Using Aurora to Solve the Mystery of an Underperforming PV System

An underperforming solar PV system is a solar customer’s nightmare. With a big investment like solar, customers want to know that their PV system is producing as much energy as forecasted and they’re getting their money’s worth.

Unfortunately for a residential customer in Cathedral City, California, for unknown reasons, their system was consistently producing less energy than expected. Fortunately, their system was monitored by Omnidian, a solar operations and maintenance (O&M) firm that is dedicated to managing the performance of solar systems.

Omnidian remotely monitors solar system performance and interprets the data with its proprietary software to identify systems that are experiencing problems–as it did in this case–and coordinates field service work with local service partners where needed.

To determine what has gone wrong in these cases, “the industry often responds with an initial truck roll to go out to the site to investigate what’s going on,” explains Omnidian Chief Operating Officer David Kenny. But truck rolls can be costly, so if there were a way to remotely identify the problem and determine the necessary service before sending someone out, reaching a resolution could be more efficient and affordable.

To this end, Omnidian turned to Aurora’s simulation tools to determine if it would be possible to remotely diagnose the cause of this system’s underproduction. In this article, we explore how Aurora’s software allowed Omnidian to identify design flaws and make adjustments that significantly increased energy production.

Remote Diagnostics with Aurora

Most users know Aurora for solar design and sales software that makes it possible to remotely design a solar system, accurately forecast its energy production, and estimate the monetary value of that production. In this case, however, Aurora’s simulation engine–which calculates how much energy a PV system will produce based on its design and location–would be used to retroactively determine why there was such a discrepancy between actual and predicted production.

To do this, the project site was first modeled precisely in Aurora. An accurate 3D model of the project site is essential to determining how much energy a PV system will produce because it provides the basis for determining the amount of shade that will fall on the modules at different times throughout the year.

The PV system was then recreated in Aurora according to how it was installed–with the same module and inverter specifications, panel placement, and stringing configuration. The existing system consisted of 30 modules and two string inverters, each of which were connected to three strings of five modules in series.

An underperforming solar installation that was diagnosed by Aurora Solar and OmnidianThe underperforming solar installation as originally designed, recreated in Aurora for modeling purposes.
Recreating the existing system design made it possible to simulate how much energy the system “should” be producing and identify factors that were reducing power output. Aurora calculated that the existing design would produce approximately 12,000 kWh per year. A preliminary comparison of Aurora’s simulation results to recorded production data from the system revealed that Aurora’s simulation was within 1% of the system’s actual annual production.
Estimated energy production by Aurora solar software was within 1% of the actual system performance Aurora’s simulated system performance (monthly values in blue) compared to the actual system performance. Aurora’s simulation was within 1% of the actual annual production of the system.
From there, a review of Aurora’s simulation warnings helped to determine why production was less than it could have been. Aurora performs a wide array of checks on every system design, testing to see if there are any violations of engineering principles, the National Electric Code, or the capabilities of the components used.
Aurora Solar precisely forecasts solar energy production. Simulation of this design revealed a design flaw. Aurora’s performance simulations, which forecast how much energy a PV system will produce, also include alerts about potential design errors and other factors that will limit energy production or violate codes. (Note that the customer’s energy usage was not modeled in this diagnostic exercise, which is why there is no energy offset data.)
One alert in particular raised a big red flag about the system design: 72% of the time the string voltage for the system was falling below the inverter’s minimum operating voltage! This information provided a helpful starting point for determining what needed to be fixed for the system to perform optimally.

Pinpointing the Performance Problem

Further exploration revealed that shading was part of the issue; the shading from two neighboring palm trees at the site frequently caused the string voltage to be too low for the inverter. Another contributing factor was the fact that the strings were short (voltage is additive when panels are connected in series, so longer strings result in higher voltage). Because the voltage was frequently falling below the inverter’s minimum starting voltage due to these factors, there were periods of time when the system wasn’t exporting any energy at all.

As the Aurora team member who led the diagnosis explains, “The shade [at this site] plays a big part in the performance issue since the strings are at the very lower limit of the acceptable range. This means any shade will drop the string out of the voltage range and stop production. In a heavily shaded situation like this, it is imperative that strings as long as possible are used with a traditional string inverter.”

With the cause of underproduction identified, it was determined that it would be necessary to restring the system, making the strings longer so that they wouldn’t drop out of the voltage range of the inverter when shade was present.

Finding a Solution

Once it was determined that the system needed to be restrung with more modules on each string in order to keep the voltage within the inverter’s operating range, Aurora was used to model possible solutions. Simulating new potential stringing configurations made it possible to quantify the resulting production changes and ensure that there were no design flaws before service work was initiated.

The underperforming solar installation as redesigned Aurora Solar and OmnidianThe new proposed stringing configuration with longer string lengths to increase string voltage; it consists of two strings of seven panels in series connected to the first inverter and two strings with eight panels in series on the second inverter.

The stringing configuration proposed for the repair of the system included two strings of seven panels in series connected to the first inverter and two strings with eight panels in series on the second inverter. Aurora’s performance simulation revealed that the redesigned system would produce 3,433 kWh more per year than the existing design!

A key reason for this is that the string voltage would only fall below the inverter limit for 0.91% of daylight hours throughout the year under the new stringing configuration, compared to 72.39% of daylight hours when this occurred with the installed design.

The new solar design by Aurora Solar would produce thousands more kWh per year!Aurora’s performance simulation results for the new proposed stringing configuration, showing monthly and annual production, as well as alerts. (Note that the customer’s energy usage was not modeled in this diagnostic exercise, which is why there is no energy offset data.)

Accounting for Inverter Clipping

Once a restringing solution was identified, there was another question to resolve: whether the stringing configuration was compatible with the existing inverters. The new, longer string lengths meant that the DC voltage was oversized compared to the inverter rating and could sometimes exceed the max DC power of the inverter during times of high irradiance. This raised a potential red flag because it could limit energy production as a result of inverter clipping.

However, because Aurora models the expected impacts of inverter clipping based on local weather patterns and shading at the site, it was possible to determine that inverter clipping impacts would actually be very limited (1% of annual production, as noted in the simulation results above). As a result of this insight, the costly need to replace one or both of the inverters was avoided.

Achieving Performance Gains

With a resolution identified remotely, Omnidian coordinated service work to restring the system with the new stringing configuration. Initial production data showed marked improvements in performance: comparing three sunny days before the service work to three sunny days after the service work, the max power (kW) increased 16% and kWh generation increased 24%!

This solar customer was fortunate to have Omnidian monitoring their system, to identify the performance problems with their PV system and find a resolution without the need for the customer’s involvement. With Aurora’s top-of-the-line performance simulations, it was possible for Omnidian to diagnose the problem without a site visit, and then coordinate service to resolve the underproduction issue.

Although not a typical use of Aurora’s solar design software, this initiative highlighted the value of Aurora’s accurate performance simulations. It also shows how solar contractors can avoid design flaws through the use of robust design and performance simulation software. Had the installer of this system used Aurora they would have been alerted that the string voltage would frequently fall below the inverter’s operating voltage range, avoiding production losses for the customer and the need for system repair.

Thankfully, the resolution Omnidian arrived at with Aurora restored the system to its optimal performance, allowing the customer to generate thousands more kilowatt hours each year!

Gwen Brown

Gwen Brown is the Senior Content Marketer at Aurora Solar, managing the development of educational solar resources like blog posts and webinars. Previously, she was a Senior Research Associate at the Environmental Law Institute. She graduated Phi Beta Kappa from Gettysburg College.