ARENA lobs $2m at hyper-accurate solar yield software

A company that began as a free solar data website in 2011 has landed a $1.97 million grant to fine tune the product that, 13 years later, it calls “the world’s most advanced solar simulation engine”.

The Australian Renewable Energy Agency (ARENA) is funding PV Lighthouse to improve the accuracy of its solar yield forecasting software, with the company matching the investment. 

“Before SunSolve existed, developers did not have a modelling solution that could accurately represent their structures and modules. This meant they had to guesstimate several of the loss factors required for their yield forecasts,” says PV Lighthouse and SunSolve founder and CEO Keith McIntosh.

“Now developers can rapidly calculate these factors with SunSolve, giving them greater confidence in their forecasts – a key advantage when negotiating project financing.”

The SunSolve Yield software works by creating a digital twin, or a 3D digital model, of a proposed project. It then calculates a variety of factors, from the direction of billions of rays of light to the effect of how shading and light reflections interact, to forecast what each panel will be able to yield over time.

A physics-based optical engine, or ray tracer, tracks light rays as they hit the ground. 

Aggregating these in the billions means McIntosh’s software can forecast all of the possible permutations and turn that into a relatively simple model in minutes, rather than days, thanks to the power of cloud computing. 

But it’s the “fudge factors” where they can make a big impact, the areas where in most cases analysts and software need to take a punt on what the impact of, say, reflected light between rows of solar panels.

“When we were first asked to help the industry, the question was about bifacial factors,” McIntosh says.

“When the light reflects off the ground and it comes up onto the rear of the module, there’s things that get in the way, the torque tubes, the frames.

“In every other software to estimate yield, you have to guess that number. The default was 10 per cent but nobody really knew what it was. But with our software, because we simulate actual physical structure, we could tell you what that number was. It could be anywhere between 5 and 20 per cent, depending on the system.”

This kind of accuracy is something developers can take to the bank. 

McIntosh says nine of the top 25 global module manufacturers use SunSolve for their research, including Qcells and Maxeon Solar Technologies, to optimise cell design and drive development. Systems developers followed and now developers are beginning to buy in. 

In Australia, the company’s customers include 5B, US-based NexTracker, and several unnamed developers.

The software was created in 2019 in partnership with Cypress Creek Renewables in a research project funded by the United States Department of Energy. 

But its true birthplace was PV Lighthouse’s ‘Grand Unified Model of Photovoltaics’, a goal to bring all of the data around solar power into one place.

“It began with the Internet revolution, because at that time I was working at Sunpower in the US and later at ANU, and I realised just how slowly information was being transferred from researchers into industry,” McIntosh says. 

“The fastest way for that to come through was through the Internet, and combining that into one single model on our website.”