CDE LIghtband has used a Demonstration of Energy & Efficiency Developments (DEED) grant from the American Public Power Association to conduct and refine an analysis of the effects of solar power penetration on the Tennessee public power utility’s operations.
The DEED project had three goals: develop simple models for forecasting solar adoptions at the local level, compare the modeled solar output with actual solar power output, and evaluate the forecasts of those models on utility revenues and load curves.
The first step in the process was to simulate solar generation. The majority of that work was done by using the National Renewable Energy Laboratory’s System Advisor Model (SAM), modeling software that NREL makes available for free as an aid to decision-makers in the renewable energy industry.
NREL’s software enables planners to model how much power a particular solar array can produce while factoring in variables, such as latitude, seasonal weather, cloud cover, and even the effects of shade from trees at different times of day.
The second step asked the “big question, ‘how accurate is the model for Clarksville?’” Jared Combs, business intelligence analyst at CDE Lightband and author of the DEED report, said.
To do that the CDE Lightband team modeled the output of a 6-kilowatt (kW) photovoltaic solar array in Clarksville using the NREL software. The 6-kW size was chosen because it is the average size of a residential solar array, according to the Lawrence Berkeley National Laboratory.
CDE Lightband also built an actual 6-kW solar array and collected the data on its output and compared those results of the simulated and actual data. “The biggest finding,” Combs said, was that using average weather data as opposed to actual weather data yielded “potentially wildly inaccurate” results. “The simulated results do not align with the actual results unless you use specific year weather data,” he said.
The comparisons show “the importance of using hourly data, modeled on a specific year weather file when conducting a utility solar economic analysis,” Combs wrote in the DEED report. “This means that to analyze demand charge and time-of-use effects specific weather files must be used in the SAM model.”
One of the driving factors behind the analysis made possible by the DEED grant was to come up with a more accurate assessment of the effect of solar penetration and solar output on utility revenues.
Even if a customer is getting most of their power from solar panels on their roof, the utility is still bearing a fixed cost in personnel and equipment such as transformers. “To be fair to customers, we need to find out how much our customers with solar arrays affect our revenues,” Combs said.
Combs ran the modeled solar output data against CDE’s 2018 rates and demand intervals to determine revenue effects. A positive revenue effect is the result of solar output reducing utility wholesale demand charges by more than the amount lost on utility energy revenues less power costs.
Taking reductions in demand charges paid to wholesale power suppliers into account, total utility revenue effects from solar generation were 42 percent lower than when demand charges were not factored in, the analysis showed.
“One of the key findings,” Combs said, was that the revenue affect of a solar array is affected by its orientation. A Southwest facing array, because it catches solar energy in the evening as utility demand is rising, can pay for itself because it can reduce demand charges at a time when the utility is often selling power to customers at a loss, Combs said.
“Understanding of hourly economic interactions between utility wholesale costs and solar array generation might inform strategic planning and various models for distributed energy resource (DER) market participation,” Combs wrote in the DEED report, adding that utilities developing community solar programs “might consider the effect of solar arrays on their wholesale demand charges when calculating investment metrics and pricing models.”
Combs also recommended that “rate adjustments intended to ensure that solar array owners cover their portion of the fixed costs of electric distribution might more accurately consider the value of customer generated solar power.”
Combs cautioned, however, that conditions such as solar irradiance and the relationships between retail and wholesale rate structures often vary hourly and are different from one utility to the next.
One of the benefits of understanding the revenue effects of a single solar array is that that data can be applied as a multiplier against anticipated solar arrays to estimate future net revenue effects of solar adoption on a utility system, Combs said.
As part of the DEED grant, CDE Lightband also analyzed census data to gauge where rooftop solar arrays would be installed. CDE looked at several census traits, such as house size, education levels, and house ages. CDE was aided in its analysis by analytics firm DNV GL. The analysis found that in Tennessee the most predictive characteristic was income. Higher income correlated with higher solar penetration.
Combs cautioned, however, that those results might not hold true for utilities in other areas. It also said it is “advisable that utilities consult expert analysts before taking business action on the results of any PV adoption models or the models intended for economic analysis.”
The cost of building the solar array was about $25,000. CDE Lightband submitted and received a DEED grant for about $10,000. Expenses for data collection, analysis and reporting were assumed by CDE Lightband and Tennessee Valley Authority, which provided the Clarksville utility with assistance, and were not included in the budget.
The DEED grant gave CDE the freedom to collect and thoroughly analyze the solar data in a way that would have been hard to justify otherwise, Combs said. “It was good, not just because it was good for us, but because of the exponential benefit of being able to share the results with other utilities,” he said.