Detailed analysis of suitable rooftops suggests solar in some western cities may meet 100% of midday load within 20 years.
Utilities that aren’t planning for that could be in trouble. Fortunately, forward thinking analysts are doing the work and running the numbers.
“If you are a distribution planner you are now clutching your coffee mug and thinking, ‘Oh, no! What are all these people doing putting generation on my distribution network?’” Black & Veatch (B&V) Solar PV Director Ric O'Connell asked during a panel discussion at Solar Power International 2014. “The holy grail is merging system planning with new tools so they can plan their distribution networks in ways they never did before.”
Consultants like B&V and ICF International (ICFI) are using automated analysis of residential and commercial buildings to create new tools for utilities to better understand solar.
The Sacramento Municipal Utility District (SMUD) is one of the first utilities to take advantage of such analyses. Understanding what is coming in distributed generation will help SMUD make critical decisions about policy, rate redesign, and infrastructure investments, explained Project Manager Obadiah Bartholomy. “We are racing to get this data into our systems and our planning processes.”
Geographic Information Systems (GIS) analysis of rooftop potential, combined with datasets profiling demographics and energy consumption patterns, reveals patterns in solar growth, O’Connell explained.
Utilities like APS and TEP are entering the space as owners and developers, O’Connell said. But instead of just asking “who is going to buy my product?” like a solar company would, they are asking “where is the best place to put this generator on the system?”
Regression analysis showed one of the strongest correlations is between a home’s age and solar growth. New homes tend to have the potential for much bigger systems than older homes with lots of trees and shading.
The next step, O’Connell said is integrating feeder information into the tool and merging it with distribution planning so utilities can be proactive. Instead of “just raising their hand and saying ‘I want solar,’” he said, “they can know which roofs will give them the biggest bang for the buck.”
ICFI has long advised utilities on when and where to develop new generation, explained Associate Ankiit Saraf. But central station generation involves a relatively small player pool, 5-10 year planning and construction timelines, and a decision to install based primarily on load.
For distributed generation, planning and construction takes weeks to months. The players, including developers and installers and households and business owners, can be hundreds of thousands or even millions. And the decision to install is based not on load growth but on things like customers’ demand for solar, tax credit policies, and utility incentives.
“The most important question utilities are facing today is when and where distributed energy resources (DER) are going to pop up,” Saraf said. “Imprecise targeting is no longer an option.”
ICF’s SolarFlare modeling tool determines a building owner's “propensity to install,” Saraf explained. Like other tools, it incorporates economic factors and technical potential. But of its millions of details, “the driving factor is demographics,” he said. “When you layer in demographic factors like occupation, children, and wealth, and compare that with other customers, you can better predict which customers are more likely to go solar."
A key insight obtained from ICFI’s sophisticated algorithm is that “solar is not growing uniformly. It is patchy. There are clusters of high installation.”
Just the numbers
“Sun Number is a component to the models that others develop,” explained Chief Operating Officer David Herrmann. Its lidar-driven automated system, which has built a database of approximately 35 million U.S. buildings, can analyze the PV potential for each rooftop in a utility’s territory.
It reads and understands pitch, orientation, shading and roof size. It also incorporates whatever existing demographic, economic, and roof age data it can obtain. “With that, it is pretty easy to calculate the suitable system size and generation capability,” Herrmann said. At pennies per roof, it is cost effective.
Sun Number begins with the 100% insolation of clear sky conditions and then factors in real world conditions like shading from vegetation and tall buildings. Maps render hypothetical systems that can be specific all the way to the panel level.
“We then ground truth the data,” Herrmann said. A portion of the calculations are post-processed to actual measurements. After ground truthing against 800 measured buildings in the SMUD territory, “we did analysis of 250,000 buildings in the Sacramento area,” Herrmann said.
The data can also be aggregated by arbitrary boundaries like utility feeder networks.
SMUD digs into the data
Based on Sun Number data and B&V and ICFI analyses, SMUD is building planning scenarios through 2050. The development time for new generation and transmission-distribution infrastructure makes very long-term planning vital, Bartholomy explained.
Meeting California’s renewables and efficiency mandates and SMUD's own carbon reduction target will take the utility from approximately 50% natural gas to less than 10% by 2050. But that will leave it with “an energy gap from projected retail demand,” Bartholomy said. “We are now trying to understand how we will fill that gap and what role distributed generation will play.”
Modeling the technical, economic, and market potentials of distributed generation raises questions. In resource planning, the questions are about the need for balancing generation. In policy, they are about net energy metering (NEM) and rate design. In infrastructure, they are about system constraints.
Most utilities are not planning for distributed generation’s impact on the distribution system, Batholomy said. SMUD is starting to get answers to some of the hard questions.
The modeling shows increasing locational diversity could be important. There is a strong correlation between PV growth and adoption of electric vehicles, demand response, and energy efficiency, Batholomy said. That clustering can worsen the impacts of variability and voltage impacts on feeders and require upgrade investments.
The modeling also suggests “distributed solar could meet or exceed our aggregate daytime minimum load, based on a market uptake scenario, within the next 13 years to 20 years,” Bartholomy said.
Sun Number data shows that with panels “in every suitable space,” Herrmann confirmed, solar could meet 40% of Palo Alto’s overall annual load and almost 100% of Denver’s load.
That will likely attenuate solar’s growth curve because it will affect the value of solar, Batholomy said. “It will also impact what you can sustain in an NEM policy when you are paying someone at net retail for something that has a potentially negative value in a wholesale market.”