The Kenics static mixer is widely used in the food and chemical industries for in-line blending of liquids, usually under laminar conditions. The mixer consists of a long, cylindrical pipe containing a number of helical elements. While these mixers have been in use for over 30 years, experiments to probe the interior are difficult to perform.

In 1992, André Bakker, then an employee at Chemineer Inc., the company that produces the Kenics mixer, and now Regional Consulting Manager at Fluent's office in Lebanon, New Hampshire, modeled one of these devices using FLUENT 4. The structured, hexahedral, body-fitted grid was built with FLUENT's preprocessor at that time, preBFC, and contained about 100,000 cells. Flow conditions were characterized by a Reynolds number of 10. A tracer species with the same properties as the background bulk liquid was added in the center of the pipe inlet, and the blending of the two liquids was studied as a function of position along the mixer.

The initial calculations, performed on a UNIX workstation, took about a week of CPU time. The number of interior cells was then doubled in two of the grid-coordinate directions within FLUENT, resulting in a refined mesh of 350,000 cells. The calculation at this point was resumed on a Cray C90 computer, and was among the largest problems solved with FLUENT at that time. The solution reached convergence after about 1,000 more iterations during an overnight run.

The large case file was pulled out of retirement recently and run in FLUENT 5.5 on a two-processor UNIX workstation, where it took only about 30 minutes to converge! The grid was then adapted in FLUENT, resulting in a mesh of 2.2 million cells. The solution with the new grid was continued on the two-processor workstation, and converged in about 5 hours.

Isosurfaces of helicity show regions where longitudinal vortex structures exist in the mixer

Comparing the results, the refined grid produces better resolution in species contour plots that show the blending pattern. It allows for the capture of more details in isosurface plots of derivative quantities, such as vorticity and helicity. The exercise also demonstrates how improvements in solver technology and computer speed and capacity have made routine work of solutions such as the most recent one. For FLUENT veterans, this means that some of the modeling strategies that were adopted in the old days to make solutions more quickly attainable can now be either partially or completely eliminated with only increased expectations from the results.

FluentNEWS