FROM ROCKET CONTROL TO VIRTUAL DESIGN
Olivier Pironneau · Advances in computer science and engineering · 2009
Applied mathematicians, such as John Von Neumann, have had a great influence on the design of computers and have adapted them to address their problems such as the solution of optimal control problems, game theory and the numerical study of fluid flows. Ordinary differential equations and optimization problems were among the first to be solved after the Fortran language came into being, but computational fluid dynamics took much longer to mature, and one cannot still say that the partial differential equations of fluid mechanics are solvable using today’s computers. For this reason such open problems still influence computer architectures. Nevertheless it is now possible to design a virtual aircraft prototype entirely and fairly accurately, and the impact of this technology is felt even more on software development for Computer Aided Design systems. This paper is a short history of the interaction between numerical simulations in engineering, and the development of hardware and software over the past thirty years based on the author’s personal experience. 1 Computational Fluid Dynamics From the very early age of computing up to the end of the cold war, Computational Fluid Dynamics (CFD) was given top priority because of its applications to the design of fighter aircraft, and to the simulation of nuclear explosions. The fundamental equations of fluid dynamics, the Navier Stokes equations, are non linear partial differential equations (PDE). They admit a number of simplifications corresponding to potential flow, boundary layer flow, Euler flow, etc. For airplanes each is useful in its own regime but ultimately the full compressible Navier-Stokes equations with turbulence modelling are necessary, together with sensitivity analysis, in order to carry out design improvements and optimizations for advanced aircraft. By 1970 computers could solve fairly accurately large classes of optimization problems and ordinary differential equations. I had myself solved two of ∗[email protected]