Fig. 2 Modified shell design velocity with higher air speed around windings.

The use of computer simulation made it possible to develop a smaller than normal jetted tub motor in eight months instead of the two to three years that would have been required using the traditional build and test method. Using the old methods, it was often necessary to build 10 to 15 prototypes to find a design with the right thermal characteristics. While an adequate ventilation system could usually be developed, it was rarely an optimal design. Each of these prototypes take about one month to build, and require additional time for testing and modifying the design based on the results. In this application, Emerson Motor Co., St. Louis., used computational fluid dynamics (CFD) to simulate the flow of air through the motor and provide detailed information on airflow velocity, direction, and temperature at every point in the design. CFD provided far more detailed information than can be obtained from physical testing and allowed engineers to evaluate performance of new designs without building prototypes. In response to changing design requirements, engineers had to completely redesign the motor three times. CFD provided insights that made it possible to respond quickly and re-optimize the design.

Thermal challenges

Fig. 3 Open midriff design of the motor.

When management made the decision to create a new 3.7 in. diameter motor, Emerson engineers knew from the beginning that their toughest challenge would be keeping temperatures within the motor low enough to avoid breaking down the insulation which could eventually cause the motor to fail. The company normally uses a Class B insulation system, which requires a maximum temperature rise of 85?C above ambient. When the design team assembled together their first prototype using a traditional ventilation system design, as they expected, it tested out well above the specification at 125?C above ambient. At this point, the normal approach would have been to change the fan geometry or the shape of the components that guide the airflow through the motor, build a prototype and test it to see if it worked any better. One problem with this approach is that a large number of resources are required to build each prototype including a machine shop, foundry and assembly personnel. For this reason building a prototype is expensive and takes about a month. In addition, engineers would have had to work largely by the seat of their pants since the test results would provide very little guidance other than whether the motor passed or failed, and the change in temperature achieved. This is the primary reason why the most challenging new motor designs used to take several years to bring from concept to production.

In recent years, Emerson has moved to a new design methodology that offers dramatic improvements over the old methods. It is based on the use of CFD to create a virtual prototype of each iteration of the motor design. CFD involves the solution of the governing equations for fluid flow, heat and mass transfer, and chemistry, at 100,000 to 250,000 discrete points for this motor model, on a computational grid in the flow domain. The original model can usually be created in a matter of days compared to a month for a physical prototype. After the initial model has been produced, new design iterations can typically be created in hours or even minutes if they represent simple changes. Another advantage of this approach is that CFD provides engineering results such as airflow velocity, direction and temperature at all points in the model, compared to the relatively few points that can be measured in a physical test. The result is that engineers have a much better understanding of what went right and wrong in their design and can meet or exceed requirements in fewer iterations. The bottom line is that Emerson has been able to substantially reduce the amount of time required to bring new motors to market using this method.

Initial design

Fig. 4 Open midriff temperature indicating reduced winding temperature.

In this example, Peter Bostwick, Engineer Specialist (CFD/Heat Trans-fer) for Emerson, modeled the initial prototype using CFX CFD software from AEA Technology, Waterloo, Ontario. Bostwick said that Emerson selected this package because they have previously had considerable difficulties in finding software that would consistently converge to a solution on the low Reynolds number problems involved in electric motor design. Reynolds number is proportional to inertial force divided by viscous force and is used in momentum, heat, and mass transfer to account for dynamic similarity. “On the other hand, we have found that CFX always converges,” Bostwick says. “We have also found that its results correlate very closely to our test measurements. CFX also makes it possible to create geometrically complex models in a minimum amount of time.”

The original design looked much like the company’s larger motors. It was configured so that the fan would draw air axially through holes in the fan cover. The air hit the fan and was blown in a radial direction towards the fan cover that enclosed the fan. The fan cover redirected the air around the shell that enclosed the windings of the motor, cooling the windings and other internal components. Bostwick modeled this design by taking advantage of CFX’s multiple frame of reference model which uses a rotating frame for the spinning fan and a stationary frame for the rest of the motor. The solution proceeds with a steady transfer of information across a pre-defined interface between the two frames. A separate frame is not required for the rotor because its geometry is so uniform that it does not have a significant effect on the airflow, and can be modeled with a moving wall.

Optimized design

Fig. 5 Prototype of actual tub motor.

Because of several simplifying assumptions that he made to reduce run-time, the first iteration had to be calibrated to the test results. From this point on, the simulation matched physical testing results within 5 percent. Bostwick said that he has had similar high levels of correlation nearly every time he has used CFD in motor design. By depicting the airflow throughout the motor design, the CFD results provided insights that helped Bostwick improve the design. He changed the end shields so that air could flow through the space around the windings. He modified the fan geometry to boost the flow through the stator slots and air gap. The new fan geometry had a second set of blades with the first intended to provide circulation around the outside of the motor shell and the second set promoting flow inside the motor. Through multiple iterations, Bostwick adjusted the geometry of each of these components to optimize cooling. He succeeded in developing a design that met thermal requirements, which was confirmed by building a single prototype.

“Everyone was happy with the new design, but management determined that a change was required in order to reduce manufacturing costs,” Bostwick says. “They decided to remove the shell that enclosed the windings. The design had to be substantially modified because without a shell much of the component attachment changes and the end shields had to be totally rethought.” But because he had already gained a clear understanding of the airflow through the motor, he was able to develop a new way to solve the cooling problem within two weeks. The biggest challenge was that the shell was no longer there to confine the airflow around the windings. So Bostwick designed a concentric cylinder for the end shield that that serves the same purpose by drawing the air into the motor along the exterior of the stator and across the windings. Analysis showed that the fan could be made considerably smaller and that it would work more efficiently by changing to pull rather than push air into the motor. CFD analysis showed that the temperature rise was only 57¿C at the rated load using this new approach. Because the temperature rise was well below the upper limit, prototype motors for ½ HP and a ¾ HP ratings were also built to prove the new ventilation system could cool these higher power ratings. This gave Emerson a substantially larger product offering than originally anticipated.

Several additional changes were suggested by the manufacturing group, and Bostwick adapted by changing the design to accommodate them. He changed the fan design back to a double-sided version similar to the way it was two major revisions ago. CFD results were used to guide the entire design process and prototypes were built only at key decision points to validate and the analysis. “CFX dramatically reduced our time to market by providing insights that we couldn’t have determined from experiments and by eliminating the time that would have otherwise been spent waiting for prototypes to be built,” Bostwick says. “We completed this project in about eight months while the trial and error method would have taken several years. Although it’s impossible to prove it, I think that the superior design insight provided by CFD helped us to produce a considerably better design. The virtual prototyping method clearly provides a superior way to develop electrical motors.”