January 16, 2018 Optimization Methods & Software AD2016Paper To appear in Optimization Methods & Software Vol. 00, No. 00, Month 20XX, 1{18 Algorithmic Differentiation of the Open CASCADE Technology CAD Kernel and its coupling with an Adjoint CFD Solver Mladen Banovi´ca∗, Orest Mykhaskivb, Salvatore Auriemmac, Andrea Walthera, Herve Legrandc and Jens-Dominik M¨ullerb aUniversit¨atPaderborn, Warburger Str. 100, 33098 Paderborn, Germany bQueen Mary University of London, Mile End Road, London, E14NS, UK cOpenCascade, 1 Place des Fr`eres Montgolfier, 78280 Guyancourt, France (v1.0 released February 2017) Computer Aided Design (CAD) tools are extensively used to design industrial components, however contrary to e.g. Computational Fluid Dynamics (CFD) solvers, shape sensitivities for gradient-based optimisation of CAD-parametrised geometries have only been available with inaccurate and non-robust finite differences. Here Algorithmic Differentiation (AD) is applied to the open-source CAD kernel Open CASCADE Technology using the AD software tool ADOL-C (Automatic Differentiation by OverLoading in C++). The differentiated CAD kernel is coupled with a discrete adjoint CFD solver, thus providing the first example of a complete differentiated design chain built from generic, multi-purpose tools. The design chain is demonstrated on the gradient-based optimisation of a squared U-bend turbo-machinery cooling duct to minimise the total pressure loss. Keywords: Algorithmic Differentiation, CAD kernel, Adjoint CFD method, Gradient-based optimisation. ∗Corresponding author. Email:
[email protected] January 16, 2018 Optimization Methods & Software AD2016Paper 1. Introduction Computer Aided Design (CAD) systems carry out an essential role in engineering de- sign workflows. In a multi-disciplinary product development cycle, the so-called CAD/- CAE/CAM workflow, CAD systems are used extensively to construct and manipulate the geometric representation of the design.