Integrating renewables into the grid: An energy dispatch model
This paper presents a mixed-integer quadratic program that models electric grid behavior with very high wind and solar capacities to predict US grid CO2 emissions reductions and the overall cost of generation under different scenarios. Specifically, the energy dispatch model mimics a perfectly competitive electricity market by dispatching generators to minimize the hourly operating cost of electricity generation for representative two-week periods throughout the year, the results from which are annualized. Unlike previous approaches, the model accounts for realistic generator operation, by considering the efficiency corrections for part-load operation, ramping constraints, minimum operating rates and costs of cycling for different technologies, and also accounts for the effects of the uncertainty in electricity production from wind and solar. The Arizona - New Mexico - South Nevada grid (the AZ-NM-SNV grid) is used as an empirical starting point because it is representative in multiple dimensions of the US grid as a whole and its actual historical performance can be used to verify the model’s predictions. The model correctly predicts the emissions and overall cost of generation of the AZ-NM-SNV grid under current operating conditions (i.e. no renewables) within 8% maximum error and the dispatch composition by generation technology within 7% maximum absolute error. The model is used to predict the results on grid operation of very large expansions in wind and solar capacity independently, in combination with one another and in combination with a modest increase in nuclear capacity. Both wind and solar individually can moderately reduce grid CO2 emissions, but neither can practically reduce emissions beyond 50% relative to 2008 levels by themselves and would increase the overall cost of generation by 50% and 164% at 100% capacity, respectively. Wind and solar’s offsetting variability does not meaningfully extend this limit, but does enable some minor cost savings. A modest expansion in conventional nuclear capacity (roughly doubling current capacity) extends the maximum practical reduction in CO2 emissions to 72% while only increasing costs of generation by 63%. Ultimately, however, this limit only represents a 50% reduction in CO2 emissions relative to 1990 levels. Even very large expansions in wind and solar capacity and increasing conventional nuclear capacity to meet base-load demand cannot alone meet the commonly held target of an 80% reduction in grid CO2 emissions relative to 1990 levels.