Simulation of flow and mixing in stirred tank reactors

Jeffrey M. Zalc, Mario M. Alvarez, Fernando J. Muzzio

Research output: Contribution to journalConference article

2 Scopus citations


The design of industrial mixing devices is often based on heuristics and on existing equipment and often has very little scientific basis. Experimental investigation of industrial mixers can be quite time-consuming and yields only limited information. The work presented here involves implementation of new solver technology (a finite element Galerkin variational formulation with equal-order interpolation for all field variables) and efficient post-processing algorithms to obtain high-quality quantitative data concerning laminar flow and mixing in single-impeller stirred tanks, one equipped with a 45° pitched-blade turbine and the other a Rushton turbine. For each case, the exact impeller geometry is modeled and flow in the entire tank is simulated without the need to exploit the four-way (pitched-blade impeller system) or six-way (Rushton turbine system) symmetry that exists for the particular systems studied. Velocity field determination and mixing simulation require less than 24 hours of CPU time on a desktop workstation for meshes comprised of 1-2 million tetrahedral elements. When validation of simulation results for the velocity field are performed using particle image velodmetry, agreement is far better than previously attained. Excellent agreement is also observed between numerical and experimental results concerning the sizes and locations of toroidal-shaped segregated regions generated by both types of impellers. The simulations are even able to resolve small-scale details of these persistent poor-mixing regions that are observed in flow visualization experiments. The simulated velocity fields are used to gather quantitative data concerning mixing processes that would be difficult to quantify experimentally. Post-processing software is used to investigate mixing performance by computing the trajectories and stretching values of a large number, O(104), of infinitesimal vectors placed in the flow. Stress distributions along tank walls are also computed to address the issue of cleaning stirred tank bioreactors.

Original languageEnglish (US)
Pages (from-to)233-239
Number of pages7
JournalAmerican Society of Mechanical Engineers, Pressure Vessels and Piping Division (Publication) PVP
Volume397 II
StatePublished - Dec 1 1999
EventProceedings of the 1998 IEEE International Electron Devices Meeting - San Francisco, CA, USA
Duration: Dec 6 1998Dec 9 1998

All Science Journal Classification (ASJC) codes

  • Mechanical Engineering

Fingerprint Dive into the research topics of 'Simulation of flow and mixing in stirred tank reactors'. Together they form a unique fingerprint.

  • Cite this