Vertical Bifacial Solar Panels: A Winning Solution for Agrivoltaics and Farmers

In 2022, we developed our experimental design and research methods, grew crops in full sun in a reduced version of the full experimental layout, and collected reference data from those crops. In 2023, we designed shade fence structures to mimic the conditions expected in three different locations within the agrivoltaic system. We grew beets and carrots next to the shade fences and in randomized full-sun locations without shade fences. We collected data on microenvironments and crop yield in each shade level. In 2024, the full experiment will be conducted, including both an agrivoltaic and a full-sun reference area, and a repeat of the shade fence experiment. We will compare crop outcomes at five east-west locations between the solar arrays, which will be shaded during different portions of the day. The main experiment will compare beets and carrots in all five locations. A secondary experiment will investigate saffron production in all five locations, and directly under the rows of solar modules. We will also compare bean and pea outcomes when trellis-grown in the empty space directly under solar modules. This will be in the northmost section of the array, which deviates from the standard stacked upper and lower solar module design by having only upper solar modules. A fourth agrivoltaic experiment will investigate lettuce production in the east center, center, and west center locations at the northmost edge of the solar arrays.

In 2024, before the start of the growing season, a solar array comprising three rows of vertical bifacial photovoltaic modules will be installed by iSun, Inc. at the University of Vermont (UVM) Horticulture Research and Education Center. The arrays will be in a certified organic field maintained by the UVM Catamount Educational Farm. The bifacial solar modules will be installed vertically in a landscape orientation using pounded steel posts and steel racking (Next2Sun Technology GmbH, Dillingen/Saar, Germany), and the modules will face east and west. The rows of modules will run north and south, each looking like a tall fence. There will be a 9.14 m (30 ft) distance east or west between each of the three rows. Each row will consist of eleven pairs of stacked solar modules with two additional single modules at the northmost end of the row, each with trellis-like racking below them instead of another solar module. Solar modules will be ~1.8 m (~6 ft) long and ~1.0 m (~40 in) high. There will be ~0.61 to 0.91 m (~2 to 3 ft) of empty space below the lower modules and ~15.2 cm (~6 in) between stacked modules, resulting in an array ~2.79 to 3.15 m (~9.2 to 10.3 ft) tall. The power capacity of this installation will be 50 kW.

In 2024, we will grow certified organic ‘Boro’ beets (Beta vulgaris ‘Boro’) and ‘Negovia’ carrots (Daucus carota ‘Negovia’) (High Mowing Organic Seeds, Wolcott, VT) in a 2x5 strip-plot design with two levels of crop type (beet and carrot) and five levels of crop bed location (west, west center, center, east center, and east) between two rows of solar modules. Replications will consist of five adjacent blocks (south, south center, center, north center, and north) independently randomized for crop type. Each block will be the length of two solar modules, ~3.66 m (~12 ft), and will contain all five crop bed locations. The crop beds will be located in the 9.14 m (30 ft) distance between two solar arrays. Each crop bed will be ~1.22 m (~4 ft) wide, with ~0.30 m (~1 ft) of pathway between them, leaving ~0.76 m (~2.5 ft) of space between each solar module row and its neighboring crop bed. Half of each block will be planted with beets, and half with carrots, with the same crop type aligned within the same block across each of the five crop bed locations. Beets and carrots will be planted in three rows within each crop bed, each with its own line of drip tape for regular irrigation. We selected beets and carrots as the main research crops due to their relatively high yield per land area and market price by weight, and the moderately high number of farms and land area in Vermont growing these crops. In other words, these crops may be profitable for many farmers in Vermont, so it will be useful to evaluate their compatibility with a vertical agrivoltaic system in Vermont. We will follow similar methods for lettuce, beans, peas, and saffron. The aforementioned research design is statistically valid as approved by Dr. Takamaru Ashikaga, Professor Emeritus of Statistics, UVM.

In 2022, to support our 2024 agrivoltaics experiment, we grew specialty crops in three crop beds in the full-sun reference field. We otherwise followed the randomization plan for future years with three blocks, each randomized separately to determine which crop type would be on the north or south end of the block. We direct seeded beets and carrots on July 8 and harvested a random sample of beets on September 2 and carrots on October 7. We also grew saffron in the same three crop beds, to the south of the beets and carrots. We planted saffron corms on August 24 and harvested flowers as they emerged between October 28 and December 8. We grew beans and peas on trellises at the southmost end of the outer two research crop beds. We direct seeded beans and peas on July 26 and harvested all mature beans on October 2 and peas on October 2, October 21, and October 31. We grew lettuce at the southmost end of the center research bed. We planted lettuce seeds in plug trays on July 7, transplanted seedlings to the field on August 3, and harvested mature heads on September 30.

As planned for future years, in 2022 we collected yield-related data from a random subsample of individual plants of each crop type in each full-sun reference subplot (each crop bed in each block). We recorded the total number of beet roots per randomly sampled area and collected data for individual beet roots with leaves removed, including mass, diameter at widest point, and a subjective damage rating. We recorded the total number of carrots per randomly sampled area and collected data for individual washed carrots with leaves removed, including mass, diameter at widest point, length, and subjective damage and quality ratings. For saffron, we observed mass as the total combined dry weight of all stigmas collected from each subplot (each crop bed in each block) over the entire harvest period. We counted bean and pea pods, and collected data for each pod including mass, width at widest point, length, number of beans or peas per pod, and subjective damage and quality ratings. We counted lettuce heads, and collected data for each washed head including mass, circumference at widest point, length, and a subjective damage rating.

In 2023, to mimic conditions predicted for our 2024 agrivoltaics experiment, we grew beets and carrots next to shade fences in a 2x4 factorial completely randomized block design with two levels of crop type (beet and carrot), four levels of shade (no shade, brief morning shade, brief afternoon shade, and moderate duration afternoon shade), and three blocked replications (south, center, and north). We installed shade fences in two crop beds ~1.22 m (~4 ft) wide and spaced ~4.88 m (~16 ft) apart to limit shading of each crop by fences in the other crop bed. Each crop bed had three rows of crops, with predicted shade treatments applied to only two of those rows due to the short height of the shade fences, which was required for structural stability. Each “morning shade” fence was offset to the east of the two main crop rows, with the third row to its east. Each “afternoon shade” fence was offset to the west of the two main crop rows, with the third row to its west. Shade fence sections were 3.66 m (12 ft) long between two 2.13 m (7 ft) steel t-posts buried 0.91 m (3 ft) deep. Shade fence panels were made of 90% UV light-blocking woven black plastic (HDPE) shade cloth (Coolaroo USA, GALE Pacific USA, Inc., Charlotte, NC), ~3.20 m (~10.5 ft) long, with a 1.22 m (4 ft) fiberglass stake in a long sleeve at each end of the panel for structural support. Fence panels were centered between end posts, attached with polyester paracord using adjustable knots, and attached to a 1.83 m (6 ft) t-post at the center of each panel, buried 0.61 m (2 ft) deep. For brief shade treatments, fence panels were 0.61 m (2 ft) tall, installed with 0.61 m (2 ft) of above-ground empty space below the panels. For moderate duration shade treatments, panels were 0.91 m (3 ft) tall, installed with 0.15 m (0.5 ft) of above-ground empty space below the panels.

We selected the shade fence parameters using the Dual-Use Shading Analysis Tool from the Massachusetts Department of Energy Resources, which predicts shade on crops in custom theoretical agrivoltaic installations. We designed the shade fences so that the shade cast on the two crop rows of interest would match the range of shade levels (in duration and time of day in shade) predicted for various crop beds in the vertical agrivoltaic system in 2024. To reduce wind damage to the shade fences, we cut four 42.55 cm (16.75 in) diameter semicircular wind vent flaps in a line across the brief shade treatment panels and eight flaps in two lines across the moderate duration shade treatment panels. We added larger secondary flaps on the panels next to carrots to limit sunlight leakage in low-wind conditions. We will also add these secondary flaps to panels next to beets when we repeat this experiment in 2024.

For the 2023 shade fence experiment, we direct seeded beets and carrots on June 15 and harvested beets on August 23, August 29, and September 6 (south, center, and north blocks, respectively) and carrots on September 12, September 21, and October 10 (south, center, and north blocks, respectively). We harvested all beets from the two rows with applied shade treatment (or all three rows for full-sun controls) in the northmost 1.83 m (6 ft) of each 3.66 m (12 ft) experimental section, not counting a 0.30 m (1 ft) buffer on the north end. We harvested all carrots from the two rows with applied shade treatment (or all three rows for full-sun controls) in the northmost 1.22 m (4 ft) of each 3.66 m (12 ft) experimental section, not counting a 0.30 m (1 ft) buffer on the north end. We collected yield-related data from all harvested beets and carrots, except only from a randomly selected two-thirds of the harvested carrots in control sections, using similar measures as used in 2022 for full-sun reference crops. We also measured the total fresh weight of leaves removed from beets or carrots from each shade level in each replication. We washed both beets and carrots before collecting data.

Starting on August 11, 2023, we conducted microenvironmental monitoring using sensors that automatically logged data every five minutes. We measured sunlight as PAR (photosynthetically active radiation), soil moisture, soil temperature, ambient temperature, and relative humidity for each replication of each shade treatment for carrots. In 2024, we will use this complete set of sensors in the west, center, and east crop beds of the agrivoltaic research area. We will use only PAR sensors in the west center and east center crop beds. We will use this sensor arrangement in three of the replications for the agrivoltaic experiment, in carrot sections. Also in 2024, we will collect all environmental measurements within all carrot sections of the shade fence experiment on a rotating basis: we will measure two shade treatments at a time in a single block and will move sensors to different pairs of shade treatments every few days. Finally, we will also collect all environmental measurements in the carrot sections of the center block of the full-sun reference area for the west, center, and east crop beds, and we will measure only PAR in the west center and east center crop beds. We will also collect insect samples weekly using sweeping nets, with the main intention of identifying pest incidence in each crop in each location between the rows of solar modules.