At many conferences and simulation events I attend people ask: “What is next for CFD?” or “What is the future of CFD simulation”. Often people discuss topics like pre-processing, meshing, solver speed, etc. Some venture into the applification of simulation tools or even broader cloud computing capabilities. While all those points are valid, I would argue that those are well-known topics. Those topics are related to a very short term future.
Here is how I like to answer when speaking about the future of CFD. While this vision won’t be realized in the next year, I am convinced that this is where CFD, and simulation in general, are heading. CFD and simulation have a very bright future, here is my vision.
Spend all your time analyzing results
This one is an obvious one but often dismissed as a dream because it may be too futuristic or too far in the future. I would disagree. If you want to get to a certain target, you should — everyday— think about how to get there.
The value of CFD is in analyzing results. Pre-processing, meshing, and setup are non-value-added tasks. I would go as far as saying they are an engineering cost. The CFD software should be so easy to use that one only has to enter boundary conditions and specific models (combustion: yes/no, particles: yes/no, etc). The mesh is created automatically and numerical setup are done automatically thanks to built in best-practices. And those are adjusted as the simulation goes. A certain simulation requires a y+ of 1: the mesh is automatically adjusted to reach this condition. The simulation does not converge? The numerical setup is automatically adjusted for more robust convergence. That is what I consider to be the “ease-of-use” holy grail. Too often, I see people equating “ease-of-use” with a cool GUI. I would argue that the GUI is only 20% of the “ease-of-use”. The rest is about having a solver smart enough to know the best setting, adjust them automatically to reach convergence, etc. This way engineers will be able to spend all of their time interacting with the software for results analysis, not having to monitor simulations, create mesh, etc.
Speed is all about optimization
We all want a faster solver. But the key is that speed is not just about getting results faster, it’s about getting results so fast that doing 100 simulations to get the optimal solution is the norm. In the future, “doing a CFD study” will mean running hundreds of simulations and have an optimized design, tested for a wide range of operating conditions, a design which is optimized for manufacturing uncertainties, etc.
Virtual reality meets CFD
CFD is about complex 3-D phenomena. But mostly we still use the same graphical analysis tools from 15 years ago: planes, isosurfaces and particles or flow path. In the future, engineers will be able to interact with the model in virtual reality — a fully immersive 3-D experience. And as they do, they will be able to modify the design on the fly. Want to see what happens if you change a geometry parameter? remove a geometry feature? No need to go back to your desk. Simply push this wall, remove this feature in the virtual reality environment (thanks to smart gloves) and the simulation starts right away.
CFD… what is this thing?
Last but not least, in the future, no engineers will really know what CFD is, or structural simulation, or low frequency simulation for that matter. They will only know it is one technology embedded in their complete product virtual prototyping platform. They will simulate the products in its entirety. CFD will be so mature that there is no need to even know what it does.
When thinking about the future of such technology, let us not limit ourselves to current thinking or simulation steps. We should think big. What is your future vision for the future of CFD simulation?
Post by Guilles Eggenspieler