Glenn Dorsch
Vice President of Engineering
Kent Keeran
Senior Engineer
Vaughan Co. Inc.
Montesano, Wash.

Edited by Paul Dvorak

The pump design dates from the 1960s, but it chops rope, copper wire, cloth, and even 1-in.-diameter plastic rods. Our recent adoption of CFD software lets us look at details of the blade’s shape and improve the pumps operation. And because simulation costs so little, we can test many subtle variations of impeller blades to find the most efficient one. The blade redesign works so well that the work done by an older pump and a 30-hp motor is now done by a newer pump with a 25-hp motor.

In addition, correlation between simulations and physical tests is so good that we no longer use physical prototypes. R&D prototypes are now modeled, simulated, and improved, and then go straight to production castings.

The chopping requirements and suction arrangement of these pumps make it difficult to apply standard impeller-design practices. Performance calculations were rough estimates, and efficiency was almost impossible to predict. What’s more, building test impellers by hand or machining them on a five-axis NC mill is expensive and time consuming. Traditional cast patterns cost about the same as machined prototypes. And while rapid prototyping an impeller is about half the cost of machining, it is still more expensive than building digital models and running simulations.

Our CAD program, Pro/E, makes it relatively easy to separate the fluid domain from the solid model, and include volumes upstream from impeller suction and downstream from casing discharge. The impeller and casing domains are meshed separately and assembled in the preprocessor in CFX software from Ansys Inc., Canonsburg, Pa. (ansys.com), where we apply boundary conditions. Although the software lets us apply a viscosity to simulate a thick liquid, most simulations are run with water, making them comparable to results from our test facility. Solid chunks in the flow are not accounted for, nor is the action of cutting large pieces into smaller ones. However, the shapes of the cutting features of the suction plate and impeller are included.

The CFD solver performs the calculations, and results are viewed with pump performance calculated in the CFD’s postprocessor. A cluster of about four computers takes 4 to 8 hr to compute head and efficiency for one flow point on a performance graph.

Impeller blade shapes were originally generated in CAD and based on our best guesses for sweep and cross sections. That provided some success, but the process was cumbersome. A better way has been to use Ansys BladeModeler software. It more easily generates the complex blade shapes based on impeller flow equations and our experience. The blade modeler then exports curves to the CAD software to construct a solid model.

It is now a simple matter to extract a wide variety of information, including flow rates, pressures, and component forces, to better design the whole pump. For example, with more accurate impeller forces, we’ve selected bearings better equipped to handle loads.

See a real chopper in action
For a look at a Vaughan Chopper Pump in action, go to chopperpumps.com and look under Info and pick Videos. The dozen or so there show what a pump can do to 1-in. rope, plastic rod, blue jeans, even a whole turkey.

The CAD model and photo of a chopper pump show straight blades on the suction plate and curved blades on the impeller. The pumps are used when a liquid that contains solids must pass through the pump without clogging.

The CAD model and photo of a chopper pump show straight blades on the suction plate and curved blades on the impeller. The pumps are used when a liquid that contains solids must pass through the pump without clogging.

The CAD model and photo of a chopper pump show straight blades on the suction plate and curved blades on the impeller. The pumps are used when a liquid that contains solids must pass through the pump without clogging.

Performance curves are for a recently redesigned 6-in. pump. The simulation slightly underpredicts efficiency because conservative simulation parameters were used which slightly overpredicted required power.

Performance curves are for a recently redesigned 6-in. pump. The simulation slightly underpredicts efficiency because conservative simulation parameters were used which slightly overpredicted required power.

Arrows in the CFD results are streamlines or particle paths. Colors represent speeds in feet/second. The CFD software calculates efficiency, head, and loads on pump parts.

Arrows in the CFD results are streamlines or particle paths. Colors represent speeds in feet/second. The CFD software calculates efficiency, head, and loads on pump parts.