Agrivoltaics: the hidden airflow beneath the panels

Agrivol­ta­ics, which involves installing photo­vol­ta­ic pan­els above agri­cul­tur­al land, is increas­ingly emer­ging as a prom­ising solu­tion for com­bin­ing renew­able elec­tri­city gen­er­a­tion with agri­cul­ture. This tech­nique provides clean energy without the need for new land and helps pro­tect crops from what can some­times be excess­ive sun­light. Pub­lished in early 2026 in the journ­al Energy Nex­us¹, our study expands on this under­stand­ing of agrivol­ta­ics by examin­ing a pre­vi­ously under-explored aspect: the way in which the pan­els alter air cir­cu­la­tion, and con­sequently the exchange of water and energy between plants, the soil and the atmo­sphere. Far from call­ing into ques­tion the value of agrivol­ta­ics, this research provides valu­able tools for fur­ther refin­ing the design of such install­a­tions and max­im­ising their benefits.

Too much sun can dam­age plants. Dur­ing the 2003 drought in Italy, maize yields plummeted by 36 %. In 2012, a sim­il­ar drought in the United States res­ul­ted in the loss of around 26 % of the pro­jec­ted maize har­vest. With glob­al warm­ing intensi­fy­ing droughts, and glob­al food demand set to rise by 45 % by 2050, pro­tect­ing crops is becom­ing a major chal­lenge. And this is pre­cisely where agrivol­ta­ics has a key role to play.

By par­tially shad­ing crops, sol­ar pan­els reduce evapo­tran­spir­a­tion, i.e. the water that plants release through their leaves. This water is thus used more effi­ciently, as it remains avail­able in the soil for longer. As such, the risk of soil dry­ing out dur­ing dry peri­ods is reduced, as is water stress in plants. These com­bined effects enable crops to be grown with min­im­al losses but redu­cing sol­ar radi­ation in this way to lower evapo­tran­spir­a­tion can only be achieved through a loc­al­ised and tem­por­ary reduc­tion in plant pho­to­syn­thes­is, and a reduc­tion in evapo­tran­spir­a­tion caused by reduced wind does not neces­sar­ily achieve this. What’s more, lim­it­ing evapo­tran­spir­a­tion too severely can be harm­ful if the plant needs these pro­cesses to cool itself dur­ing very hot periods.

The real expert­ise lies in find­ing the right bal­ance of shade depend­ing on the cli­mate and the crop.

Nev­er­the­less, when con­di­tions are optim­ised accord­ing to the spe­cif­ic crop being grown, some stud­ies have shown that, under drought con­di­tions, beans or toma­toes grown under pan­els cap­ture more CO₂ and yield bet­ter res­ults than those grown in open fields. The real expert­ise there­fore lies in find­ing the right bal­ance of shade depend­ing on the cli­mate and the crop. This is a bal­ance that the most suc­cess­ful agrivol­ta­ic pro­jects have already mastered, and which this study helps to fur­ther refine.

To estim­ate evapo­tran­spir­a­tion, agro­nom­ists have been using well-estab­lished for­mu­las for dec­ades, the best known being the so-called “PM-FAO56” equa­tion, pop­ular­ised by the Food and Agri­cul­ture Organ­isa­tion (FAO) of the United Nations. These for­mu­las are based on simple meas­ure­ments: sol­ar radi­ation, air tem­per­at­ure, humid­ity and wind speed. They have proven their worth in open fields. When applied dir­ectly to innov­at­ive applic­a­tions such as agrivol­ta­ics, they need to be veri­fied and, if neces­sary, adjus­ted. The for­mu­las are based on three assump­tions: con­sist­ent ground con­di­tions, a dir­ect cor­rel­a­tion between wind speed and tur­bu­lence, and wind meas­ure­ments taken at a ref­er­ence height that are rep­res­ent­at­ive of con­di­tions at plant level. Until this study, no one had dir­ectly meas­ured the tur­bu­lence of the air­flow beneath agrivol­ta­ic panels.

To get a clear­er pic­ture, we installed son­ic anem­o­met­ers (sensors cap­able of meas­ur­ing the three wind com­pon­ents ten times per second, and there­fore tur­bu­lence) on SIRTA’s agrivol­ta­ic exper­i­ment­al plat­form, a research site near Par­is. Three sensors were deployed: one in a con­trol area without pan­els, and two at the heart of the agrivol­ta­ic install­a­tion: one beneath the pan­els and the oth­er above them. These meas­ure­ments, com­bined with soil mois­ture and radi­ation sensors, were col­lec­ted over sev­er­al months between Novem­ber 2024 and June 2025. In addi­tion, we util­ised state-of-the-art numer­ic­al sim­u­la­tion soft­ware (code_saturne, developed by EDF) to recon­struct the three-dimen­sion­al air­flow around the pan­els and estim­ate energy and water exchanges at vari­ous points with­in the install­a­tion, to assess their spa­tial vari­ations. This com­bined exper­i­ment­al and numer­ic­al approach clearly illus­trates the level of tech­nic­al soph­ist­ic­a­tion now being applied to agrivol­ta­ics, a sec­tor bene­fit­ing from con­sid­er­able research invest­ment to optim­ise each indi­vidu­al design parameter.

The wind speeds under the pan­els are gen­er­ally lower than in the con­trol area, some­times by as much as 70 %. This is actu­ally good news for crops sens­it­ive to drought, as a lower wind speeds mech­an­ic­ally lim­it water losses through evapo­tran­spir­a­tion. Fur­ther­more, air tur­bu­lence is, in some places, sig­ni­fic­antly stronger beneath the pan­els, with dif­fer­ences exceed­ing 300 % com­pared with the con­trol area under cer­tain con­di­tions, a phe­nomen­on that demon­strates the extent to which the pan­els cre­ate a dis­tinct and rich micro­cli­mate, which can now be pre­cisely char­ac­ter­ised thanks to this study. This find­ing chal­lenges one of the assump­tions under­ly­ing con­ven­tion­al evapo­tran­spir­a­tion equa­tions, namely that tur­bu­lence always fol­lows wind speed.

This study shows that the rela­tion­ship is more nuanced in agrivol­ta­ic envir­on­ments, requir­ing even more pre­cise mod­els, tail­or-made for these install­a­tions. We have also observed that wind speed, depend­ing on the dir­ec­tion and tilt of the pan­els (which, in some install­a­tions, can pivot through­out the day to track the sun’s path), can be reduced by between 10 % and 60 % loc­ally. This there­fore provides an addi­tion­al lever that design­ers can util­ise to fine-tune crop expos­ure accord­ing to their spe­cif­ic needs. The ver­tic­al wind pro­file, which is more com­plex beneath the pan­els than in open fields, reaches a loc­al max­im­um just below the pan­els, then decreases in the wake zone they cre­ate, before rising again above them. This is excel­lent news, as it indic­ates pre­cisely where and how to take meas­ure­ments to obtain data that is truly rep­res­ent­at­ive of what is hap­pen­ing at crop level, rather than rely­ing on meas­ure­ments taken high­er up, which do not always reflect the real­ity on the ground.

Numer­ic­al sim­u­la­tions con­firm and refine these field obser­va­tions. They show that con­di­tions vary between loc­a­tions: dir­ectly beneath a pan­el, between two rows, or near the edge of a photo­vol­ta­ic install­a­tion. Far from being a draw­back, this diversity of micro­cli­mates with­in a single plot can become an asset: it makes it pos­sible, in the long term, to devise plant­ing strategies tailored to spe­cif­ic zones, tak­ing advant­age of the most favour­able con­di­tions at each loc­a­tion with­in the install­a­tion. This same diversity is reflec­ted in soil mois­ture levels. In spring, the soil loc­ated dir­ectly beneath a row of pan­els retains sig­ni­fic­antly more water (around 30 %) than that situ­ated between two rows (which drops to 18 % under the same con­di­tions), fur­ther con­firm­a­tion that the pan­els effect­ively ful­fil their role in con­serving soil mois­ture, with the added bene­fit of nat­ur­ally col­lect­ing rain­wa­ter that runs off their surface.

By com­bin­ing advanced numer­ic­al sim­u­la­tions with arti­fi­cial intel­li­gence, the research­ers envis­age even more effect­ive design tools.

These new find­ings have prac­tic­al implic­a­tions. Depend­ing on the cal­cu­la­tion meth­od used and the exact loc­a­tion of the meas­ur­ing sensor, estim­ates of evapo­tran­spir­a­tion under the pan­els can vary sig­ni­fic­antly, where­as in open fields, the dif­fer­ent meth­ods tend to yield sim­il­ar res­ults. It is pre­cisely by identi­fy­ing this mar­gin of uncer­tainty that the study greatly bene­fits the sec­tor. It enables pro­ject developers to fine-tune their fore­casts right from the design stage, thereby ensur­ing that install­a­tions are set up in a way that allows for even bet­ter con­trol of their agro­nom­ic and energy potential.

In the longer term, the study opens up an excit­ing pro­spect: the devel­op­ment of evapo­tran­spir­a­tion mod­els spe­cific­ally designed for agrivol­ta­ic sys­tems. By com­bin­ing advanced numer­ic­al sim­u­la­tions with arti­fi­cial intel­li­gence, the research­ers envis­age the pos­sib­il­ity of devel­op­ing even more effect­ive design tools, cap­able of incor­por­at­ing the geo­metry of the pan­els, their tilt and loc­al weath­er con­di­tions. This will enable future agrivol­ta­ic install­a­tions to sim­ul­tan­eously optim­ise both elec­tri­city gen­er­a­tion and agri­cul­tur­al yields. This break­through con­firms, once again, the enorm­ous poten­tial of this sec­tor, which lies at the cross­roads of the energy trans­ition and agri­cul­tur­al resilience.