[p_l]

While not unknown in the west in the past, "better algorithms" are in a way a dying breed in aerospace - these days a student will easily have a CFD package brute force through the problem overnight and be done.

My father, upon taking over his promotor's office in Warsaw Institute of Technology, found among other things a book in Russian that described analytical, symbolic methods allowing "good enough" approximated results for many airfoil/wing design tasks that are now exclusively done through CFD.

As described, in 30 minutes with pen & paper you had maybe a more coarse, but valid and in desired precision result that takes overnight run in ANSYS CFX.

[orbital-decay]

CFD only looks like brute force when compared to purely analytical methods. In practice, you can't brute force a chaotic system as you quickly hit the wall. Anything complex pretty much requires you to get clever with simulation just like it did before computers with analytical approach, and/or rely on experimentation. It's especially true for rocket engine design; there's a reason why nobody made a practical RDE yet.

[p_l]

True, but for a large class of problems in aviation, the analytical methods can provide faster iteration before going into CFD for deep optimization, at least that's my understanding (I'm the computer science person in the family, not aerodynamic physicist :) )

[jacquesm]

Past a certain minimum grid size it certainly looks like brute force to me, am I missing something?

[p_l]

Sometimes you can't get that grid size ;)