Great video! I am going to share your videos with my marine craft manufacturing clients. Our CFD solution is very effective for boats, personal marine crafts, etc. (i.e. the smaller craft). It is also very good with propulsion units (propellers, waterjets, etc.) and extremely accurate with cavitation.
The data we use for research is DNS. It is extremely detailed and extremely large. Our professor got it made at the University of Austin in Texas using their super computer. It has great quality and the time steps are small but good enough to emulate turbulent flow at the boundary layer.
A small simulation of aerodynamic profile with DNS takes more than 50 million cpu hours, the reason we need may be 100 years to complete one real simulation!!
I appreciate this discussion, but I hoping for a more in depth overview of cfd methods given the title. nowadays there’s also FEM, spectral methods, high order FD, and Lattice Boltzmann.
thank yuu Datawave Marine Solutions for your online commitment, but how can be reduced the running time. or any some simplifiaction methods of the software?
Most things in CFD are a trade-off between run time and accuracy. The best thing you can do to reduce run time is optimize the mesh. Remove any unnecessary details and really reduce your mesh count. Minimize you domain size. Do everything you can to minimize the cell count. Also look at your simulation and see if you need all the equations running. Is turbulence a critical component, or can you add that in afterwards with an empirical equation? One trick that we use for preliminary studies is to perform a mesh independence study to determine the relationship between mesh resolution and our variable of interest. The mesh independence study typically doesn't require a full simulation, or all the simulation cases. After that is completed, you can perform the production runs at a lower mesh resolution and use the results of the mesh independence study to extrapolate up to a higher mesh resolution. But I only recommend this for preliminary study. Not final design. Hope that helps.
Do you recommend a CFD program(s)? (Specifically some of the open source software options you mentioned in the earlier portion of the video.) Thank you for your help! Cheers, Peyton
The major open source CFD code is openFOAM. You can learn more at www.openfoam.com. But be warned, this is not an easy software to learn, and there are multiple variations on the code available. Plan to spend at least a month learning the software, and another month on validation studies to confirm it's accuracy.
@@DatawaveMarineSolutions you scared me ama new urban hydraulic engineer and i really want to learn alot about CFD univ courses in master's degree aren't that much i think but when u said 2 months i think u mean 2 months for someone who already understand about CFD right ? :p i need books to to learn about it too if possible :p finite elements 2d we used to do it with excel :D
maybe a stupid question, but actually, what are games doing? I know that it is not CFD, that's for sure, but I have been always surprised as at least in simple qualitative matter they approach reasonable comportment of liquids...
For "LES" you state it "don't ignore all the turbulence" but RANS has turbulence? Do I understand correctly that RANS only has an abstract average turbulence but doesn't simulate eddies and that is what LES adds? Thanks for the overview! I'd like to learn OpenFOAM to see if a solar powered hydrofoil cruising trimaran might be possible. Can you predict lift and drag for 3D hydrofoils with openFoam somewhat accurately? I used JavaFoil but it gives way too optimistic numbers.
Yes, you have the essential idea of LES vs RANS. RANS creates an average effect of turbulence. For most of the RANS turbulence models, this also eliminates any directional information. You have the same turbulence applied in all directions. That can work perfectly fine for many cases. But it tends to create averaged, smoothed out flow. LES captures those eddies and can propagate the turbulence in far more detail further down the flow stream. Very important when the turbulence influence the flow direction. Main applications would be cases of boundary layer interaction: flow over foils or detachment at the edges of round bodies. Think wings, wind turbine blades, jackets in the offshore industry, submerged pipes. All large industrial structures. As to your question about OpenFOAM accuracy for a 3D hydrofoil. I have heard good things about OpenFOAM. My default answer to these questions: Nothing beats a validation study to test the accuracy of the software.
I only wear them like that for the video. It's a common video trick to prevent the lighting from reflecting into the camera and creating glare. One of the many small details that goes into these videos, and an example of the quality that DMS strives for. (I will admit that the angle of the glasses is fairly extreme. That was the only angle they would stay on my head. This is actually a great metaphor for engineering. Sometimes we all need to make concessions to physical constraints of reality.)