By Kumar M. Dhanasekharan, Ph.D.
Philadelphia Mixers has used a new mixing analysis software package to solve a baffled tank's pH adjustment problem in record time. The complete analysis, which required several simulations, took only one day. This quick resolution took considerably less time than the week or more required for general-purpose computational fluid dynamics (CFD) software and grid generation tools.
In the past, building the model would have taken much longer because of the complex system geometry, which contained both stationary and moving parts. But, the mixing-specific software quickly set up the model automatically from user-entered parameters.
Identifying the Problem
A steel mill was having a pH adjustment problem in part of its wastewater pretreatment system. Although it purchased the mixer from another supplier, it called Philadelphia Mixers.
Philadelphia Mixers, based in Palmyra, PA, manufactures and supplies mixers for chemical processing, water and wastewater treatment, mining, food, pharmaceuticals and more. The company also offers its services for consulting and mixing design, scale up and testing, on-site start up, and employee training, inspection and repair.
The mixing system consisted of a vessel and two impellers, an axial impeller and below it, a radial impeller. The influent port was located near the radial impeller. The system was not delivering a constant pH.
"The first thing we wanted to do was determine the mixing efficiency of this design," said Wojciech Wyczalkowski, director of technology at Philadelphia Mixers. One option was to simulate mixing in the existing system with general-purpose CFD software.
CFD, Pros and Cons
A key advantage of CFD is the flexibility to change design parameters and evaluate new configurations quickly.
A major challenge in using CFD to model this type of problem is the continual motion of the impellers in the baffled vessel. In recent years designers have met the challenge by using the multiple reference frames (MRF) method to simulate the spinning agitator in a rotating frame while the remainder of the tank (including the baffles and inflow port) is simulated in a stationary frame.
The solution proceeds with a steady transfer of information across a pre-defined interface between the two frames (rotating and stationary). As with any CFD simulation, this approach requires creating the geometry of the impellers and vessel, and then creating an analysis mesh. It also requires the user to specify other information about the problem such as fluid properties and performance parameters like shaft speed. Setting up a complicated mixing problem in this manner can take up to a week.
Dedicated Mixing Software
Philadelphia Mixers staff overcame these limitations by using dedicated mixing software to determine the mixing efficiency of the malfunctioning pH system. The software they chose, MixSim, is available from Fluent Inc., Lebanon, NH, also the supplier of Fluent, a popular CFD program.
The software's user interface simplified creating the mixing analysis model.
"One of the main advantages of MixSim for us is its ability to create the geometry and mesh automatically according to specifications entered by the user," Wyczalkowski said.
MixSim specifically addresses mixing and related flow phenomena and simulates the hydrodynamics of agitated mixing vessels. It predicts the complete blending and motion in batch, semi-batch, and continuous stirred tanks, and incorporates additional transport phenomena.
Simple Setup
Wyczalkowski decided to start with a 2D analysis of the existing pH system. Because he had laser Doppler velocimetry data of the system, he decided to use the velocity data method initially.
He simply entered the laser data into MixSim, specified the liquid properties, and ran the analysis.
The solution time took only 15 minutes. A particle flow plot clearly showed why the steel mill was having problems with the mixer. A design flaw caused a short circuit in which the pH adjustment fluid exited the vessel before mixing with the entire volume.
"CFD is an excellent visualization tool that helps you understand a problem," said Wyczalkowski. "Once you understand it, you can apply physics to it to solve it."
Fixing a Faulty Design
Using the velocity data method once again, Wyczalkowski made several design changes and was able to evaluate them quickly. Keeping the vessel configuration fixed, he added a third impeller and changed the direction of the pumping action of the original axial impeller.
These changes were easily implemented through the MixSim interface by entering parameters such as impeller location, size, and pumping direction. For each design change, the software built the geometry and created the analysis mesh automatically.
"The entire process of setting up each model took five minutes," Wyczalkowski said. "After I entered a few specifications, the software created the geometry and the analysis mesh in less than five minutes."
After the first analysis showed the reason for the fluctuating pH, Wyczalkowski ran more analyses to improve the design. By the end of one day, he had an optimized solution that the customer later successfully implemented.
"My goal was to configure the system so that the pH adjustment stream was pushed into an impeller instead of the wall," he said.
When he achieved one that thoroughly mixed the pH adjustment material with the entire contents of the vessel, he repeated the analysis in 3D to capture any 3D effects and make it easier for the client to visualize. The solution time for the 3D analysis was three hours.
The customer was convinced by the analysis results that Philadelphia Mixers' new impeller configuration would work. The steel mill replaced the existing impeller configuration with Wyczalkowski's new design. Since then, the pH adjustment system has achieved the constant values it was designed to produce.
"Mixing-specific software makes CFD practical in a situation such as this," said Wyczalkowski. "It is so fast that it you can use the computer to evaluate many different designs."
About the Author: Dr. Kumar M. Dhanasekharan specializes in mixing applications in his work at Fluent. He obtained his Ph.D. from Rutgers and has researched modeling food extrusion and laminar mixing using viscoelastic models. More information is available at www.fluent.com.