Inverse design of glass forming process simulation using an optimization technique and distributed computing

In this paper, an inverse problem of heat transfer in the glass forming process is studied to determine a number of unknown heat transfer coefficients. An analysis program for transient heat conduction is developed to simulate the forming and cooling processes. The analysis involves the contact and separation of glass and molds and it is repeated until the process reaches a periodic state. Temperature measurements are taken at several available times and spatial positions of glass and molds during the forming operation. Heat removal by the cooling water from the plunger is also recorded. An optimization problem is formulated to determine the heat transfer coefficients that minimize the difference between the measured data and the analysis results. Significant time savings are achieved in computing finite difference-based sensitivity during the optimization by employing a distributed computing techniques over a cluster of computers. The analysis results of the optimum heat transfer coefficients are found to agree well with the measured data.