Photovoltaics

A software company from California is making some remarkable inroads in predictive modelling for the solar industry. We look at how it is using weather variables, such as the effect of snow on solar generation and predictive data to make solar energy output and O&M more cost effective.

The larger the solar project, the greater the need to maximise a system’s production. This means siting judiciously, monitoring closely, and knowing the best time to plan O&M downtime.

For developers: risk management

For solar developers, software from California’s Clean Power Research can help in choosing a particular site, by providing the data on how efficiently irradiance can be translated into energy production, based on thorough environmental modelling of potential energy generation.

Even wind speeds are accounted for, because wind can impact solar performance, due to the cooling effect of a brisk breeze on panel efficiency.

“SolarAnywhere Data provides the necessary historical information to accurately assess energy risks,” says Skip Dise, SolarAnywhere Product Manager, at Clean Power Research. “It’s a database of solar irradiance, wind and temperature measurements spanning a 15-year time period that provides the user feedback on the expected energy at a site, down to hourly granularity.”

“Historical irradiance data is constantly updated based on satellite sources. From all of these data points, the user can make some determinations about how much solar energy has been available in the past at a particular site. This determination is then used to project what future production and year-to-year variability can be expected, helping to reduce overall project risk.”

Project owners: maximising production

Knowing why your plant is not performing as expected is crucial for owners of systems. If your production dropped precipitously last month, you need to know whether this is due to failures within the system, that will need to be addressed, or was simply due to a run of unusually cloudy days.

Cosmos Corbin is the software designer who founded Deck Monitoring. He recommends a fairly granular level of monitoring, but not for every component. Most of DECK Monitoring’s customers opt for monitoring at the inverter or the combiner level.

“You’re producing energy, that energy is worth money, and depending on your monitoring choices you may have a blind spot in your ability to see if your system is producing is as much as it could be,” says Corbin. “So the more granular you make the monitoring the smaller that blind spot becomes.”

His firm recently began offering Clean Power Research’s SolarAnywhere as an add-on to their system monitoring - for the aspect of performance that basic system monitoring cannot measure: the weather. He describes the need for this as “pretty critical.”

“If you don’t have environmental data then there’s a lot that you don’t know,” Corbin explains. “It’s about this blind spot. If you look at just the performance of a site that’s performing at 50%; if you don’t have weather data, you don’t know if that’s because it’s really cloudy, or if it’s more of a problem with one of the components.”

Of course, a weather station on site can provide this information, but is a more expensive option. He recommends that - even if a customer does have a weather station on site - using the SolarAnywhere software as a system check, because it can provide the calibration to ensure that it is accurate.  “You can basically see if the weather station is operating properly with a secondary data source,” he points out.

O&M is where monitoring software becomes crucial for solar project owners. “There’s soiling, shading, inverter failures, there’s component failures, module failures, wiring issues,” says Corbin. “There are hundred things that can happen.”

“This solar industry is still pretty young. So you have a lot of people who, ten years ago, basically were only installing systems and didn’t really care about how the systems performed over time. So you’ve got lots of systems out there, performing terribly. They’re not producing the power that they should be,” he explains.

“And that’s just lost investment. People spent hundreds of thousands of dollars - I’m sure they got their tax credits, but now their systems are operating at a fraction of their capacity, because they haven’t been maintained.”

Plant operators: forecasting when to schedule O&M

But scheduling maintenance can mean down time for a plant. A new option for solar plant operators who need to know when they will lose the least amount of money by shutting off production to repair wiring or replace failed components is SolarAnywhere Forecast, a new offering from Clean power Research.

“Excessive site maintenance can negatively impact profitability,” says Dise, explaining that his software also helps plant operators address operational risks once a project is complete.

“From a daily operations standpoint, forecasted power output available with SolarAnywhere Forecast can be used to more intelligently plan plant operations to model, assess and mitigate over- and under-production situations and grid-wide impact.”

The new forecasting product will go from the next hour to seven days ahead, letting plant operators find the best time window for maintenance.

Richard Perez, the Research Professor at the Atmospheric Sciences Research Center at the University of New York at Albany, is sponsored by Clean Power Research and has been contracted for modelling following his contribution to the development of the National Solar DataBase with the DOE and NREL.

“For the first few hours we know exactly where the clouds are and where they are moving,” says Perez. “However clouds tend to evolve, dissipate, and change direction, so the cloud motion model loses precision as the time horizon increases.” At that point, they transition to the numerical weather prediction models which are the backbone of all the forecasting that’s done globally with satellite data.

For policymakers: determining the value of a feed-in tariff in a snowy region

“The most interesting thing for me is not in the resource data per se, but in all the analytical work that you can do with the data,” he says. As an example, he mentions the interest in trying a value-based feed-in tariff in northeast states like New Jersey, and how the software can be used to determine just what that value is - in a region that gets snow.

According to Perez, Clean Power Research is now close to releasing a new product that includes the effect of snow conditions. They have recently done modelling for Canada, Minnesota and New York.

“One application that requires the detailed underlying resource data is the determination of the value of PV-generated electricity to ratepayers and tax payers,” he explains. “As an extension of this, we came up with the concept of the value-based feed-in-tariff which is receiving growing interest in the US. This value is dependent upon three key factors: location, resource penetration and PV system specifications.”

And to determine this value, high resolution and time-correlated resource data is needed - such as the effect on solar generation when there is snow.

The north-eastern states have more solar insolation than Germany, but they do get snow, unlike the more traditional solar hotspots of the southwest states.

So part of determining the value of a potential feed-in tariff in America’s northeast would have to require accurately forecasting solar energy generation in snowy conditions.