How much energy does a misaligned tracker row lose?

It depends on the fault and the latitude, but the range is wide and often surprising. A tracker frozen flat can give up a mid-to-high-teens percentage of its annual energy or more, because it forfeits the morning and evening capture that justifies tracking in the first place. A subtle encoder drift might lose only a few per cent. The critical point is that the loss concentrates in the shoulder hours, so a check based on midday output will badly understate it — the real figure comes from integrating across the whole day.

What is backtracking drift?

Backtracking is the deliberate early-morning and late-evening manoeuvre where trackers turn away from the ideal sun angle to avoid shading the row behind them in a densely packed array. Backtracking drift is when that manoeuvre goes wrong — from a miscalibrated algorithm or a drifting position sensor — causing either self-shading between rows or over-rotation. Because it happens specifically in the shoulder hours, its signature is a dip in early-morning or late-evening production that correct backtracking would not produce.

Why do stow errors go unnoticed for months?

A stow fault leaves a tracker in its defensive weather position when it should be tracking, usually after a high-wind or snow event where the stow command never released. It goes unnoticed because the row still produces power — just far less than it should — and no offline alarm fires. Unless someone is comparing that row's daily production against its neighbours, a stowed tracker looks like a row having a quiet month, and quiet months do not get investigated.

Single-Axis Tracker Faults: The Losses Nobody Alarms On

Ellume Engineering

A single-axis tracker that stops moving, stows incorrectly, or drifts out of alignment usually throws no fault code — it keeps producing power, just from the wrong angle. The loss hides at the edges of the day and can reach double digits over a year while every midday peak looks healthy. Detecting it rarely needs tracker telemetry; the signature is in the string data you already collect.

A stuck tracker is one of the most expensive faults in solar because it does the one thing a fault is not supposed to do: it stays quiet. When a string goes offline, something trips and someone gets an alarm. When a tracker stops tracking, nothing trips. The modules are still connected, still generating, still exporting power to the meter. The tracker has simply stopped following the sun — frozen flat, stuck at an angle, or drifting a few degrees off where it should be. There is no offline state to detect, because the equipment is not offline. It is misaimed, and misaimed does not have a fault code. So the loss accumulates in silence, in the part of the day nobody scrutinises, until an annual review notices the block underperformed and nobody can say for how long.

A tracker frozen flat can lose a fifth of its annual energy while every one of its daily peaks looks perfectly healthy.

Why tracker faults produce no alarm

The economics of single-axis tracking come from geometry. A tracker earns its cost by keeping the modules facing the sun through the morning and evening, capturing energy a fixed-tilt array gives up at the edges of the day. At solar noon, a tracker and a flat array are nearly the same — the sun is overhead and the angle barely matters. This is exactly why a stuck tracker is so hard to see. Frozen flat, it still produces close to its potential at midday, when the array's output is highest and most visible. The loss lives in the mornings and evenings — lower-power hours that contribute a smaller share of the daily total and attract almost no attention on a dashboard. The daily peak, the number people glance at, looks fine. The shoulders of the curve, where the money quietly leaks, do not. Compound that across a year and the shape of the loss becomes clear: a tracker stuck flat gives up a meaningful fraction of its annual energy — often in the mid-to-high teens or beyond as a percentage, depending on latitude and season — while never once producing a midday number that looks wrong.

Four fault modes and their data signatures

Tracker faults are not all the same. Each has a distinct signature in the production data, which is what makes detection possible without instrumenting the tracker itself.

Frozen or stuck

The tracker stops moving entirely — a motor failure, a seized bearing, a controller fault. Its production curve loses its broad, shouldered shape and narrows toward what a fixed array at the stuck angle would produce. Compared to healthy neighbours, the stuck row shows a symmetric, narrower daily profile with depressed mornings and evenings.

Stow fault

Trackers stow — move to a defensive position — for high wind, snow or night. A stow fault leaves the tracker in the stowed position when it should be tracking, often after a stow event that never released. The signature is a row that produces far below its neighbours across the whole day, sometimes near flat, following a weather event.

Backtracking drift

In dense arrays, trackers 'backtrack' in the early morning and late evening — deliberately turning away from the ideal angle to avoid shading the row behind. Backtracking errors, from a miscalibrated algorithm or a position-sensor drift, cause either self-shading (rows shading each other) or over-rotation. The signature appears specifically in the shoulder hours, as a dip that healthy backtracking would not produce.

Encoder or position-sensor error

The tracker moves, but to the wrong angle, because the sensor telling it where it is has drifted. This is the subtlest fault: the tracker is doing exactly what it is told, and what it is told is wrong. The loss is smaller than a full freeze but persistent, and it shows as a consistent, modest underperformance against neighbours that does not correlate with weather.

Detecting angle deviation from string data alone

The key insight is that you do not need to measure the tracker's angle to know it is wrong. You need to compare the shape of its production against what the sun did and against its peers. A correctly tracking row produces a broad, flat-topped daily curve. A misaimed row produces a narrower or distorted curve. Three comparisons expose the fault:

1.Shape against the sun. The daily production profile should broaden and shift with the seasons in a way that follows the sun's path. A row whose profile has narrowed or stopped shifting is not tracking.
2.Shape against peers. Neighbouring rows see the same weather and the same sun. A row whose morning and evening production diverges from its neighbours, while its midday matches, is misaimed rather than degraded — degradation would depress the whole curve, not just the shoulders.
3.Symmetry. A healthy tracked curve is roughly symmetric around solar noon. A row with a distorted morning but a normal evening (or vice versa) points to a directional problem — backtracking or a partial obstruction — rather than a full freeze.

These comparisons use only inverter- or string-level power, irradiance and time. No tracker telemetry required — which matters, because tracker controllers often do not surface position data to the monitoring system in the first place.

Quantifying the loss across a block

Detection tells you a tracker is misaimed. Quantification tells you whether to send a crew. The method is to reconstruct what the row should have produced — from the sun's position, the irradiance, and the row's own healthy behaviour before the fault — and difference it against what the row actually produced. The gap, integrated across the affected period, is the recoverable energy, and multiplied by the price captures the revenue at stake. This is where the shoulder-hour concentration matters commercially. Because the loss is small at midday and large at the edges, a naive check that looks only at peak output will badly understate it. The correct quantification integrates across the whole day, which is where the real number lives.

When it is worth instrumenting trackers directly

Detection from string data is powerful, but it has a resolution limit: it works best where production is metered at a granularity fine enough to isolate individual trackers or small groups. On plants where many trackers sit behind a single combiner, string-level detection localises the problem to a zone rather than a specific row. Direct tracker instrumentation — position feedback surfaced to the monitoring system — is worth it where tracker faults are frequent, where the plant is large enough that manual inspection is impractical, or where the block granularity is too coarse to localise faults from production alone. For many operators the pragmatic answer is a hybrid: string-data detection to flag which zones are losing energy, and targeted inspection or instrumentation where the flags cluster.

Frequently Asked Questions

How much energy does a misaligned tracker row lose?: It depends on the fault and the latitude, but the range is wide and often surprising. A tracker frozen flat can give up a mid-to-high-teens percentage of its annual energy or more, because it forfeits the morning and evening capture that justifies tracking in the first place. A subtle encoder drift might lose only a few per cent. The critical point is that the loss concentrates in the shoulder hours, so a check based on midday output will badly understate it — the real figure comes from integrating across the whole day.
Can you detect tracker faults without tracker telemetry?: Yes, and often you must, because tracker controllers frequently do not surface position data to the monitoring system. The detection works on production shape: a correctly tracking row produces a broad, symmetric daily curve, while a misaimed row produces a narrower or distorted one. Comparing a row's morning and evening production against its neighbours — which see the same sun and weather — isolates misalignment, because a genuine degradation would depress the whole curve rather than just the shoulders.
What is backtracking drift?: Backtracking is the deliberate early-morning and late-evening manoeuvre where trackers turn away from the ideal sun angle to avoid shading the row behind them in a densely packed array. Backtracking drift is when that manoeuvre goes wrong — from a miscalibrated algorithm or a drifting position sensor — causing either self-shading between rows or over-rotation. Because it happens specifically in the shoulder hours, its signature is a dip in early-morning or late-evening production that correct backtracking would not produce.
Why do stow errors go unnoticed for months?: A stow fault leaves a tracker in its defensive weather position when it should be tracking, usually after a high-wind or snow event where the stow command never released. It goes unnoticed because the row still produces power — just far less than it should — and no offline alarm fires. Unless someone is comparing that row's daily production against its neighbours, a stowed tracker looks like a row having a quiet month, and quiet months do not get investigated.

The bottom line

Tracker faults are the losses that hide in plain sight — no alarm, no offline state, full production at the one hour anyone looks. The energy leaks at the edges of the day and adds up to real money over a year. The good news is that the fault writes its own signature into the production curve, and reading that signature needs only the string data you already collect, not new instrumentation. The question is whether anyone is looking at the shape of the day, or just its peak.