Imagine a system consisting only of wind turbines and natural gas combined cycle generators. The initial configuration starts with all natural gas, then wind is added to reduce CO2 emissions. The solid blue line in the figure below shows the resulting cost as a function of grid emissions. Cost increases sharply as wind penetration increases (to low emissions) due to curtailment. At certain times with low load, wind is generating too much power. At that time all the natural gas generators are shut down, and wind needs to be curtailed (shut down some turbines). Also note that we cannot reach zero emissions at any cost with just wind and natural gas generators.
The analysis uses hourly wind and load data for the PJM system in 2012. Cost data is from the US Department of Energy’s Energy Information Agency’s Annual Energy Outlook (AEO) for 2015 based on new 2020 installations. The red square reference (100% CO2 emissions) is the cost of an all natural gas system assuming that the system needs 15% reserves. The black X is existing PJM grid, 25% dirtier than all natural gas and at today’s retail price slightly higher than AEO’s cost estimate for the cost of a new 2020 all natural gas installation. If we had perfect storage (zero cost, 100% efficient, infinite size) we would get the blue dashed line. The line slopes slightly up to the right because with AEO2015 costs, the discounted capital cost of the wind is slightly greater than the cost of natural gas saved. But real storage is expensive. More realistic storage (red line) is based on Bath County Pumped Hydro Storage, $200/kWh and 80% efficient. Storage reduces system costs modestly but only at very high wind penetration. The green square is a system consisting of nuclear base load (AOE2015 costs) plus natural gas to manage diurnal variations and reserves. The green diamond is a system consisting of nuclear plus domestic hot water storage (like the French system) to manage diurnal variation plus natural gas reserves. These cost/performance tradeoffs are being documented in more detail in a series of papers starting with an ASME conference in 2014.