SCRO / Journées de l'optimisation

HEC Montréal, 29-31 mai 2023


HEC Montréal, 29 — 31 mai 2023

Horaire Auteurs Mon horaire

OPMII Operations Planning and Management II

29 mai 2023 15h30 – 17h10

Salle: Saine Marketing (vert)

Présidée par Benedikt Kern

4 présentations

  • 15h30 - 15h55

    Operator assignment to non-identical parallel machines in a real-life environment

    • Nazanin Haghjoo, prés., 181152

    The aim of this study is to propose a model and solution for the operator assignment problem, which involves assigning a single operator to parallel machines. This problem is specific to non-identical parallel CNC machines, and the objective is to find the optimal allocation of the operator to some defined tasks. The tasks include supervising CNC machines as regular tasks and performing manual operations as parallel tasks.
    Constraint Programming (CP) is a practical tool for modeling scheduling problems. To address this problem, we utilize CP to create a new CP model that considers both regular and parallel tasks. This study is defined in a real-life environment based on the need of our industrial partner, JITbase company. JITbase faces the challenge of assigning a single operator to these tasks and we provide a solution that allocate a single operator to the machines while minimizes the makespan, or the total time to complete all tasks. We tested the model on a real dataset provided by JITbase and provide sensitivity analyses to help improve their operations.

  • 15h55 - 16h20

    Safety Stock Estimation Based on Forecasted Demand Distribution using Recurrent Mixture Density Network

    • Mahya Seyedan, prés., Concordia Institute for Information Systems Engineering, Concordia University, Montreal, Canada
    • Fereshteh Mafakheri, École nationale d’administration publique (ENAP), Université du Québec, Montréal, Canada
    • Chun Wang, Concordia Institute for Information Systems Engineering, Concordia University, Montreal, Canada

    Calculating safety stock is essential for any business to maintain adequate inventory levels and ensure customer satisfaction. Accurate demand forecasting is critical for reliable estimation of safety stock and proper inventory management. Over the past few decades, a significant amount of research has focused on improving demand forecasting. While some studies have focused on estimating the average and variation of demand with point forecasting, the prediction of demand distribution has received limited attention. However, one critical factor affecting safety stock calculation is demand distribution which helps companies predict the likelihood of certain demand levels occurring and adjust safety stock accordingly. Recently, it has been shown that recurrent mixture density networks (RMDNs) are effective in modeling complex and nonlinear patterns in time series data by predicting the distribution of data. Thus, the main contribution of this study is to propose a new framework for demand forecasting using RMDN to calculate safety stock accordingly. In this approach, retailers can make informed decisions about different aspects of the supply chain by modeling the probability density function of the target variables. A comparison of the proposed approach to other well-known time series forecasting models is conducted using data from retailers. The study's findings suggest that RMDN is a superior alternative to accurately forecasting demand distribution in complex and dynamic environments. This practical solution has the potential to help retailers predict the probability of experiencing stockouts or overstock situations and estimate the amount of safety stock more accurately which effectively reduces inventory management costs and enhances customer satisfaction.

  • 16h20 - 16h45

    Sustainable Production Planning in the Plastic Industry: An Innovative Timed Route Approach Considering Recycling Opportunities

    • RAZIEH LARIZADEH, prés., University of British Columbia
    • Babak Tosarkani, University of British Columbia

    The diverse application and affordability of plastic-based products have resulted in their widespread consumption in recent years. This has given rise to global concern about the disposal of large amounts of non-biodegradable plastic waste discarded in the environment. To address this issue, recycling plastic waste is a promising solution. Therefore, policymakers are encouraging manufacturers to comply with recycling programs and replace virgin raw materials with recycled materials. However, the process of collecting, segregating, and recycling plastic waste is time-consuming and requires significant investments. In this study, a timed route multi-objective mixed-integer linear programming (MILP) model is proposed to support recycling programs in sustainable production planning in the plastic industry. The findings of this study have significant implications for the manufacturers and contribute to the integration of sustainability practices in production planning.
    keywords: Production Planning, Plastic Recycling, Sustainability

  • 16h45 - 17h10

    Operational production planning in smart factories

    • Benedikt Kern, prés., Concordia University
    • Masoumeh Kazemi Zanjani, Concordia University

    In this study, we propose a Benders Decomposition (BD) algorithm to solve an operational-level production planning problem in a smart factory manufacturing setting. More specifically, we consider modular-structured products produced in a two-echelon manufacturing network comprising the final production and sub-assembling facilities, which are convertible factories with modular production lines. We also assume that machines are equipped with smart sensors to track their deterioration level. A mixed-integer programming model is formulated that seeks the optimal decisions on production quantities, maintenance actions, and workforce training with the objective of minimizing total costs. Considering that the size of this model grows exponentially, we apply a BD approach to efficiently solve it for real-size problem instances. Several acceleration techniques, such as valid inequalities and heuristics, are embedded into the BD framework to ensure efficiency. Extensive computational results show the improvement of the algorithms’ performance.
    Keywords - Operational production planning, Industry 4.0, Smart factory, Mixed-integer programming