⚡ electrical engineering

Process Parameters and Geometric Accuracy in Robotic Pellet Extrusion of Wood-Based Polylactic Acid (PLA) Composites

This study investigates the effects of robot travel speed, extruder rotation speed, and nozzle temperature on the geometric accuracy of wood-based PLA composites in robotic pellet extrusion, revealing that travel speed is the dominant factor while the other parameters can be independently tuned within a stable regime to achieve precise dimensional control.

Original authors: Osama A. Q. Ziada, Ares A Gomez-Gallegos

Published 2026-09-21

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Osama A. Q. Ziada, Ares A Gomez-Gallegos

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a factory floor where giant robotic arms, usually reserved for welding car frames or moving heavy cargo, are being taught a new, delicate skill: building objects from the ground up, layer by layer. This is the world of robotic additive manufacturing, a field that promises to create massive structures far larger than what traditional 3D printers can handle. Instead of using thin, pre-made plastic strings, these advanced machines often use raw pellets of plastic, melting them down and squeezing them out through a nozzle to form solid shapes. To make this process even more sustainable, scientists are mixing these plastics with tiny wood fibers, creating a material that is part renewable plant and part synthetic polymer. However, there is a catch. Wood fibers make the melted plastic behave differently than pure plastic; it becomes thicker, more sensitive to heat, and harder to control. If the robot moves too fast or the heat is wrong, the material might not stick together properly, or the object might collapse before it is finished. The challenge is finding the perfect rhythm between how fast the robot moves, how fast the machine pushes out the material, and how hot the nozzle gets.

In a recent study, researchers at Linnaeus University in Sweden set out to solve this puzzle for a specific type of wood-based plastic. They wanted to know exactly how to control the size and shape of the objects these robots build. To do this, they built a test setup using a standard industrial robot arm equipped with a specialized extruder designed to handle wood-filled pellets. They chose to print hollow cubes, a simple shape that is easy to measure, to see how well the final product matched the digital design. The team tested three main variables: the speed at which the robot arm traveled, the speed at which the screw inside the extruder rotated to push the material out, and the temperature of the nozzle. They printed twenty-four different cubes, changing one setting at a time while keeping the others steady, to see which factor had the biggest impact on the final result.

The results revealed a clear hierarchy of control. The most critical factor turned out to be the speed of the robot. When the robot moved at a moderate pace, the cubes were built successfully. But when the speed was increased to a specific high level, the entire structure failed. The robot moved so quickly that the material could not keep up, causing the layers to skip and the cube to shrink to less than a quarter of its intended height. This happened because the robot was moving faster than the melted plastic could be deposited, breaking the connection between the machine's motion and the flow of material. Within the safe, slower speeds, however, the researchers found a surprising order. They discovered that the three settings they were adjusting acted almost independently of one another, each controlling a different part of the cube's shape.

Specifically, the temperature of the nozzle was the main driver for the total height of the cube. If the heat was too high, the material became too runny and the walls could not support themselves, causing the cube to sag or collapse. The speed at which the extruder screw rotated controlled the thickness of the walls; turning the screw faster pushed out more material, making the walls thicker. Finally, the speed of the robot arm determined the height of each individual layer. By understanding these separate roles, the team realized that operators could tune the machine in a simple, step-by-step process: first set the temperature to get the right overall height, then adjust the screw speed to get the right wall thickness, and finally fine-tune the robot's travel speed to get the layers just right.

The study also highlighted that not all wood-based plastics behave the same way. They tested two different materials, one with a higher concentration of wood fibers and one with less. The material with fewer fibers was much more sensitive to heat and speed. At high temperatures and low screw speeds, this lighter material would completely lose its structural integrity and collapse, shrinking dramatically. The material with more wood fibers was more stable, holding its shape better across a wider range of temperatures. This finding suggests that there is no single "perfect" setting for all wood-based plastics; the recipe must be customized for the specific mixture being used.

Ultimately, this work provides a practical roadmap for using robotic arms to build large, sustainable structures from wood-plastic composites. It proves that while the process is sensitive, it is not chaotic. By keeping the robot speed within a safe limit and understanding that temperature, screw speed, and travel speed each control a specific dimension, engineers can reliably produce accurate parts. The researchers noted that their findings are specific to the machine and materials they tested, but the method they developed offers a clear starting point for others to calibrate their own systems. The path forward involves testing these settings on more complex shapes and with different material blends, but the fundamental lesson is clear: success in robotic 3D printing with wood fibers comes from respecting the limits of the material and finding the precise balance between heat, flow, and motion.

Technical Summary: Process Parameters and Geometric Accuracy in Robotic Pellet Extrusion of Wood-Based Polylactic Acid (PLA) Composites

Problem Statement
Robotic Large-Scale Additive Manufacturing (RAM) using pellet extrusion offers a sustainable pathway for manufacturing with wood-based Polylactic Acid (PLA) composites. However, systematic data regarding the dimensional accuracy of this specific material–system combination is limited. Unlike filament-based Fused Deposition Modelling (FDM), pellet extrusion involves direct extrusion of raw granules, which introduces challenges related to melt flow heterogeneity, moisture sensitivity, and the synchronization between robot motion and material feed. Existing literature often focuses on unfilled thermoplastics or filament systems, leaving a gap in understanding how process parameters influence geometric fidelity in wood-based PLA pellet extrusion. Specifically, the boundaries between stable and unstable deposition regimes, and the interplay between robot travel speed, extruder rotation, and nozzle temperature, remain under-characterized.

Methodology
The study employed a One-Factor-At-a-Time (OFAT) experimental design to map the process window for two wood-based PLA grades: KCL Formi® 3D Nordic Birch PLA-20 (20 wt% cellulose) and PLA-10 (10 wt% cellulose).

System: An ABB IRB 1600 industrial robot coupled with a KFM Eco 3D 18 screw-based pellet extruder (3 mm nozzle).
Specimens: Twenty-four hollow cubes (100 × 100 × 100 mm nominal) were printed using a continuous spiral toolpath to eliminate start/stop artifacts.
Parameters:
- Robot travel speed ( $V$ ): 15, 30, and 45 mm/s.
- Extruder rotation speed ( $N$ ): 3, 4, and 5 rpm.
- Nozzle temperature ( $T$ ): Four discrete levels ranging from 150/155 °C to 180/185 °C (Zone 1/Zone 2).
Metrics: Geometric accuracy was evaluated using mean deviation, Mean Absolute Error (MAE), standard deviation, and percentage error against nominal dimensions for cube height, wall thickness, and layer height.
Analysis: The dataset was analyzed to identify stable vs. unstable regimes. Within the stable regime ( $V \le 30$ mm/s), Pearson correlation matrices and second-order polynomial models were developed to characterize parameter interactions and decoupling.

Key Results

Dominance of Travel Speed: Robot travel speed was identified as the primary determinant of geometric stability. At 45 mm/s, both materials experienced severe under-extrusion, resulting in cube heights below 25% of the nominal value (e.g., 15.47 mm for PLA-20). This failure mode manifested as a loss of synchronisation where material accumulated laterally (increasing wall thickness) rather than vertically.
Stable Regime Decoupling: Within the stable operating window ( $V \le 30$ $V \leq 30$ mm/s), the three process parameters exhibited near-separable effects on specific geometric quantities:
- Travel Speed primarily governed layer height.
- Extruder Rotation primarily governed wall thickness.
- Nozzle Temperature primarily governed cube height.
  This decoupling supports a sequential calibration approach rather than a fully coupled multi-variate optimization.
Material-Specific Thermal Sensitivity:
- PLA-20 (20% fibre): Exhibited a sharp drop in cube height from 113.35 mm at 150 °C to 77.59 mm at 160 °C, followed by a thermal plateau where height remained relatively stable (77–83 mm) across the 160 to 180 °C range.
- PLA-10 (10% fibre): Showed a critical interaction between temperature and rotation. At low rotation (3 rpm), increasing temperature from 160 °C to 180 °C caused a catastrophic collapse in cube height (a 67.4% reduction), attributed to the lower melt viscosity and lack of fibre reinforcement leading to wall starvation. However, at 4 rpm, PLA-10 maintained near-nominal height even at 180 °C, indicating that higher material flow can compensate for thermal thinning.
Systematic Layer Height Overshoot: All specimens exhibited layer heights exceeding the nominal 0.5 mm (ranging from 0.55 to 1.86 mm). This systematic deviation was strongly correlated with travel speed within the stable regime, suggesting it is governed by kinematic control and die swell rather than slicer settings alone.

Predictive Modeling
Second-order polynomial models were fitted to the stable-regime data:

PLA-20: Well-described by a separable quadratic model (no interaction term required, $R^2_{adj} = 0.68$ ).
PLA-10: Required a temperature–rotation interaction term to accurately capture the thermal collapse at low rotation ( $R^2_{adj} = 0.91$ ). The interaction coefficient ( $a_5 = +2.43$ mm·°C⁻¹·rpm⁻¹) quantifies how increased rotation counteracts the height loss induced by elevated temperatures.

Significance and Claims
The authors claim this work establishes a quantitative framework for evaluating dimensional accuracy in robotic pellet extrusion of wood-based PLA. The primary contributions are:

Identification of the Operating Envelope: Defining the upper boundary of the stable regime at 30 mm/s for this material-system combination, beyond which geometric fidelity collapses.
Decoupled Control Strategy: Demonstrating that within the stable regime, speed, rotation, and temperature can be tuned independently to control layer height, wall thickness, and cube height, respectively. This supports a proposed three-step calibration procedure.
Material-Specific Calibration: Highlighting that parameter selection cannot be universal across different wood-based PLA grades; specifically, PLA-10 requires a coupled temperature-rotation strategy to avoid thermal collapse, whereas PLA-20 is more thermally robust.

Limitations and Future Work
The paper explicitly notes several limitations:

The polynomial models are based on a small dataset ( $n=10$ per material) without external validation; the reported $R^2$ values reflect training fit rather than out-of-sample predictive accuracy.
The OFAT design prevents the extraction of a numerical global optimum via surface extrapolation.
The calibration is specific to the ABB IRB 1600/KFM Eco 3D 18 system and the cube geometry; coefficients are not expected to transfer to other systems without recalibration.
Future work is identified as necessary to validate the models with external data and to employ multi-parameter designs (e.g., Taguchi or Response Surface Methodology) to resolve interaction effects more rigorously.

More like this