3. DAKOTA-OpenFOAM optimization loop | Gradient-based optimization

The blunt body shape optimization case - Part 1. In this demo case, we use geometry generation and mesh generation with Salome, a black box solver OpenFOAM 11, automatic post-processing with bash scripting and Python, plotting with Python, and the DAKOTA 6.19.0 optimization and orchestration tool. Gradient‑based optimization relies on a well‑formulated problem: initial point, step size, tolerances, goal, linear and nonlinear constraints, how to compute the gradients (forward, differences, central differences, complex step), Hessian information if needed, and the type of design variables (integers, floats, cardinal). The video walks through the preliminaries, the case base, a reminder of what we are doing, DAKOTA’s output format, documentation remarks, input file, simulator script, launching the optimization case in real time, the final solution, post‑processing, quantitative data manipulation with bash scripting, the params.in and results.out files, final remarks, cleaning up the case directory, and simulator script execution permission. 00:00 The blunt body shape optimization case - Preliminaries 00:45 The case base 03:20 A reminder of what are we doing 06:24 DAKOTA's output format - How to feed information to DAKOTA 08:40 Some remarks on DAKOTA-s documentation 09:40 DAKOTA's input file 18:20 On the simulator script 25:20 Let's launch the optimization case - Real-time simulation 30:00 The final solution 33:20 Let's do some post-processing 38:24 On the quantitative data manipulation and post-processing | Bash scripting 40:20 The params.in and resluts.out files | DAKOTA's input and output files 40:44 Final remarks - Main takeaways 42:42 Cleaning up the case directory 43:25 Simulator script execution permission