I...am astonished that people don't understand the difference between engineering, experimentation, and modelling sometimes. It is fun to discuss, so please forgive me the diversion.
Folding at home is taking known items, modeling them, and reverse engineering their structures with finite rules. It's modelling complex systems...not developing new stuff. As such it's only useful for modelling and thereby extrapolating data from. Once you get that protein structure mapped you still need it to do something.
Experimentation is defining those rules of modelling, by defining processes. This seems to be what the OP was actually asking about years ago. The problem is that the final structure of a thing can be greatly influenced by how it's processed. Others cited Roman concrete, which has had half a dozen "break throughs" in the last decade. Some of it was material input, other bits were processing, and some was simple testing. I find it funny that people think this is the realm of a supercomputer, where for the last several hundred years it's actually been patents and protected trade secrets which make things like brand name metal alloys and composites so valuable. Somebody out there tested thousands of alloys before settling on 1095 as "cabon steel" and even more before they started manipulating grain sizes and the like with annealing. All of which is so complex that modeling it stupid when a test is cheaper and faster.
Finally, engineering should be back of the bus. Engineers solve problems, using specified materials. They don't, largely, care if you find a 10% more carbon neutral solution, they care if the compressive strength remains the same. The problems usually stem from materials displacing design. Think some of the earliest porus parking lots getting destroyed by traffic in months rather than years.
So no, I don't think that supercomputers model concrete at any substantive level. It's not about money...it's about a black box with dozens of incompletely understood variables...which is a bad black box. It's also a black box whose output is devoid of meaning without finite goals...as anybody trying to sequester fly ash can make basically disintegrating foamy (but highly insulative) concrete or a surprisingly dense block with little more than process changes.
