Milling difficult-to-machine ductile materials comes with issues, such as formation of continuous chip, built-up edge, higher cutting forces, tool wear, thermal deformation, and microstructural changes in both cutting tool and target materials. Abrasive waterjet (AWJ) milling has proven its uniqueness in machining ductile materials due to material removal by local erosion by particles and exertion of low cutting forces. Furthermore, freeform surfaces can be generated by manoeuvring the jet strategically. Irrespective of these unique advantages, the stochastic behaviour of the jet results in non-uniform surfaces along different regions of the surface milled. On the other hand, the controlled surface generation demands an understanding on the blind kerf formation so that the desired geometrical surface can be achieved by suitably varying the process parameters. Based on that a suitable model that can predict the material removed can be developed. Furthermore, a model for the prediction of the cavity generated at a given degree of overlap of the jets is required for prediction of the local surface generation. Apart from these models, to address the nonlinear material removal in various zones on the milled surface, a model to predict the influence of jet acceleration/deceleration on the material removal is required. Finally, very limited attempts have been reported on prediction of the comprehensive milled surface.
To address the above, in present work, an analytical model for prediction of milled surface is proposed by considering non-linearities involved in the material removal in various regions of the milled surface by the AWJs, and validated by milling flat surfaces in ductile material, Al6061-T6 as a case study. Experimentally obtained kerf profiles and milled surfaces by varying AWJ process parameters were analysed to understand the physical phenomenon involved. From the understanding gained, a modular approach was proposed with building blocks, such as (a) trench formed in single pass erosion, (b) cavity formed in overlapping pass, (c) kerf geometry over jet start/stop region, and (d) jet direction change region. The model considered the effect of abrasive particle velocity distribution, effective jet plume divergence, spatial abrasive mass distribution, and varying traverse rate profile, along with particle erosion theory. Towards capturing the non-linearity during overlapped pass, the initial non-flat surface and its effect on the variation of local particle impact angle, uneven jet deflection during subsequent passes around the jet axis were incorporated. From the experimental results, the non-uniform depth was observed at jet start/stop and jet direction change region, whereas, relatively uniform depth was observed at central portion of milled pocket. The mean erosion depth at jet start/stop region was higher than mean erosion depth over jet direction change region, which was further less than the mean erosion depth over central region. The proposed model predicts the kerf CP geometry with a maximum MAE of 36 μm and 49 μm, and maximum erosion depth with a maximum error of 7%, 11%, in single and overlapped passes, respectively. The model predicted 3D kerf geometry over the jet start/stop and jet direction change region captured the depth variation trends accurately. Finally, the predicted milled surface topography was found in good correlation with the experimentally obtained surface. With the variation in the overlap, a maximum absolute error in 3D waviness was predicted under 22 μm.