Doctoral Dissertations

Orcid ID

https://orcid.org/0000-0003-4028-1575

Date of Award

5-2025

Degree Type

Dissertation

Degree Name

Doctor of Philosophy

Major

Energy Science and Engineering

Major Professor

Mingzhou Jin

Committee Members

John Kobza, Tony Schmitz, Sachin Nimbalkar

Abstract

This research aims to guide energy management improvements using smart manufacturing concepts to achieve net-zero 2050 goals. The global manufacturing industry is undergoing a fourth industrial revolution. At the same time, the United States has committed to a goal of net-zero greenhouse gas emissions by 2050. Navigating these two concurrent revolutions in manufacturing is the motivation for this research.

This multi-part dissertation focuses on understanding smart manufacturing for energy management through a review of the applicability of smart manufacturing maturity models, an application of a machine learning framework for energy prediction to show a real-world case of smart manufacturing for energy management, and an analysis of the Better Plants program to determine policy recommendations for achieving net-zero 2050.

Smart manufacturing is a priority research area for many organizations, but there is no clear understanding of the terms and dimensions across industries, including energy management. This gap was addressed by reviewing twenty-four maturity models across ten categories of dimensions. The focus areas for companies and areas with weak guidance were identified.

Energy consumption predictions are typically performed using basic linear regression based on production or often naïve approximations. Better data fidelity and analytic methods are envisioned with the fourth industrial revolution. No existing study was identified that predicts energy consumption at the facility level using machine learning or at a time scale shorter than a year. A framework with machine learning and an automated data collection framework for energy prediction at the facility level is proposed to address this gap. The framework is used to test increasingly complex machine learning models on daily, weekly, and monthly datasets. The complex models can be utilized for increased prediction accuracy. Still, significant challenges need to be addressed to automate the process, including reliable and quality data and much initial human labor.

The BP program was reviewed and accomplishments were outlined according to year and subsector. The BP program has been crucial for emission reductions, but acceleration is required to meet further goals. To maximize the effectiveness of the BP program, the US government should consider adopting additional incentives and combine them with disincentives.

Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS