CFD Validation
Key Benefits Include:
- identify regional changes in process conditions over time for comparison with CFD predictions
- provide process information to improve CFD models
- determine repeatability of process conditions to characterise inherent process variability
Introduction
Computational Fluids Dynamics (CFD) is an increasingly important technique for modelling and understanding the behaviour of fluids. For example observing the flow of air over an aircraft wing. It is very computationally demanding but with the growing power of computers it is increasingly used as a more cost effective route than practical experimental rigs. In the process industries its use is growing to model and predict batch mixing, polymerisation reactions, polymer extrusion and processing and crystallisation.
However the many variables changing during a process (concentration and turbulence) and the different length scales make CFD very challenging for many processes.
While the mathematical techniques and tools for CFD are advancing rapidly the Achilles’ heel is always the validation of the computer model against the real behaviour and determining the “constants” and initial conditions of the CFD model.
Good process measurements can help CFD developers optimise their models. Data can also be used to validate whether a model is reliable.
Process Tomography Solution
Process tomography provides concentration measurements through a process volume. This data can be used to test and validate CFD models.
In applying tomography to CFD challenges, a key aspect is the accuracy of the tomography data and the confidence levels in the CFD model.
Process tomography can be used to test and validate a wide range of CFD models, including:
-
gas-liquid mixing
-
solid-liquid mixing
-
multi-phase flow
-
packed bed reactors
Typically our p2000 instrument is used for cold models where water is used as a surrogate material. Depending on the nature of the process, a variety of different reconstruction methods can be used, including SCG (Sensitivity Conjugate Gradients) and parametric reconstruction.
