OwlSolar production estimator
Describe your system from the dropdowns below, add your ZIP code, and this free solar production calculator will model an annual production range for a system matching that description in that location. Nothing here needs your paperwork, your login, or an account.
This is a model, not a measurement of your roof. It says what a system described like yours would be modelled to produce at your ZIP code. It is neither a finding about your system nor a shortfall, and it cannot tell you whether your system is working properly. That last question is settled by measurement, not by this page.
How to read this number
Open your monitoring app, find your total production over a full calendar year, and compare it to the range above. Only measurement settles it.
Be careful in both directions. The range is wide because panel model, shading and inverter losses all move it, so a system that lands inside this range may still have a dead string or a shaded array, and one that lands below it may be fine for a reason this page cannot see. A single year, compared once, decides nothing on its own.
Every assumption behind it
Each of these moves the number, and this page had to pick a value for all of them.
Model and data
Email yourself a copy
We will send this one report to the address you enter, once, because you asked for it. There is no follow-up sequence.
Why the range is wide
A production model needs about forty inputs to be precise. Six dropdowns give it six. The rest (panel model and its temperature coefficient, string layout, inverter sizing, wiring runs, soiling, snow, and above all shading) have to be assumed, and every assumption widens the band. A narrower number here would be a more confident-looking lie.
The two ends of the range are built from named, checkable things: the dimmest and brightest of five measured years of irradiance at your ZIP code, and a system-loss band from 10% to 22% around the 14% NREL PVWatts default. Shading sits outside both ends, because this page cannot see your trees.
Why an independent estimate is worth having at all
The projection in your original solar proposal was produced by the company selling you the system. That does not make it wrong, and a careful installer models it properly. But it was assembled before anything was built, from assumptions about your roof that may not have survived the installation, and nobody has revisited it since.
This estimate is built from a different direction. It uses five complete calendar years of measured irradiance at your ZIP code (what the sky did, not a typical-year average) run through the same physics engine that produces OwlWatt's customer analysis: pvlib with an Ineichen clear-sky model and Faiman cell temperature, and NREL's standard 14% system-loss assumption at the mid-point.
We deliberately do not fall back to a clear-sky ceiling when data is missing. Clear-sky output overstates a real year by roughly a third, and doing that silently would produce a confident number that flatters your system rather than measuring it. If fewer than two complete years of irradiance are available for your location, this tool declines to produce a figure instead of guessing.
What this estimate cannot see
Three things, stated plainly, because an estimate whose limits are hidden is worse than no estimate:
- Shading. Nothing here models trees, chimneys, dormers or a neighbour's roofline. A shaded array can land below the low end of the range and still be behaving exactly as its geometry dictates. This is the single largest reason a real system misses a modelled figure.
- Your actual installed layout. The dropdowns describe one orientation and pitch. Many systems span two or three roof planes facing different directions, and a mixed array behaves differently from any single plane you could pick here.
- Equipment faults and soiling. A failed microinverter, a dirty array or a string that has been offline for a month all reduce output, and none of them are visible to a model that only knows your system size and location.
That is why this number should be used as a screen, not a verdict. If your actual annual production sits inside the range, the model has nothing to tell you. If it sits well below, you have a question worth pursuing, and the three items above are the first places to look.
What to do with a number that looks low
Nothing, yet. A single modelled range is a starting question, not an answer. If your actual annual production sits well under the low end, the useful next steps are to check whether any panels or microinverters reported zero for extended periods, whether production dropped after a particular date, and what your contract guarantees, which is often written as a percentage of an estimate made by the company that also sold you the system. Our guide to how production guarantees work covers what that promise does and does not cover, and expected versus actual production covers how the comparison is properly made.
Measuring it properly, month after month, against your own contract terms and against measured weather rather than a model of it, is what OwlWatt does. See how it works.