Bifacial modules collect light on both faces. The rear-side contribution is real and well documented. The question is not whether it exists but how much of it you get on your roof, which is a completely different question and one the datasheet cannot answer.
Three things, in order of importance:
Albedo – how reflective the surface beneath the array is. This dominates everything else. White TPO membrane reflects perhaps 0.6-0.7 of incident light. Aged grey EPDM or bitumen sits around 0.10-0.15. That is a factor of five, and it translates almost linearly into rear-side yield.
Height above the surface. Rear irradiance needs room to diffuse and reach the cells. Modules mounted 150 mm off a roof see far less rear light than the same modules 800 mm up on a ground-mount frame. Commercial rooftop mounting is usually low, for wind and ballast reasons.
Ground coverage ratio. Densely packed rows shade each other's backs. High packing density, which is what you want on a constrained roof, directly suppresses bifacial gain.
| Installation | Typical bifacial gain |
|---|---|
| Ground mount, white gravel, high frame, wide rows | 9 - 15% |
| Ground mount, grass, standard spacing | 5 - 8% |
| Flat roof, new white membrane, tilted racks | 5 - 9% |
| Flat roof, aged dark membrane, low tilt | 2 - 4% |
| Flush-mounted on a pitched roof | ~0 - 1% |
That bottom row deserves emphasis. Bifacial modules laid flush on a pitched roof, with a few centimetres of air behind them and a dark surface underneath, are monofacial modules that cost more. This gets sold anyway.
Bifacial carries a premium – it varies with market conditions, but budget somewhere in the region of 3-8% on module cost, occasionally more when supply is tight. Glass-glass construction also adds weight, which matters on roofs where structural capacity is the binding constraint.
So the test is straightforward: does your expected gain exceed the premium plus any structural cost? At 6% gain against a 5% premium, marginal. At 3% gain against a 6% premium, no. At 9% gain on a bright new membrane with tilted racking, comfortably yes.
People sometimes propose painting or recoating a roof to raise albedo specifically to boost bifacial yield. Occasionally that stacks up, but usually only when the roof needed recoating regardless – the bifacial gain becomes a bonus on a job you were already funding.
Recoating purely for solar yield rarely pays. And be careful: reflective coatings degrade and soil, so the albedo you model at commissioning is not the albedo you will have in year eight.
Two things worth noting even when the bifacial gain is modest.
Glass-glass construction is generally more durable than glass-backsheet – better resistance to moisture ingress, PID and mechanical stress. Several manufacturers offer longer warranties on glass-glass for that reason, and the degradation curves tend to be flatter.
Also, bifacial modules are increasingly just what is available at competitive prices in mainstream product lines. In some procurement situations you end up with bifacial because that is the current generation, not because you chose it. Nothing wrong with that – just do not let anyone model 9% gain on a roof that will deliver 3%.
Request the albedo value used in the simulation, and ask what it is based on. If the answer is a default library value rather than the actual membrane on your building, the yield model needs redoing before you rely on it.