
By: Alvaro Zanon, a photovoltaic applications engineer at REFUsol. He designs and plans commercial and utility-scale PV projects. He can be reached at alvaro.zanon@refusol.com.
In an ideal world where design and environmental constraints do not exist, all photovoltaic (PV) systems would look almost identical. However, real PV projects are typically subject to a variety of factors that significantly affect physical configuration and performance — usually in a negative way. Such factors include shading, soiling, module mismatch, different module types and orientations, or large temperature fluctuations.
While central inverters have been the common choice in the U.S. market for PV systems for decades, a distributed design approach using string inverters is gaining increasing popularity. Among string inverters, transformerless (TL) solar inverters offer several advantages not present in their transformer-based counterparts — regardless of size — such as higher efficiency, lighter weight and more compact form.
This study had two main goals:
To investigate the benefits from a technical and economic standpoint of designing PV systems using central inverters (centralized approach) versus using TL string inverters (distributed approach).
And to evaluate which approach is the most potent tool for combating the negative effects previously mentioned
System Definition
To provide a holistic comparison between the two design approaches, three array configurations of sizes 100 kW, 260 kW and 500 kW were created for each (detailed description provided in Table 1). For the purpose of defining realistic temperature profiles, the study assumed Newark, N.J., as the location of the project. All systems assumed a south-facing roof at a 20° tilt to accommodate 60% of the modules with minimal obstructions and shading, along with a west-facing carport at a 5° tilt large enough for the remaining 40% of the modules.
The three array sizes were created, simulated and evaluated separately using both central inverters and REFUsol 20K-UL string inverters. Furthermore, the response to ›voltage and current mismatch was analyzed by adding inter-row shading on the rooftop portion, and shading from trees over the carport portion. Additionally, a constant module-mismatch factor, as well as a varying soiling factor over the course of the year, was introduced in the simulations to evaluate their effect on both inverter approaches.
Cost Analysis
The study showed that when considering all system costs, the centralized approach was found to be more expensive than the distributed approach by upwards of $0.16/W (15% gross margin added). An interesting trend arose when plotting the total system cost versus the size of each array configuration (Figure 1). Under this scenario, it’s clear that string inverters are a more cost-effective solution for systems up to 500 kW. Although larger system sizes were out of the scope of this study, it seems apparent from Figure 1 that system cost no longer becomes a deciding factor between central and string inverters. Other factors such as energy production, reliability or uptime (which will be studied later on) need to be brought into consideration.
When considering only balance of system (BoS) costs, the distributed concept reduced the cost by up to 25%. Breaking down the results into each individual system size, the results showed that total BoS costs (material and labor) were reduced by 25%, 15%, and 9% respectively with distributed inverters compared to 100-kW, 260-kW, and 500-kW central inverters. The results displayed in Figure 2 show these differences in more detail.
The higher BoS costs of the centralized arrays are originated by the amount and price of DC wiring and equipment. Proportionally, the cost of DC wiring and equipment of the centralized arrays is higher than the cost of AC wiring and equipment of the distributed arrays.
Energy Production
In the distributed approach, the PV arrays are broken up into smaller sub-arrays, each with its own TL inverter and, therefore, its own MPPT tracking system. This system granularity translates into a higher energy production and reduced effect on output from individual inverter failure. Energy modeling of these designs show distributed inverters outperformed central inverters by approximately 1.5%. Although the distributed system had slightly higher ohmic losses than the central system by 0.2% of 0.3%, the high AC losses of the distributed inverter were comparable to the high DC losses of a central inverter. More importantly, the string inverter efficiency losses were noticeably lower (~ 2%) than the central inverters (~4%) used in this study. Table 2 shows the energy production values for each array along with the sources of energy losses.
Inverter Response To Negative Effects
The south-facing array on the roof and the west-facing array mounted on the carport(s) experienced different amounts of irradiation throughout the year because of their different orientations and shading conditions. Since current is directly proportional to irradiance and additive inside the solar inverter, the performance of either the central and string inverters will not be significantly altered. As opposed to having multiple string inverters, however, all parallel circuits of a central inverter are held at the same voltage by the inverter. If this is different than the arrays’ maximum power point voltage, there will be a power loss. This was confirmed by energy modeling analyses in PVsyst.
Over a year, hourly data modeling showed a 0.24% power loss due to voltage mismatch on a single inverter. As such, an additional 0.24% mismatch loss has been added to the central array solar inverters.
Additional Benefits
The light weight of TL inverters allows for direct mounting on either the carport or on one of the building walls, eliminating the need for a concrete pad and a lifting crane. Time and cost can be further saved since most TL inverters have built-in DC-fused combiners and disconnects.
Finally, maintenance and servicing is much easier and inexpensive with the distributed concept. If there is a problem with one of the string inverters, it can be easily exchanged for a new one in the field by a general electrician.
In the United States, the most popular approach is centralized solar inverters, but the distributed concept does offer the system designer a new approach with added flexibility and performance expansion. SPW
I think its great more solar power designs and technology keep showing up. But I think more home owners should take the initiative and use solar power for themselves. I am already saving on my electric bill. tim lee