An MILP-based framework to the logistical problem of an industrial gas supply chain

BONINO, SERGIO G.; ZEBALLOS, LUIS J.; MOOLYA, AKASH; LAÍNEZ AGUIRRE, JOSÉ; PINTO, JOSÉ; GROSSMANN, IGNACIO; MÉNDEZ, CARLOS A.

Buenos Aires

Congreso; 11th World Congress of Chemical Engineering; 2023

The work presents an effective MILP-based heuristic framework to deal with the distribution scheduling of industrial size problems of gases. The paper introduces new procedures for obtaining efficient routes for product distribution taking into account an aggregated production scheduling. The addressed problem considers a predetermined time horizon, a set of production facilities, a set of clients with daily product demands, a trucks fleet to deliver gases, min/max production rates in plants as well as min/max tank levels in plants and customers. As part of the problem it is considered that trucks can make multiple visits per route, routes can have a duration of one, two or three days, every vehicle can make multiple routes per day. In addition, the trip duration depends on the distance to the customers taken into account in the trip, the order in which customers are visited, and the loading and unloading times of the products. An efficient mathematical discrete time MILP-formulation was developed to represent the production routing problem (MPRP) with multiple production plants with different features. Initially, the logistic model was tested considering a case study involving 50 clients, 2 production plants, 6 vehicles and 2 products. Good results were achieved when the approach was applied to a case study with a time horizon of three days and direct shipments to customers. Due to the large size of the industrial instance of the problem and the need to deal with an extended time horizon and several customers per route, an iterative procedure based on the efficient discrete time MILP-model was developed. The iterative technique contemplates one day in each iteration and the whole problem is faced by subsequently solving and combining multiple discrete time route generation MILP models. In each iteration, the model updates the inventory levels of plants and customers. It also creates a ranking of promising customers to visit on the next day and selects the most critical of them. Once these clients are selected, the possible routes to be used are enabled. The main decisions of the iterative framework are a) the amount of each product manufactured in each plant per day, b) the customers visited per day, c) the customers visited in the same trip, d) the order in which customers are visited, e) the trips that trucks make each day, f) the amount of products loaded to the trucks per trip, and g) the amount of product delivered to each customer included on the trips. The procedure was tested considering a real instance of supply chain of industrial gases with 5 production plants, approximately 500 monthly deliveries to 120 customers, 14 vehicles and a 30-day time horizon. The computational results demonstrate the effectiveness of the proposed approach. Thus, efficient solutions can be generated with a robust solution strategy and stable CPU times.