Two different questions get asked with the same words, and they have opposite answers.
Does a panel produce anything when the sky is overcast? Yes, comfortably. Does a panel produce anything useful when a chimney casts a hard shadow across one corner of it? That depends on the module's internal wiring, and the cost is far higher than the shaded area suggests.
Overcast is not shade
Sunlight reaching a panel arrives two ways. Direct beam radiation comes straight from the disc of the sun. Diffuse radiation is scattered by cloud and atmosphere and arrives from the whole sky dome. On a heavily overcast British day the direct component is close to zero and the diffuse component is doing all of the work.
That is why the reference array in these guides still delivers 134 kWh in December. The sun is barely above the horizon, the days are short, and much of the month is cloud. It generates anyway, at roughly 29% of its July output.
A panel under general cloud is uniformly lit. Every cell in it is receiving about the same reduced amount, so the module behaves normally at a lower power. Nothing pathological happens.
Hard shade is the problem, and the reason is wiring
Cells inside a module are wired in series. Current has to pass through every cell in turn, which means the current through the string is set by the cell passing the least. Shade one cell heavily and, without protection, it throttles every other cell wired with it. A module that is 2% shaded can lose far more than 2% of its output, and the shaded cell heats up as it absorbs the power the others are trying to push through it.
Bypass diodes exist to stop that. A typical 60 or 72 cell module carries three, each spanning roughly a third of the cells. When one cell is shaded enough to hold back its group, the diode across that group conducts and routes current around it.
The diode turns a disproportionate loss into a proportionate one, and it stops the hot spot. It does not give you the third back.
What it costs the rest of the array
With bypass diodes working, a shaded module in a string of ten drops its own contribution by a third. The string keeps running. The array loses roughly one thirtieth, not a tenth and certainly not everything.
Two things can make it worse than that.
- Partial shading distorts the string's current-voltage curve so that it has several local power peaks instead of one. The inverter's maximum power point tracker has to find the highest. Good equipment sweeps the full curve periodically and finds it. Weaker equipment settles on a lower peak and stays there, and the loss is invisible unless someone looks at the data.
- Shade moves. A chimney shadow crosses the array through the day, so the affected sub-strings change hour by hour. Averaged across a year the total can be substantial even though no single moment looks disastrous.
This is the case that per-module electronics are sold for. Microinverters and DC optimisers give each module its own maximum power point, so a shaded module's problem stops at that module instead of influencing the string. Whether that is worth the extra cost is set out in microinverters against a string inverter.

Where shade comes from, in rough order of cost
| Source | Character | Why it costs what it costs |
|---|---|---|
| Chimney or flue | Hard, close, moves through the day | Sweeps across sub-strings for hours; present every day of the year |
| Neighbouring building | Hard, fixed geometry, seasonal | Worst at low winter sun, when output is already scarce |
| Deciduous tree | Hard, seasonal, growing | Lighter in winter when bare; gets worse every year as the tree grows |
| Evergreen tree | Hard, year-round, growing | No seasonal relief at all |
| Overhead cable or aerial | Thin, hard, sweeping | Small area but can hold back a sub-string as it crosses |
| Bird droppings, lichen, leaf litter | Hard, static, localised | Opaque over a few cells; the one case where cleaning is clearly worth it |
| General cloud | Soft, uniform | Reduces output smoothly; no sub-string effect |
Everything above the last row is opaque and localised. That is what triggers a bypass diode. Cloud does not.
Working out what your own shade costs
The only honest way is measurement, and an installer who is not doing it should be asked why.
- 1A shading survey from the roof plane, with an instrument that maps the skyline and the sun paths across the year. This produces a percentage, not an adjective.
- 2A yield model with that shading applied. PVGIS accounts for the terrain horizon from elevation data, so it already knows about hills. It does not know about your chimney, and the difference between those two is exactly the number you need.
- 3Module-level monitoring once the system is running, if the array has per-module electronics. It shows which modules underperform and when, which is the only way to catch an MPPT settling on the wrong peak.
One practical point on layout. If part of a roof is reliably shaded at the times of day that matter, putting fewer panels on it is often better than putting panels there and adding electronics to manage the consequences. That decision belongs in the design, before anything is bought.
Our calculator models an unshaded roof and says so. If your array has significant object shading, the figure it produces is an upper bound rather than an estimate.
Monthly output figures modelled with PVGIS 5.2 (European Commission Joint Research Centre), PVGIS-SARAH2 radiation database, 2005–2020 average. 4 kWp crystalline silicon, roof-mounted, 14% system loss, Birmingham, UK. Retrieved 2026-09-12. Bypass diode counts and module wiring are standard for 60 and 72 cell crystalline modules; confirm against your own module datasheet. Source