Home › Guides

How WattSketch calculates its results

This is a short, plain-language summary of the models behind WattSketch's numbers, with real sources — not a black box. For the full detail, every formula, and every disclosed limitation, open the in-app Methodology page once you're using the tool; it's reference material, written to be read on its own, not just a list of citations.

1. Where the solar resource data comes from

You pick a location on a map, and WattSketch fetches real historical solar irradiance and temperature data for that point from a public dataset — either NASA POWER's Climatology API (monthly averages, fetched directly from your browser) or PVGIS's typical-year hourly dataset (fetched through a small, stateless relay, since PVGIS's own API rejects requests sent directly from a browser). You can also type in your own values by hand if you prefer. Every number shown is labeled with exactly where it came from.

2. How solar yield is estimated

The yield calculation follows the same family of physical models behind NREL's PVWatts: given your system size, mounting tilt and azimuth, and the fetched irradiance, it works out how much sunlight actually lands on a tilted panel (direct beam, sky-diffuse, and ground-reflected light, each converted from the dataset's horizontal-plane measurement to the tilted-plane equivalent), then derates that for module temperature and system losses to arrive at AC energy output. There are two calculation paths, matched to the two solar resource sources: an hourly path (PVGIS) using Muneer's (1990) anisotropic sky-diffuse model, Martin & Ruiz's (2001, 2013) angle-of-incidence reflection-loss model, and a Faiman (2008) module-temperature model combined with Huld et al.'s (2011) power-rating model; and a monthly path (NASA POWER) using a simpler isotropic sky-diffuse model with the standard Liu & Jordan (1962) / Klein (1977) monthly beam-tilt conversion and a NOCT-referenced cell-temperature model. Both paths apply a flat 96% DC→AC inverter-efficiency factor (PVWatts' own default) — this does not model a specific inverter's clipping behavior or DC:AC ratio, a level of detail beyond what a preliminary tool like this attempts.

3. Suggested system size and default orientation

The suggested starting system size divides your audited average daily energy use by how much energy the real yield model above says one kilowatt would actually produce at your site's own tilt, azimuth, and losses setting — not a generic rule of thumb. Default tilt and azimuth for most mounting types are picked the same way: a real search over your site's own annual yield at every allowed angle, not a fixed formula. Both are only ever starting points you can override by typing your own values.

4. Self-consumption, export, and battery dispatch

Once hourly (or monthly-typical-day) generation and load are both known, WattSketch works out, hour by hour: how much generation is used directly on site (self-consumption), how much surplus is exported (if your grid connection allows it) or curtailed, and how much remaining demand is drawn from the grid. If you add a battery, it charges from surplus solar and discharges against remaining demand, within a round-trip efficiency (85%, sourced from NREL's 2024 Annual Technology Baseline for lithium-ion storage) and a power rating derived from its energy capacity over a duration you choose. For a site with an unreliable grid and a generator, the dispatch rules change — see sizing solar and battery for an unreliable grid.

5. Financial results

Financials.tsx converts the same energy numbers into standard financial figures: net present value (NPV), simple and discounted payback period, internal rate of return (IRR), and levelized cost of energy (LCOE) — all computed with standard discounted-cash-flow formulas, not anything invented for this tool. LCOE is shown two ways: per kWh generated (unaffected by whether you add a battery, since a battery changes how generation is used, not how much of it there is) and per kWh actually used on site, which can improve with a battery that raises self-consumption.

What's disclosed as an assumption, not a measurement

Several figures are the tool's own reasonable, disclosed defaults rather than site-specific measurements — battery dispatch assumes one repeating "typical day" per month rather than simulating your real specific days, a generic ground-reflectance (albedo) value stands in for one measured at your site, and a generator's fuel use comes from one published fuel curve rather than your exact generator's spec sheet. None of this is hidden — every one is spelled out, with its source where one exists, on the Methodology page.