Application of systems engineering principles in the analysis, modeling, and development of a DoD data processing system
Date
2023
Authors
Fenton, Kevin P., author
Simske, Steven J., advisor
Bradley, Thomas, committee member
Carlson, Ken, committee member
Atadero, Rebecca, committee member
Journal Title
Journal ISSN
Volume Title
Abstract
In support of over 1000 military installations worldwide, the Department of Defense (DoD) has procured contracts with thousands of vendors that supply the military with hazardous materials constituting billions of dollars of defense expenses in support of facility and asset maintenance. These materials are used for a variety of purposes ranging from weapon system maintenance to industrial and facility operations. In order to comply with environmental, health, and safety (EHS) regulations, the vendors are contractually obligated to provide Safety Data Sheets (SDSs) listing EHS concerns compliant with the requirements set forth by the United Nations Globally Harmonized System of Classification and Labeling of Chemicals (GHS). Each year chemical vendors provide over 100 thousand SDSs in a PDF or hard copy format. These SDSs are then entered manually by data stewards into the DoD centralized SDS repository – the Hazardous Material Management Information System (HMIRS). In addition, the majority of these SDS are also loaded separately by separate data stewards into downstream environmental compliance systems that support specific military branches. The association between the vendor-provided SDSs and the materials themselves was then lost until the material reaches an installation at which point personnel must select the SDS associated to the hazardous material within the service-specific hazardous material tracking system. This research applied systems engineering principles in the analysis, modeling, and development of a DoD data processing system that could be used to increase efficiency, reduce costs, and provide an automated solution not only to data entry reduction but in transitioning and modernizing the hazard communication and data transfer towards a standardized approach. Research for the processing system covered a spectrum of modern analytics and data extraction techniques including optical character recognition, artificial neural networks, and meta-algorithmic processes. Additionally, the research covered potential integration into existing DoD framework and optimization to solve many long-standing chemical management problems. While the long-term focus was for chemical manufacturers to provide SDS data in a standardized machine-encoded format, this system is designed to act as a transitionary tool to reduce manual data entry and costs of over $3 million each year while also enhancing system features to address other major obstacles in the hazard communication process. Complexities involved with the data processing of SDSs included multi-lingual translation needs, image and text recognition, periodic use of tables, and while SDSs are structured with 16 distinct sections – a general lack of standardization on how these sections were formatted. These complexities have been addressed using a patent-pending meta-algorithmic approach to produce higher data extraction yields than what an artificial neural network can produce alone while also providing SDS-specific data validation and calculation of SDS-derived data points. As the research progressed, this system functionality was communicated throughout the DoD and became part of a larger conceptual digital hazard communication transformation effort currently underway by the Office of the Secretary of Defense and the Defense Logistics Agency. This research led to five publications, a pending patent, an award for $280,000 for prototype development, and a project for the development of this system to be used as one of the potential systems in a larger DoD effort for full chemical disclosure and proactive management of not only hazardous chemicals but potentially all DoD-procured products.
Description
Rights Access
Subject
chemical management
occupational health
artificial intelligence
safety
meta-algorithmics