Adjoint Shape Optimization using OpenFOAM
8 likes22,274 views
The document describes an OpenFOAM shape optimization example where the goal is to modify a geometry to match a target parabolic velocity profile at the outlet by running the adjoint solver. Over 20 iterations, the geometry is modified based on sensitivity maps from the adjoint method, with inward displacements where sensitivity is positive and outward displacements where sensitivity is negative. This improves the match of the calculated outlet velocity distribution to the target profile by 64%.
Adjoint Shape Optimization using OpenFOAM
- 1. 1 Adjoint Shape Optimization Example #1 using OpenFOAM - Target Velocity Profile at Outlet - Fumiya Nozaki Last Updated: 11 August 2014 English Keywords： • Continuous Adjoint • Shape Optimization • OpenFOAM
- 2. 2 Optimization objective Inlet Outlet Fixed velocity: Parabolic profile Fixed pressure Target velocity profile @ outlet： Parabolic profile Change shape to obtain the target velocity profile @ outlet • Laminar flow • Air @ 20℃
- 3. 3 Objective function 𝐼 = 𝑐 2 𝒖 − 𝒖 𝑡𝑎𝑟𝑔𝑒𝑡 2 𝑑𝛤 𝑜𝑢𝑡𝑙𝑒𝑡 Difference between the calculated velocity and target velocity profile Calculated by simpleFoam Target velocity profile Repeat shape modifications to get smaller value of 𝐼 Adjoint method gives this information
- 4. 4 Sensitivity map Positive sensitivity area ⇒ Inward surface displacement improves the objective Sensitivity distribution Negative sensitivity area ⇒ Outward surface displacement improves the objective
- 5. 5 Shape comparison between baseline and modified geometry baseline modified Shape modifications 20 times
- 6. 6 Outlet velocity distribution baseline modified Improved by 64%
- 7. 7 Thank You!