Turning a CFD model that runs for hours into one that answers in milliseconds — fast enough to run alongside the real equipment as the basis for a digital twin.
CFD is run across a range of operating conditions; the ROM learns the relationship between inputs and results and interpolates for any condition inside that range.
Because it answers almost instantly, a ROM can be coupled to sensor data to estimate quantities that cannot be measured directly inside the equipment.
Outside the trained range, the results are no longer reliable; designing that range is therefore the most important part of a ROM project.
The source of training data — a set of CFD results across the condition range.
View details →ROM is usually supplied as part of a project and requires an existing calibrated CFD model as its data source. CFDWAYS takes the whole chain: building the base model, training the ROM, and handing over the finished tool.
Without a calibrated CFD model there is no ROM yet. CFDWAYS helps determine which step you are at.
Feasibility assessmentIt does use data-driven learning, but the training data are physical simulation results. That is why a ROM keeps the physical constraints of the original problem.
It depends on the number of input parameters. A few parameters need a few dozen runs; many parameters require a careful design of experiments so the number of runs stays manageable.
As long as the equipment geometry is unchanged and operation stays inside the trained range. A retrofit means retraining.
If you already have a CFD model of the equipment, a ROM is the next step — CFDWAYS assesses feasibility first.
