Menu
Our client reached out to us because their supply chain operations were not optimal, leading to inefficiencies in production and increased costs. The goal of our mandate was to optimize the manufacturing process given the client’s specific constraints for production volume, working capital, inventory buffers, and more. This is a classic example of the minimum-cost routing problem – which involves finding the lowest-cost solution for producing a fixed quantity of a product in a manufacturing chain. What makes the problem particularly challenging, however, is the limitations in data completeness and quality, due to sourcing issues among the client’s subcontractors. The solution must therefore be adaptive to these issues and remain useful to the client even when data is missing or inaccurate.
Our custom-made tool gives the client the ability to optimize their manufacturing process, thereby saving the client potentially millions of dollars in production costs over the next several years of operation. In addition, the tool supplies key decision-making insights as the competition landscape rapidly evolves and changes must be made in phasing out or consolidating specific factory sites.
The graph modeling and optimization routine are packaged together in a web application, which allows the user to adjust high-level settings, reconfigure factory sites and routes within the graph, add or remove additional constraints, run sensitivity analyses and the optimization routine, and interpret the results
Our team started by developing a thorough understanding of the client’s products and supply chain. Through a close collaboration with the client lasting over several weeks, we built a clear roadmap for the mathematical modeling and integration of our proposed solution. This research gave us insight into the necessary data inputs, possible pitfalls, and the necessary deliverables central to the client’s needs.
Supply chain problems are best modelled mathematically using directed graphs, wherein factories are represented by nodes of the graph and transport routes as edges. Optimization of the supply chain graph may be run under a variety of user-specified constraints. Given the particulars of the problem, a custom optimization routine needed to be devised, building upon the classic network simplex algorithm. This approach ensures the best solution is found in an efficient time.
Upon completion of our mission, we scheduled an in-person meeting with the client to hand over the final application and train their employees with the operation of the tool and the interpretation of its outputs. Our relationship with the client continued after the final delivery to supply application hosting services and technical support.
Python, R, CSS, HTML
Docker
Azure Cloud
Github CI/CD
The implementation team faced several challenges during the implementation:
Newsletter
