Statistical Leanring Applied to Climate-Water-Energy Nexus for Quantification and Characterization of Severe Drought Risks in Adaptation Models
Restricted (Penn State Only)
- Author:
- Nguyen, Duc
- Graduate Program:
- Energy and Mineral Engineering (MS)
- Degree:
- Master of Science
- Document Type:
- Master Thesis
- Date of Defense:
- June 14, 2023
- Committee Members:
- Mort D Webster, Thesis Advisor/Co-Advisor
Renee Obringer, Committee Member
Jeremy Gernand, Program Head/Chair
Karen Fisher-Vanden, Committee Member - Keywords:
- Climate
Energy
water
power systems
statistical learning - Abstract:
- As the global average temperature has been increasing in the past few decades, the impacts of climate change have become more transparent in different sectors. In the climate-water-energy nexus, climate change impacts have been observed to cause various disruptions to the operations of power systems. In particular, as the frequency and intensity of severe heat waves and droughts increases in the United States, the power systems flexibility to meet electricity demand at a spatial and temporal level decreases. The majority of power generation units require cooling water (to maintain thermal efficiency accounts for 91% of the United States total power generation in 2021. Higher ambient air and water temperature from severe drought events reduces cooling water availability for these thermoelectric generators resulting in an increased number of outages from significant reductions in power reserves. In order to mitigate and prevent the impacts from climate change, efforts in decarbonizing the power sector are necessary. Retirements of conventional power generating units for renewable technologies are often considered an optimal mean to meet the carbon emission reduction targets. However, this transition further challenge to the power grid to meet electricity demand due to the high intermittency of renewable generating units. In this thesis, I propose a statistical modeling framework for characterizing the compounding effects of high water temperature and planned retirement of conventional power generation capacity on the operations of Western Electric Coordinating Council (WECC) power system. This statistical framework utilizes both unsupervised and supervised machine learning methods to detect common failure patterns from a wide range of climate forcing scenarios and identifying key interactions between water temperature and retirements that lead to high impact scenarios. Building from the insights from the statistical framework, I test several adaption strategies by mitigating water temperature impacts, retirement impacts, and adding transmission capacity for an example scenario to observe the systems response in order to develop more detailed adaptation model that can be applied to a wider range of scenarios facing a common failure pattern.