Renewable electricity generation in the U.S. continues to grow, stimulated by favorable public policies and falling costs for renewables. However, the transition to a low-carbon electricity grid will not happen overnight. We are in the midst of a multi-decade transition toward a cleaner grid. During this transition, fossil fuel electricity generators will operate alongside wind farms and fleets of solar photovoltaic panels. A key challenge in this transition is intermittency of renewables – generation from wind turbines and solar panels can be highly variable, with weather-driven fluctuations that may be difficult to forecast. As penetration of intermittent renewables increases, conventional generators must startup and shut-down more frequently to offset fluctuations in renewable generation.

In a recent IJIO publication, Market Dynamics and Investment in the Electricity Sector, Joseph Cullen and Stanley Reynolds develop and analyze a dynamic competition model of investment incentives and operation costs when conventional and renewable electricity generators co-exist. A key innovation of their analysis is incorporation of output change costs (called ramping costs) for natural gas generators. Incorporation of ramping costs leads to significant changes in predicted volatility of wholesale energy prices, profits of conventional generators, and investment incentives for conventional generators.

Data from the Electric Reliability Council of Texas system is used to estimate and parametrize the model. Texas has the highest wind turbine capacity of all U.S. states and has experienced rapid recent growth in solar panel capacity. The parameterized model is the basis for counterfactual policy experiments that evaluate the impact of renewable investment subsidies and carbon taxes. The policy experiments are designed to achieve CO2 reduction targets of 20%, 40%, and 60%, relative to a baseline case. The cost of achieving a 60% CO2 reduction under a carbon tax is estimated to be $1.1 billion/year, versus $1.4 billion/year under a renewable investment subsidy; a $300 million/year gap in CO2 reduction cost between the two policies. These costs should be weighed against climate benefits of reduced CO2 emissions.

A common alternative approach for evaluating environmental policies in the electricity sector is to use a merit order model. Under merit order, electricity generators are dispatched in ascending order of their marginal generation cost. Cullen and Reynolds show that their dynamic competition model yields dramatically different results than a merit order model. Wholesale electricity prices are predicted to be more volatile with the dynamic competition model compared to the merit order model. In addition, the dynamic competition model predicts much more costly CO2 reduction compared to the merit order model. For example, 60% CO2 reduction under a carbon tax and a renewable investment subsidy are predicted to be 40% more costly with a dynamic competition model compared to a merit order model. This in turn implies a smaller predicted gap between CO2 reduction costs for carbon tax and investment subsidy policies under a merit order model (a $209 million/year gap for the merit order model versus a $300 million/year gap for the dynamic model).

Why are policies more costly in the dynamic model than in the merit order model? Several forces are at work. Wind turbines are most productive at night and solar panels are most productive mid-day. There are lower equilibrium energy prices and more renewable curtailment for these hours in the dynamic model compared to the merit order model. As renewable penetration increases, combined cycle units have more frequent ramps, leading to higher ramping costs in the dynamic model. More frequent ramping of combined cycle units changes investment, shifting the plant mix toward units with high operating costs and away from lower cost combined cycle units in the dynamic model.