The key to operating a successful PV power plant is maximizing energy harvest while keeping a vigilant eye on minimizing inefficiencies that lead to reduced profits or even underperformance.
Array Technologies’ SmarTrack™ energy optimization software platform is a new offering available to ensure asset owners are using the best tracking strategy possible to optimize energy production.
Optimizing yield with innovative machine learning algorithms, SmarTrack minimizes row-to-row shading and increases production in diffuse light.
Backtracking: An Imperfect Initial Solution
One of the clearest obstacles to efficient energy harvesting is row-to-row shading. Simply put, as trackers allow modules to follow the sun, panel-on-panel shading can occur during early morning and late afternoon hours.
The shading of one cell on a module can cause a loss of production from that entire panel, not just the shaded area. This means that the overall dip in energy production during row-to-row shading can be significant.
Traditional backtracking solves this problem––to an extent.
Backtracking, which has been in use for over two decades, refers to a tracking strategy where modules move to shallower angles during the “shoulders” of production hours to avoid row-to-row shading––the morning and evening hours of energy harvest. This strategy increases energy production during those early morning and late afternoon shoulder hours.
A backtracking strategy can exhibit issues in terms of practicality. On an ideal site where row height is entirely equal and the slope of the site’s terrain is perfectly flat, backtracking can be very successful. However, these ideal sites rarely exist in today’s utility-scale PV landscape, and conservative or inefficient backtracking on less-than-perfect sites can lead to serious production loss. And as utility-scale solar power production popularity increases, more sites will be built on challenging, undulating terrain and in regions of lower insolation. So backtracking efficiency in these sites becomes more important for profitability.
“In the real word, site design is never perfect,” said Dr. Kyumin Lee, Array Technologies Director of Product Innovation. “Our research indicates that conventional, conservative backtracking doesn’t work for most sites.”
To account for row height variation and terrain grade, Lee says traditional backtracking models had “one knob to turn” in the form of ground coverage ratio. By adjusting this parameter, the traditional tracking algorithm can be forced to backtrack more aggressively to avoid row-to-row shading.
“Tweaking the conventionally accepted backtracking model can help, but it’s not the optimal solution,” Lee says.
How SmarTrack Elevates Backtracking and Diffuse Light Strategies
With these practical concerns in mind, Array Technologies developed SmarTrack.
With SmarTrack, PV solar plants can take row height variation and undulating terrain into account, ensuring that row-to-row shading is minimized and that a site isn’t losing valuable production by excessive backtracking.
“With SmarTrack, we account for the site slope,” Lee says. “You are not wasting sunlight. It’s counter-intuitive, but the revelation is that we’re always backtracking the minimum amount to maximize the power production. The angle is fundamentally different from what you can calculate with a conventional model. We are using a backtracking model, leveraging machine learning, that’s more reflective of typical real-world limitations at a site.”
Even a slight extension in the sun angle when backtracking begins – say, from 64 degrees to 67 degrees under certain site conditions – can have a significant impact on broadening the “shoulders” of energy production at a site.
“You cannot just walk away from backtracking, because if you were to just ignore backtracking altogether, you’d lose too much energy when the sun is low,” Lee says. “But you do not want to backtrack more or longer than you absolutely have to. The idea of SmarTrack is to produce more energy by backtracking less.”
Utilizing Machine Learning for Simple, Efficient Implementation
One of the primary advantages of the model SmarTrack uses to “learn” the optimal tracking strategy for a site is that it isn’t only applicable to the period when the learning is completed; rather, the parameters learned by the model are valid for the lifetime of the plant, making overall operations simpler.
The learning process is also extremely fast, Lee said, often taking a matter of days to complete depending on weather conditions during SmarTrack’s learning process.
Kendra Conrad, Array Technologies Principal Engineer and a veteran solar industry professional, said SmarTrack offers unparalleled benefits in terms of cost as well.
In particular, while a manual commissioning optimization can indeed get a significantly improved result, this process is both time-consuming and costly.
“You can actually optimize energy production relatively well when manually commissioning a power plant,” Conrad said. “However, it’s a very slow process. It takes personnel, it takes time, and you’re not going to get as good a result – even if you do it perfectly – compared to SmarTrack. The fundamental equation of traditional backtracking does not allow you to factor in the row-to-row height variation or slope.”
SmarTrack has been validated by independent engineering firms like DNV GL to continue to prove its effectiveness and help calculate project bankability.
Contact us to learn more about optimizing production with more intelligent backtracking and smarter diffuse light strategies, or to get a quote for your plant.