Final report for LNE22-454R
Project Information
Our hypothesis was that vertically-positioned bifacial solar arrays are a promising new technology for use in agrivoltaics, where agriculture is combined with electricity production via photovoltaic solar panels. The incorporation of vertical bifacial solar arrays into farming can conserve valuable agricultural land for food production, produce abundant energy, and save farmers money on electrical costs. Each row of solar modules in a vertical photovoltaic array occupies just 4 inches of land, and the ample space between rows facilitates the planting and harvesting of crops with heavy equipment if needed. The bifacial solar modules collect sunlight on both sides. This type of system, currently used in Europe, has only recently been installed on a small scale in the United States, where it has never been tested with vegetable crops. Effective applications of vertical solar arrays on farmland allow for crop production within their boundaries while supporting farmers’ financial stability. Additional research on the agronomic and economic impacts of vertical agrivoltaics for different crops, climates, and economies will support successful widespread adoption of this novel technology.
Our original research plan comprised a detailed evaluation of a 50 kW vertical bifacial solar array scheduled for installation at the University of Vermont (UVM) Horticulture Research and Education Center by iSun Energy, Inc., a Northeast US solar contractor. Due to many factors including state permitting, contract negotiations between iSun and UVM, supply chain issues in part related to the novelty of the system in the US, and financial challenges of our solar partner, this vertical solar array was not installed as projected in 2022. This has given us additional insights into the challenging process of installing a vertical bifacial photovoltaic array on farmland. We still believe in the benefits of this planned UVM vertical bifacial solar array for research and demonstration purposes, and we have continued to pursue options to arrange for its installation by another company. Final negotiations are underway between UVM and Norwich Technologies, and it is hoped the array will be erected by late 2026, after which the original crop research will commence. Without our dogged determination and continual urging, this would never happen.
Given the extended delays to our original research plan, a new approach to address the need for field-based data on vegetable crop growth in vertical agrivoltaic systems was developed without an actual vertical array. Our solution was to create “shade fences” (Fig. 1) that simulate the microclimatic conditions crops would experience within a vertical solar array. By growing plants around these “shade fences” we could evaluate crop production under conditions comparable to the vertical arrays. This would provide insights into the suitability of vertical arrays for farming in our Northeast US climate. We focused on beet and carrot crops, both commonly grown in Vermont with relatively high yields and market prices by weight. In 2022, we collected full-sun reference data, then conducted a shade fence experiment in one field in 2023 and repeated the shade fence experiment in two fields in 2024. The shade fence experiments tested three levels of shade expected within a vertical solar array, randomized with full-sun areas as controls. If our hypothesis of robust crop yields is correct, then limits to solar power development on agricultural lands will be minimized, providing opportunities for farmers to adopt agrivoltaics to improve their financial security, produce electricity, and maintain high levels of agricultural production.
A vertical bifacial solar array (50kW capacity), compatible with mechanized agriculture and never tested in the US, will be evaluated at the UVM Horticulture Research Center. A variety of high-value specialty crops (vegetables, herbs, etc.) will be grown in and outside the array to determine the suitability of this system for Northeastern farms. Data on energy generation, crop yield and quality, light conditions, environmental impacts on the crops, and economic aspects will be compiled to demonstrate the agricultural opportunities and drawbacks of this vertical system for farmers compared with conventional arrays, which occupy 10 times the space.
Vertically-positioned bifacial solar panels for agrivoltaics represent a new technology and a system never before tested for vegetable cultivation in the United States. We hypothesize that this system will conserve agricultural land, produce energy greater than that produced by traditional fixed angle arrays, be suitable for crop production, and supply farmers with additional income. When installed vertically in rows running north-to-south with bifacial solar modules facing east and west, vertical bifacial solar arrays capture the sun’s energy as it rises in the east and sets in the west, shifting some electricity production to late afternoons, a critical time of day for energy usage. The vertical position of the modules prevents snow buildup from blocking sunlight, and because the system captures sunlight on both sides of the panels, a maximum amount of electricity can be produced. Of equal importance is the fact that vertical systems occupy only a very small fraction of agricultural land. While, by design, conventional fixed-tilt solar arrays limit the usable land between rows, vertical panels occupy an absolute minimum of land. Thus, agricultural fields are conserved for further crop production. It is absolutely necessary to determine if crops grown around the panels will generate high yields of marketable produce during the growing season in our area. This was one of our original research objectives. If agricultural yields are maintained in vertical agrivoltaic systems, farmers' income will benefit from combining electricity and crop production.
The parties involved in this research effort included farmers, the University of Vermont (UVM), the UVM Horticultural Research Farm, undergraduate and graduate students, and solar installers (originally iSun Energy, Inc. and recently Norwich Technologies, Inc.). Our original plan was to grow high-value vegetable crops in and outside our planned research and demonstration array at UVM to determine the suitability of this system for Northeastern farms. Because of unanticipated delays in the installation of our vertical system, vertical “shade fences” were constructed and put up to simulate the proposed vertical agrivoltaic system. Beets and carrots were selected as high-value crops to test. Data on crop yield and quality, light conditions and environmental impacts on the crops will demonstrate the agricultural opportunities and drawbacks of this type of vertical system for farmers.
Since the termination of the vertical solar contract with iSun, we have worked with UVM administration to collaborate with another solar company to install a vertical array. Negotiations between Norwich Technologies and UVM are in their final stages, and it is anticipated that planning for installation of the system will commence in April 2026.
This project was directly linked with the academic program for a UVM Ph.D. candidate. The original research design was developed as part of this project. Members of the grad student studies committee were directly involved with fine tuning the research plan. The unforeseen delays caused by the original solar installer required major redesign of the research. These redesigns were developed by the graduate student, the studies committee, and project leaders. Considerable time was invested over the project period to urge completion of the solar array installation. This involved project leaders and UVM administrators responsible for the contract negotiations. The final termination of the iSun contract was carried out by UVM legal counsel. Alex DePillis was responsible for locating an alternate solar contractor (Norwich Technologies). Decisions on proceeding with this company to achieve our goal of establishing the vertical array at UVM were made by UVM administrators from multiple departments and offices. This complicated process was led primarily by the PIs of this project. Without our urging, there would have been no hope of getting the system installed ever.
Cooperators
Research
In 2022, we developed our experimental design and research methods for the full agrivoltaic experiment, which was planned to begin after the array was installed. The plan involved three rows of solar modules running north-to-south with inter-row spacing of 30 ft east-to-west. Each row would be ~9-10 ft tall, including ~2-3 ft of above-ground empty space underneath two stacked solar modules.
In 2023, because the installation was delayed, we designed “shade fences” to mimic conditions expected in three different locations within the planned agrivoltaic system. Shade fence panels were made from 90% UV light-blocking knitted black plastic (HDPE) shade cloth We selected the shade fence parameters based on output from the Dual-Use Shading Analysis Tool from the Mass. Dept. of Energy Resources, which predicts shade on crops in customized agrivoltaic installations. Shade fences were designed so the shade cast on the two crop rows of interest would match a range of shade levels (by duration and time of day) predicted for different crop beds in the 30 ft space between two rows of our planned vertical agrivoltaic system.
Each shade fence panel was ~10.5 ft long with a 4-ft fiberglass stake in a long sleeve at each end for structural support. For “brief shade” conditions (fewer hours of the day in shade), fence panels were 2 ft tall, installed with ~2 ft of empty space below the panels. For “moderate duration” shade conditions, panels were 3 ft tall, installed with ~0.5 ft of empty space below the panels. To reduce wind damage to the shade fences, semicircular wind vent flaps were cut in the fabric (16.75 in diam.). “Brief shade” panels had four wind vent flaps in one row across the panel, while “moderate duration” panels had two rows of four flaps. We added larger secondary flaps on the panels to limit sunlight leakage in low-wind conditions. These were installed on panels next to carrots only in 2023, and on all panels in 2024. Shade fences were spaced ~16 ft apart to limit shading of crops in one bed by fences in the other bed.
We tested four shade conditions (no shade, brief morning shade, brief afternoon shade, and moderate duration afternoon shade) for two crops (beet and carrot). Treatments were blocked with three replications (south, center, and north). Each block contained one plot for each combination of crop type and shade level. We collected data on microenvironments and crop yield in each shade level. The experiment was repeated once in 2023 and twice in 2024.
For Trial 1 in 2023 in our north field, we direct-seeded ‘Boro’ beets and ‘Negovia’ carrots on June 15, and installed shade fences by July 3. Microenvironmental monitoring was initiated in August. In 2024, we seeded crops after the installation of shade fences, on June 28 for Trial 2 in our north field and on July 19 for Trial 3 in our south field. However, for Trial 3, secondary flap installation over wind vents was completed on July 26. In 2024, we conducted microenvironmental monitoring for Trial 2 only, initiated on July 7. 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. We collected yield data from 6 linear ft of two crop rows per plot. Yield data included root count, mass, diameter, length of carrots, and subjective quality and damage ratings.
In this section, we refer to the four experimental shade fence conditions as “None” (no shade), “Early” (brief morning shade), “Late” (brief afternoon shade), and “Late2” (moderate duration afternoon shade). These descriptions refer to the predicted crop shading at ground level. We reviewed environmental outcomes by trial and month as mean hourly values and included here only one representative hourly graph for each measurement, as relationships between shade conditions tended to be similar among trials and months.
Sunlight (PAR). The largest environmental differences among shade conditions were in the level of PAR (photosynthetically active radiation). Sensors showed greatly reduced PAR in the morning for Early shade (~60% maximum shade ~9 am); greatly reduced PAR in the afternoon for Late shade (~55% maximum shade ~4 pm); and moderately reduced PAR in the later afternoon for Late2 shade (~40% maximum shade ~5 pm) (Fig. 2). Compared to the unshaded (None) conditions, observed mean reductions in total daily PAR were ~20% for Early, ~15% for Late, and ~5% for Late2 shade (Fig. 3). This matched our predictions of ~15% daily shade for Early and Late shade fences, but not our prediction of ~30% daily shade for Late2 shade fences. One possible explanation for the Late2 discrepancy is the height of the PAR sensors above ground (31 in.), because our shade predictions were for ground level.
Ambient Temperature. Differences in ambient air temperatures among shade conditions were moderate, with the greatest in the late morning, and appeared to be driven by hourly shading (reducing temperatures) and the shade fence panels (increasing temperatures). Mean ambient temperatures varied by at least 8°C daily. However, the mean difference among shade conditions was less than 2°C at any given hour and was generally much lower (Fig. 4). In the late morning, the highest air temperatures tended to be in Late2 shade, followed by Late, None, and Early shade. In the mid-to-late afternoon, differences were smaller, with slightly higher temperatures in None and Early shade compared to Late and Late2. Ambient temperatures near shade fences may be higher than in vertical agrivoltaic arrays because of the proximity to shade fence panels and the shorter shadows cast by shade fences, which affect more localized areas.
Soil Temperature. Shade fences appeared to have a moderate cooling effect on the soil, particularly in the mid-to-late afternoon. Mean soil temperatures varied by ~4 to ~7 °C daily, while the largest mean difference among shade conditions was ~3.5 °C and typically lower (Fig. 5). Soil temperatures tended to be highest in the unshaded (None) plots and lowest in Late2 plots, while Early and Late plots were in between. Soil temperatures in Late2 plots increased in the morning slightly faster than for other conditions and decreased sooner; ~2 pm compared to ~4 pm for the others. This suggests that Late2 fences may have shaded the ground earlier in the afternoon and at a greater intensity than indicated by PAR measurements taken above crop height. Late2 conditions at ground level may have been closer to the predicted ~30% daily shade than the measured ~5% daily shade.
Relative Humidity (RH). Moderate differences in relative humidity among shade conditions were observed, with the greatest differences in the late afternoon. Humidity appeared to be more correlated with hourly air temperatures than shading. Mean air moisture levels varied by at least 25% RH daily. However, the mean difference among shade conditions was less than 5% RH at any given hour and was generally much lower (Fig. 6). The largest difference among shade conditions was in the late afternoon through early evening, when Late and Late2 had higher RH than None and Early shade, with the latter experiencing slightly higher air temperatures at that time. In the late morning, RH was instead slightly higher for None and Early than for Late and Late2 shade, which experienced higher air temperatures at those hours. Shade fences may impact relative humidity differently than vertical agrivoltaic solar arrays: the difference between None and Early shade was generally negligible, which does not support the hypothesis that shade increases RH.
Soil Moisture. The daily range in mean soil moisture was typically ~0.01 m3/ m3 within a given shade condition. We did not present a graph of soil moisture because the standard error tended to be very high between the three replications, causing most of the graphed lines to overlap. The timing of irrigation events may have also been responsible for the wide variations observed. We cannot make a conclusion about any differences in soil moisture between shade conditions due to the high variability of our measurements. In our future work collecting soil moisture data, we will reduce variability by applying sensor calibrations and carefully installing and monitoring the sensors. We might also calculate mean soil moistures by number of days since the last irrigation event.
Crop Yields. We evaluated crop yields primarily by bulk mass of saleable product harvested per experimental plot. We found no clear trends for differences in beet or carrot crop yields among shade conditions for the three experimental trials. However, variability between plots was substantial, which may have masked treatment differences (Fig. 7 and 8). Results were similarly variable for the other crop outcomes: bulk counts per plot, individual root mass, root diameter, and length of carrot roots. In general, our results among trials suggest that the shade conditions created with the shade fences did not have a large effect on saleable beet or carrot yields.
Shade Fences. Overall, our results suggest that shade fences may be appropriate to use as a faster, lower-cost alternative for researching crop suitability for vertical agrivoltaics. They can produce shade that matches our predictions for vertical agrivoltaics while reducing soil temperatures. However, ambient temperature and relative humidity may be affected somewhat differently by shade fences than by actual vertical solar arrays. We predicted 14-33% daily shade within our planned vertical solar array, depending on the proximity of each crop bed to the rows of vertical solar modules. The Early and Late shade fences were designed to produce ~15% daily shade, and our results suggest this was successful. The Late2 shade fences were designed to produce ~30% daily shade, and while our PAR results from 31 in above ground suggest this was not achieved, our soil temperature results indicate that ground-level shading may have been higher than measured by PAR. We recommend further research to compare crop yields and microenvironments between shade fences and vertical solar arrays to develop a better understanding of the quality of shade fence outcomes as a representation of true vertical agrivoltaic effects.
Crop Yields. Our results suggest that beets and carrots may be appropriate crops to grow in vertical agrivoltaic systems in the Northeast without major yield reductions from increased shade. However, to validate these results and obtain more certainty that vertical agrivoltaic shade conditions do not adversely affect saleable crop yields for beets and carrots, we recommend and plan to conduct further research with more replications for each crop type and larger sampled areas per plot within an actual vertical agrivoltaic array. This should produce stronger, more actionable conclusions.
Future Work. We will share these conclusions along with further experimental details, statistical analyses, and additional agrivoltaics-related outreach in a University of Vermont doctoral dissertation, a scientific journal article, a reference document for farmers, and updates to our laboratory’s agrivoltaics research website. We continue to move forward with a new solar developer and revised plans to install a vertical agrivoltaic system at the University of Vermont Horticulture Research and Education Center. We intend to secure research funding to continue evaluating crop suitability for vertical agrivoltaics in the Northeast within that agrivoltaic system. Throughout this funded project, we have developed our knowledge of agrivoltaics and have made mutually beneficial connections with researchers, educators, farmers, solar developers, government officials, and farmland owners. We have additionally planned and practiced experimental designs, protocols, and techniques for farm work, crop yield quantification, microenvironmental monitoring, and data analysis appropriate to studying crop outcomes in shade. These preparations, skills, and knowledge will be directly transferable to our future research on crop suitability in vertical agrivoltaic systems.
Education & outreach activities and participation summary
Educational activities:
Participation summary:
Our main outreach activity to date has been an online workshop on the topic of agrivoltaics in the Northeast US for farmers, solar energy providers, researchers, and other interested parties. The workshop was held on 11/28/23 and was titled “Agriculture & Solar Energy for the Northeast: A Winning Partnership? A Virtual Workshop Exploring Solar Power Opportunities for Farmers”. Here is a link to the recorded set of presentations: https://www.youtube.com/watch?v=4V8WSLHrw1E. A total of 55 people registered for the event, including farmers, undergraduate and graduate students, solar installers, and university faculty. Speakers included solar installers from the US and Europe, a UVM graduate student, a US farmer, and faculty from Rutgers University. Evaluations were received from only 3 of the participants. They rated the event as very useful. One said he gained insights on both the state of play among agrivoltaics practitioners and the state of knowledge of farmers.
In 2023, we demonstrated our shade fence research with an on-site farm tour to a group that included Vermont state government officials and individuals from UVM administration and iSun who were involved with the vertical solar array installation project. We have also shared research details on-site to a touring group of individuals who were focused mainly on integrated pest management at the UVM horticulture farm, to undergraduate farm interns, and to a local farmland owner. We have given oral presentations on our research project locally and internationally, and presented a poster at the 2024 Solar Farm Summit in Illinois, where our project plan was a finalist for Dual-Use Plan of The Year for the 2024 North American Agrivoltaics Awards. We have also consulted with other vertical agrivoltaics researchers regarding their related projects, and have been interviewed by three news reporters and by undergraduate students inside and outside of UVM.
Finally, we plan to publish our research results and a review paper in scientific journals and to present summaries of our results and conclusions to public media outlets and interested groups including farmers, solar energy providers, researchers, and local government officials. We started a website, https://site.uvm.edu/agrivoltaics/, which we will use to showcase those outreach efforts and this vertical agrivoltaics research project.
Learning Outcomes
Key areas in which farmers/and or educators reported changes in knowledge:
- Farmers: They learned that it is complicated to install an agrivoltaics system for their operation and the technology is evolving. It may be several years before vertical agrivoltaics is ready for most conventional farmers. A few pioneers may consider it, esp. if rebates are available to make it worth the time and cost.
- One Ph.D. candidate was involved with the project throughout. She has learned firsthand the challenges of installation of an agrivoltaics system, especially one that represents new technology as the vertical array we proposed.
Description of Others listed above include:
- Students: 2
- Solar installers: 14
- University researchers: 16
Project Outcomes
Over the life of the project, we submitted 11 proposals or letters of intent, of which 6 were awarded. We were also able to secure assistantship support for the graduate student working on this project through UVM. It was difficult to seek funding given that the solar array was not yet installed. With the likely installation expected in the spring of 2026, we are hopeful that we can secure funding to conduct the research.
Laura Eckman was awarded additional funding ($29,758) from a 2023 University of Vermont (UVM) Sustainable Campus Fund (SCF) Innovation Research Project Award for additional microenvironmental monitoring equipment and for compensating summer undergraduate research assistants for 2023-2024. From a 2022 InSPIRE project ASTRO Seed Grant through the U.S. Department of Energy Office of Energy Efficiency and Renewable Energy (EERE) Solar Energy Technologies Office under award DE-EE00034165, Gwenneth Sletten was awarded $3,000 for compensation of her work as an undergraduate research assistant for 2022-2024 and for additional microenvironmental monitoring equipment, and Laura Eckman was awarded $2,987 for additional microenvironmental monitoring equipment.
Laura Eckman was additionally awarded $3,342 from the University of Vermont (UVM) College of Agriculture and Life Sciences (CALS) Robert L. Parsons Fund for Professional Development in Production Agriculture for a professional development trip in 2023 to visit vertical agrivoltaic research sites, meet vertical agrivoltaics researchers in Germany, and give presentations on this research project. To attend the 2024 Solar Farm Summit and North American Agrivoltaics awards in Illinois to present a poster and represent our research group as finalists for the 2024 Dual-Use Plan of the Year Award, Laura was also awarded $730 from the UVM CALS Graduate Student Enhancement Fund and $500 from Next2Sun GmbH in addition to receiving complimentary tickets for attendance from the event organizers (a $525 value). These supplemental awards represent a significant effort on the part of UVM’s undergraduates and Laura Eckman, our PhD graduate student.
The commitment to solar power, and specifically agrivoltaics, by the Federal government has declined in the past year. However, the need to assist farmers in their effort to increase revenues has never been greater. For that reason, we have continued to advocate for installation of the vertical solar array at the UVM Horticulture Research Center, even as hopes were crushed by the dissolution of iSun. Our original research plan was sound and when the system is installed by Norwich Technologies, we are hopeful funding will be secured to conduct the work we had planned for when we applied to SARE for support.