A Simple Approach for Clustering Common Insolation Profiles in Agrivoltaic Systems

Abstract

Heterogeneous insolation distribution in agrivoltaic systems (AVS) impacts plant growth beneath solar panels via shading and perturbed evapotranspiration profiles. Most agricultural systems models, meanwhile, assume uniform irradiance patterns across an entire field when simulating biomass production, meaning that they cannot readily account for spatiotemporal trade-offs between agricultural production and energy generation pertaining to AVS. We develop a simple approach for enumerating trade-offs between crop/pasture production and energy generation that accounts for spatial heterogeneity in insolation that typifies most AVS fields. First, long-term spatially explicit daily insolation profiles at the ground surface are produced for several layouts, including variations in PV panel orientations, row spacings, heights and tilt angles. A clustering technique was then applied to all insolation profiles to group them into rationally bounded cluster groups. The insolation profile of each cluster group was set as an input to a conventional point-based agricultural systems model to determine agricultural production under heterogeneous insolation profiles. The proposed approach is applied to a case study near Hobart, Australia, to determine an optimal layout that maximizes energy generation and plant growth associated with 81 AVS layouts. We find a manageable number (19 clusters) of point-based agricultural model scenarios capture much of the variance in insolation variability associated with varying AVS layouts. Compared with open fields, we show that AVS can amplify pasture growth rates during late spring and early summer. The optimal layout for our case study region enhanced land productivity by 47% while maintaining 80% of agricultural production compared with open-field agriculture.

Keywords

• Agrivoltaic System (AVS)
• AVS Layout Optimization
• Livestock-Based AVS
• K-Means Clustering
• Land Equivalent Ratio