DescriptionOne in eight deaths globally is due to air pollution. Exposure to high concentrations of atmospheric fine particulate matter (PM2.5) has negative health consequences. Air quality models, such as the Community Multiscale Air Quality (CMAQ) model, are employed to evaluate effectiveness of air pollution abatement strategies partly designed to minimize PM2.5 exposure and protect human health. Energy production and consumption is the largest controllable source sector contributing to ambient PM2.5 mass. The highest electricity sector (energy subdivision) emissions occur on hot, stagnant summer days, when energy demand is highest and the atmosphere is most conducive to photochemical production and PM2.5 accumulation. Electricity generation is positively correlated with peak PM2.5 concentrations. CMAQ consistently underpredicts these peak values. Accurate prediction of peak pollutant concentrations is critical to develop strategies that protect human health. This dissertation works to reduce underprediction of peak PM2.5 concentrations from an energy sector and heat wave event perspective in the Northeast U.S., where PJM Interconnection governs the electricity transmission for 61 million people. Temporal representation of electricity sector emissions is improved in CMAQ during a heat wave, and episodic increases in peak PM2.5 at the surface and aloft are predicted. PJM EGU emissions, especially sulfate, impact not only the PJM region, but also outlying areas. Monitored and controlled peaking units, EGUs used during highest electricity demand, contribute up to 87% of maximum hourly PM2.5 concentrations. Urban areas experience the highest potential exposure (calculated as population-weighted concentrations (PWCs)) from peaking unit emissions, regardless of the location of predicted peak ambient concentrations. Peaking units contribute substantially to exposure potential on the worse air quality days, but are historically exempt from Federal air quality rules. Eight sensitivity experiments indicate CMAQ-predicted PM2.5 PWCs are most sensitive to uncertainty in onroad primary organic carbon emissions, while ambient PM2.5 concentrations are most sensitive to planetary boundary layer height. Model development strategies optimized to protect health may look different than traditional evaluation-focused strategies optimized to match annual averages in measured PM2.5 mass. This dissertation provides issues to consider for prioritizing model development to address peak air quality events that drive non-attainment and threaten human health.