Pietro Elia Campana

Agrivoltaic systems increasingly combine dual-axis tracking with bifacial modules, yet field validation of the underlying simulation tools remains scarce. Four platforms, PVSOL, SAM, SISIFO and the in-house UCSC model, were evaluated against 8,373 quality-controlled hourly records from a 506.97 kWp overhead horizontal dual-axis bifacial agrivoltaic plant in Piacenza, Italy, from August 2025 to July 2026. UCSC achieved the lowest mean absolute error, 0.0317 kWh/kWp, against 0.0343 for PVSOL, 0.0357 for SISIFO and 0.0478 for SAM. Campaign bias ranged from -5.06% for SISIFO through +4.10% for UCSC and +8.77% for PVSOL to +19.72% for SAM, with regression slopes between 0.936 and 1.135. UCSC ranked first on all five metrics included in the campaign ranking, with the ordering of PVSOL and SISIFO behind it reversing with the metric chosen. Errors were seasonal and associated with how each platform represents tracker geometry and operating constraints. SAM, which accepts a rotation limit but does not apply it under two-axis tracking, overestimated in every month. PVSOL overestimated by 34.9% in December and underestimated by 2.0% in July, crossing zero in late spring. SISIFO, alone in reproducing the horizontal dual-axis kinematics, reversed sign with season, underestimating by 15.6% in May and overestimating by 13.6% in November. Its exported output showed no observable response to the bifacial settings. Anti-tracking, tested on one clear-sky and one overcast day, reduced yield by 59.9% and 17.8% relative to simultaneously operated standard-tracking sectors of matched baseline yield. Rotation-limit enforcement, array azimuth and inter-tracker geometry were identified as configuration sensitivities.

CONTEXT: Agrivoltaic (APV) systems are increasingly promoted as a dual land-use strategy for combining photovoltaic (PV) electricity generation with agricultural production. However, many APV assessments treat row spacing as a freely optimisable design variable and rarely account for the operational constraints imposed by crop distribution, farm structure and agricultural mechanisation.OBJECTIVE: This study develops a spatially explicit framework for assessing APV suitability across Sweden by linking crop-yield responses to county-level crop composition, farm-size structure and machinery-derived row-spacing constraints. The aim is to identify where APV systems are not only agronomically promising, but also compatible with real-world mechanised farming systems.METHODS: Crop-yield responses were simulated for vertical and single-axis tracking APV systems at one representative location in each Swedish county using ERA5 climate data for 2014–2023. Three crop-response models were evaluated: original EPIC, EPIC-adapted and regression-based models. County-level statistics from Jordbruksverket (the Swedish Board of Agriculture) were used to represent crop shares, farm-size distributions and machinery-use patterns. Machinery-width classes were translated into feasible APV row-spacing constraints, allowing yield-preservation and layout-density suitability to be assessed under regional agricultural and operational conditions.RESULTS AND CONCLUSIONS: The simulations showed a strong nonlinear crop-yield response to row spacing. Under the EPIC-adapted model, vertical APV performed better at narrow row spacings, especially for cereals, where relative yield at 5 m was approximately 0.88 under vertical APV compared with about 0.53 under tracking. Around 10–13 m row spacing, crop yields were generally close to or above open-field levels while remaining compatible with common machinery-derived spacing classes. At the national scale, vertical APV was selected as the preferred system in all counties, but suitability varied substantially. Kronoberg and Jönköping achieved the highest combined suitability scores, 0.918 and 0.914, respectively, whereas Skåne had the lowest score, 0.130, despite favourable solar resources.SIGNIFICANCE: The results show that realistic APV assessment must account for crop response, regional cropping patterns, farm structure and mechanisation constraints. The proposed framework complements Geospatial Information System (GIS)-based APV screening by adding an agricultural-systems and operational-feasibility perspective, and can support more practical national-scale APV planning.

Agrivoltaic (AV) systems offer a promising solution to mitigate land-use competition between agricultural production and renewable energy conversion. However, photovoltaic (PV) modules and related supporting structures modify the local microclimate, influencing evapotranspiration (ET) processes. Despite recent advances in computational fluid dynamics (CFD) applications for AV systems, ET remains a complex microclimatic parameter that has received limited validation. This study develops a hybrid framework coupling a three-dimensional CFD model and field measurements with the Penman-Monteith FAO 56 approach to estimate the spatial distribution of reference evapotranspiration (ETo) in AV systems. The CFD model solves RANS equations, k-ε standard turbulence model, discrete ordinate radiation model, and it applies solar heat flux using solar load model with ray tracing approach. The proposed method is validated using field measurements from a vertically-mounted AV (VMAV) experimental facility at Kärrbo Prästgård, Sweden. The ETo is measured using an in-house developed weighing micro-lysimeter for three consecutive clear sky days. The validated model is then applied to investigate edge effects and compare ETo distributions among VMAV, ground-mounted AV (GMAV), and overhead AV (OAV) configurations. The results demonstrate that the hybrid framework accurately captures ETo variations that are otherwise overlooked by standard weather station data. By providing high-resolution spatiotemporal mapping, this approach serves as a decision-support tool for precision irrigation and the strategic optimization of monitoring networks to capture microclimatic heterogeneity. At the daily scale, comparative simulations show that, relative to the control area, GMAV, OAV, and VMAV configurations reduce ETo by 22%, 12.8%, and 13.1%, respectively.

Saffron is a bulbous herb with edible flowers and is considered a high-value crop. However, cultivating saffron at northern latitudes presents challenges, primarily due to adverse weather conditions during its productive phase. Therefore, this study tested whether microclimate modifications caused by a single-axis tracking agrivoltaic (AV) system could enhance saffron production. The experiment was carried out near Västerås, in central Sweden. A field trial was set up to cultivate saffron both under open-field conditions and beneath the photovoltaic (PV) modules of a single-axis tracking system, utilizing the area that is typically unused for cultivation. Beneath the AV system, the maximum density of 24.3 flowers/m2 was observed, compared with a maximum of 0.5 flowers/m2 in the open field. In total, 226 flowers were counted beneath the AV system, yielding 1.09 g dry stigma, whereas only three flowers were observed in the open field, corresponding to 0.02 g dry stigma. Results from this small-scale experiment correspond to a potential yield of up to 203.4 g/ha of dry stigma under a single-axis tracking AV system, compared with 90.1 g/ha in open-field conditions. In addition, intercropping saffron within AV systems improved land-use efficiency by reducing land loss due to PV panel structures from 10% to 3%. As a high-value, low-input crop, saffron cultivation under AV systems therefore offers potential economic benefits for farmers. Overall, these results demonstrate that AV systems can positively influence saffron productivity at northern latitudes. However, further research is required to assess the long-term productivity of saffron under AV systems, as the crop is cultivated for a single year in this study but can be maintained for 5 to 10 years.

Agrivoltaic (AV) systems offer a promising pathway for energy conversion on agricultural land while supporting crop cultivation. However, their widespread adoption is hindered by agronomic uncertainties, including the lack of quantitative and qualitative crop performance data across regions and system configuration. In this context, the present study investigated the growth, yield, and quality of oat cultivated under an AV system. The experiment was conducted near Fellingsbro, in central Sweden, under rainfed conditions. In May 2024, oats were cultivated under a vertical AV system and in a reference area without photovoltaic (PV) modules. Leaf area index (LAI), dry matter yield, thousand kernel weight (TKW), and crude protein and fat content were measured to evaluate morphological, productive, and qualitative parameters. The results show that LAI was significantly lower under the AV system. An 8.4% reduction in total biomass (kernel + straw, in dry matter kg/ha) was observed under the AV system; however, this difference was not statistically significant. TKW, crude protein, and crude fat contents (% of dry weight) under the AV system showed no significant differences compared to the reference. These findings highlight that moderate shading under vertical AV systems (with a seasonally average reduction of 41.8% in photosynthetically active radiation in our case) can sustain stable oat yields both quantitatively and qualitatively. Additionally, relative crop yields varied across the AV system row, where yield in the west, middle, and east part were 110.1%, 85.9%, and 78.8% with PAR reductions of 42.3%, 36.6%, and 46.6%, respectively. It highlights the importance of diurnal shading dynamics. This is likely due to favorable west-side shading, which provides full sun during peak photosynthesis and beneficial afternoon shade in accordance with the peak of vapor pressure deficit.

This study evaluated the economic feasibility of various agrivoltaic (APV) systems in Sweden using Monte Carlo simulations, focusing on two different scenarios: dynamic electricity prices (Wholesale Price Scenario) and fixed electricity prices (Power Purchase Agreement (PPA) Scenario). The assessment of small-scale systems (1-ha) revealed that using average PPA prices (0.058 €/kWh) can lead to unprofitable installations, particularly in the absence of government incentives. In contrast, simulations assessing the wholesale price scenario show that all systems achieve positive net present value (NPV), highlighting the importance of accounting for multi-year price variability. By exploring economies of scale for vertical APV systems, the Capital Expenditure (CAPEX) decreases significantly from 0.06 €/kWh for 1 ha to 0.043 €/kWh for 5 ha, representing a 28% reduction. At a scale of 5 ha for the vertical APV system (equivalent to 1.9 MWp installed capacity), the mean levelized cost of electricity (LCOE) approaches the PPA price of around 0.052 €/kWh. This suggests that even at a fixed contractual price near the PPA, a somewhat larger-scale APV system can yield positive income. The profitability of APV systems improves if self-consumption is considered. The one-axis APV system performs best, with NPVs rising from 63 k€ to 133 k€ and 229 k€ at 0%, 25%, and 50% self-consumption, respectively. The vertical APV system becomes viable at 25% and 50% self-consumption, with NPVs of 15 k€ and 91 k€, respectively. The elevated APV system is the least attractive option for the investigated location and assumptions.