The first article uses CFD to match tank, impeller, and baffle geometry as well as additive feed location. The simulations were performed to determine the mixing time which is critical to mixing efficiency. The case is particularly interesting due to the use of three different impeller models. One, a simple velocity data method, does not explicitly model the impeller, but enforces known velocity and turbulence values (taken from experiment) in the impeller region. The other methods model the impeller explicitly, but differ in their treatment of the frame of reference change. One uses the true transient sliding-mesh interface and the other employs averaging at the interface. In addition, the accuracy of k-e and RSM turbulence models for this geometry was examined. Accuracy is quoted to be within experimental error.

The design team used CFD to analyze the non-Newtonian fluid and found that shear thinning (a drop in viscosity when the fluid is under shear) was the cause of the inefficient mixing. The original impeller had been designed for a constant viscosity fluid. CFD analysis showed that a small change to increase the RPM of the current mixer would correct the problem.

The small change saved the plant $15,000 over the cost of a new mixer. CFD proved vital to the solution to this problem as the proprietary nature of the latex material prevented experimental testing.

I find these two success stories to be indicative of the expanded use of CFD in engineering problem solving. Problems such as these would only have been undertaken with experimental methods or crude approximations and correlations a few years ago. Recent advancements in modeling multiple frames of reference, turbulence, and non-Newtonian flows have resulted in the ability to problem-solve using these methods. The accuracy of the results shows that these methods are applicable to engineering flows, their use in an engineering problem solving context shows their maturity.