Predictive analysis of installation and operational qualification issues vs. process severity events

Abstract

As the retail industry grows more popular, ABCD, a world-class electronic commerce (e-commerce) business, is increasingly building new Fulfillment Centers (FCs) to support this rapid demand growth. It is integral for ABCD to validate the installation quality and functionality of Material Handling Equipment (MHE) in these newly built FCs so operations can avoid errors. To achieve this objective, ABCD introduced the Installation and Operational Qualification (IOQ) process in late 2014. While the IOQ process reduces early operational failures, it does not completely eliminate them. Inadequate IOQ and tighter installation timelines are leading to degraded installation quality, resulting in operational issues and costs for ABCD. As the FC network continues to grow, there is a need to improve installation quality to reduce early operational issues and enhance the FC start-up experience.

This project is a part of the ABCD Operation Engineering teams' effort to improve the existing IOQ process and the FC start-up experience. This initiative consists of three main phases. The first phase - the research phase - is dedicated to understanding the current processes and problem statement. It also includes a study of available data sources to discover failure patterns across different FCs. The second phase involves developing analytical frameworks and machine-learning models to uncover the most problematic equipment in the FC. The third phase focuses on evaluating the effectiveness of the current IOQ process based on Phase 1 and 2 findings, and identifying opportunities to better the process. The thesis summarizes the outcomes from all of these phases. The project focuses on improving IOQ coverage, efficiently reprioritizing the testing schedule, introducing threshold metric for installation quality, and exploring predictive and preventative maintenance opportunities.

This thesis also includes recommendations for refining the data-gathering process to improve future model outcomes. The ultimate goal is to improve FC installation quality and enhance the IOQ process to eliminate start-up issues. The approach taken and the recommendations proposed seek to approximate the ideal state as closely as possible. Incremental adoption of these recommendations will help deliver better-installed FCs, reduce early operational issues, improve start-up experiences, and strengthen ABCD's infrastructure.