The long read: Model progress

Recent developments suggest that the industry is in the midst of a major shift toward bifacial technology, with two-sided modules on display at virtually every trade show booth, and major announcements promising gigawatt-scale production from Longi Solar, Canadian Solar, and Jolywood; to name but a few.

On the installation side, tracker suppliers Soltec and Nextracker both have around 2 GW of bifacial projects installed or under contract. And there is certainly no shortage of interest from project developers attracted by the prospect of higher energy yields from modules at a similar price point. “We are receiving an increasing number of requests to assess bifacial project developments,” says Nicolas Chouleur of energy consultancy Everoze. “Most are for large ground-mounted single axis tracker systems in North Africa and the Middle East. In Europe, especially in merchant markets such as Spain or Italy, developers are also ready to adopt bifacial for all types of systems, from roof-top, facade or other BIPV installations to ground-mounted PV farms.”

The energy yield boost that bifacial modules can provide in the field is often quoted between 5% and 20%, or within a similar range. While this has been enough to spark significant interest, project developers and investors demand far more specific figures on annual energy yield. And the additional complexity inherent to calculating the yield of a bifacial installation still holds the technology back from mainstream acceptance. “With monofacial modules, the accuracy is considered so good that even a fraction of a percent on particular losses can be highly debated,” notes Chouleur. “Currently, the uncertainty on bifacial is noticeably higher. This has a direct impact on project finance.”

Modeling the yield

In a monofacial installation, engineers need to consider two types of light hitting the cells – direct sunlight and diffuse irradiation on the frontside. Speaking at the BifiPV Conference in Amsterdam in September, Itai Suez – VP product development at Silfab, a Canadian PV manufacturer – explained that with bifacial, this is increased to six: direct sunlight on the rearside, diffuse irradiation on the rear side, direct reflection from the ground and diffuse irradiance reflected from the ground, in addition to the first two. Combine this with the fact that the four types of light hitting the rear side rarely do so uniformly, and understanding bifacial performance becomes a complicated affair.

“Modeling bifacial yield is complex due to the variability in the irradiance hitting the back of the modules,” explains Joshua Stein, PV performance lead at Sandia National Laboratories. “This variability is caused by local obstructions (e.g., racking and other structures), albedo that varies spatially (plants, bare patches of ground, microtopography on the ground that casts shadows, etc.), and the fact that different parts of the module are different heights and thus have a different ‘view’.”

When it comes to developing a comprehensive model for bifacial output, characterizing the albedo of the surface below the installation is of particular concern. Companies and research institutes around the world are working to develop sophisticated models that take all of these factors into account. And while able to take into account the complexities of light conditions in a bifacial installation, so far many are held back by the lack of field data from actual installations needed to validate assumptions on which the models are based. A wide range of approaches to modeling bifacial yield was presented at the BifiPV conference in September, and virtually every presentation closed with a request for project owners to share more field data. And there was broad consensus that the development of a shared dataset will be essential to improving the modelling accuracy. “One possible approach is to generate high fidelity and high-quality irradiance measurements (both front and rear facing) within a bifacial PV array and provide these data as part of a well-documented standard dataset to model developers,” comments Stein. “If several such datasets were available, these could be used to benchmark bifacial PV optical models.”

For now, the models rely largely on satellite data for their albedo measurements, and while these provide a broad assumption of regional albedo, they do not typically account for local shading/microtopography or seasonal changes to the land – things that might better be termed the “local irradiance profile” of the site. For this reason, most recommend taking on-site measurements in order to accurately model the output of a given project.

Multiple industry players report that, as long as this onsite data is measured accurately, existing modeling solutions – such as the industry standard PV Syst software – can provide good results. “To keep bifacial simulation relatively simple, real hourly simulation with a 3D shading scene added to the PVsyst interface could be the best solution,” says Chouleur. “It is worth mentioning that there are interesting software developments that allow for very accurate ray-tracing simulation. But it is challenging to undertake the multi-MW simulation without programming skills and using only a standard laptop.”

Optimizing output

Gaining a better understanding of bifacial performance will allow developers to find cost effective ways to optimize the system in order to maximize energy yield. Many of the ‘first order’ optimizations have already been made – from tweaks inside the module such as moving the junction box, to optimized tracker designs. “We have done the first round of high-impact, first-order drivers in optimizing for bifacial,” says Greg Beardsworth, director of product management at Nextracker. He explains that the company’s one-in-portrait Horizon tracker features an increased distance between the modules and the torque tube, gaps at all bearings and no control system components underneath any panels, while its newly launched two-in-portrait Gemini tracker can be built with a gap above the torque tube, also to minimize shading on the module.

In the field, bifacial optimizations are so far focusing on system size and layout. “Total DC system size is being reduced relative to the bifacial gain, and projects are utilizing a smaller DC system size to reach the same MWh/year generation,” explains Beth Copanas, director of solar energy at EPC provider RES Americas. “The reduction in DC system size for the same acreage allows for additional decreases in the ground coverage ratio leading to increases in system yield.”

Copanas goes on to note that bifacial project stakeholders also need to consider inverter loading (DC/AC ratios) and clipping impacts, bifacial gain assumptions used for cable sizing, fuse sizing and number of strings paralleled per DC inputs, meteorological station equipment selection and station/sensor placement, increased weight of glass-glass modules, DC cable/wire management to avoid rear side shading, ensuring capacity test methodology is modified to account for bifacial gain, and that front and rear side irradiance sensor placement is indicative of site spatial and albedo variation.

Yield vs. capital

The U.S. National Renewable Energy Laboratory (NREL) calculates that, based on standard PERC technology and single axis tracking, the cost of a bifacial system is around $0.03-$0.06/W higher than a similar monofacial installation. According to Michael Woodhouse – senior energy technologies, economics and policy analyst at NREL – this breaks down to $0.01 or $0.02 on the module side, and $0.02 to $0.04 at system level. He further notes that NREL’s analysis assumed the same O&M costs for both monofacial and bifacial, which may be an oversimplification.

The lab’s calculations also find that every 3% boost to a project’s annual energy yield can offset a $0.04/W increase to the upfront capital cost. And though it is not yet clear what can be achieved within this price range, developers will seek the highest energy yield possible. “As FIT rates get lower, project developers are looking for every missing kilowatt-hour to make financial justification. So trying to capture energy yield benefits in single digit percentages, they would want to include that in their model,” explains Woodhouse. “The big question for the total system economics is, whatever the yield gains are, do they overcome the currently slightly higher module price and BoS costs?”

Many of the bifacial test installations springing up around the world utilize albedo-enhancing materials on the ground below modules, and report significant –in some cases, double-digit percentage – energy yield gains from this. pv magazine has seen examples of small installations with white gravel, white painted surfaces, plastic sheeting, and even light-colored flowers added underneath the modules to enhance the rearside output. In a utility-scale project, however, altering the ground would be a major undertaking which, factoring in the need to keep the surface clean and its albedo high for the project’s lifetime, could prove costly.

It is becoming clear to the industry that adopting bifacial technology necessitates a rethink at almost every level of the value chain. But the potential for bigger energy yields cannot be denied, and the optimization cycle for bifacial is only at the beginning. “Our overall experience is that these initial complications can be and are being addressed,” notes Copanas. “Owners who have invested in bifacial already are continuing those investments on future projects.”