Parallel computing for simulating nanoscale temperature behaviour on laser glass interaction

Abstract

Glass materials are widely used in optical, optoelectronic and windshields. Glass is very brittle because the molecules are arranged closely and have strong bonds between them. To break up these bonds and produce fine quality glass, the glass needs to be cut by using a high temperature machine which uses the laser technique. Experimental method involves high expertise and costly and due to this an analytical method is preferred. The objective of this research is to develop a parallel algorithm to simulate the temperature behavior of laser glass cutting. Partial Difference Equation (PDE) mathematical model is used to represent the numerical simulation and this equation is discretised before solving by using multicore platform. The methodology used in the study is the parallel computing platform that is based on masters and workers concept. The parameters and initial values are input to the simulation which uses Alternating Group Explicit (AGE) BRIAN Three Dimensional method to solve the problem. In order to develop the parallel algorithm for the simulation, a sequential algorithm is developed initially and instructions that can be parallelized from this sequential algorithm are identified using the Microsoft Parallel Studio. Based on this, the parallel algorithm is developed using OpenMP language. The results from both sequential and parallel algorithms are recorded, analyzed and compared using Amdahl’s law. The results proved that the simulation using parallel computing algorithm is faster and cost effective. Furthermore, the time execution to simulate the program is reduced by 53% and the speed up is boosted up to 11%. The research illustrated that the analytical simulation using parallel algorithm is cheaper and faster and thus proves that parallel programs are best in simulating the temperature behavior of laser glass cutting.