#coatML project, part of the #Mind4Machines consortium, uses high fidelity computational fluid dynamics (#CFD) simulations for the data-driven models of #MultiFluidX.
We have developed a series of simulations for our data-driven model and then tested the our machine learning (ML) approach for obtaining a CFD analysis.
The direct benefit of the developed approach is that, the ML-CFD results here, are obtained within a couple of seconds, contrary to the pure CFD approach that generally required 32 CPUh (4 CPUs for 8 hours) for each one of the selected cases. This means that the ML-CFD model was thousand times faster and completely bypassed the CFD intermediate steps such as meshing. This speed-up in the calculations, makes it possible to use the ML model for predictions in a variety of hardware including devices not necessarily having graphics cards. It also makes its integration in 5G technologies and Internet of Things (IoT) applications or emerging technologies such as VR/XR really appealing, since it enables near real-time analysis using edge devices and cloud computing.
The method was tested and validated for various industrial scenarios although it can naturally extended to predict the flow for any liquid or two-phase flow. It considers all the necessary models for the underpinning physics of the fluid flow. The impact of the storage conditions and nozzle geometry on flow was also illustrated. The presented findings suggest that the presented method can accurately calculate the pressure, and velocity of the produced two-phase jets. The novel machine-learning-CFD approach achieves thousand times faster results when compared to the classical CFD.
You can learn more about #coatML and #M4M and how their #Industry4.0 solutions are transforming the manufacturing industry at mind4machines.eu
#M4M #Mind4Machines #EITManufacturing #simulations #EuropeanManufacturing #manufacturing #ictsolutions #coatingthickness #measurement
