Site
Time
Panel
Weather and losses
Camera
Playback
Quality
Model checks
The science
What is simulated
A rooftop array of three 60-cell modules (1.6 m² each) sits at the latitude you choose. For the chosen day and solar time the sun's position follows from the declination δ = 23.45° sin(360°(284+n)/365) and the hour angle ω = 15°(t − 12): sin α = sinφ sinδ + cosφ cosδ cosω. The beam hits the panel at incidence angle cos θ = sinα cosβ + cosα sinβ cos(γs − γp) for tilt β and panel azimuth γp.
Light on the panel
Beam irradiance uses the Meinel clear-sky law DNI = 1361 E0 0.7AM0.678 with the Kasten-Young air mass. Cloud cover c scales global irradiance by 1 − 0.75 c3.4 (Kasten-Czeplak) and moves energy from beam to diffuse light. The plane-of-array irradiance is beam × cosθ + isotropic sky diffuse × (1 + cosβ)/2 + ground reflection (albedo 0.2). In the 3D page a ray from the panel to the sun is cast against the Blender tree; when it hits, the beam is removed and only diffuse light remains.
The solar cell: single-diode model
Each module obeys I = IL + I0 − I0 exp((V + I Rs)/a) with a = Ns n kT/q, photocurrent IL ∝ G, and saturation current I0 ∝ T3 exp(−qEg/nkT). With shunt resistance neglected, the implicit equation has the exact solution I = IL + I0 − (a/Rs) W( (RsI0/a) e(Rs(IL+I0)+V)/a ) where W is the Lambert W function. The maximum power point is found by golden-section search. Cell temperature is Tc = Ta + (NOCT − 20)/800 · G with NOCT = 45 C, so hot sunny days lose about 0.4 % of power per kelvin.
Independent numerical checks
(1) The Lambert-W current is compared with a bisection solve of the implicit equation at 400 random operating points and the maximum power point with a 4000-point scan. (2) The daily extraterrestrial energy on a horizontal plane from the closed-form sunset-hour-angle formula is compared with a 1-minute numerical integral of the same sun geometry. (3) Day length from the closed form is compared with a bisection of altitude = 0.
Honest simplifications
The sky is a clear-sky model plus a constant cloud fraction, not weather data, so the annual number is an indication, not a yield forecast. Shunt resistance, spectral and angle-of-incidence losses, snow, inverter clipping, mismatch and partial-shading bypass diodes are ignored. Panel and module data are generic, not a specific product. Solar time is used, so no equation-of-time or time-zone correction.