Thesis: M.B.A., Massachusetts Institute of Technology, Sloan School of Management, 2019, In conjunction with the Leaders for Global Operations Program at MIT / Thesis: S.M., Massachusetts Institute of Technology, Department of Civil and Environmental Engineering, 2019, In conjunction with the Leaders for Global Operations Program at MIT / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 72-76). / Printed circuit boards (PCBs) are core components of virtually every modern electronic device, from smartphones to servers. Accordingly, printed circuit board assembly (PCBA) has become core to Flex, a leading electronics manufacturing services (EMS) company. As the EMS industry continues to automate the PCBA process, it captures more data and creates opportunities to leverage this data and to generate value through analytics. One such promising opportunity is using defect prediction to improve downstream yields. For instance, x-ray inspection, which mostly detects solder defects, has a yield of about 97% for one of Flex's automated PCBA lines, and an improvement even to just 98% would create significant cost savings. Given this opportunity, this project aims to use the new data captured by the first steps in the automated PCBA process to predict solder defects that are usually identified during inspection, several days after the board begins the PCBA process. Specifically, the proposed boosted trees model uses data on 20,000 solder pads to predict whether an entire board will fail a downstream x-ray test. Other, more granular models are also studied, as well as other predictive models such as logistic regression and convolutional neural network models. The model is able to identify defective PCBs with an AUC of 0.74 and improve x-ray inspection yields from 97% to 98%, using one PCBA line at Flex as a case study. A second additional use case would reduce the number of x-ray inspection machines needed. Furthermore, a pilot implementation demonstrated that the model works well enough to enable these savings to be realized in practice. At the site where this study was conducted, these two use cases are estimated to produce significant savings over the seven-year useful life of the PCBA machinery. Since Flex has over 1,500 PCBA sites, the results of this case study suggest that there is potential to scale these analytics and related savings across the company. / by Andrew Wallace Foster. / M.B.A. / S.M. / M.B.A. Massachusetts Institute of Technology, Sloan School of Management / S.M. Massachusetts Institute of Technology, Department of Civil and Environmental Engineering

Thesis: M.B.A., Massachusetts Institute of Technology, Sloan School of Management, 2019, In conjunction with the Leaders for Global Operations Program at MIT / Thesis: S.M., Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 2019, In conjunction with the Leaders for Global Operations Program at MIT / Page 75 blank. Cataloged from PDF version of thesis. / Includes bibliographical references (page 67). / This project focuses on the creation of a novel tool to detect and flag potential errors within Amgen's capacity management forecast data, in an automated manner using statistical analysis, artificial intelligence and machine learning. User interaction allows the tool to learn from experience, improving over time. While the tool created here focuses on a specific set of Amgen's data, the framework, approach and techniques offered herein can more broadly be applied to detect anomalies and errors in other sets of data from across industries and functions. By detecting errors in Amgen's data, the tool improves data robustness and forecasts, which drive decisions, actions and ultimately results. Flagging and correcting this data allows for overcoming errors, which would otherwise damage the accurate allocation of Amgen's human resources to activities in the drug pipeline, ultimately hampering Amgen's ability to develop drugs for patients efficiently. A user interface (UI) dashboard evaluates the tool's performance, tracking the number of errors correctly identified, the accuracy rate, and the estimated business impact. To date the tool has identified 893 corrected errors with a 99.2% accuracy rate and an estimated business impact of $77.798M optimized resources. Using the paradigm of intelligent augmentation (IA), this tool empowers employees by focusing their attention and saving them time. The tool handles the human-impossible task of sifting through thousands of lines and hundreds of thousands of data points. The human user then makes decisions and takes action based on the tool provided output. / by David Goldberg. / M.B.A. / S.M. / M.B.A. Massachusetts Institute of Technology, Sloan School of Management / S.M. Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science

Thesis: M.B.A., Massachusetts Institute of Technology, Sloan School of Management, 2019, In conjunction with the Leaders for Global Operations Program at MIT / Thesis: S.M., Massachusetts Institute of Technology, Department of Civil and Environmental Engineering, 2019, In conjunction with the Leaders for Global Operations Program at MIT / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 105-107). / This thesis suggests a method to characterize congestion alarms triggered by the Emergency Department (ED) at Massachusetts General Hospital, attempts to predict the incidence of these alarms using logistic regression, and proposes operational recommendations for the mitigation of congestion events termed Code Help. In order to characterize Code Help alarms, we begin by identifying a set of relevant operational features that allow us to describe them objectively and proceed to clustering Code Help observations using k-means. We regress these features on binary variables indicating Code Help incidence to predict, at 7AM in the morning, whether or not Code Help will occur on a given day. Based on this analysis, we suggest a set of recommendations to operationalize a more effective response to Code Help. Our characterization uncovers three main classes of Code Help: those exhibiting a high level of ED arrivals in the hour preceding the alarm with a relatively low operational utilization of inpatient beds, those exhibiting a low level of ED arrivals in the hour preceding the alarm with a relatively high operational utilization of inpatient beds, and those exhibiting high arrivals and utilization. The logistic regression identifies two statistically significant predictive features: ED Census at 7 AM and the Number of Boarders in the ED at 7 AM, scaled against same time of day and day-of-week observations. Moreover, we identify discharge orders and outpatient pharmacy orders as early discharge indicators that can be used to prioritize Medicine patients in terms of their readiness to be discharged when Code Help is called. / by Christian Adib. / M.B.A. / S.M. / M.B.A. Massachusetts Institute of Technology, Sloan School of Management / S.M. Massachusetts Institute of Technology, Department of Civil and Environmental Engineering

Thesis: M.B.A., Massachusetts Institute of Technology, Sloan School of Management, 2019, In conjunction with the Leaders for Global Operations Program at MIT / Thesis: S.M., Massachusetts Institute of Technology, Department of Civil and Environmental Engineering, 2019, In conjunction with the Leaders for Global Operations Program at MIT / Cataloged from PDF version of thesis. / Includes bibliographical references (page 67). / As consumer preferences shift towards online shopping and utilizing their homes as fitting rooms, traditional brick and mortar retailers are faced with the challenge to adapt. Many retailers are experiencing a growing number of returned merchandize, many of which cannot be easily resold to consumers due to various supply chain challenges. This thesis explores the opportunities to improve the consumer returns process and presents methods for modeling the supply chain process for reverse logistics in the retail industry derived from case studies. The model then allows for hypothesis testing. By changing parameters in the model, this thesis further explores the scenarios in which the supply chain process may be improved to increase margin and decrease cost. The primary recommendations include specific modifications to the current reverse supply chain flow, enabling new channels that improve speed and margin, as well as developing the decision tool further for better accuracy and integration into the supply chain. / by Nina Hughes. / M.B.A. / S.M. / M.B.A. Massachusetts Institute of Technology, Sloan School of Management / S.M. Massachusetts Institute of Technology, Department of Civil and Environmental Engineering

Thesis: M.B.A., Massachusetts Institute of Technology, Sloan School of Management, 2019, In conjunction with the Leaders for Global Operations Program at MIT / Thesis: S.M., Massachusetts Institute of Technology, Department of Civil and Environmental Engineering, 2019, In conjunction with the Leaders for Global Operations Program at MIT / Cataloged from PDF version of thesis. / Includes bibliographical references (page 51). / The primary goal of this project is to link critical product quality attributes in biological medicines with patient outcomes as evidenced by clinical trial data. This project adopts a novel methodology of applying statistical and machine learning models on a dataset combining data on clinical outcomes, patient population characteristics, and product attributes. Understanding the mechanisms driving patient outcomes will help us develop risk-based control strategies for product attributes. Ultimately, our project can lead to better patient usability, greater flexibility in the manufacturing process, and improved competitive positioning in the market. We focus on de-risking one product attribute, high molecular weight species (HMW) and one patient outcome, development of anti-drug antibody (ADA) response in patients given consistent regulatory authority focus on the role of HMW in determining patient safety associated with the use of biological medicines. Although there is some correlation between ADA response and HMW exposure in patients, our findings strongly suggest the the importance of factors beyond HMW exposure, such as patient conditions and exposure to other attributes, in driving patient outcomes. As such, we have provided evidence to suggest that HMW exposure does not by itself increase the likelihood of a patient developing an ADA response. / by Shalini Shreekant Singh. / M.B.A. / S.M. / M.B.A. Massachusetts Institute of Technology, Sloan School of Management / S.M. Massachusetts Institute of Technology, Department of Civil and Environmental Engineering

Thesis: M.B.A., Massachusetts Institute of Technology, Sloan School of Management, 2019, In conjunction with the Leaders for Global Operations Program at MIT / Thesis: S.M., Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 2019, In conjunction with the Leaders for Global Operations Program at MIT / Cataloged from PDF version of thesis. / Includes bibliographical references (pages 70-71). / As the popularity of online retail expands, world-class electronic commerce (e-commerce) businesses are increasingly adopting collaborative robotics and Internet of Things (IoT) technologies to enhance fulfillment efficiency and operational advantage. E-commerce giants like Alibaba and Amazon are known to have smart warehouses staffed by both machines and human operators. The robotics systems specialize in transporting and maneuvering heavy shelves of goods to and from operators. Operators are left to higher-level cognitive tasks needed to process goods such as identification and complex manipulation of individual objects. Achieving high system throughput in these systems require harmonized interaction between humans and machines. The robotics systems must minimize time that operators are waiting for new work (idle time) and operators need to minimize time processing items (takt time). Over time, these systems will naturally generate extensive amounts of data. Our research provides insights into both using this data to design a machine-learning (ML) model of takt time, as well as exploring methods of interpreting insights from such a model. We start by presenting our iterative approach to developing a ML model that predicts the average takt of a group of operators at hourly intervals. Our final XGBoost model reached an out-of-sample performance of 4.01% mean absolute percent error (MAPE) using over 250,000 hours of historic data across multiple warehouses around the world. Our research will share methods to cross-examine and interpret the relationships learned by the model for business value. This can allow organizations to effectively quantify system trade-offs as well as identify root-causes of takt performance deviations. Finally, we will discuss the implications of our empirical findings. / by Anjian Wu. / M.B.A. / S.M. / M.B.A. Massachusetts Institute of Technology, Sloan School of Management / S.M. Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science

Thesis: M.B.A., Massachusetts Institute of Technology, Sloan School of Management, in conjunction with the Leaders for Global Operations Program at MIT, May, 2020 / Thesis: S.M., Massachusetts Institute of Technology, Department of Civil and Environmental Engineering, in conjunction with the Leaders for Global Operations Program at MIT, May, 2020 / Cataloged from the official PDF of thesis. / Includes bibliographical references (pages 56-59). / As the U.S. electric grid continues to experience an increase in the penetration of distributed energy resources (DER), electric utilities are evaluating new approaches for utilizing DER to help cost-effectively maintain grid resilience and reliability. One such approach is to create a transactive market for DER to provide grid services, which are services required to support reliable grid operation. Though work has been done to understand some of the technical mechanisms of this type of market, gaps still exist in understanding the value and market opportunity of ancillary services at the distribution level. One type of ancillary service - reactive power - is of particular interest because of the theoretic ability to source from existing assets on the distribution network. This paper aims to build understanding of the value of procuring reactive power from one of these assets: Combined Heat and Power (CHP) systems. The value of procuring reactive power from a CHP system will be quantified by 1) characterizing CHP systems' capacity to produce and absorb reactive power, 2) assessing the annual cost of procuring reactive power from CHP systems, and 3) comparing the CHP system technical capability and cost to the utility's conventional solution: capacitor banks. This study finds that, while there are promising scenarios in which CHP systems can technically and economically provide reactive power in a comparable or slightly advantaged manner to capacitor banks, the overall statistics for the 29 CHP systems analyzed in the New York fleet do not conclusively demonstrate an advantage that supports outright replacement of capacitor banks. Further assessment of CHP systems as a complementary source of reactive power and site-specific case studies are recommended to inform the next step in the decision making process for determining whether this path should be pursued as a source of reactive power. / by Monica Harnoto. / M.B.A. / S.M. / M.B.A. Massachusetts Institute of Technology, Sloan School of Management / S.M. Massachusetts Institute of Technology, Department of Civil and Environmental Engineering

Thesis: M.B.A., Massachusetts Institute of Technology, Sloan School of Management, in conjunction with the Leaders for Global Operations Program at MIT, May, 2020 / Thesis: S.M., Massachusetts Institute of Technology, Department of Civil and Environmental Engineering, in conjunction with the Leaders for Global Operations Program at MIT, May, 2020 / Cataloged from the official PDF of thesis. / Includes bibliographical references (pages 66-68). / Pharmaceutical companies traditionally manufactured drugs in-house, and have only recently been increasingly outsourcing production to contract manufacturing organizations (CMOs). In this work, we use inventory modeling to explore the trade-off between the attractive purchase price and the hidden costs of outsourcing for two active pharmaceutical ingredient (API) supply chains at AstraZeneca (AZ). We assess the inventory levels recommended by a base stock policy with deterministic purchase order lead times at each contracted stage of the supply chains. The single-echelon calculations reveal that safety stock levels are not systematically inflated at individual stages. The current inventory costs and service levels vary widely across products studied, as performance is hard to track over long periods of time and different inventory types. However, we demonstrate with a multi-echelon inventory optimization that a fully integrated API supply chain would yield savings compared to a purely external chain. Today, AZ's organizational processes allow it to partially coordinate with CMOs and capture up to 60 % of the value left on the table by not being able to optimize the full chain due to outsourcing. We propose using cost premium frontiers to prioritize further improvements at strategic outsourced nodes and align incentives. Partnering with CMOs to shorten lead times and increase flexibility is set to become a key advantage in a changing pharmaceutical environment with exacerbated volatility. / by Audrey Bazerghi. / M.B.A. / S.M. / M.B.A. Massachusetts Institute of Technology, Sloan School of Management / S.M. Massachusetts Institute of Technology, Department of Civil and Environmental Engineering

Thesis: M.B.A., Massachusetts Institute of Technology, Sloan School of Management, in conjunction with the Leaders for Global Operations Program at MIT, May, 2020 / Thesis: S.M., Massachusetts Institute of Technology, Sloan School of Management, Operations Research Center, in conjunction with the Leaders for Global Operations Program at MIT, May, 2020 / Cataloged from the official PDF of thesis. / Includes bibliographical references (pages 73-74). / Retail manufacturers often expedite inbound freight shipments from contract manufacturing bases to their distribution centers in destination markets at high cost to improve service levels to their wholesale partners and retail arm. The current process around these decisions has yielded lower than anticipated improvements to service level. This thesis (1) reframes the goal of expediting inbound freight in quantitative, measurable terms that more directly impact the business outcomes, (2) develops an optimization model to select a set of freight shipments to expedite and best improve service, and (3) uses the optimization model to estimate potential improvement magnitudes with strategic changes. / by Colin Alex McIntyre. / M.B.A. / S.M. / M.B.A. Massachusetts Institute of Technology, Sloan School of Management / S.M. Massachusetts Institute of Technology, Sloan School of Management, Operations Research Center

Thesis: M.B.A., Massachusetts Institute of Technology, Sloan School of Management, in conjunction with the Leaders for Global Operations Program at MIT, May, 2020 / Thesis: S.M., Massachusetts Institute of Technology, Department of Civil and Environmental Engineering, in conjunction with the Leaders for Global Operations Program at MIT, May, 2020 / Cataloged from the official PDF of thesis. / Includes bibliographical references (pages 74-75). / Johnson & Johnson Vision (JJV) is a leading manufacturer of contact lenses offering a variety of vision correction products. They have a strong commitment to innovation--defining new product categories and improving existing ones--to better patient outcomes, but this poses a challenge for machine production capacity and long-term planning. As medical devices, contact lenses must first be qualified and validated to run on lines. Additionally, capital equipment has a multi-year lead time from design and order to onsite implementation. Taken together, these constraints add great complexity to JJV's supply chain. The JJV team has a strong capability in aggregate demand planning but determining the right product mix can be difficult as consumer tastes change, new products are uncertain, and the future cannot be predicted. / This complexity faced in manufacturing contact lenses along with forecasting product mix highlights the importance of having the right capacity at the right time in maintaining high customer service standards. Strategic capacity planning, looking out 3-5 years, is currently viewed deterministically, meaning that a single number is decided upon for each product line for both demand and supply. An aggregate production plan using the various machines is then built around this deterministic forecast. This thesis attempts to address strategic capacity planning through quantification of risk relative to a plan of record using various techniques, specifically looking at risk factors as inputs to demand and production planning. The focus of this research is to probabilistically model the risk in manufacturing line variability as an input to production capability and planning at JJV. / A proof-of-concept was developed for each technique with a focus of Monte Carlo simulation as a model for uncertainty in production, which was then expanded to all other lines where appropriate conditions were met. Under the analysis assumptions, 75% of the fleet across both manufacturing sites was able to be analyzed and able to be identified as high risk in their current plan for both over- and under-production, which helps to inform capital needs. Ultimately, the results of this project intend to smooth out Long Range Financial Planning and challenge existing forecasting methods and metrics. / by Margaret E. Neff. / M.B.A. / S.M. / M.B.A. Massachusetts Institute of Technology, Sloan School of Management / S.M. Massachusetts Institute of Technology, Department of Civil and Environmental Engineering