Poster
Client Context
Cisco Systems is a multinational technology conglomerate that delivers software-defined networking, cloud, and security solutions, such as switching, routing, wireless, and data center products. Cisco has been unable to fulfill the demand for its products due to a combination of factors, including the global semiconductor shortage and the increased demand for networking devices during the COVID-19 pandemic. The Component Supply Planners (CSP) introduced a process to allocate limited component supply across different products produced in various locations, where each CSP manages a subset of the components. To aid allocation decisions, the Data Science & Innovation team created a decision-support tool, namely, the Minimum Viable Product Tool (MVPT).
Project Objective
The current allocation process has several shortcomings, including: (i) it does not enable CSPs to coordinate the planning of the components that feed into the same products; (ii) the MVPT generates a plan for one week, without considering future demand forecasts or component availabilities; (iii) the allocation plan generated by the MVPT requires many hours of manual adjustments. Opportunities to aid the component allocation decision-making include: (i) building a new centralized and automated system, and (ii) a comprehensive, interactive Tableau dashboard to reveal performance metrics and graphical insights on how allocation decisions for each component affect the overall system.
Design Strategy
The design strategy consists of (i) a multi-period, multi-objective component allocation decision-support tool that assists CSPs to make informed decisions, and (ii) an interactive visualization tool that allows CSPs to evaluate the allocation plans. The tool creates a 26-week production plan and component allocation plan based on the chosen objectives. Additionally, the team developed a stochastic optimization model. The stochastic extension introduces the potential for the Data Science & Innovation team to develop a comprehensive plan that can mitigate risk from uncertainty. Validation for the model (and its extensions) was done using multiple samples of randomly selected products (from multiple time periods) from historical data.
Deliverables
Deliverables to improve the component allocation process include a new decision-support tool with an easy-to-use graphical user interface and an interactive Tableau dashboard. The model, namely, the Component Allocation Optimization Model (CAOM) behind the tool is a linear, multi-objective, multi-period optimization model solved by Gurobi. The CAOM recommends component allocation (coordinated across multiple products and production sites) with consideration for demand forecasts and supply availability over a 26-week planning horizon. The allocation plans generated by the CAOM along with the associated performance metrics increase the quality of and the visibility into the allocation process.
Project Information
Student Team
Anjana Anandkumar, Udisha Bhattacharyya, Grace Gilpatric, Katie Landers, Kat Pospichel, Briana Sims, Tan Tanthien, Anna (Tu) Vu
Faculty Advisor
Faculty Evaluator
