Modelagem estocástica de um sistema produtivo com Factory Physics e simulação de Monte Carlo

Abstract

This work develops and applies a stochastic approach grounded in Factory Physics and Queueing Theory, implemented through Monte Carlo simulation, with the objective of quantifying and decomposing lead time in critical resources of a discrete production system. The research was conducted in an industrial plastic injection molding facility, focusing on two real injection molding machines, Machines 10 and 11, classified as bottleneck resources in the process. Operational variables such as cycle times, setup times, failures, repairs, and defect rates were statistically characterized and incorporated into a mathematical model calibrated using real data from the manufacturing environment. The simulation enabled the estimation of probabilistic distributions for performance metrics such as lead time, throughput, and work in process, as well as the evaluation of improvement scenarios related to availability, reliability, and setup reduction. The results indicated that variability has a multiplicative and nonlinear effect on production performance, being especially sensitive to the level of resource utilization. It is concluded that the integration of statistical modeling, applied physics, and production engineering provides a consistent quantitative basis for diagnosing industrial systems, prioritizing improvement actions, and supporting decision-making.