As renewable energy continues to play a larger role in China's electricity grid, obstacles for increasing grid flexibility and reliability need to be addressed. Improved grid flexibility could be achieved by institutional changes, adding flexible supply, improving the reliability and efficiency of the transmission sector, or enhancing demand side flexibility. This dissertation focuses on the potential of demand response (DR) to provide flexibility in times of extreme need in a future with large penetration of renewables. It develops two different DR models that can be used to estimate DR impact for commercial buildings at the building level and at the national level, respectively.

Using a national level model, this dissertation characterizes the impact of DR events across the commercial sector to provide flexibility across four netload cases in 2030: hours of high and low netload, and hours of high down and up ramping. I describe the methodology to actuate DR by managing the buildings' HVAC system and explain the expected impact statistics. I find that for a 2030 basecase netload scenario with installed renewable capacity of about 1,200 GW, DR from the commercial sector can decrease peak netload between 7 and 12 GW, saving the system between 7 and 12 billion dollars in deferred capacity expansion. Extending demand response to the highest 1% of netload hours in the year, an average of 3 to 4 GW decrease in needed capacity during those hours is possible. This decrease is equivalent to a 276 to 377 GWh reduction in demand, and additional operation savings of between 21 to 28 million dollars. Calculating the cost and benefits of increasing netload at the lowest hours requires further analysis on the potential generation portfolio in 2030 that is likely to benefit from an increased operational baseload. In addition, DR can provide flexibility to alleviate extreme ramping down and up throughout the year. Actuating DR on the highest 1% ramping hours of the year can provide, on average, between 6 to 11 GW per hour and -14 to -27 GW per hour for the extreme down and up ramping hours of the year.