Enhancing reliability in bifacial PV modules: a novel approach to albedo estimation

In bifacial solar PV systems, albedo is a critical parameter, representing the percentage of radiation reflected by a surface. Any error in albedo estimation can result in a performance estimation error ranging from 2% to 8%. Enertis Applus+ adopts a customised approach to albedo characterisation, tailored to specific market requirements, ensuring the highest level of accuracy in performance estimations. To this end, various complementary methodologies are proposed for measuring albedo, including surface albedo measurements or estimates derived from satellite data.

Recent research indicates that, due to the high spatial variability of albedo, the most reliable measurements are obtained through long-duration surface campaigns with high spatial resolution (areas less than 100m²), supplemented by satellite measurements when surface data is limited.

However, particularly in new PV system locations, conducting long-term albedo measurements may not be feasible. In these cases, albedo values need to be estimated from short-term measurements lasting a few hours or days. A recent study by the Enertis Applus+ R&D team, in collaboration with the Institute of Microelectronic Technology of the University of the Basque Country (TiM-UPV/ EHU), has demonstrated that characterising albedo using a single short-term measurement campaign results in significant variability due to meteorological or seasonal changes. Furthermore, albedo variations of more than 60% have been observed within the same day.

As an example, Figure 2 depicts the distribution of specific albedo measurements, recorded every minute over the course of one year at a meteorological station in Sabinal (Texas, USA). These measurements are correlated with the clarity index (Kt), representing the degree of cloudiness, and are further analysed based on various levels of daily precipitation. This data, made publicly accessible through the DuraMAT consortium coordinated by the US National Renewable Energy Laboratory (NREL), underscores the complexity of albedo calculation.

Various factors, including high levels of cloudiness and precipitation, contribute to significant dispersion in albedo values, as illustrated in Figure 2. This high variability highlights the necessity of considering multiple environmental factors when calculating albedo and it underscores the potential for significant errors in long-term albedo estimation if based solely on single short-term measurements. Therefore, it is critical to take temporal and spatial variability into account when interpreting and using albedo data for solar energy and climate-related applications.

Figure 3 depicts a daily-scale analysis conducted during the month of January, following the same methodology as Figure 2 It reveals significant variation in albedo values, even within consecutive days or throughout the same day.

The considerable variability in albedo values poses significant challenges for accurately predicting its long-term behaviour based solely on short-term measurements. For instance, in scenarios where measurement campaigns run for only one day per month, fluctuations exceeding 20% in albedo values are common and this is not only limited to monthly scales. Fluctuations of more than 10% within a single day are also observed.

Recognising this complexity, Enertis Applus+ and TiM-UPV/EHU have conducted joint research to address albedo variability using a rigorous statistical approach. We have carried out a detailed study to model the statistical distribution of albedo variability over time, enhancing understanding of observed fluctuations and establishing correlations between albedo measurements from short-term campaigns, long-term campaigns and satellite data.

The objective of this research is to determine conditions under which Enertis Applus+ can offer short-term albedo measurement campaigns that provide data equivalent to long-term ones. This approach would enable replacing expensive long-duration campaigns, which are limited to specific locations and require weather stations, with short-duration campaigns that use lightweight equipment and that allow measurements to be made over larger areas and with better return on investment, as illustrated in Figure 4.

The proposed model that establishes the conditions under which short-term campaign measurements are equivalent to long-term measurements relies on an exponential function denoted by Equation 1. This function categorizes short-term albedo measurements based on climate and surface type. The functions obtained from Equation 1 are based on complex variables in which parameters such as cloudiness, temperature, wind, climate, humidity, etc. intervene, and they are collected within the function β.

The function aims to determine the percentage of valid albedo measurements in short-term campaigns relative to long-term measurements, within a defined margin of uncertainty. Thus, Enertis Applus+ can define the required number of measurements and conditions necessary to determine the albedo value with confidence. Figure 5 illustrates this concept, depicting valid albedo values against their respective uncertainty levels, categorised by surface type from different DuraMAT stations (grass, gravel and sand).

The figure presents both raw station data (depicted by squares) and the model adjustment using Equation 1 (illustrated by a solid line).

The proposed model provides a flexible tool that can be adapted depending on the type of soil or climate to define the admissible percentage of albedo measurements within an acceptable margin of uncertainty.

When applying this model in a practical scenario, we illustrate how its predictive capacity can be used to evaluate the validity of albedo measurements in different environments. For example, when considering albedo measurement on gravel surfaces with a maximum margin of uncertainty of 5%, equivalent to an albedo variation between 0.19 and 0.21 for a base albedo of 0.2, the model reveals that up to 90% of these measurements could be considered valid (Valid albedo). This suggests high reliability in the precision of measurements on this type of surface, with only a minimal fraction of measurements outside the acceptable margin of uncertainty. In contrast, when applying the model on sand surfaces, valid measurements would not reach 60%. This discrepancy implies that, statistically, at least four out of 10 albedo measurements on sand surfaces could have an uncertainty greater than 5%, suggesting greater variability or difficulty in the precision of measurements in this specific environment.

The model, as presented in Table 1, has been adapted for a wide range of conditions, covering different types of surfaces and climates. Furthermore, it has been evaluated when comparing albedo values from short-term and long-term measurement campaigns, and when comparing short-term and satellite data. The data obtained by satellite correspond to the Solargis database. This platform provides a single monthly albedo value obtained from multiple satellite measurements for each location.

Results suggest that albedo variability follows a non-random pattern and that the percentage of valid albedo values can be modelled as a function of the admissible uncertainty using the exponential function presented in Equation 1. Therefore, we can predict how albedo measurements vary with a certain degree of precision, depending on the admissible margin of uncertainty.

The developed model allows us not only to understand the nature of this variability, but also to establish an indicator of the minimum number of measurements required for each type of surface or climate. This ensures that, within a predefined admissible uncertainty, the obtained albedo values correspond to long-term measurements or satellite data.

It is worth noting that while surface measurements and satellite data yield comparable results, satellite data exhibits greater scatter. This suggests that while satellite data provides valuable information, adjustments may be necessary to enhance ts accuracy for specific applications.

This scatter is influenced by the homogeneity or heterogeneity of the terrain, as Solargis data has a spatial resolution of 1 x 1km². Figure 6 displays three satellite images of different PV plant sites, illustrating variations in terrain homogeneity. This suggests that the variability in satellite data can be related to the variability in the homogeneity of the terrain, confirming observations made by Enertis Applus+ in its albedo measurement campaigns (Figure 6).

Moreover, it is important to acknowledge that while satellite data offers a quick and cost-effective option for preliminary studies, additional adjustments may be needed to improve the accuracy of this data. The proposed model helps estimate uncertainties associated with albedo measurement during campaigns, optimising them to provide better performance estimates for PV systems in specific locations.

About the authors

Eneko Ortega is assistant professor in the Department of Electricity and Electronics at the University of the Basque Country (UPV/EHU) and member of the Technological Institute for Microelectronics (TiM). His research is focused on the field of PV systems monitoring and characterisation.

Sergio Suarez is the global technical manager of testing and optimisation at Enertis Applus+. He is currently taking a PhD in electronics at UPV/EHU (Spain), an institution involved in the manufacturing of bifacial PV cells, participating in R&D projects as principal researcher and developing new testing
methodologies.

Juan Carlos Jimeno has been a professor at the University of the Basque Country since 1995 and is director of its Technological Institute of Microelectronics. His R&D focus since 1981 has been on bifacial technology and screen printing manufacturing.

Mario Martínez is the global technical deputy of testing and optimisation at Enertis Applus+. He is currently taking a PhD in PV Solar Energy at UPM/IES (Spain), participating in research into the development of new materials and concepts for the PV industry.

Ignacio J. Fernández is the head of testing and optimisation at Enertis Applus+. With more than 11 years of experience in the PV industry, Ignacio has managed TDD processes, contract negotiations, expert works and QA/QC campaigns related to PV plants worldwide.

Sofía Rodríguez is the R&D manager at Enertis Applus+. She has worked on the development of innovative defect detection systems in PV modules, coupled with categorisation and classification techniques.

The results of the study carried out by Enertis Applus+ and TiM-UPV/EHU have been published in the journal “Renewable Energy”, in volume 221 of February 2024.