Simulating a spinning 82-in.-diameter pulper rotor in CFD software let engineers finetune designs for a 25% energy savings without loosing performance.
"Many pulping rotors used in the paper industry were developed decades ago and are relatively inefficient," says John Egan, chief technologist at Kadant Black Clawson, Mason, Ohio. Pulper rotors churn slurries of recycled paper or cardboard into fiber for recycling. "Building and testing physical prototypes would be tedious and expensive. So we used computational fluid dynamics to evaluate and predict energy consumption. It worked, and with a high degree of accuracy," he says.
Comparisons to physical experiments show the software's power prediction to be within 3% of the actual value.
"We first analyzed existing designs to understand their fluid-flow characteristics, then modeled and evaluated alternate designs on the computer, and made changes until we hit at least a 20% energy savings," says Egan.
"Because the churning flow in pulpers is difficult to visualize, engineers do not fully understand how the mechanical components contribute to defibering a slurry. And older designs perform reliably, so there was little demand to improve them," he adds.
Egan first developed a simplified rotor-performance model to screen rotor configurations for analysis. It captured the effects of rotor geometry such as diameter and vane-sweep angle. Although simplified assumptions limit prediction accuracy, analyses can still gage relative capabilities of new designs. "Several promising new ' firstcut' designs were developed this way," says Egan. Meshed files were imported into the Fidap CFD from Fluent Inc., Lebanon, N.H.
Engineers applied rotation velocities to all rotor surfaces while tub-wall velocities were set to zero. Top-surface normal velocities were also set to zero while tangential velocities were set to frictionless flow conditions. The top fluid surface was modeled as quasi-free to shorten run times. Using water as the model fluid provided performance comparisons within reasonable periods. The so-called mixing-length model was used to calculate turbulent eddy viscosity because the k-emodel would have taken too long.
"There are theoretical calculations for power levels, but we calculated a net value by summing power contributions from each element face on the rotor surface with a program we wrote. It uses a file from the CFD results that includes the rotor-element faces and total-fluid stresses for pertinent element faces," says Egan.
Based on results, the team built and tested a series of rotors to evaluate the accuracy of CFD predictions. They created a mold master for a series of vanes directly from the 3D rotor models using stereolithography. "This process let us cost-effectively run through a round of physical validation tests confirming our predictions. The net-power predictions correlate well with physical experiments. For instance, the 82-in.-diameter rotor was built and tested in a 20-ft-diameter tub. Computer models predicted 310 hp, while the actual figure was 319 hp, an error of less than 3%. Actual power savings is 25%," says Egan. "
Having validated the CFD model, we can use it to design a series of production rotors to find how rotation speed, tip height, sweep angle, and other design features interact with one another to determine pulping efficiency."