VIINotes and limitations

VII.1 Putting the results into context

Screening, not an expert opinion. The analyses are a software-based, model-driven screening and do not replace an expert opinion. Simulation results are a model-based approximation of actual conditions. Before use, in particular in permitting and planning procedures, they must be reviewed by qualified professionals.

Direction of the simplifications. Most simplifications of the model work in the direction of more glare: clear sky, continuous module surface, capture margin of the hit check, no soiling. Some work in the opposite direction or have an open direction, such as the homogeneous luminance per zone; they are named in the following sections. The reported values are therefore not an upper bound in every respect.

No assessment. The report states glare durations, intensities and directions. It contains no limit lines, traffic lights or assessment verdicts; classifying the results against a guideline is the task of the responsible expert (Ch. V.8, Ch. VI.1 to Ch. VI.5).

VII.2 Weather and atmosphere

• No cloud cover. A clear sky is assumed throughout. The real glare duration is usually considerably shorter due to clouds, haze and precipitation; for a typical site in the Swiss Plateau, the EnergieSchweiz/Swissolar guide states about half (guide 2023, p. 41).
• Direct irradiance from a clear-sky model. The direct normal irradiance follows the model of Ineichen and Perez with a climatological turbidity per month (Ch. V.1). The calculation uses one site altitude per simulation, the mean terrain height of the PV corner points; if the PV areas of a project lie at very different heights, the same altitude applies to all. Simulations before calculation state 2026.10.11-5 were calculated for sea level; at higher sites their luminance and illuminance were too low, the glare duration practically not.
• Luminous efficacy as a mean value. The conversion of radiation into light follows a mean relationship with solar elevation according to Littlefair (1988). Individual measurements scatter around this relationship; Littlefair states a scatter (RMS) of about 17 %.
• Reflection path without attenuation. The reflected ray from the module surface to the observation point is applied without atmospheric attenuation, scattering or refraction.
• Low solar elevations. The calculation runs down to an apparent solar elevation of 0°. A minimum solar elevation can be set in the report with the filter "Min. sun elevation ≥" (Ch. V.8).

VII.3 Module and installation

• Continuous module surface. Each PV area is modelled as a continuous body or a continuous surface. Real installations consist of module tables and rows with row gaps, racking and substructure; reflections may therefore occur more intermittently than in the model. In this respect the calculated glare duration tends to be too high. With a steep view, for example from high observation points or from the air, the gaps between the rows are visible; the body then shows more glare minutes and a higher illuminance than real table rows. The difference is quantified by the group "Model limits" of the quality assurance (Ch. D).
• What reflects. For ground-mounted and racked roof installations, the top of the body and its side band from the bottom to the top edge reflect; whether the observer sees the front or the back there is decided by the module normal. Underside and plinth from the ground to the bottom edge occlude but do not reflect. Parts that the observer sees from behind or at more than 89° to the normal do not reflect; for facades and roof-parallel areas only the view from behind counts. Mutual occlusion of individual module rows within one area is not represented (Ch. V.2).
• Module glass only. Reflections from frames, cell connectors, substructure or the ground are not calculated.
• Simplified reflection behaviour. The module profile describes the lobe with reflection profile, major and minor axis and reflection offset δ in 10° steps, circular or elliptical, with up to three zones of homogeneous luminance. In reality, brightness falls off continuously. The calculation uses the luminance stored per zone in the module type; whether the steps over- or underestimate the brightness depends on this value. It tends to be overestimated if each zone carries the luminance at its inner edge (Ch. 8.4). With strongly scattering glass the lobe can be irregularly shaped. The model does not represent such shapes (Ch. V.3).
• Standard profiles are reference values. The standard module types supplied represent typical glass classes with representative curves and do not replace a product-specific data sheet. The highest accuracy is achieved with measured reflection data (BRDF) of the module used as a custom module type (Ch. 8).
• Perfect mirror. The module type "Perfect mirror" is an idealised upper bound of the luminance: the reflection has the luminance of the sun, its size is that of the solar disc. A luminance filter practically never takes effect for this module type.
• No soiling, snow or frost. A covered surface reflects less or differently; in this respect the real exposure tends to be lower.
• No mutual occlusion of PV areas. Each PV area is considered on its own and can neither shade another nor block its reflection path. Where PV areas actually occlude each other, their contributions appear combined in the result, for duration as well as intensity. An area in front can be modelled as an obstacle (Ch. 10.1).
• Dual orientation. PV areas with "Dual Orientation" (east-west) are calculated as two PV areas with the same outline, one per orientation. Each orientation is thus applied over the whole outline, although in reality only about half of the modules face that way. Solid angle and illuminance therefore come out too high for each orientation; the glare duration is hardly affected (Ch. 7.3).
• One orientation per area. Each PV area calculates with a uniform module orientation (azimuth, tilt, cross-tilt), with two for dual orientation and with the time-dependent tracking motion for trackers. Differing orientations can be represented by several PV areas, differing cross-tilts with the differentiated cross-tilt using zones; within a zone a uniform value applies, so the real terrain shape is approximated in steps (Ch. 7.6).
• Trackers. The tracking angle is calculated per time step from the solar position. Backtracking assumes level terrain across the axis; no cross-tilt is applied to trackers. Park, wind, fault and special positions of the tracking system are not represented; the key points paper of the Federal Highway Authority names them as states to be considered (p. 9).

VII.4 Surroundings and terrain

• Obstacles only as modelled. Buildings, trees, hedges, embankments or terrain edges only take effect insofar as they are created as an obstacle (Ch. 10.1); terrain is additionally captured by terrain occlusion (Ch. 6). Obstacles are vertical walls between the points entered, with top and bottom edges running linearly; roofs, overhangs, gaps in tree rows or the permeability of foliage are not represented. Shading of the module surface is checked per 0.5° cell at one point, the centroid of its calculation points.
• Terrain occlusion only with a terrain model. Terrain blocks reflections and shades PV areas only if terrain occlusion is switched on and a terrain model is available in the project (Ch. 6). The calculation uses a 2 m grid; buildings and vegetation are not contained in the terrain model. Where the model has gaps, the missing piece does not block; terrain before and behind the gap still blocks. If a gap lies between observation point and PV area, the report marks the point as incomplete. Shading of the module surface by terrain is checked at sampling points spaced 25 m apart at the height of the module top edge and reaches 2 km; lower module parts and terrain in between therefore tend to be shaded too little. Older simulations do not calculate shading by terrain.
• Far-field shading only via the horizon line. The horizon line that can be activated per PV area (PVGIS, terrain model with a grid of about 90 m, 48 directions) is retrieved at the centre of the area's corner points and shades only the direct sun; it does not block lines of sight to the observation point. Buildings and vegetation are not included, and narrow terrain edges can appear smoothed. The horizon line is meaningful mainly for terrain from about 1 km away; this is a guide value derived from the grid and angular resolution, and PVGIS itself states no minimum distance. Closer terrain is captured by terrain occlusion on both paths, provided a terrain model is available (Ch. 6). For very large PV areas, retrieval at the centre can distort the result, for example if a ridge shades only part of the area; the area should then be split into sub-areas, each with its own horizon line (Ch. 10.2).
• Heights from elevation data. The heights of module surfaces, obstacles and observation points follow from the stored ground levels (GL) and the heights entered. For ground-mounted installations, the module surface follows the terrain formed from corner and support points. Automatically retrieved heights come from elevation data providers or official terrain models; their currency varies by provider and region, and later changes to the terrain, such as fills or new buildings, are not included. No guarantee is given for the currency and correctness of these data. Deviations act directly on the result; elevation data must always be checked independently (Ch. 6).
• Excluded areas only as modelled. Sub-areas without modules are only left out insofar as they are created as an excluded area. Façades with two corner points are not cut and always enter the calculation as a continuous surface (Ch. 7.7).
• Plane calculation geometry. The calculation between PV area and observation point takes place in a local plane coordinate system without Earth curvature. Earth curvature lowers a distant point relative to the tangent plane by about 0.08 m at 1 km, 0.7 m at 3 km, 2.0 m at 5 km and 7.8 m at 10 km. For observation points several kilometres away, the height of the PV area is therefore set slightly too high. The near horizon of terrain occlusion, by contrast, does take Earth curvature into account; at large distances, terrain therefore tends to block the area slightly too little in the calculation, and the glare duration tends to be slightly too high.

VII.5 Observer and perception

• Fixed, point-shaped observation points. The assessment applies to one point in space at a time. Extended areas such as window fronts, several rooms or storeys can be represented by several observation points, moving observers on roads or approach paths by points along the route (Ch. 9).
• Field of view as an angular window. A field of view is a rectangular window in azimuth and elevation around the viewing axis ("H±", "V-Sky", "V-Ground"), not a circular cone. Eye movements can be represented by larger fields of view or several points with different fields of view (Ch. V.6).
• No perception model. The calculation provides luminance and illuminance at the eye. Adaptation, age, glare sensitivity, veiling luminance or a hazard to the retina are not calculated (Ch. V.5).

VII.6 Time, grids and calculation accuracy

• Time step. The calculation runs in steps of 1 minute for a reference year without a leap day. The sun moves by about 0.25° per minute; reflections that would be visible for less than a minute are usually covered by the extent of the solar disc and lobe. The sun path of a specific calendar year deviates only slightly from this.
• Standard time. All clock times are given in the project time zone as standard time. During daylight saving time, local clock time equals the stated time plus one hour.
• Assessment grid. Hits and filters are decided on fixed cells of 0.5° × 0.5°. So that no cell is missed, the lobe is widened by half the cell diagonal (0.354°), and a hit cell counts with all its visible calculation points. For wide lobes this amounts to a few percent and works in the direction of more glare; narrow lobes are checked point by point (Ch. V.2).
• Point grid. Solid angle and vertical extent arise from a point grid with about 40 points per axis and area, at most 0.125°. The resulting error of the solid angle is of the order of a few percent.
• Older simulations. Some evaluations require data that are only stored from a certain version on, for example the filter "Illuminance ≥" (simulations from 18.08.2026), the vertical extent per azimuth column (from 25.09.2026) and combining simultaneously glaring areas into one source (from 01.10.2026). For older simulations, such filters are greyed out or work according to the earlier method; a re-simulation brings them up to date (Ch. V.8).
• Recalculated days. For very large data volumes, the report's view of individual moments recalculates a day when needed. If the calculation state has changed since the simulation, this day differs from the annual values; the report points this out, and a re-simulation brings both back in line (Ch. V.7).

VII.7 Presentation in the report

• Combined glare times. The glare duration per observation point combines the contributions of all PV areas; a minute counts once, even if several areas cause glare. Under "PV areas" in the filter settings, the contributions of individual areas can be traced separately (Ch. V.8).
• Glare locations as approximations. The reflection map shows the directions from which reflections are visible at the observation point, the 3D perspective the position of the glare sources as seen by the observer. The glare locations shown on the PV area are geometrically derived approximations on the cell and display grid; the position and extent of real glare spots may differ (Ch. 14).

VII.8 Input data and responsibility

• Input data. The input data are the responsibility of the creator. Values derived from map material, aerial images or terrain models, as well as automatically calculated or suggested values, in particular positions, heights, orientations, tilts, cross-tilts and obstacle dimensions, must be checked before the results are used and corrected where necessary.
• Module parameters. Module parameters based on manufacturer information must be confirmed by the manufacturer with regard to suitability and currency; this applies to custom module types as well as to stored profiles. No liability is accepted for the correctness, completeness and currency of data provided by manufacturers.
• Representability of the reflection behaviour. Before the results are used, it must be checked whether the reflection behaviour of the modules used can be represented with sufficient accuracy by the model of reflection profile, lobe shape and reflection offset δ, in particular for strongly scattering or low-glare special glass. In case of doubt, measured values from the manufacturer (BRDF) should be used.
• Guidelines. The assignment of filters to criteria of individual guidelines (Ch. V.8) describes what can be represented. Whether a guideline applies in an individual case and whether its criteria are met is judged by the responsible expert.
• Retention. Currently no time limit applies; a time limit remains reserved and will be announced in advance. Simulation results are removed when the version with the result, the project or the account is permanently deleted. Deleted projects stay in the trash for 30 days (Ch. 4). For an own archive, reports and exports must be downloaded and backed up by the user.
• Irregularities. If results appear implausible or a malfunction is suspected, the results must not be used and support should be informed (info@pv-glarecheck.com).

See also

• Ch. V Methodology
• Ch. V.8 Filters
• Ch. 6 Terrain and elevation data
• Ch. 8 Module types
• Ch. 9 Observation points
• Ch. 10 Obstacles and shading
• Ch. 14 Report