Down 11% in Two Years With No Alarms

Correct answer: C) A cause that accumulates gradually, such as vegetation, progressive soiling or degradation, because a threshold that compares each day against expectation cannot see a loss arriving a fraction of a percent at a time

A threshold compares today against what today should have been. That catches anything which arrives as an event: an inverter drops out, a string opens, a breaker trips, and the day falls off a cliff. It is structurally incapable of catching a loss that arrives in slices too thin to breach it. If a site loses half a percent a month, no single day is ever 15% below expectation, and the alarm is working exactly as designed while the site quietly gives away a ninth of its output.

Put briefly: alarms detect steps, not slopes. The causes the JTA lists for production losses divide neatly along that line. Weather and curtailment tend to be events. Vegetation growing up a ground-mount row, soiling building through a dry season, an animal population establishing itself, pollution settling from a new neighbour, cells degrading: all slopes, all invisible to a daily threshold, and all of them cumulative.

The only thing that sees a slope is a comparison against a fixed point far enough back, which is what the year-on-year figure in the scenario is. That is why the shortfall was found at all, and it was found by somebody looking rather than by something alarming.

The O&M point is about who set the threshold and what they were thinking about. Alarm limits get configured at commissioning, by people whose job that week is to catch equipment failures, and they are almost never revisited. So the class of loss that costs the most over the life of a contract is the class the alarm system was never designed to notice. Nobody is going to tell you about it. It belongs to whoever decides to go and compare this year with the first one.

Why the other options are wrong

A) An inverter fault the monitoring failed to catch; the thresholds need tightening

This blames the detector for missing something it was never watching for, and the proposed fix makes things worse. Tightening a daily threshold until it catches a slow loss means setting it so close to expectation that ordinary weather and ordinary scatter will trip it constantly, and an alarm that cries wolf gets ignored or switched off. The answer to a slope is not a tighter threshold on steps, it is a different comparison.

B) A metering error — an 11% shortfall with no alarms at all is more likely a measurement problem than a real one

Worth checking and not the first conclusion. A metering error is a real possibility and it is also the explanation that ends the investigation most comfortably, because it turns a production problem into a paperwork problem. The pattern argues against it too: a metering fault usually appears as a step when something was changed or recalibrated, and this is a drift over two years.

D) Normal module degradation, which is roughly what two years looks like

Right family, wrong size. Degradation is genuinely a gradual cause and belongs in the correct answer's list, and it runs at a fraction of a percent a year, which is far short of the eleven being looked for. Reaching for it is tempting because it explains the shape of the problem perfectly and requires nobody to do anything, and that combination is worth being suspicious of in any diagnosis.

References

• NABCEP OMAT JTA v.2026.5 — Domain 4, Task 4.D: "Production losses and their causes (e.g., weather, pollution, shading, animals, soiling, system degradation)". The list itself divides into causes that arrive as events and causes that accumulate, which is what the question turns on.
• No standard or degradation figure is quoted. That degradation runs at a fraction of a percent per year is a property of the modules installed and appears on their own performance warranty rather than in any code.