This study aims to develop a joint optimization model for production planning and preventive maintenance in a crude oil refinery. The purpose is to minimize total costs (production, storage, shortage and maintenance) while considering stochastic demand, a specified service level and equipment corrosion. It integrates a production optimization model with a hybrid maintenance model that uses Markov chains and a Cox model to simulate corrosion-related failures, ultimately seeking a cost-effective maintenance plan that ensures system reliability.
The approach combines two modeling techniques. Production planning is formulated as a stochastic optimization problem to minimize costs subject to capacity, storage and service-level constraints under random demand. For maintenance, a hybrid model is developed: a homogeneous Markov chain simulates the system's evolving operating conditions, while a Cox proportional hazards model estimates the time-dependent failure rate due to chemical corrosion. These models are integrated, and the combined framework is evaluated using a case study with simulated data, including a sensitivity analysis on the number of preventive interventions.
The results demonstrate a direct correlation between the production rate, corrosion phenomena and the frequency of equipment failures. This interdependence significantly impacts total costs. The integrated model successfully identifies an optimal preventive maintenance schedule that effectively balances system reliability and expenditure. Sensitivity analysis confirms that the number of preventive interventions is a crucial factor; a well-calibrated plan minimizes both unexpected breakdowns and overall costs, thereby limiting service interruptions while maintaining the specified service level.
This study's originality lies in its integrated modeling of production and maintenance, specifically addressing chemical corrosion in a refinery. It combines a stochastic production planning model with a novel hybrid maintenance approach, using Markov chains for operational states and a Cox model for corrosion-driven failure rates. This methodology provides a more realistic framework for managing interdependent production and degradation processes. The value is a practical decision-support tool that enables managers to jointly optimize production schedules and maintenance plans, effectively balancing cost, service level and equipment reliability in corrosive environments.
