Although some of the evidence presented in this chapter is experimental (i.e., based on incomplete measurements over limited time spans and limited climatic ranges), the results clearly state that: 1. We can predict the variability of solar resources given the speed of cloud structure that causes variability; ② We can fully estimate the variability of any solar power installation, from a single small system to a system of any spatial distribution and scale swarms, including extensive private solar farms.
In particular, we can be fairly certain that fluctuations of 20s are not a problem for solar power plants spread over 500m (even if the cloud velocity reaches 50km/h). Figure 6.10 compares the variability of a single location measured on a day of high variability in New York City with the city-wide electricity generation network.
Short-term fluctuations and slow rate of change of less than 20s will affect small individual systems, but this effect will be minimized when the system group covers an area of several square kilometers. At the system level, these fluctuations can (but rarely) lead to localized voltage disturbances and can cause the system to go off-grid. The best way to address these issues is at the interconnect hardware level, including the use of appropriate buffers to increase the electrical inertia of the system and eliminate the above risks. It’s like a car that, with the right suspension, can navigate rough roads without having to anticipate every bump that might come, and without thinking about the effects of those bumps.
For a cluster of distributed systems of several square kilometers served by a substation or a very large central power plant (hundreds of megawatts), fluctuations of the order of a few minutes over an area of this size are sufficient for our attention. Nonetheless, in the case of a cluster of distributed systems, these fluctuations should have minimal impact on generators with a wide range of utility. This level of buffering involves the interconnect buffers mentioned above and some degree of voltage and power regulation, including short-term storage lasting several minutes in large centralized arrays.
By feeding a large amount of power into the grid to “buy time” for the associated combined cycle gas turbines to ramp up and down, it currently accommodates ramp up times of close to 5 minutes. At the upper limit of this time-space scale, predicting the exact timing of this variability is extremely valuable, especially when the area is a separate grid (such as on an island). Continuing with the car analogy, the driver must concentrate fully on adjusting the power input to maintain the vehicle speed when going over a short incline.
0.5~lh or even longer passive time will affect the application system. In this case we need load tracking based on reserve (or worst case, contingencies) generation, load management and storage. Fortunately, the temporal and spatial scales involved (over half an hour and tens of kilometers) and the accuracy of the solar radiation (predicted) resources available at these scales allow us to manage these fluctuations effectively. Based on an upper bound on the time scale, a locally balanced system serving some regional utilities only needs to be concerned with fluctuations over lh.
In practice, by combining the results presented in this chapter with historical solar resource satellite detection data, utilities or developers can estimate any proposed thermal photovoltaic configuration that is concentrated or dispersed within 1km or more variability. The equation provides guidance for picking At over the range of interest (the smaller the range, the higher the frequency required). Since current satellite sounding irradiance models produce data at frequencies close to 1 min and 1 km resolution, any variability beyond the 1 km range can be inferred from the satellite data time series. In addition, Hoff (2011) proposed and patented a method to infer variability at any temporal and spatial scale from a known reference point (eg, 1km/1min), thereby extending the application of satellite data to a single system with a relevant time interval of the order of a few seconds.
Read more: What are the fundamentals of satellites?
