High penetration challenges and strategies: case studies from Hawaii and Ontario

A particular challenge to steady power production from solar energy plants is steep power ramps caused by the shadows of fast moving, low-level cumulus or stratocumulus clouds. Observations of power ramps at small photovoltaic (PV) facilities provide guidance in estimating the range of power ramps that can be expected at larger utility-scale projects. However, the proportional scaling of power and project size may not provide the best estimate of the effect of cloud shadowing on systems with high penetration PV plant power output. Thus, to quantify the potential effects of fast-moving clouds on large-scale PV plants or markets with deep penetration of solar power, a detailed analysis of the surface light environment is required. 

The temporal and spatial spectrum of the cloud-light environment can vary considerably depending upon the local, regional, and synoptic-scale climate. For example, within the subtropics, trade wind cumulus clouds dominate, and the cloud fraction distribution peaks at 0.5. In the mid-latitudes, clouds fraction tends toward a bi-modal distribution--most of the time, it is either clear or cloudy (although there are seasonal and regional exceptions).

To accurately depict the long-term time-space cloud environment, high quality, high frequency, and high density measurements of irradiance are a necessity. As such observational networks are virtually non-existent, sophisticated statistical and modeling techniques are required to create the needed data. The methods described here focus less on matching the absolute insolation at a site at a given time, but rather attempt to match the relative changes in insolation over time. Thus the statistical characteristics (i.e. ramp distributions, power spectral densities) are comparable. Here, we present methodologies used to create representative irradiance and power output data sets for the two distinct climate regimes referenced above, more specifically, Hawaii and Ontario, Canada.