2D-Only Simulation in IST PVSolar Simulator 9.0.10

Not every project needs a drone photo traced roof-by-roof. A simple rooftop with a clear sky view, or an early-stage feasibility estimate, often just needs — module, inverter, tilt, azimuth, row spacing typed in by hand — with the 3D Shadow Analysis tool never opened. The question worth asking honestly is: how much accuracy does that "2D-only" path actually give up compared to the fully 3D-traced workflow?

The answer, once you look at what's actually running underneath, is: more than you'd expect stays real physics either way — the 3D tool doesn't add a second simulation, it just replaces a handful of specific inputs with site-measured ones. Here's the honest breakdown.

Every one of the following runs at full fidelity in the pure 2D workflow:

• Real measured weather — PVGIS TMY (true ISO 15927-4 typical meteorological year with measured DNI).
• Full solar position for every timestep.
• PoA transposition - plane-of-array irradiance, split into beam, sky-diffuse, and ground-reflected components.
• Single- or two-diode I-V modeling of the actual selected module (not a flat efficiency number), including irradiance-level and low-light behavior.
• A real cell-temperature model (IEC 61724-1 Uc/Uv coefficients) reacting to the site's actual ambient temperature and wind.
• Air-mass spectral correction for the module's specific technology.
• The inverter's actual datasheet efficiency curve (piecewise interpolated across its 5–100% load points) plus its real MPPT voltage window checked at every timestep's actual cell temperature.
• Martin-Ruiz angle-of-incidence loss, front and rear.
• The full bypass-diode partial-shading electrical model, when row-to-row shading is present.
• Bifacial rear-irradiance via a proper infinite-sheds view-factor model — front/rear beam, sky-diffuse, and ground-reflected components, not a flat "+X% bifacial bonus."
• Degradation, aging, uncertainty (P50–P95), CO₂, and every downstream financial calculation.

None of that is a "lite" version. A 2D-only run and a 3D-informed run of the same module, inverter, and site weather will produce the same core annual energy number for an unobstructed, single-orientation array — because that number comes entirely from the physics list above, not from the tracing tool.

Where the 2D-only workflow substitutes an analytical model for a measured one

The differences are narrower than "half the accuracy" — they're specifically about three inputs:

1. Row-to-row (near) shading. Without 3D-traced geometry, the simulator falls back to a closed-form analytical formula: given your entered row pitch and module height, it computes the exact shadow one infinite row of modules casts on the next at the sun's current elevation and azimuth, and derives the shaded fraction directly. This is genuinely correct physics for a perfectly uniform, unobstructed field of parallel rows — which describes a large share of real ground-mount and simple rooftop layouts. What it can't do is account for a real obstacle: a chimney, a neighboring building, an oddly-shaped roof that isn't uniform rows. That's specifically what tracing real geometry in the 3D tool replaces it with — an hour-by-hour shading profile computed from your actual traced obstructions and panel positions, rather than an idealized infinite-row assumption.

2. Far-horizon shading. In the 2D-only path, this comes from a short list of Site Condition presets (0.3–3.5%, e.g., "Suburban — mixed 1–2 storey, some trees" or "Hill town — terrain obstruction") — a categorical estimate, not a site-specific number. The 3D tool's horizon-fetch pipeline instead pulls a real terrain/building horizon profile for your exact coordinates. For a genuinely open site, the two converge; for a site tucked into a valley or against a hillside, the preset is a coarser stand-in.

3. Bifacial geometry inputs. The infinite-sheds bifacial model itself is identical either way — but in 2D-only mode, its pitch, GCR, and mounting height come from what you type into Unlimited Sheds 2D Model, rather than from the 3D tool's measured layout. The 2D card runs the exact same geometry (row pitch, GCR, mounting height, albedo) as a live preview — it's a real calculator, not a rough guess — but it reflects the numbers you enter rather than a traced site.

So — what's the "% of perfection"?

There isn't an honest single number here, and a blog post that hands you one (say, "94.7% accurate") would be manufacturing false precision. What can be said honestly:

• For a simple, unobstructed, single-orientation site — open sky, uniform row spacing, no nearby structures — the 2D-only workflow and the fully 3D-traced workflow converge to essentially the same energy estimate, because every difference above collapses to zero (no near-shading beyond the idealized formula, no meaningful horizon, and correctly entered bifacial geometry).
• For a shaded, irregular, or multi-roof site, the gap is exactly the size of the three substitutions above — and that gap is the reason the 3D tool exists at all. A hillside site, a rooftop with a chimney throwing an afternoon shadow across two strings, or a parcel next to a tree line will show a real, sometimes material, difference between the analytical estimate and the traced one.
• The financial and lifetime layers (P50–P95, degradation, LCOE, IRR, DSCR) are equally rigorous either way, since they consume whichever energy number the physics layer produced — they don't independently add or remove accuracy.