Array Energy Estimator
Every number this page uses comes from you. There is no irradiance database behind it, no regional solar table, no assumed module efficiency and no equipment prices. It knows the definition of standard test conditions, the number of days in a year and the number of hours in one; everything else is yours to supply, because everything else depends on where you are and what you are installing.
The values it opens with are round numbers chosen so the shape of the output is visible on load. They are not typical, recommended or representative of anywhere in particular. Replace all of them.
The two ways to state a size
An array can be described by its physical extent or by its rating, and the two are connected by a definition rather than by a measurement.
Standard test conditions specify an irradiance of one kilowatt per square metre. A module’s efficiency is the fraction of that it converts, so a square metre of module rated at some efficiency has a peak DC rating of that fraction of a kilowatt. Multiply by the module area and you have the array’s rating in kilowatts peak.
Two cautions. The area that matters is the active module area, not the roof, the field or the fenced site — those are larger, sometimes much larger once row spacing is included. And efficiency belongs to the module, which is always lower than the efficiency of the cells inside it, because a module contains frame, spacing and interconnection that convert nothing.
If you already know the rating from a datasheet, use it directly and skip the question entirely.
Peak sun hours is a compressed year
Peak sun hours is the day’s total irradiation expressed as the number of hours it would take to accumulate at standard test conditions. A site receiving four kilowatt-hours per square metre on the plane of the array over a day has four peak sun hours, whether that arrived as a short bright day or a long dull one.
Two things about it are easy to get wrong.
It is specific to the plane of the array, not to the location. Tilt, azimuth, tracking and shading all change it, sometimes substantially, at an unchanged address. A horizontal figure quoted for a city is not the figure for a south-facing pitched roof in that city, and neither is the figure for a single-axis tracker in the field next door.
And an annual average conceals the seasonal swing, which is larger at higher latitudes. Annual energy computed from an annual-average figure is reasonable. Anything that depends on the worst month — off-grid sizing, a load that must be met without import, a battery specified to carry a system through winter — cannot be derived from the annual number and needs the monthly profile.
You must bring this figure yourself, from an irradiance dataset for your own location and array geometry. It is the single input this page most conspicuously cannot supply.
What the derate factor is hiding
The derate factor is where all the losses between the module rating and the meter are collapsed into one multiplier. In reality it is a product of several independent effects with different behaviour:
- Temperature. Module output falls as cell temperature rises above the test condition, and cells run hotter than the air around them. This is usually the largest single loss, and it is worst precisely when irradiance is highest.
- Soiling and snow. Site-specific, seasonal, and dependent on rainfall and cleaning regime.
- Shading. Near shading from objects and self-shading between rows. Its effect is disproportionate to the shaded area, because of how cells and strings are connected.
- Mismatch and wiring. Modules in a string are not identical and the weakest constrains the rest; conductors and connections dissipate a little of what is left.
- Conversion. Inverter efficiency, which varies with load, plus transformer losses where present.
- Clipping. If the DC array is deliberately oversized relative to the inverter rating, output above the inverter’s limit is discarded at the peaks. This is often a sound design choice and it always costs some energy.
- Availability. Time when equipment is out of service, plus any curtailment imposed by the network.
Compressing these into one number is a legitimate first-pass method and a poor final one. Several of them interact with each other and with the weather, so the combined effect is not stable across the year, and a single derate applied to an annual total quietly assumes it is.
Capacity factor, and a check worth remembering
Capacity factor is annual energy divided by what the array would have produced running flat out at its rating for every hour of the year. It is the usual way to compare very different kinds of generation on one axis.
For this calculation it reduces to something you can do in your head: capacity factor equals peak sun hours multiplied by the derate factor and divided by twenty-four. The array size cancels out entirely. That identity is a useful sanity check on any yield figure you are given — if the implied capacity factor is not consistent with the site’s irradiation and a plausible loss factor, one of the numbers is wrong.
It also makes the ceiling obvious. Nothing can exceed twenty-four peak sun hours, so a fixed array has a hard upper bound on capacity factor that is set by the resource and cannot be engineered around. Comparisons between technologies with very different capacity factors are comparisons of resource availability at least as much as of equipment quality.
Degradation and the assessment period
Modules lose a little output each year. This page applies your degradation rate geometrically — each year produces the previous year’s output multiplied by one minus the rate — and totals the series over the assessment period you enter.
That is the conventional simplification, and it is a simplification twice over. It ignores the higher loss commonly seen in the first year, and it treats degradation as smooth when a real fleet also loses output in steps as individual modules and strings fail. It also says nothing about the inverters, which have a shorter life than the array and will need replacing inside a long assessment period.
Use the rate from the manufacturer’s warranty if you have nothing better, and remember that a warranty is a floor rather than an expectation.
What this is not
This is an annual-average energy estimate. A yield assessment used to raise money is a different exercise and produces a different number.
It uses a time series of irradiance and temperature at sub-hourly resolution rather than one averaged figure, so that temperature and clipping losses are computed when they actually occur. It models the array geometry, including row spacing and near shading, rather than assuming a single loss. It uses the specific electrical characteristics of the modules and inverters chosen. It accounts for inter-annual variability in the weather, and it reports not a single figure but a distribution — a central estimate together with the levels that would be exceeded with stated probability, because a lender is interested in the bad years.
The gap between the two is not a rounding error. Treat what this page produces as a way of checking whether a proposal is in the right region and of seeing how the answer moves when an input changes, and not as evidence for anything.