As an Amazon Associate, Akermin earns from qualifying purchases. Product links on this page are affiliate links.
You cannot describe a bifacial module from the front alone, because its output responds to an equivalent irradiance that adds the rear contribution weighted by the module bifaciality coefficient. With one front facing sensor the three defensible options are to shade the rear so the measurement becomes a conventional front side one, to estimate the rear contribution as a range and report the result as a band rather than a value, or to compare the module only against itself and its neighbours under matched conditions so that the unknown rear term largely cancels. What is not defensible is quoting a front side irradiance beside a bifacial output and treating the pair as a performance figure.
Front irradiance alone cannot describe a bifacial module
A bifacial laminate has cells that generate from either face, and the back is transparent or framed with a clear backsheet so light that reaches it is converted too. The rear is less effective than the front, because of cell metallisation, the optics of the rear encapsulation and the structure behind the module, and the datasheet expresses that with a bifaciality coefficient: the ratio of rear performance to front performance under identical irradiance. Module families differ, so the figure comes from the datasheet rather than from a rule of thumb.
The standard way of handling this collapses the two faces into one number. Equivalent irradiance is the front irradiance plus the rear irradiance multiplied by the bifaciality coefficient, and the module behaves as though it were a conventional module receiving that total. The arithmetic is simple. The problem is that it needs a rear reading, and almost every handheld irradiance instrument has one sensor on one face of its body.
Which means a front only reading is not slightly incomplete, it is missing a term that can be a meaningful fraction of the total. Pairing it with an electrical measurement and calling the result a performance ratio produces a figure that is wrong in a consistent direction: the module always looks better than it is, because it is being credited with output from light you did not count.
What the rear side actually collects
Three sources feed the back of a module. Ground reflected light is usually the largest, and it is governed by the reflectance of whatever lies beneath: a white roof membrane and fresh snow return a large fraction of what lands on them, bare soil and grass return much less. Diffuse sky radiation reaches the rear wherever the module can see open sky past the mounting, which favours high clearance and wide row spacing. Reflection from neighbouring rows adds a smaller third contribution.
None of those arrive evenly. The rear of a module sees more light near its lower edge, where the ground is close and the view is open, and less near the upper edge and anywhere the torque tube, rails or junction box shade it. A single rear reading taken at one point is therefore already an approximation of a varying field, which is worth remembering before treating a rear sensor as the complete answer either.
The uneven field also means that repeating a measurement on different days without matching the ground condition is not a repeat at all. Dew, frost, a wet surface or mown grass all shift the reflectance underfoot. The general problem of deciding whether two readings count as the same measurement is set out in what repeatability means in a measuring instrument.
Bounding the error when you only have one sensor
If the rear cannot be measured, it can still be bracketed. Take a plausible low and high value for the rear irradiance from the ground conditions in front of you, apply the bifaciality coefficient to both, and carry the pair through the calculation. The result is a band, and a band that is honestly labelled is more useful than a single figure that is quietly wrong.
Work it through. Front irradiance reads 820 W/m2, the datasheet gives a bifaciality coefficient of 0.75, and the ground conditions suggest rear irradiance somewhere between 60 and 140 W/m2. The low case gives an equivalent irradiance of 865 W/m2 and the high case gives 925 W/m2. The expected current therefore spans about seven percent, which immediately tells you something useful: a module that comes in five percent under the central expectation has not been shown to be faulty, because the uncertainty you brought with you is larger than the discrepancy you found.
| Input | Low rear case | High rear case |
|---|---|---|
| Front irradiance measured | 820 W/m2 | 820 W/m2 |
| Rear irradiance estimated | 60 W/m2 | 140 W/m2 |
| Bifaciality coefficient from datasheet | 0.75 | 0.75 |
| Rear contribution after weighting | 45 W/m2 | 105 W/m2 |
| Equivalent irradiance | 865 W/m2 | 925 W/m2 |
| Spread carried into the expected current | About seven percent | About seven percent |
A second option removes the unknown instead of bounding it. Hold an opaque board close behind the module, let the reading settle, and the rear contribution drops to something negligible. The measurement is then a conventional front side one and can be compared directly against a front side rating. It works when there is room behind the module and fails when there is not, which on a close mounted installation is most of the time. Getting any view of that gap at all sometimes calls for the kind of instrument described in borescopes for side view inspections.
Air movement behind the module changes the temperature too
Bifacial installations are usually mounted with clearance so the rear can see the ground, and that clearance also lets air move across both faces. Convection is the main route by which a module sheds the energy it does not convert, so the same module in the same sunshine runs cooler on a breezy day than a still one, and cooler than a close mounted module with a dead air space behind it. Since cell temperature is one of the two conditions every output figure must be read against, wind speed is part of the explanation for a temperature you record.
Measuring it is straightforward. The BTMETER hot wire anemometer states air velocity measurement from 0.2 to 67 mph with a quoted accuracy of plus or minus 5 percent, airflow volume in cubic feet and cubic metres per minute, a temperature range of 32 to 113 degrees Fahrenheit, a telescopic probe for reaching into confined spaces and maximum, minimum and average values on the display. The averaging function is the one that matters outdoors, since a single instantaneous gust is a poor description of the cooling the module experienced over the preceding minutes.
A thermal look at the laminate is tempting as a second observation, and it needs its limits stated. The AMG8833 thermal imaging array module states an 8 by 8 infrared sensor with a detection range quoted up to 7 metres, power over USB at 5 V, a setup time of around 50 ms and a body weighing 17 grams. It publishes no temperature accuracy and no thermal sensitivity figure, so both are: not stated in the listing bullets. An 8 by 8 array is 64 pixels in total, which across a full size module is far coarser than a single cell, so it can show a broad temperature gradient across the laminate and cannot resolve an individual hot cell. That is a real limitation rather than a criticism, and knowing it prevents a coarse image from being read as evidence it cannot supply.
Reporting a front side measurement honestly
Write the record so that someone else can tell what was and was not measured. State the front irradiance with the instrument and how it was held. State explicitly that rear irradiance was not measured, and give the range you assumed and why. State the ground surface and its condition, because that is the single biggest determinant of the term you estimated. State the cell temperature and the wind speed beside it. Then give the result as a band.
Instruments that bring several of those quantities together reduce the chance of a mismatched pairing. The Fluke 393FC clamp meter with irradiance meter states a CAT III 1500 V rating, a thin jaw for crowded combiner boxes, direct current power measurement with audio polarity and visual continuity, MC4 test leads, and measurement of solar irradiance, ambient and module temperature together with array orientation and tilt. It gathers the electrical side and the conditions in one instrument, and it still has one irradiance sensor facing one way, which is the constraint this whole page is about.
Finally, resist the pull to act on a number that moves for legitimate reasons. A bifacial module measured from the front will read differently on a wet morning and a dry afternoon over the same ground, and neither reading is a fault. Treating an expected variation as a defect is the same mistake as blaming a monitor for a reading that a normal activity caused, described in whether cooking can affect a monitor reading, and the discipline of setting thresholds that tolerate normal movement is covered in setting useful alerts without overreacting.
Frequently Asked Questions
What is the bifaciality coefficient?
It is the ratio of the rear side performance of a module to its front side performance, measured under the same irradiance and published on the datasheet. A coefficient of 0.75 means light arriving on the back is worth three quarters as much as the same light arriving on the front. It is the factor that lets a rear irradiance reading be folded into a single equivalent figure.
Can I just shade the rear and measure normally?
Yes, and it is the cleanest option when it is physically possible. An opaque board held close behind the module reduces the rear contribution to something negligible and turns the measurement into a conventional front side one that can be compared with front side ratings. The difficulty is practical rather than theoretical, since most mounting arrangements leave no room for it.
How much does the rear side typically contribute?
It depends on ground reflectance, mounting height, row spacing and how much structure sits behind the module, so a single figure would be misleading. The honest approach in the field is to estimate a range rather than a value, propagate that range through the calculation and report the result as a band. A number without a band implies knowledge nobody has from the front.
Does the ground surface really change the reading?
Substantially. A white membrane roof, bare concrete, grass and fresh snow reflect very different fractions of the light that lands on them, and the rear side sees only reflected and diffuse light. Two identical modules on the same site over different ground will legitimately produce different outputs under the same sun.
Why does wind speed belong in the record?
Because it sets how fast the module sheds heat, and cell temperature is one of the two conditions any output figure has to be interpreted against. A bifacial module usually has clearance behind it, so air moves on both faces and the module runs cooler than a close mounted one in the same sunshine. Recording the wind speed explains a temperature that otherwise looks anomalous.
Is a thermal image a substitute for a rear irradiance reading?
No. A thermal image shows where the module is hot, which can reveal shading or a failed cell, but temperature is a consequence of many things and does not recover the irradiance that caused it. It is a useful second observation alongside the electrical measurement rather than a replacement for the condition you could not measure.