CFD vs DNS — why hypersonics “should” need supercomputers

27:37>> I love it. to show that that this is equivalent. >> Yes. >> Right. Right. >> That like we can assume these things. >> We've just sort of been like ah >> yeah probably >> probably. >> Right. >> We got other things to do. >> And then and and then came some experimental data that suggested not. >> Okay. >> Okay. So there was a conflict. Experimental data came out of something called PIV which I'm going to go into. But there on the bottom you're seeing the red is the experimental data. >> Yeah. Yeah. >> And the green is the DNS, which is the direct numerical simulation, the really high thing. >> And there's a discrepancy. >> You see that discrepancy? >> And it's it is it is sign statistically significant. >> It is it is statistically significant.

28:18You're going from something like 0.4 to to like 1.1 on the y- ais, right? That's that's a 2x almost. >> So, >> okay. >> Right. >> Yeah. So, now we're in a conundrum. >> There's a problem. >> There's a problem. Right. If we want to trust this thing, if we want to trust Marovven's hypothesis to beyond Mach 5, yes, in this hypersonic regime, >> yes, >> we have to answer the question one, >> is Marovven's hypothesis correct or >> Mhm. >> is >> is is the experimental data correct? Right. Right. One of those two >> is the thing is the thing >> is the is true. Right. Right. So that was the question that these guys at

29:00Stevens Institute were trying to handle. >> Shout out Jersey. >> Shout out to Jersey because here's what they did. They said, "Okay, direct numerical simulation is like kind of the gold standard." >> Sure. Understandably. >> How did they get the experimental data? >> Mhm. >> They use something called PIV, which is particle image velocymmetry. Okay. It's a state-of-the-art optical experiment. And here's what's happening. You got some fluid. You put in some particles can be dust particles, oil particles, aluminum dust. And those dust are going to track where the fluid is moving. >> Moving, right? Got it. >> That's the idea.