California, USA — As states across the country set higher and more aggressive renewable energy goals, our nation’s grid operators and utility planners have responded with increasing concern about the impact that those resources may have on electric grid systems. Unlike the fossil fuel generators around which our existing energy infrastructure has been planned and built, solar generation fuel is inherently variable. Both the daily movements of the sun and short-term changes in weather conditions both have real, direct impact on a PV system’s electric output.
This mismatch between the renewable energy resources being integrated onto the grid and the models built to plan for new generation has significant implications. Perhaps most concerning to those in favor of renewable energy build-out is that the traditional models call for significant levels of new natural gas generation development to shape the variable renewable resources being brought online – an outcome that would directly undermine our nation’s renewable energy progress.
But if we are able to fundamentally shift the planning paradigm to include variable renewables as a non-negotiable piece of our energy mix rather than a risky outlier, the prospect of integrating high levels of solar onto the grid becomes eminently manageable.
The conventional view of grid operations presumes that load creates grid instability and that generation is dispatched in a manner to address that instability. From this view it is easy to see how renewables might raise a number of concerns from the people responsible for keeping our lights on and electricity available.
However, this view fails to recognize that all generation presents at least the potential for some type of grid instability or impact. So that means that solar generation is disproportionately penalized simply because solar is the last resource to be added to the planning “stack.” In other words, solar generation is subject to cumulative historic generator impact or, to add another acronym to the already the acronym-heavy world of grid operation, CHGI.
By way of example, in the California market, certain categories of power — imports, nuclear, Qualifying Facility (QF), and some types of hydroelectric generation — are actually so inflexible that they have deleterious grid impacts. Similarly, even the most flexible gas resources have limited operating ranges and can be subject to unexpected outages. Despite their limitations, these generator resources do not raise emphatic concerns about grid impacts like their renewable counterparts.
For the most part, these generator limitations have been operationally addressed through balancing area coordination of the entire generation fleet. The related costs have been rolled into general concepts of grid reliability, and then attributed, both conceptually and financially, to load. Solar generation, because it is the “last in the stack,” risks being both conceptually and actually charged with the impact of not only its intrinsic solar variability, but also the CHGI that has traditionally been assessed to load.
Existing scheduling, dispatch and settlement protocols further exacerbate the CHGI issue. Because they were developed based on the requirements of non-renewable generation such as gas, nuclear and coal, the existing protocols are not well suited for solar generation. The historic paradigm assumes baseload generators, such as nuclear and coal, run all the time, serving the minimum level of expected load. Additional load and fluctuations of that additional load, both expected and unexpected, are usually “followed” by more flexible resources such as gas.
From this paradigm protocols such as hourly scheduling were established. Hourly scheduling makes sense for scheduling traditional generation, such as nuclear, coal or gas, because it is not designed to change output quickly or because it is not generally susceptible to sub-hourly fuel fluctuations. Solar variability, on the other hand, can occur on a sub-hourly basis, and thus solar generation is poorly served by traditional generation protocols such as hourly scheduling.
Furthermore, a general lack of understanding regarding how solar generation will perform on a balancing-area-wide basis is causing solar variability impacts to be overestimated. The research from Thomas E. Hoff and Richard Perez as well as many others, compellingly demonstrates the importance of solar generator geographic diversity. As demonstrated in their geographic diversity research, a dispersed portfolio of solar electricity systems spread across the balancing area “smoothes” the impact of weather–based solar variability to the point that it can be cost-effectively managed. These two experts will be presenting their findings as part of Vote Solar’s free Get Some Sun webinar series: PV Output Variability: the Sheep in Wolf’s Clothing.
Taken as a whole, the concerns identified above hold considerable potential for chilling the continued development of solar energy and for driving unnecessary development of fossil fuel generation to purportedly back up the solar. To ensure the continued progress of solar energy development, the following should be proactively pursued:
- establishment of solar generation integration requirements that are fact-based, informed, well vetted, and accurate;
- development and implementation of elegant, efficient, and cost-effective solutions to minimize solar variability, and
- embodiment of fair and equitable principles of cost allocation in all related policy decisions.
By separating the impacts of solar variability due to the daily movement of the sun (called DMV – diurnal movement variability) from the weather change impacts (WBV – weather based variability), grid planners can begin to address their intermittency concerns. The former is predictable and known, such that it can be addressed ex-ante, meaning that its grid impacts can be effectively eliminated in a least cost manner. The latter, WBV, however, is more likely to require ex-post solutions, such as requiring grid operators to consider solar generation on a fleet wide basis, rather than assessing performance on each individual unit. Thus, while WBV cannot be entirely avoided, it can certainly be significantly minimized.
Again by way of example, the current California Public Utilities Commission (CPUC) long-term planning proceeding does not distinguish DMV and WBV from each other. This lack of separation could potentially cause the CPUC’s integration model to overestimate the amount of new gas resources needed to firm, follow or back-up solar generation.
Once isolated, DMV can be addressed ex-ante through policies that allow DMV to be shaped in a manner that best neutralizes this type of variability (DMV Shaping). Take, for instance, the case where the most desirable solar generation shape is one that ramps consistent with load. To the extent that solar generation ramps up faster than load, compensated curtailment can be explored. Likewise, to the extent solar generation ramps down faster than load, acquisition of clean shaping energy, such as hydroelectric generation or energy storage, can be explored.
FERC’s Integration of Variable Energy Resources Notice of Proposed Rulemaking (NOPR) is currently considering three elements, including sub-hourly scheduling, that relate directly to making DMV Shaping a reality. When combined with advocacy directed at balancing area authorities and state commissions, the NOPR holds considerable promise for advancing this issue.
Once DMV is addressed, policies and planning should focus on minimizing the impacts of WBV. These weather-driven impacts can be largely addressed by incorporating the smoothing impacts of geographic diversity into grid planning and operations and advancing policies that support such dispersal.
If a cloud passed over a specific solar generator, the momentary change in output of that solar generator can and should be smoothed by other solar generators that are located where the sun is still shining. It sounds optimistic, but analysis of geographically dispersed PV system output indicates that this smooth effect does indeed effectively minimize the impact, on a fleet-wide basis, of the WBV.
Nevertheless, depending on the manner in which any particular balancing authority operates, on a meter-by-meter basis, individual generators could be charged for WBV that, in effect, had only a minuscule impact on the grid. Exploration of protocols to address or avoid this possibility should be pursued with balancing area authorities and, to the extent necessary, before FERC and appropriate state commissions.
With DMV Shaping enabled and WBV smoothing of geographic diversity properly incorporated into grid operations and planning, that leaves a final area of advocacy to effectively manage solar variability: policies that address the costs related to any remaining WBV impacts.
To the extent that Residual WBV impacts are entirely attributable to solar generation, the costs should be appropriately allocated. On the other hand, careful attention must be given to recognizing the unaccounted for CHGI, and ensuring that solar generation, simply because it is the last in the stack, is not solely responsible for mitigating, and paying for, the impacts.