Forge Physics

Mirroring reality is infinitely better than simulating it.

Our inviscid approach to physics unlocks a completely different level of engineering.

[Forge Physics @Forge_physics]

@grok Haven't posted may results lately, been cookin this up!

CRM Wing-Body-Tail @ Mach 0.85

Grid: 512×128×256 | 3000 steps | dt = 5e-4
Runtime: 7h42m (RTX A4000, float64 16GB VRAM)
U∞ ≈ 300 m/s | ρ∞ = 1.225 kg/m³
q∞ ≈ 55 kPa
ν = 0 (solver)
Re ≈ 9.42e32
CFL ≈ 15 (stable)
Mach_max ≈ 1.08

Cd ≈ 9.8e-5
Drag ≈ 3.26 N
Transonic shock captured
Coherent wake structure maintained

Watching this was the first run in a while that I was absolutely fascinated by!

The live PNGs coming out every step had me locked in. Next thing I knew it was midnight! Then, just 'to be sure it was running alright' I had to watch a couple more...

Sent the request off for the Drag Prediction Workshop this morning.

We didn't know about it until a couple weeks ago unfortunately, but hopefully it works out!

@aiaa @NASAJPL
#DPW #CFD #Aerodynamics

[Forge Physics @Forge_physics]

@elonmusk

[Forge Physics @Forge_physics]

@Cadillac_F1 Best of luck!!

(We definitely didn't run a model of your competition through our solver. Nope. Just a Model we found laying around)

[Forge Physics @Forge_physics]

@ASMEJournals That would be a great sim to run! I’m looking for ideas, so I’m trying to keep my eye out!

Is there EMF data available? It’d be cool to see how that affects the air circ

[Forge Physics @Forge_physics]

@mathelirium Several! Most of my .h5’s are 65-80GB though. Tracking 77 scalars at float64 for each sim.

The Fluid Dynamics are at 66, but the multi-physics runs are 77

[Forge Physics @Forge_physics]

@Scivf4 When I have a chance I’m going to run this through my solver. I have a pretty good idea of how that works. Just a matter of running the sim!

[Forge Physics @Forge_physics]

@ASMEJournals That reminded me of a vortex sim I ran a few months back! Thanks for sharing!!

[Forge Physics @Forge_physics]

@MaxkwTet @mathelirium

[Forge Physics @Forge_physics]

@MaxkwTet @mathelirium Like a Calabi-Yau slice…?

[Forge Physics @Forge_physics]

@2Rad116564 @mathemetica I ran this on a 256x128x128 grid. Only 3 depths, but I’m running the 512x256x256 grid soon!

[Forge Physics @Forge_physics]

@paratroxinator @mathelirium Basic Cylinder in flow

[Forge Physics @Forge_physics]

@tk50562007 I’d be happy to help if you need

[Forge Physics @Forge_physics]

@grok
More details to follow, but here is the 512^3 Sierpinski pyramid depth = 5

More to follow, I'll start showing some of the vortex runs I've been doing. THOSE are fun!

The implications of a purely inviscid solver cannot be overstated, @elonmusk Especially when EMF and FEM are integrated within 1300 lines of Python.

[Forge Physics @Forge_physics]

@grok @grok It's running now, I'll probably stop it at 100. I've got a bit of a backlog.

The wave pool to simulate the Spike Wave is next.
Dual F1 cars soon
Vortex Ring
Leap Frogging Vortices
Tokamak Field
Double Sheer Layer
Vortex Sphere
Double Vortex Sphere
Fast/Curve/Knuckleball

Started 15 min ago, at step 24 now... so like an hour, give or take

[Forge Physics @Forge_physics]

@grok Picking up from our previous conversation. Starting the 512^3 Sierpinski pyramid (depth 5) now.

Step 500 (256³ Sierpinski pyramid, depth 4):

u: mean = 19.9008214487 m/s, range = -4.89972259736 to 25.5068900387 m/s
v: mean = -0.00369140211303 m/s, range = -5.58499450587 to 5.74956643601 m/s
w: mean = -0.000547937189999 m/s, range = -7.08240429283 to 8.50502336173 m/s
Speed magnitude: mean = 19.9088244679 m/s, max = 25.6342157133 m/s
Speed p95 = 20.9721093198 m/s, p99 = 21.2745848004 m/s
Pressure: mean = 8.67361737988e-19 Pa, range = -0.581225623402 to 0.575360347184 Pa
Vorticity magnitude: mean = 7.45315278625 /s, max = 786.23739411 /s
Inlet speed: 15 m/s
Viscosity: ν = 0 (inviscid Euler)
Grid: 256³, float64

@CFDFoundation @FluidsLeeds

[Forge Physics @Forge_physics]

@grok @CFD_UniBw_M @PhysRevFluids @grok
Step 200 visuals, scalars coming soon

[Forge Physics @Forge_physics]

Menger Sponge, depth = 4 running on a 512^3 grid.

@grok Can you give a quick description and analysis? What scalar data would you want/need?

[Forge Physics @Forge_physics]

@grok @CFD_UniBw_M @PhysRevFluids @grok Oh, it's only at step 200. I'll resume it later if I need to.

For the Spieramid at 512, I'll just keep the H5 data every 100~ steps.

Easy peasy!

[Forge Physics @Forge_physics]

@grok @CFD_UniBw_M @PhysRevFluids @grok Visually it looks like it stabilized around step 90 and hasn't wavered for the past 100 steps. I can just stop it now to get you the data if you want. Resuming from a checkpoint later is just as easy as starting at step 0.

[Forge Physics @Forge_physics]

@grok @CFD_UniBw_M @PhysRevFluids @grok Next I was going to re-run Spierinski pyramid on a 512^3 with a depth of 5