Skip to main content

Variability of Renewable Energy Sources

Wind and solar energy are referred to as variable generation sources because their electricity production varies based on the availability of wind and sun. However, they are not the only source of variation in a power system. The demand for electricity, or load, also varies, and the power system was designed to handle that uncertainty. Short-term changes in load (over seconds or minutes) are generally small and caused by random events that change demand in different directions. Over longer periods (several hours), changes in load tend to be more predictable. For example, there is a daily pattern of morning load pickup and evening load drop-off highly correlated with human endeavors. The key difference is that load variations are better understood than wind and solar variations. Uncertainty arises when levels and variabilities of power supply and demand must be sequentially matched.

Some aspects of renewable energy variation are easily predicted. For example, the electricity production of an individual wind turbine is highly variable. But the aggregate variability of multiple turbines at a single site is significantly less variable. The aggregation of multiple wind generation sites over a large geographic area results in even less variability. Harnessing the "law of large numbers," variability smoothing over large areas, yields enhanced prediction. Variability also decreases as the timescale decreases. The variability of large-scale wind power over seconds or minutes is generally small. Over several hours, however, it can be great.

Two charts of minute-to-minute variability of wind production over a 9-hour period. The first graph, which represents a wind farm of 200 turbines, shows a comparatively smooth production curve with small peaks and valleys. It has a standard deviation of 14.89 and standard deviation divided by mean of 0.126. The second graph, which represents a wind farm of 15 turbines, shows a jagged production curve with pronounced peaks and valleys. It has a standard deviation of 1.21 and a standard deviation divided by mean of 0.184.

A 9-hour comparison of second-to-second variability of wind production between a wind plant with 15 wind turbines and a wind plant with 200 wind turbines

Similarly, some aspects of solar variability are predictable (for example, sunrise and sunset). Other aspects, such as intermittent cloud cover, are much less so. However, the same reduction in variability is observed for the aggregation of solar photovoltaic plants over a broad geographic area.

Graph of normalized solar power output for one PV plant, two PV plants, six PV plants, 25 PV plants, and all PV in Southern California. As more PV plants are added, overall production increases and the production curve smooths.

A comparison of solar power production variability in Southern California.

All types of variability must be managed by the electric power system operator. With low penetrations of variable generation, the related impact and response are small because the wind and solar variability is much less than the load variability. At high penetrations, however, the renewable variability may be more challenging to respond to.

NREL is addressing variability issues on the transmission system through its work on:

For additional information, see Long-Term Wind Power Variability.