Modelling and Optimization of Surface Roughness in Wood Powder/PLA Composites

Abstract

Fused filament fabrication (FFF) of wood-filled polylactic acid (Wood/PLA) offers an alternative for manufacturing components with a wood-like appearance. However, achieving consistent surface quality remains a challenge, particularly for components incorporating both internal and external surfaces. This study investigates the effects of flow rate, layer height, and printing speed on the surface roughness of FFF-manufactured Wood/PLA specimens. A three-factor, three-level full factorial experimental design comprising 27 parameter combinations was employed. Hollow cylindrical specimens were fabricated using a Bambu^TM Lab A1 Mini printer, and internal and external surface roughness were measured using a stylus profilometer. Separate quadratic regression models were developed, and their statistical performance was evaluated using analysis of variance and regression diagnostics. The models exhibited coefficients of determination of 99.60% for internal roughness and 99.58% for external roughness. Layer height was the dominant factor associated with both responses, with increasing layer height from 0.20 to 0.40mm accompanied by substantial increases in measured roughness. Multi-response optimization identified a model-predicted candidate combination comprising a flow rate of 90.9%, a layer height of 0.20mm, and a printing speed of 116.0mm/s. The corresponding predicted roughness values were 13.648μm for the internal surface and 13.228μm for the external surface, with a composite desirability of approximately 0.998. Experimental confirmation was subsequently performed using three additional specimens corresponding to the multi-response and individual-response optimization solutions. The measured roughness values showed good agreement with the model predictions, with relative deviations below 3.2% across all comparisons. These findings highlight the dominant influence of layer height and support the applicability of the developed regression models for optimizing internal and external surface roughness within the investigated parameter range.