To graduate successfully with a master’s degree in supply chain management from MIT, each student needs to work on a capstone/thesis project. A capstone project includes application of theoretical skills learnt into practical use by collaborating with companies or organizations. A thesis on the other hand, is conceptualizing new scenarios in the field of supply chain management and providing theoretical evidence to back it. A few select capstone/thesis projects are presented below. Check out more such projects in the CTL publication website (this year’s projects will be added over the summer).
A Natural Language Processing Approach to Improve Demand Forecasting in Long Supply Chains – William Teo
Information sharing is one of the established approaches to improve demand forecasting and to reduce the bullwhip effect, but it is infeasible to do so effectively in a long supply chain. A new Natural Language Processing (NLP)-based forecasting model, known as NEMO, was proposed in this thesis to forecast the demand of B2B commodities in long supply chains. NEMO uses modern NLP techniques to extract information from lengthy news articles to forecast the demand of such commodities without requiring downstream companies to share information. NEMO’s performance fared comparatively well to a statistical model and a gradient boosting model. NEMO can be used alongside other forecasting models to provide invaluable information about upcoming demand volatility.
Development and Application of an Immunization Network Design Optimization Model for UNICEF – Henrique Carretti & Yuto Hashimoto
This capstone project, explored the potential benefits of applying an optimization model in the design of vaccination networks. The developed model focuses on the last-mile vaccine distribution, where one-day outreach clinics are commonly used to provide immunization to remote areas. Using the case of Gambia, the developed modelling approach was validated to increase immunization access and generate meaningful insights.
How to Plan and Schedule for Profit: An Integrated Model and Application for Complex Factory Operations – Alessandro Silvestro
Optimization of factory operations is a fundamental aspect of any manufacturing company. However, planning and scheduling is a challenging and complex task, often very demanding in resources, investment and training. The research project relied on a large-scale MILP model for accurately evaluating, simulating and/or optimizing the internal manufacturing supply chain, in order to balance competing production/SCM cost goals while maximizing profit in the (short-term) planning horizon.
Modeling the Location-Routing Problem with Ancillary Modes – Amr Taiyeb & Kelly Doan
Combining vehicles and drones in the last mile transportation of small parcels can achieve significant improvements in cost and speed. This research project focused on the Location-Routing Problem with Ancillary Modes (LRPAM), which involved identifying the most strategic locations for distribution facilities, optimal trucks and drones’ delivery routes. However, the increased optimization complexity that comes from the integration of drones makes the exact optimization intractable for large realistically-sized operations. To overcome this challenge, a model was developed and solved using the Multiple Ant Colony Optimization algorithm. When applying the model to a real road network, with 200 customers and 5 candidate depot locations, the model confirmed a 24% saving in daily distribution costs from adding 3 drones to every delivery truck.
E-Commerce based closed loop supply chain for plastic recycling – Saikat Banerjee
Right now most of the plastic waste is dumped in the landfills and the ocean, and there is a dire cost to the environment because of that. This thesis aimed to contribute to science by finding a novel way to manage the take-back of plastic from the consumers to the recycling plants using the existing e-commerce reverse logistics network. This thesis focused on data from CPG companies about the sales of products contributing to the plastic waste, and considered location data of US counties, Amazon warehouses and Material Recovery Facilities (MRFs), to calculate an optimized route for plastic take-back. The research also assessed various costs and different sensitivity analyses based on scenarios planned for the take-back mechanism. As a result of this, a model and a cost equation were formulated to understand the feasibility of the process.
IoT-Based Inventory Tracking in the Pharmaceutical Industry – Andrew Kerr & Tony Orr
Inventory visibility has been a primary concern for corporate supply chains for decades. Utilizing inventory location and time data is particularly important for pharmaceutical companies and, until recently, archaic tracking processes created inaccuracies and mismanaged inventory for pharmaceutical manufacturers. However, recent Internet of Things (IoT) innovations provide potential solutions for pharmaceutical companies to manage and protect retail inventory levels while mitigating consumer risk and existing corporate financial waste streams. Through technology research, real-world experimentation, and cross-functional supply chain analyses, this capstone project proposed a Bluetooth IoT network infrastructure and business approach to meet traditional pharmaceutical visibility needs.
Blockchain Adoption: Aligning Incentives of SC Actors – Vijay Krishnan & Zhehao Yu
Blockchain Technology can revolutionize the supply chain industry, but the technology faces several impediments to an industry wide roll out. This capstone project proposed blockchain technology for Walmart’s small parcel supply chain. A system dynamics model was built to map goods, information, financial flows. The model was then validated on the basis of data collected and stakeholder’s feedback to simulate the scenario that maximized the benefit for the supply chain. Incentives that could drive blockchain adoption were identified using the model.
Decarbonizing Road Freight Transportation with Carbon Offsets – Abdelrahman Hefny & Cat Dame
Carbon offsets present a mechanism to leverage corporate sustainability commitments to accelerate investment in green transport systems through projects like fleet renewal programs. This study evaluated the feasibility of this approach from a financial and logistical perspective, analyzing the potential market size and emissions avoidance, quantifying the costs, and synthesizing best practices in fleet renewal programs. The analytical frameworks developed can be utilized to support the design and implementation of such a program that has the potential to drive significant impact in global carbon emissions reduction.
Unearthing the Hidden Treasure with Inventory Risk Pooling – Hari Kishan Sharma & Angelica Bojorquez
Rising inventory costs is an ongoing challenge for any firm. While inventory is necessary, it holds up capital, occupies premium storage space, and hazards higher obsolescence cost. An important portion of this inventory is held as safety stock to safeguard against risk of uncertainties. Risk pooling is known to reduce this uncertainty by centralizing the inventory and lowering associated cost without compromising service levels. However, it is not adequately leveraged.
Why can’t firms just reconfigure their network to unearth this value? How does one evaluate whether implementing pooling would save cost for them? Through this capstone project, the students developed a simple decision support system that can help firms evaluate the potential of introducing risk pooling. The research demonstrates how introducing risk pooling could significantly reduce supply chain costs for a leading retailer in Mexico, without affecting service levels.
Omnichannel Retail For A Seamless Grocery Shopping Experience – Wassim Aouad & Nikhil Ganapathi
With the rising adoption of e-commerce and online shopping, many retailers are facing the challenge of transitioning across channels to offer a seamless customer experience. One way of addressing this challenge consists of leveraging omnichannel retailing. The sponsor company of this project, a large US grocery retailer, has been operating in a multichannel environment by utilizing distinct networks for its online and offline channels. The objective of this project was to develop an omnichannel distribution model by leveraging the existing network infrastructure of the company. A mixed integer linear program was formulated to determine the omnichannel network model, and multiple scenarios were simulated to highlight the robustness of the model as well as the potential savings that can be realized.
Intermittent Demand Forecasting for Inventory Control: The Impact of Temporal and Cross-sectional Aggregation – Ngan Ngoc Chau
Managing intermittent demand is a challenging operation in many industries since this type of demand is difficult to forecast. This challenge makes it hard to estimate inventory levels and thus affects service levels. The purpose of this study was to examine the impact of multiple levels of data aggregation on forecasting intermittent demand, and subsequently, on inventory control performance. In particular, this thesis proposed a procedure that integrated lead-time and customer heterogeneity into forecasting using temporal and cross-sectional aggregation. Using data from a real-world setting and simulation, the analysis revealed that when high service levels were important for the company operations, the forecasting approach using temporal aggregation that incorporated lead-time information yielded a higher level of inventory efficiency in terms of both the holding cost and the realized service level. It appeared that when forecasts using temporal aggregation were augmented with information about customer behavior, their purchase patterns might be a helpful consideration for enhancing inventory performance.