Date of Publication :2nd March 2017

Abstract: Increased competition in the manufacturing industry requires performance improvement at any enterprise level. Optimizing performance in terms of metrics such as production costs requires understanding and optimization of cost-inducing variables from product design and manufacturing. Nonetheless, the number of variables influencing production costs is very high, so it is time-intensive so computationally difficult to optimize all variables. Therefore, it is important to recognize and optimize select few variables that have high cost inducing potential. To this end, a dimension reduction approach is proposed, incorporating the mathematical modelling paradigm for dimensional analysis and the principle of graph centrality. The proposed approach incorporates current cost-inducing knowledge of variables, their interactions, and input-output relationships for various functions or device behaviour, in the form of a causal graph. To recognize conflicting effects on the variables in the graph, the propagation of optimization goals in the causal graph is verified. Following the study of the paradox, the principle of graph centrality is used to rank the different regions within the graph based on their relative importance to the problem of optimization and to classify the variables into two classes of optimization, namely less important variables and the most important variables in relation to cost optimization. The question of optimization is designed to address less relevant variables at their highest or lowest levels based on their cost interaction, and to optimize the larger variables to minimize costs. The proposed approach for dimension reduction is seen for an optimisation problem, to reduce the bladder manufacturing costs and the main mechanism for a high-field superconducting CERN magnet capable of generating 16 Tesla magnetic fields. The graph region representing the electromagnetic force and resulting stress produced during magnet energizing was found to be ranked highest for impact on the bladder and main cost of manufacture. Using a genetic algorithm solver in MATLAB, an optimization of the stress and its associated variables is performed to reduce production costs.

Reference :