• Da Silva, Lfm
• Paiva, Rmm
• Antonio, Cac

Abstract

On the footwear industry, the composition of adhesives has a high contribution for the product quality. This paper aims to develop a model capable to predict and optimize the creep rate using the composition of the adhesive joints. The proposed mixed numerical and experimental approach is based on following stages: the planned experimental measurements, the learning procedure aiming to obtain the optimal artificial neural network (ANN) configuration, and the optimal design procedure for adhesive joint composition. The design variables are the weight percentages of the solid raw material constituents of the adhesive, such as polyurethanes (PUs), resins, and additives. Considering the experimental results obtained for Taguchi design points as input/output patterns, the ANN is developed based on supervised evolutionary learning. In the last stage, the optimal solution for adhesives composition considering minimum creep rate is investigated based on ANN and genetic algorithm. The optimal results for creep rate minimization based on proposed approach are reached when large quantities for PUs and for some additives are considered, and when colophony and vinyl resin are not considered on the formulation. The sensitivity of the structural response of footwear adhesives to composition constituents is also studied based on Sobol indices obtained from ANN-Monte Carlo simulation procedure. The performance measured by creep rate of the adhesive joint is very sensitive to the influence of some polyurethanes, and a particular sensitivity to caprolactone types with extremely high crystallization is observed. The sensitivities of the creep rate to the resins colophony and coumarone-indene are also important.

Topics

• impedance spectroscopy
• simulation
• resin
• crystallization
• creep