
This contribution focuses on how the current design of rooftop photovoltaic (PV) modules, in particular thickness and tempering of the glass, affects snow load carrying capacity and consequently the module lifetime. The issue is urgent since many publications report on the growing number of PV module glass breakages from the field. At high-latitude and high-altitude locations, nonuniform snow load represents a particular challenge for PV systems since this load is typically present in conjunction with cold temperatures. The aim of this work was to understand the insufficiently studied case of combined stress situation and fill the research gap of testing uneven load in low temperature (−25 °C) as well as at normal room temperature, which is the standard procedure. When the correlation between the module's mechanical stability and glass surface compression stress on glass thicknesses of 3.2, 2.8, 2.0, and 1.6 mm was investigated, there were unexpectedly no significant differences in failure loads between the modules with different glass thickness. Similarly, the test temperature only had a minor impact on the failure loads. Unlike tempered thick glass modules, heat-treated thin glass modules showed local glass fracturing upon failure load. Thus, a novel acoustic signal analysis was developed to study the glass breakage since the fracture pattern of the new thin glasses could not be captured with conventional methods. Such local fracturing, which potentially leads to the loss of insulation resistance, may lead to higher failure rates in the field.
Snow losses can significantly reduce photovoltaic (PV) energy production in cold climates but are difficult to predict accurately. This study applies the Marion and Townsend snow loss models across the continental United States using satellite-derived gridded data (National Solar Radiation Database and Snow Data Assimilation System) at 0.2° resolution over a 22-year period, producing monthly snow loss maps that characterize typical winter conditions. The study then compares measured PV performance from multiple GroundWork Renewables, Inc. soiling measurement stations with the modeled snow losses to evaluate the accuracy of the Marion and Townsend models. Both models capture the primary structural characteristics of snow losses; however, significant regional variation and modest correlation values indicate that further model development is needed to reduce uncertainty in site-specific snow loss estimates. Despite these limitations, the results demonstrate a practical framework for generating approximate snow loss assessments using publicly available gridded data.
This article reports on the design, implementation, outcomes, and best practices of the Natural Sciences and Engineering Research Council of Canada Collaborative Research and Training Experience program “Training in Optoelectronics for Power: From Science and Engineering to Technology” (TOP–SET), a Canadian interdisciplinary training initiative in high–efficiency photovoltaics. Delivered from 2017 to 2024 by a collaboration between the University of Ottawa, Université de Sherbrooke, and McMaster University, TOP–SET combined advanced technical training, professional skills development, and industry–aligned experiences spanning photovoltaic materials, devices, systems, and techno–economic analysis. The program trained 169 undergraduate students, graduate students, and postdoctoral fellows, exceeding its original targets, and engaged a broad network of academic, industrial, and government collaborators. Training activities included Summer Schools, seminars, interdisciplinary challenges, graduate coursework, professional development workshops, and national and international mobility experiences. Program outcomes are assessed through research output, mobility, and posttraining activity, demonstrating high levels of research productivity, award attainment, and employment in fields related to trainees’ studies. Trainee feedback indicates strong satisfaction with both technical and professional components. Based on these results, we identify key best practices related to stipend allocation, trainee involvement, networking, and program governance that may inform the design of future large–scale graduate training programs in science and engineering.
Accurate solar irradiance forecasting is critical for the stability of photovoltaic (PV) systems. However, pure meteorological predictions are frequently insufficient for grid integration due to the nonlinear operational constraints of physical PV hardware. To bridge this gap, this study proposes an end-to-end physics-informed modeling pipeline. The first stage employs a novel hybrid deep learning architecture, bidirectional long short-term memory with custom attention, optimized via a Bayesian optimization algorithm, to forecast global horizontal irradiance by explicitly decoding cloud-driven atmospheric transmissivity via the clearness index ($k_{t}$). The second stage propagates these forecasts through a deterministic PV translation layer accounting for thermal derating and inverter capacity saturation to seamlessly translate raw weather data into operational alternating current (AC) power. The framework is validated across five diverse climatic zones using 15 years of meteorological data and benchmarked against five recent state-of-the-art architectures. Crucially, the model's generalizability and operational safety are further verified using measured AC power data from five real-world microgrid PV plants, complemented by rigorous empirical uncertainty calibration ($U_{95}$) and structural risk assessment via Williams plots. Results demonstrate that the proposed pipeline reliably restricts AC power translation errors to below 0.20 kW across all evaluated locations and seasonal extremes. By successfully suppressing nonlinear error amplification during hardware translation, this framework establishes a highly robust and transparent foundation for scalable PV integration studies.
A hot-carrier solar cell consists of a hot-carrier absorber (HCA) sandwiched between energy-selective contacts (ESCs); the ESCs extract photogenerated charge carriers above the energy threshold. While recent progress in the HCAs has achieved notably low carrier-thermalization rates, the ESCs remain less explored. In the present study, we conducted model calculations based on the detailed balance to identify the required ESC features and assessment criteria. It is elucidated that an appropriate value of the conductance for the carrier extraction, along with an appropriate energy threshold, is required; not only insufficient but also excessively large conductance values lower the conversion efficiency. Nevertheless, the detriment caused by deviation of one of these two characteristic quantities from its optimum is compensated by adjusting the other quantity, providing design guidelines for the materials and device architectures. Whether these two quantities of a given device are appropriate is determined by measuring the relation of the short-circuit current density (jSC) versus the illumination intensity (or solar concentration ratio, Rc). For an appropriate ESC, jSC/Rc starts to decrease with decreasing Rc from the value used in the design, whereas it remains almost unchanged in the higher Rc range. When the ESC features are inappropriate, the jSC/Rc versus Rc relation exhibits distinctive signatures. The experimental procedure for this evaluation is more straightforward and reliable than the previously adopted measurements of jSC dependent on the average energy of the absorbed photons.