Pointwise Reliable People. Reliable Tools. Reliable CFD Meshing. The Connector newsletter Pointwise Facebook Pointwise GitHub Pointwise Instagram Pointwise LinkedIn Another Fine Mesh blog Pointwise Pinterest Pointwise Twitter Pointwise YouTube Y+ Calculator
Get a Free Trial

The Connector, the newsletter for CFD Mesh Generation from Pointwise

March / April 2011

Automating Porous Flow Meshing with Glyph

by Justin Finn, Ph.D. student, Oregon State University

At Oregon State University, researchers in the Computational Flow Physics Lab (CFPL) are using large scale, high resolution simulations to better understand flow through complex porous media. Both arranged and random sphere packs, such as the one shown in Figure 1, are being studied.

Random packing of spheres in a tube.

Figure 1: A typical simulation domain consisting of a cylindrical tube packed randomly with contacting spheres. The surface mesh on the spheres is shown.

Porous materials are building blocks for many of the natural and man-made systems that influence our daily lives, such as subsurface groundwater networks and packed bed reactors. Despite their importance, we know very little about flow dynamics in porous media, particularly at Reynolds numbers much greater than one. At these higher flow rates, the effects of fluid inertia may result in complex, unsteady, three dimensional flow features, including helical vortices, jets, and backflow regions. These pore scale features can dramatically affect properties of broader interest, such as pressure gradient, mixing, and dispersion in the medium.

Creating quality meshes for these types of geometries notoriously can be tricky and time consuming. The crux of the process is dealing with the sphere-to-sphere contact points, in which elements can become unmanageably small and have high aspect ratios. One way to mitigate this problem is to bridge the contact point with a small fillet or cylinder [Kuroki, J. Chem. Eng. Japan, 2009]. By removing the small regions of fluid near the contact points, a smooth, non-skewed surface mesh can be obtained, as shown in Figure 2.

Fillets between contacting spheres improve mesh quality.

Figure 2: A smooth surface mesh transition provided by bridging two contacting spheres with a fillet.

It quickly becomes challenging to generate this type of mesh for more than a few spheres by eye using a graphical user interface because of the large number of geometric entities and high surface area-to-volume ratio. The CFPL researchers realized that their problem was ripe for automation using the Pointwise/Glyph scripting capabilities. The geometry creation and mesh generation processes were parameterized easily, and a generalized algorithm was developed to handle arbitrary sphere positions and boundary types. Glyph, which provides the full functionality of Pointwise as an extension to the Tcl scripting language, made the transition to automation surprisingly straightforward. In addition, a Tk user interface, shown in Figure 3, was designed to minimize the required user interaction. The CFPL group is now generating quality meshes for large scale sphere packs and has decreased their interaction time from days to minutes.

A graphical interface helps minimize user interaction.

Figure 3: The Tk user interface for the Glyph scripts used for mesh automation.

The flow simulations are performed using parallel research codes. Simulations utilizing up to 350 processors have been run on the NSF's TeraGrid supercomputers over a large range of Reynolds numbers. The high resolution meshes are able to capture the important unsteady dynamics and vortex shedding at the pore scale. This is shown in Figure 4, where the Lambda-2 vortex detection criteria is visualized as an isosurface for a Reynolds number of 529. Contour colors show regions of high and low pressure for a single instant in time.

Isosurfaces of Lambda-2 vortex detection criterion.

Figure 4: Flow from left to right through a simple cubic packing of spheres at a Reynolds number of 529. A single isosurface of the Lambda-2 vortex detection criteria is shown. Color contours show the local pressure variation. This image was created using Tecplot 360.

◀ Previous Article     ▲ Contents     Next Article ▶