Speeding Up Extrusion Design: Analytical Correction Factors for Non-Newtonian Flows

Competence Center CHASE
Jun 9
3 min read

Optimizing polymer machinery requires fast and accurate flow predictions. This article introduces novel analytical correction factors that calculate side-wall effects instantly, replacing slow numerical simulations to enable real-time design automation and enhanced energy efficiency.

To design highly efficient polymer processing machinery like extruders, dies, and molds, engineers must understand exactly how molten plastic behaves under pressure. Traditional numerical fluid simulations provide accurate flow predictions, but their high computational demands make them too slow for rapid design optimization loops. Our study eliminates this operational bottleneck by introducing novel analytical correction factors that calculate side-wall effects instantly, removing the need for time-consuming numerical techniques.

The Challenge of Side Walls in Polymer Processing

Modern polymer processing relies heavily on rectangular flow channels. These geometries appear everywhere—from the flight channels of single-screw extruders to shaping operations inside complex manufacturing dies and molds.

Predicting these flows introduces severe engineering challenges:

The Nonlinear Complexity: Polymer melts are non-Newtonian, meaning their viscosity changes dynamically depending on how fast they are sheared. As the shear rate increases, the molecular chains untangle, causing the material to thin out.
The Side-Wall Drag: Real rectangular ducts have side walls that create extra drag. This drag alters the flow rate and triggers intense viscous dissipation—internal friction that generates heat.
The Computational Bottleneck: Accounting for side walls transforms straightforward math into multidimensional partial differential equations. Solving them usually requires slow, iterative numerical software packages.

To bypass this bottleneck, the research team analyzed the flow of power-law fluids under fully developed, isothermal conditions. They ran extensive finite-volume numerical simulations across a parametric sweep of 126 independent combinations of channel dimensions and material shear-thinning properties.

Bridging the Gap with Symbolic Regression

Instead of settling for slow simulations, the study utilized advanced mathematical optimization tools to extract ready-to-use algebraic formulas from raw simulation data.

The team compared and synthesized three distinct layers of engineering theory:

The Parallel-Plate Baseline: Engineers frequently use simple, one-dimensional equations for infinitely wide plates because they solve instantly. However, because they neglect side-wall boundaries, they introduce severe calculation errors in realistic, boxier channels.
Traditional Shape Factors: Previous literature attempted to use simple linear or quadratic correction factors. While useful for Newtonian fluids, these classical equations break down when applied to highly shear-thinning plastics, dropping in accuracy.
Symbolic Regression via Genetic Programming: Rather than forcing the simulation data into a pre-set curve, the team used an offspring selection genetic algorithm. The software mathematically "evolved" new algebraic formulations, testing millions of combinations of arithmetic operators to find the absolute best fit.

The result is a set of compact, multi-parameter analytical correction factors. These expressions directly modify the simple parallel-plate equations to account for side-wall effects across a broad range of aspect ratios and material behaviors.

High-Precision Design Without the Computational Cost

The mathematical breakthrough delivers immediate advantages for machinery manufacturers, die designers, and process engineers:

Instantaneous Calculations: Complex non-Newtonian fluid dynamics that previously required minutes or hours of computer simulation can now be evaluated in milliseconds using basic algebraic formulas.
Exceptional Precision: Validation tests confirmed that the new formulas achieve outstanding correlation coefficients, keeping mean relative prediction errors below 1% for most industrial settings.
Optimized Thermal Profiles: The equations accurately predict viscous dissipation. This lets engineers identify and minimize localized friction hotspots that cause material degradation or discoloration.
Rapid Design Automation: Because these correction factors are fully analytical, they can be coded directly into automated optimization loops, allowing design software to evaluate thousands of virtual screw geometries or die profiles in seconds.

Scaling Extrusion Systems Faster with Simplified Thermal Modeling

The derived equations transform how developers approach temperature variations and system scaling. Instead of wasting time setting up slow 3D mesh simulations for every physical design change, engineers can now use fast lumped-parameter models:

Efficient Channel Segmentation: The software splits complex flow geometries into short, manageable segments to calculate temperature shifts sequentially.
Instant Dynamic Evaluation: Integrating the new correction factors into each segment allows for immediate calculation of local flow rates, pressure drops, and heat generation.
Rapid Performance Charting: Design software can map out accurate machine capabilities across changing operational profiles in seconds, entirely bypassing high-performance computing clusters.