How Renault reaches 1.5% optimality gaps on their packaging logistics optimization problems

renault-packaging-logistics

Renault Group is a French multinational automobile manufacturer, and one of the largest automakers in the world. Its Operations Research department is in charge of solving the various optimization problems arising in the company. The most frequently encountered issues are related to Supply Chain Management: production scheduling, transportation optimization, network design, item packing in trucks, layout optimization of workshops, etc. The Renault OR team has been using LocalSolver to solve many of these problems since 2016. In this case study, we present Renault’s packaging logistics optimization problem.

Renault’s packaging logistics optimization problem

The French carmaker uses packaging to transport its parts from suppliers to plants. The suppliers need enough packaging to ship the parts. Therefore, Renault must ensure enough packaging in circulation in its supply chain at all times. The shipment of parts between suppliers and factories is supervised by Renault’s Material Requirements Planning (MRP) systems, like SAP.

Renault also reuses most of its packaging, which plants must send back to suppliers after receiving the parts. The transportation of parts and empty packaging generates complex packaging flows inside Renault Group’s supply network, as illustrated below:

  • Orange arrows represent the parts’ shipping between the suppliers and the plants;
  • Red arrows represent the return of empty packaging to the suppliers;
  • Blue arrows represent the transport of empty packaging that must be cleaned or repaired;
  • Gray arrows represent the transport of the packaging to be discarded.
Packaging and return flows between suppliers, plants, and warehouses

Efficiently handling empty packaging is crucial for Renault. Indeed, empty packaging management represents several million euros per year in terms of transportation and renewal costs. To reduce these costs, Renault’s OR team sought LocalSolver’s help. As a result, return flows of empty packaging are now operated by the Packaging Management System (PMS), a custom software solution developed by Renault and based on LocalSolver as an optimization engine.

Mathematical model and results

The mathematical model to be solved by LocalSolver is characterized by the following decisions, constraints, and objectives:

Input data

  • 1,400 suppliers in Europe
  • 40 plants and cross docks
  • 30 types of standard packaging

Decisions

For each type of empty packaging, each day, each supplier, and each plant:

  • Number of pallets transported between the supplier and the plant

Constraints

  • Consistency between stocks and shipments
  • Minimum quantities to allow shipping
  • Maximum inflows and outflows per factory
  • Stock balancing of transfer stations to avoid activity peaks

Objectives

  • Minimize packaging shortages (immediate and in stock)
  • Minimize the number of shipments
  • Minimize the total distance traveled

This model enables Renault’s OR team to quickly solve even the largest instances of their packaging optimization problem. Indeed, a typical instance of this problem comprises

  • 45,000 integer variables with large domain ranges, typically between 0 and 10,000
  • 100,000 constraints
  • 4 ordered objectives

and is solved to near-optimality within minutes. Indeed, LocalSolver reaches an average optimality gap below 1.5% with a time limit of five minutes.

Why LocalSolver?

After a benchmark against CPLEX, Renault‘s Operations Research team chose LocalSolver to solve this packaging logistics optimization problem. The simple and powerful modeling language offered on top of LocalSolver, LSP, allowed the team to develop the mathematical model quickly. Then, the innovative primal heuristics inside LocalSolver, combined with state-of-the-art exact methods, delivered proven near-optimal solutions in minutes of running times. The dedicated and responsive support provided by LocalSolver experts was also instrumental in moving the project forward quickly.

We use LocalSolver for several optimization problems, including scheduling door manufacturing in car factories and optimizing empty packaging return flows. What we appreciate most is the incomparable ease of modeling provided by LocalSolver mathematical formalism. Besides, exchanging with the LocalSolver team is always extremely fruitful and highly pleasant.

Alain Nguyen Head of Operations Research, Renault Group

Please look at our customers’ webpage to discover more case studies in various industries. Are you interested in trying it out? Get free trial licenses here. In the meantime, feel free to contact us; we will be glad to exchange your optimization problems.

Share