I took several courses in computational fluid dynamics while in graduate school from a professor who had begun his academic career in the late 50s. He had worke

I took several courses in computational fluid dynamics while in graduate school from a professor who had begun his academic career in the late 50s. He had worked for the Army on ICBM re-entry vehicle dynamics, worked for JPL planning space missions, etc. In class he sometimes spoke about this transition. As I remember it, we were discussing the issues surrounding the design of the injector plate for the Saturn V F-1 engines. These famously took more than 1000 iterations before the designers arrived at a combination of baffles that damped standing waves within the combustion chamber. The professor, who had helped with this work, said something like (forgive me, this is as close as I can remember what he said), "You have to understand, we didn't try this 1000 times because we wanted to try it 1000 times. We had the mathematics to try to explore the workings of this combustion chamber. But we didn't have the computing power to experiment with our mathematical models! Now I can take advantage of many powerful, commercial solvers and have them run many times, tweaking initial conditions, and evaluate the results with statistics. Then, all we could do was sit around and do some very crude math and come up with another baffle configuration and try again until it worked. Because it HAD to work."