint solution for the observable of interest.
After calculating the adjoint solution and before performing mesh adaption, you can postprocess the Goal-Based Error Indicator (Raw) field variable (under the Sensitivities... category). Mesh adaption will have a greater influence in regions with higher sensitivities, where refinement may be needed.
After selecting Goal-Based Error Indicator from the Predefined Criteria drop-down list in the Manual Mesh Adaption dialog box or Automatic Mesh Adaption dialog box; the Adaption Criteria Settings dialog box opens automatically, and is used to define necessary mesh adaption settings.
Controls
(when setting up the Goal-Based Error Indicator predefined criterion under Aerodynamics...)
- Set Up Goal...
Opens the Adjoint Observables dialog box where you can select, edit, or create the desired observable.
- Skip Initial Adjoint Advancement Before Adaption
When enabled, if the adjoint solution is converged, the mesh adaption will ignore the computationally expensive initial adjoint iterations which are specified by Initial Adjoint Iterations before Adaption.
- Initial Adjoint Iterations Before Adaption
This specifies the number of adjoint iterations in the initial adjoint calculation before the first mesh adaption.
- Adjoint Iterations Before Adaption
This specifies the number of adjoint iterations during the adjoint calculation before each mesh adaption.
- Number of Modes for Adjoint Advancement
This specifies the number of modes in the Krylov subspace used to approximate the solution. A larger value will help to converge the adjoint solution at the expense of memory cost and potentially more computational time.
- Smoothing Steps Before Adaption
A larger value will lead to smoother adaption.
- Reset Adaption History
Deletes any previous adaption history.
- Print Adaption History
Prints the goal-based adaption history in the console.