Quantitative representation and parameter weight analysis of dimple in multi-point flexible die forming process

Abstract

To investigate dimple defects in the multi-point flexible die stretch forming (MPFDSF) process, this study focuses on the quantitative characterization, prediction, and control of dimples in 2024-O aluminum alloy spherical parts. Based on ABAQUS finite element simulations, the thickness-direction plastic strain component PE33 along the OR characteristic path was extracted, and the average peak-to-valley difference of the PE33 curve was used as the dimple evaluation index. The effects of sheet thickness, forming radius of curvature, punch radius, and pre-stretching ratio were analyzed using single-factor analysis, an L16 orthogonal experimental design, range analysis, and regression modeling. The results show that the dimple index decreases with increasing sheet thickness, forming radius of curvature, and pre-stretching ratio, but increases with increasing punch radius. Range analysis indicates that the influence of the four parameters follows the order: punch radius > forming radius of curvature > pre-stretching ratio > sheet thickness. To further quantify the relationship between dimple severity and process parameters, a logarithmic regression model was established. The model gave a coefficient of determination of R 2 = 0 . 9666 , and leave-one-out cross-validation gave R 2 = 0 . 9214 , indicating acceptable predictive capability within the investigated parameter range. Based on the orthogonal simulation results and the regression model, a preliminary process window for dimple control was proposed. Within the investigated range, R p ≤ 24 mm and R s ≥ 2000 mm are recommended for acceptable dimple control, while a punch radius of 40 mm should be avoided. Experimental validation showed that the simulated thickness distribution and OR sectional profile were in good agreement with the measured results. The proposed evaluation method, prediction model, and process window can provide a reference for dimple control and parameter optimization in MPFDSF.