The FlowDict module
The FlowDict module predicts effective material properties (flow velocity, flow permeability, and flow resistivity) by simulating flow experiments and post-processing the simulation results. A flow experiment in FlowDict requires the input of a:
- 3D representation of a structure or material.
- Newtonian fluid (gas o liquid) with constant density (incompressible).
- Experimental process parameters, such as mass flow rate, pressure difference and flow direction.
FlowDict can perform three categories of calculations:
- Prediction of mean flow velocity for a given pressure drop.
- Prediction of the pressure drop for a given mean flow velocity.
- Prediction of the full or partial permeability tensor.
In post-processing, FlowDict uses Darcy's law to compute the material permeability using mean flow velocity, fluid viscosity, pressure drop, and media thickness. Darcy's law only applies to very slow flows (Stokes flows) with a Reynolds number of zero. Faster flows are described by the Navier-Stokes equation. For faster flows, the relationship between pressure drop and mean velocity is not linear. For slow and faster flows, FlowDict assumes a steady flow regime, without time-dependent behavior such as turbulence. Thus, velocity and pressure drop cannot be arbitrarily high.
The FlowDict module bundles three solver technologies (Explicit Jump (EJ), SimpleFFT, and LIR), each with strengths and weaknesses:
- The EJ solver uses a uniform grid to discretize velocity and pressure. The solver is very fast for highly porous materials (e.g. filter media).
- The SimpleFFT solver also uses a uniform grid and is very fast for low porous materials (e.g. digital rocks).
- The LIR solver uses an adaptive grid structure, is extremely fast for highly porous materials, and requires very low memory.
The FlowDict module computes stationary fluid flows described by the equations:
- Stokes (with EJ, SimpleFFT, or LIR solver)
- Stokes-Brinkman (with SimpleFFT or LIR solver)
- Navier-Stokes (with SimpleFFT or LIR solver)
- Navier-Stokes-Brinkman (with SimpleFFT or LIR solver)
For very fast flows, where no stationary solution exists, the pressure drop or mean velocity can be approximated with Forchheimer Approximation. FlowDict also provides an export to perform flow simulations with third-party software:
- Navier-Stokes (Fluent)
Among many other applications, FlowDict can be used:
- to determine air and water permeability in woven fabrics.
- to study gas and liquid permeability and pressure drop in filter media.
- to predict gas permeability for the extraction of gas in reservoirs.
- to characterize flow properties of groundwater in aquifers.
- to predict the absolute permeability on digital rock, an important property for enhanced oil recovery.
Velocity of water flowing through a 3D-model of fibrous structure - computed with the LIR solver in FlowDict
Black boxes visualizing the computational adaptive grid structure of the LIR solver in FlowDict
Velocity of water flowing through a 3D-model of a Berea sandstone - data acquired by ÁCT and imported with ImportGeo-Vol
Pressure distribution of water flowing through a 3D-model of a fibrous structure - computed with FlowDict
The GeoDict Base package is needed for basic functionality.
FlowDict works on 3D (micro-) structure models that can either be a segmented 3D image (ÁCT-scan, FIB-SEM) imported with the ImportGeo-VOL module, or a 3D material model created with one of the GeoDict Modules for Digital Material Design, e.g. the FiberGeo module for nonwovens, GrainGeo for granular and sintered structures and sphere-packings, or FoamGeo for foams.
- FlowDict is required for other modules to function:
FlowDict is optionally needed for:
- SatuDict needs the Stokes solver in FlowDict for computing relative permeability.