Temperature loss

Temperature is the largest single loss in the breakdown and the only one the application works out for you. As soon as a module file is loaded, the temperature row stops being an assumption and becomes a calculation, because the number it needs — the module's temperature coefficient of maximum power — is read from that file and appears nowhere in the input panel.

That is why this loss has its own page. Every other row on The loss breakdown is a percentage you defend by judgement. This one you defend by showing the working, and the application shows it on screen.

The model

Two relations, applied in order.

T_module = T_ambient + G × exp(a + b × W)

Loss (%) = |μ_Pmpp| × (T_module − 25)

Read the first relation as a temperature rise added to the air temperature. The rise is proportional to irradiance, and the constant of proportionality is an exponential in the mounting coefficients and the wind speed. The exponential is always a small positive number, so the module always runs hotter than the air, never cooler.

The second relation is then the definition of the coefficient itself. A module rated at 25 °C and operating at T_module loses |μ_Pmpp| per cent of its maximum power for every degree of that difference. Nothing is lost when the module sits at 25 °C, which is the only condition at which the nameplate wattage is the truth.

The four mounting configurations

Mounting type selects a pair of Sandia coefficients. It does not change the form of the relation, only how much the module heats up for a given irradiance and wind.

Open Rack – Ground Mount is the default, and it is the right one for essentially every plant this application designs. Air reaches both faces of the module and both sides of the rack, so heat leaves the module in every direction. It carries the most negative a of the four, which is the smallest temperature rise per unit of irradiance in still air.

Roof Mount – Close and Insulated Back sit at the other end. A module mounted tight against a roof deck, or one with its rear face insulated, has no air path behind it at all — the heat has to leave through the front. They carry the least negative a, so the largest rise. Note that the two share an identical coefficient pair: chosen either way they produce exactly the same module temperature, and the distinction between them is descriptive rather than numerical.

Stand-off Mount sits between the two — a module held off a roof surface on brackets, with a restricted but real air gap behind it.

b is a separate property: it is multiplied by wind speed inside the exponential, so it sets how quickly the mounting sheds its temperature rise as the wind picks up. Stand-off Mount carries by far the strongest wind response of the four, so the ordering of the four configurations at any given wind speed follows from the combination a + b × W, not from a alone.

The three inputs you control

The direction of the wind effect is worth being explicit about, because it is the input most often left at its default: more wind means more cooling, a cooler module, and a smaller loss. A coastal site with a genuine annual average above the default will show a lower temperature loss than the same design inland, and that difference is real rather than an artefact.

Use annual averages in both fields, not summer peaks. The relation is applied once to produce one annual loss figure, so a peak-condition input produces a loss that is true at noon in June and wrong for the year.

Where the operating irradiance comes from

G is not an input. It is derived from the annual in-plane irradiance you have fetched or entered:

G = min(900, GTI × 1000 / (365 × 8))   W/m²

Read the division first. GTI × 1000 converts the annual figure from kWh/m²/yr to Wh/m²/yr; dividing by 365 days and by 8 hours a day spreads it across an eight-hour equivalent generating day. The result is the average irradiance the modules see while they are actually producing — deliberately not the 24-hour average, which would include the night and understate the operating temperature badly.

Two special values complete it:

• The cap is 900 W/m². However high the annual irradiance, G never exceeds 900, so the modelled temperature rise cannot run away on a very sunny site.
• When no irradiance is set, G is 600 W/m². The in-plane field ships at 0.0, so before any weather data has arrived the model still has a sensible operating point to work from rather than reporting a zero temperature rise.

The consequence is that the temperature loss changes when your irradiance changes. Fetch a higher in-plane figure and the modules are modelled hotter, so the temperature loss grows — which is correct, and is one reason a loss schedule copied between two sites does not transfer.

What you see

The computed module temperature is displayed next to the temperature loss row, in degrees Celsius, so the intermediate result is visible rather than buried inside the percentage.

Below the inputs sits a formula trace: the relation with your own values substituted into it. Between the two, the loss row is auditable — a reviewer can take the trace, the coefficient pair from the table above and the module's datasheet coefficient, and reproduce the percentage by hand.

Check the module temperature rather than only the loss. It is the quantity with a physical meaning, and it is the one that tells you immediately whether the inputs are sane: a module temperature that has not moved from the ambient temperature means no irradiance has arrived yet.

Monthly temperature loss

The annual figure is one number, but the monthly performance-ratio table is not built by repeating it twelve times. Each month gets its own module temperature from the same Sandia relation, using that month's own operating irradiance and its own ambient temperature. Monthly ambient temperature comes from your weather file's temperature column when it has one, and otherwise from a sinusoidal seasonal model whose amplitude scales with the site's absolute latitude.

This is why a hot low-latitude site shows its worst monthly performance ratio in the hottest months rather than the least sunny ones. See How yield is calculated for the monthly relations, and Weather data for what a temperature column in your file changes.

Without a module file

Typing the module dimensions and wattage by hand is a complete input for placement and capacity, but it leaves this row uncalculated. If a yield figure has to stand up to review, load the manufacturer's .PAN file for the module you are actually buying — see Module specifications. It takes one click and it converts the largest loss in the breakdown from a number you have to argue for into a number you can derive.

Where to go next