Answers are grounded in this paper and cite their sources.
Optimizing the plasticizing process in injection molding increases efficiency by decreasing (i) melt temperature, which reduces cooling and cycle times, and (ii) torque, which minimizes energy consumption. With growing computational power, digital process optimization becomes increasingly practicable, but the optimization and sampling methods available differ in accuracy, computation time, and suitability for specific problem dimensionalities. Most studies investigating numerical optimization have focused on high-dimensional sampling; polymer processing, however, typically involves only a few key variables. Thus, a systematic assessment of SHGO sampling strategies for low-dimensional polymer processing problems remains lacking. We therefore concentrated on simplicial homology global optimization (SHGO) to optimize melt temperature and specific mechanical energy input, as this method constructs a simplicial complex to explore the design space and covers it more uniformly than other optimization methods while remaining computationally feasible, especially for low-dimensional problems (≤ 10 dimensions). In contrast to existing SHGO applications, this study provides a comparative evaluation of different SHGO sampling methods specifically for low-dimensional (2D and 4D) plasticizing process optimization. Our study systematically assessed three sampling methods within SHGO, Sobol, Halton, and hypercube triangulation, in two- and four-dimensional parameter spaces, considering circumferential screw speed, mass flow rate, and axial temperature profile. Modeling used the JKU S3 Screw Simulation Software, which provides performance assessments in under two minutes per point. All methods were evaluated in terms of computational efficiency, run time, coverage of the search space and optimization performance, with Bayesian optimization as a reference benchmark. The results show that SHGO with the hypercube triangulation sampling yields the best optimization results in both 2D and 4D problems for melt temperature and SME minimization, while the optimization combine with Sobol sampling achieves comparable solution quality but requires a higher number of sampling points, and Halton sampling seems to tend to variability in optimization, leading to inferior results. These results demonstrate that optimization performance depends on both problem dimensionality and optimization objectives, providing specific guidance on method selection for low dimensional plasticizing process optimization.
No takes yet. Share an insight, caveat, or question.
Schaufler et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: