Poster
Client Context
Our project team partnered with Kinaxis to work with one of their clients, referred to as Client X for discretionary purposes, a major player in the biopharmaceutical industry. Client X specifically deals with the production of plasma-derived therapeutic products. Client X operates a global supply chain, with production facilities in Germany, Switzerland, United States, and Australia. The production process is primarily split into two main steps: fractionation, in which plasma is deconstructed into its constituent proteins (called pastes), and bulk-fill-pack, in which the pastes are mixed with stabilizing solutions, packed, and shipped to end consumers. Paste supplied by the fractionation side is insufficient to fill intermediate demands generated by the bulk-fill-pack steps. Our project specifically focused on improving the planning process for the fractionation step of the supply chain.
Project Objective
Fractionation is the bottleneck in Client X's supply chain - there is significantly more demand for pastes than the fractionation facilities can fulfill. Currently, fractionation planning is done manually by individual planners at each facility. This is due in part to the fact that, unlike later steps in the manufacturing system, fractionation is a disassembly process in which a single input is divided into multiple outputs. As a result, fractionation cannot be modeled in Kinaxis’s proprietary software, Rapid Response. This all leads to a fragile production planning process that is late on production targets and long on planning time. Due to the fact that the basis of this problem revolved around the efficiency and efficacy of certain planning decisions being made, our team felt that this would be an ideal opportunity to develop a Gurobi-based optimization model that could be integrated into the back end of a software tool.
Design Strategy
Our solution was to build a mixed integer program to generate fractionation production plans. Prior to model construction, a sizable amount of data processing is done to convert the format of the initial data to one suitable for input to the optimization model. After the model is constructed and optimized, we then convert the decisions generated into a form that matches the client’s existing production plans. This allows for easy implementation of the plan as well as direct comparison with current production plans. Production plans generated by the model were validated using an algorithmic checker. This script checked the core constraints of the system, including inventory balance and resource capacity constraints.
Deliverables
Kinaxis received a software tool that generates production plans for the fractionation stage of the manufacturing process. The optimization model developed for the project is integrated into the backend of this tool. Extensive documentation and a user guide were delivered alongside the software. This tool accepts data from Kinaxis’ proprietary supply chain planning software and outputs a production plan in a format that can be re-imported into their system. Once implemented, the tool will reduce the lateness of fractionated pastes.
Project Information
Student Team
Nosrat Chowdhury, Brice Edelman, Osman Ghandour, Aniruddh Hari, Yash Lunagaria, Alice Pagoto, Maria Yagnye
Faculty Advisor
Faculty Evaluator
