Optimizing Computational Time for Numerical Simulation of Dust-Laden Flows through Queue Management in Multi-Server Environments

With increasing environmental impact of atmospheric pollution resulting from fine dust emissions of the industrial sector, optimization of industrial equipment through computer simulations is becoming more and more important. In this study we minimize computation time for simulating dust laden flows by optimizing parallelization processes using distributed server systems. Since the movement of each particle is calculated separately, parallelization is very efficient. We adapt a linear regression-based method to run operations efficiently, avoiding unnecessary parallelism and resulting in better load forecasting for queue systems. This approach drastically reduces processing time and enhances the predictive capabilities of these simulations in dynamic systems, increasing overall efficiency. It also enables us to run multiple simulations with different parameter sets quickly, thereby contributing to better environmental safety and equipment performance. In fact, integrity and security of distributed systems is also important, as data in local networks are not encrypted, but its integrity should be kept during processing.