8.2 Shading, Loss Stack and MGD 005

8.2 Shading, the handbook loss stack and MGD 005

Quick Answer: Shade is not one number you invent on the drive home. Near shading is a close object (chimney, aerial, dormer, tree crown within about 10 m). Far shading is a distant horizon object (hill, another terrace, a stand of trees well back from the roof). Far objects often cost more in winter, when the sun stays low. Handbook v1.3 outcome 4.1.1(c) then lists the wider loss stack: cell temperature, module quality, module mismatch, DC wiring, inverter, AC wiring, AC connection, equipment failure, DC optimisation and seasonal far shading. For the shade term itself, the handbook formula taken from MGD 005 Issue 1.0 is Shading Factor = 1 − (NS × 0.01), where NS is the number of affected segments counted on the official sunpath chart, from the horizon line and from any near-object shade circles. MCS Appendix B also lets you write SF = 1 − estimated fractional loss, or SF = 1.00 if the horizon is obviously clear.

Independent OpenExamPrep teaching uses the published MGD 005 shade-evaluation procedure and MIS 3002 Appendix B wording. This section does not replace the official sunpath chart, and it does not invent elevation angles or a DIY horizon sketch held at arm's length.

Near shading versus far shading

Far shading is the silhouette of objects that sit well away from the array. MGD 005 treats objects further than 10 m by plotting them on the official sunpath chart from a position as near as practical to the base and centre of the proposed array (often an upstairs window). You need a compass and a proper elevation-measuring method. Segments that the horizon line touches, and segments that fall under that line, are counted. Each counted segment is worth 0.01 of annual yield. MGD 005 Issue 1.0 does not publish a total segment count for the chart, so do not quote one; this chapter does not reprint segment angles either. The chart in MGD 005 is the authority.

Near shading is an object 10 m or closer to any part of the array: chimney, soil stack, aerial, dormer cheek, neighbouring wall. MGD 005 does not treat that object as a thin horizon line. After you draw the ordinary far-horizon line looking due south, each near object gets a shade circle. The circle's radius equals the height of the object, and the apex sits on the object's highest point (or on the intersection with the uppermost summer sun arc if the object pokes above that arc). Every segment touched by or inside that circle joins the count. The same chimney that nicks a few far-horizon segments can swallow a large block of the chart once it is inside 10 m. MGD 005's own worked contrast is the point: one example counts 11 segments as a far-style plot (SF = 0.89) and 40 segments when the same object is treated as nearer than 10 m (SF = 0.60). Near shade is punitive because it can sit on the array for long parts of the day, not only at sunrise.

Printing the sunpath chart, holding it at arm's length and sketching the skyline is not a valid MGD 005 result. The guidance says so in plain language. If you cannot get a geometrically honest chart, say so and use a method you can defend.

Seasonal bite of far shading

A distant hill that barely clips June midday can erase a slice of December generation because the winter sun path sits lower on the same chart. That is why handbook 4.1.1(c) names seasonal far shading as its own loss line rather than folding it into a summer photograph. A survey done only on a bright June afternoon under-reports the winter horizon. Far shade is a yearly-energy problem with a winter accent, not a "the view looks open" problem.

In practice: A chimney one metre off the ridge, inside 10 m of the nearest module, is near shade. You take the reading from the worst-affected array location (often just south of the object), add the shade circle, and expect NS to jump. A hill a kilometre south is far shade. You plot it on the horizon line. If it only covers winter segments, the annual SF still moves, and the customer's December bills move more than the annual percentage suggests.

The 4.1.1(c) loss stack

Kk in the next section is not raw satellite irradiance. MCS states that the published Kk tables start from the European Commission Joint Research Centre Climate-SAF-PVGIS dataset and are multiplied by 0.8. That 0.8 is a packaged allowance for typical system behaviour. You still have to understand the mechanisms the handbook lists, because exam items and customer conversations name them one by one. You do not invent a second 0.8 on top of Kk unless a documented method says so.

• Cell temperature. Modules are rated at 25 °C cell temperature under Standard Test Conditions. UK roof tiles run hotter in still summer weather, so operating voltage and efficiency sag. The packaged Kk already reflects typical UK thermal behaviour in the 0.8 factor; a black in-roof cassette in a hot void can still be worse than a well-ventilated on-roof plane.
• Module quality. Nameplate tolerance, light-induced degradation and manufacturing spread mean the field array is not a perfect sum of brochure watts. Use the data-plate watt-peak at STC for kWp; do not upgrade the nameplate because a salesperson prefers a "flash-test" story.
• Module mismatch. Series strings follow the weakest current. Partial shade, mixed orientations on one maximum-power-point tracker, or mixed batches increase mismatch. Layout is a loss control, not a decoration.
• DC wiring. Voltage drop and connection resistance between modules, isolators and the inverter eat watts as heat. Short, correctly sized DC routes help; undersized string cable shows up as a quiet annual leak.
• Inverter. Conversion efficiency, start-up threshold, clipping above the AC rating and a poor maximum-power-point window all discard energy that the modules produced.
• AC wiring. The run from inverter to consumer unit or distribution board has its own drop. A loft inverter with a long thin AC tail is a loss, not a tidy install.
• AC connection. Metering, isolators, export limitation and Distribution Network Operator constraints can cap what leaves the inverter even when the roof is in full sun.
• Equipment failure. Downtime is a loss. A failed isolator, a tripped residual-current device or a week of inverter fault is missing kilowatt-hours. The MCS table method does not invent a custom availability factor; the handbook still expects you to name the mechanism.
• DC optimisation. Module-level power electronics can change how mismatch and partial shade behave. They are not a licence to set SF = 1.00 on a shaded chimney. MGD 005 still starts from the geometric assessment. An alternative software shade factor is allowed only if you can show it is equivalent to or better than MGD 005; one published MCS route is to model annual generation with and without the obstacles and take the ratio as SF.
• Seasonal far shading. This is the winter-low-sun horizon effect already introduced. It appears in the loss list and in the SF method when winter segments are counted. Do not ignore it because the June photo looks open.

Shading Factor arithmetic

MGD 005 Issue 1.0 turns the chart into a factor:

SF = 1 − (NS × 0.01)

NS is the shade metric in that formula: the total count of official chart segments affected — touched by, or falling under, the horizon line (objects on the horizon, near or far), plus every segment touched by or enclosed within a shade circle (objects within 10 m). Each counted segment is worth 0.01. If NS = 11, SF = 1 − 0.11 = 0.89. If NS = 0 on a honestly empty chart, SF = 1.00.

MCS Appendix B states the same idea in fractional language: where the horizon is obviously clear and there is no near or far shading, skip a detailed shade assessment and use SF = 1.00. Otherwise SF = 1 − estimated fractional loss. An 11 percent shade loss is SF = 0.89, which matches the NS = 11 example. Alternative shade software may be used only when it is demonstrated to be equivalent to or better than MGD 005.

Do not invent unofficial sunpath angles in an exam answer. Name the method, count NS on the official chart, or state the fractional loss you estimated and how.