Response surface optimization and ANFIS Prediction of abrasive waterjet machining in areca nut biofiller epoxy composites

Abstract Abrasive waterjet machining (AWJM) of epoxy-10 wt.% areca nut biofiller composite was investigated using a face-centred central composite design in response surface methodology (RSM). Waterjet pressure (WJP), traverse speed (TS), abrasive flow rate (AFR), and stand-off distance (SoD) were varied to assess surface roughness (SR) and kerf angle (KA). Measured SR and KA ranged from 3.55 to 5.43 µm and 3.08 to 4.42°. Analysis of variance identified SoD as the dominant factor for both responses. Backward elimination improved the RSM predicted R 2 from 0.8287 to 0.8553 (SR) and from 0.8954 to 0.9307 (KA). In the six-observation comparison, ANFIS achieved lower RMSE values (0.0313 µm and 0.0298°) than the backward-eliminated RSM models (0.0469 µm and 0.0462°) for SR and KA. ANFIS exhibited the lowest RMSE across all 30 observations (0.0859 µm and 0.0379°). Full-RSM desirability optimization identified 300 MPa WJP, 200 mm/min TS, 150 g/min AFR, and 1.5 mm SoD as the optimal conditions, with a desirability of 0.95. Confirmation experiments yielded 3.48 µm (SR) and 3.11° (KA), with prediction deviations of 3.53% and 2.81%. SEM observations revealed finer cutting tracks and fewer surface defects under optimized conditions. These findings highlight the complementary modelling capabilities of RSM and ANFIS for AWJM of biofiller-reinforced sustainable epoxy composites.