Quick Answer
Solar backtracking is a control algorithm that tilts single-axis tracker rows away from their ideal sun-facing angle during early morning and late afternoon to avoid inter-row shading. It trades a small cosine-angle loss for a larger elimination of electrical mismatch losses, typically recovering 1–4% annual energy on tightly spaced tracker arrays.
Single-axis trackers now dominate new utility-scale solar capacity. A horizontal single-axis tracker (HSAT) can raise annual yield by 15–25% over fixed tilt at high-direct normal irradiance sites. But that gain only materialises if the rows avoid shading each other during the long morning and evening shoulders. That is where solar backtracking comes in.
Solar backtracking tilts single-axis tracker rows away from their ideal sun-facing angle during early morning and late afternoon to avoid inter-row shading. It trades a small cosine-angle loss for a much larger elimination of electrical mismatch losses. On a typical 100 MW site with GCR around 0.40, the difference between correct backtracking and none can be several gigawatt-hours per year.
This guide explains the algorithm from first principles. We cover the geometry of inter-row shading, the energy tradeoff between perfect tracking and backtracking, and standard versus slope-aware implementations. We also explain how GCR drives the backtracking threshold, why real terrain breaks simple models, and how to verify backtracking in solar design software. For the broader tracker decision, see our tracker vs fixed tilt comparison.
Quick Answer
Solar backtracking is a control algorithm that tilts single-axis tracker rows away from their ideal sun-facing angle during early morning and late afternoon to avoid inter-row shading. It trades a small cosine-angle loss for a larger elimination of electrical mismatch losses, typically recovering 1–4% annual energy on tightly spaced tracker arrays.
In this guide:
- Why backtracking exists: the inter-row shading problem
- How the backtracking angle is calculated from row geometry
- Backtracking vs standard tracking: the energy tradeoff
- Standard, slope-aware, 3D, and advanced backtracking algorithms
- How GCR and row spacing set the backtracking threshold
- Real-site effects: terrain, bifacial modules, and controller calibration
- Suboptimal backtracking losses and how to avoid them
- How to verify backtracking settings in PVsyst and field SCADA
Solar Backtracking: What It Is and Why It Exists
A single-axis tracker rotates rows of PV modules around a horizontal north-south axis. Through the day the row tilts east in the morning, flattens at solar noon, and tilts west in the afternoon. The goal is to keep the module surface as close as possible to perpendicular to the sun vector, minimising the cosine loss that plagues fixed-tilt arrays.
The problem appears at low sun angles. At 8:00 AM, the sun may be only 10–20° above the eastern horizon. If a tracker row tilts 50° east to face the sun, its long shadow falls across the row behind it. The shaded modules receive only diffuse light. In a series-wired string, the current through the whole string is limited by the weakest module. A small shaded band on a few modules can depress the entire string’s output by 30–50%.
Backtracking solves this by deliberately not tracking the sun during those hours. Instead of tilting 50° east, the controller tilts the row to a shallower angle. Perhaps 25° east. That keeps the shadow from touching the next row. The modules receive less direct irradiance than they would at the ideal angle, but they avoid the disproportionate loss from partial shading.
This behaviour is not a failure of tracking. It is an optimisation. The controller knows the sun position from GPS and astronomical algorithms. It also knows the row pitch, module width, and tracker height. With that geometry, it computes the maximum tilt angle that keeps every row unshaded. When the ideal tracking angle exceeds that threshold, the tracker switches from normal tracking to backtracking mode.
Pro Tip
Backtracking does not increase the total solar energy intercepted by the field. It only changes how that energy is distributed across modules. The benefit comes from avoiding electrical mismatch losses that would otherwise destroy string output during shaded hours.
How Backtracking Works: The Geometry
The backtracking angle is a simple geometric problem. Imagine two adjacent tracker rows aligned north-south. Each row has a collector width (W) and the centre-to-centre row spacing is (P). The Ground Coverage Ratio is (GCR = W / P).
When the sun is low in the east, the front row casts a shadow toward the west. The length of the shadow depends on the sun’s elevation angle and the tracker tilt. The backtracking algorithm finds the tracker tilt angle (\beta_B). At this angle, the shadow from the top edge of the front row just reaches the bottom edge of the rear row.
For a horizontal single-axis tracker on flat ground, the relationship between solar transversal angle (\theta_T) and backtracking correction (\theta_c) is:
θ_c = -sign(θ_T) × arccos( |cos(θ_T)| / (GCR × cos β_c) )where (\beta_c) is the cross-axis slope angle. On flat ground, (\beta_c = 0) and the equation reduces to the familiar flat-terrain form. The tracker then operates at:
θ_B = θ_T + θ_cThe result is a rotation curve that follows the sun closely near noon but flattens out in the morning and evening. If you plot tracker angle against time, the backtracking curve looks like the normal tracking curve with its early and late shoulders clipped off.
NREL’s slope-aware backtracking study provides the derivation for terrain-corrected versions of these equations. The flat-ground simplification is accurate only when the cross-axis slope is under about 1° and row heights are uniform.
Backtracking vs Standard Tracking: The Energy Tradeoff
Without backtracking, a tracker follows the sun continuously. At low sun angles, rows shade each other. The shaded cells drop out, bypass diodes activate, and string current collapses. The energy loss is far larger than the shaded area alone would suggest because of the electrical behaviour of series strings.
With backtracking, the tracker accepts a small orientation penalty. The modules are no longer perfectly perpendicular to the sun beam, so direct irradiance falls by the cosine of the angle error. But every module in the row sees some direct light, so string current stays high.
The net effect is almost always positive for tightly spaced arrays. PVsyst documentation notes that even with backtracking enabled, a near-shading loss of 2–3% typically remains. Diffuse and albedo components are still partially shaded. Without backtracking, that loss would be much larger.
| Mode | Behaviour | Main Loss Mechanism | Typical Annual Impact |
|---|---|---|---|
| Standard tracking | Follows sun continuously | Inter-row shading + electrical mismatch | 3–8% annual loss at GCR 0.35–0.45 |
| Backtracking | Flatten angle to avoid shading | Cosine angle + IAM penalty | 0.5–1.5% net annual loss |
| Backtracking gain | Difference between the two | Avoided mismatch loss | 1–4% annual energy recovered |
The exact gain depends on GCR, latitude, climate, and electrical design. High-GHI, low-diffuse sites see larger absolute gains because the direct beam dominates and shading would have been severe. Cloudy climates with high diffuse fractions see smaller gains because diffuse light cannot be tracked anyway.
Backtracking Algorithms: Standard, Slope-Aware, and 3D
Not all backtracking is the same. The controller’s algorithm determines how aggressively it tilts away from the sun, whether it accounts for terrain, and whether it optimises for diffuse or bifacial conditions.
Standard Astronomical Backtracking
This is the default in most commercial tracker controllers and in PVsyst. It assumes flat ground, uniform row pitch, and identical tracker geometry across the array. The controller computes a common tracking angle for all rows based on the sun position and a single reference GCR.
Standard backtracking works well when the site is genuinely flat and the as-built row spacing matches the design. It fails when terrain undulates, when different zones have different GCR, or when tracker heights vary.
Slope-Aware Backtracking
NREL’s 2020 slope-aware backtracking work extended the standard algorithm to account for cross-axis terrain slope. On a slope, the shadow from one row falls at a different angle than on flat ground. A single common tracking angle cannot keep all rows unshaded.
Slope-aware algorithms calculate a terrain-corrected axis tilt and cross-axis slope angle for each row or row pair. The correction angle then depends on the local slope rather than the flat-ground assumption. This can recover 1–2% annual energy on sites with moderate terrain.
3D and Terrain-Following Backtracking
Advanced tracker systems now use 3D backtracking that considers north-south as well as east-west slopes, installation deviations, and even corner shading between staggered rows. PVH reports that 3D backtracking can deliver up to 4% more annual energy than standard backtracking and 50–300% more energy during sunrise and sunset backtracking phases.
The key insight is that eliminating every possible shadow is not always optimal. A 3D algorithm may tolerate minor corner shading if the resulting sun-facing angle captures enough extra irradiance to outweigh the small mismatch loss.
Diffuse-Optimised and Machine-Learning Variants
Some research groups have proposed backtracking strategies that optimise for total irradiance rather than just beam irradiance. In cloudy conditions, a flatter tracker orientation can capture more diffuse light from the sky dome. Machine-learning controllers use historical weather and production data to predict cloud cover and adjust the tracking strategy accordingly. These are not yet mainstream but are moving into commercial products.
GCR, Row Spacing, and the Backtracking Threshold
Ground Coverage Ratio is the single most important design input for backtracking. It directly sets the angle at which the controller must switch from normal tracking to backtracking.
| GCR | Row Spacing (m per m of collector width) | Backtracking Behaviour | Typical Use Case |
|---|---|---|---|
| 0.30 | 3.33 | Minimal backtracking; wide shoulders | Land-abundant sites, bifacial optimisation |
| 0.35 | 2.86 | Moderate backtracking | Typical utility-scale layout |
| 0.40 | 2.50 | Frequent backtracking at low sun angles | Standard Indian/USA utility projects |
| 0.45 | 2.22 | Heavy backtracking; narrow energy shoulders | Land-constrained, high energy-price sites |
| 0.50 | 2.00 | Almost continuous backtracking near sunrise/sunset | Very tight layouts, rarely optimal |
At GCR 0.30, the rows are so far apart that backtracking is rarely needed. At GCR 0.50, the controller is backtracking for hours each morning and evening. The LCOE-optimal GCR for most utility-scale projects falls between 0.35 and 0.42, balancing land cost against energy yield.
Backtracking also affects project finance. A 2% annual energy gain on a 100 MW project at a high-DNI site can add $200,000–$400,000 in revenue over a 25-year life, depending on PPA price. That is why independent engineers and lenders now ask for backtracking validation as part of the energy yield assessment.
The backtracking threshold also depends on module dimensions. Longer modules cast longer shadows for a given tilt. Taller tracker axes cast shadows that start higher on the rear row. The controller must be configured with the actual module width and tracker hub height, not just a generic GCR value.
A common commissioning mistake is to enter the wrong GCR or module width into the tracker controller. The controller then backtracks at the wrong angle. Either it allows shading that the PVsyst model did not predict, or it over-backtracks and sacrifices production during the shoulders. Both errors appear as unexplained yield gaps.
Backtracking on Real Sites: Terrain, Bifacial, and Controller Calibration
Design software assumes ideal conditions. Real sites do not. Three factors in particular change how backtracking performs in the field.
Terrain Undulation
Even sites described as “flat” often have local height differences of a few centimetres between tracker posts. Over hundreds of metres, these add up to slopes of 1–2°. That is enough to break the flat-ground backtracking assumption.
A 2026 study by the Polytechnic University of Madrid analysed more than 7,000 trackers across seven utility-scale plants with aggregate capacity of about 1 GW. The researchers found that real tracker controllers apply an overcorrection during backtracking to avoid shading on slightly undulating terrain. This overcorrection flattens tracker angles more than the ideal flat-ground model assumes, creating visible ground illumination patterns and lost irradiance.
The study quantified these “suboptimal backtracking losses” at up to 2% of tracking-related irradiance gains. The losses are economically significant. They occur during morning and afternoon hours when energy prices are often higher.
Bifacial Modules
Bifacial solar panels capture reflected light from the ground beneath and around the array. Backtracking changes the panel tilt and therefore the ground illumination pattern. It also changes the view factor to the sky dome and the rear-side shading from adjacent rows.
Bifacial systems generally prefer lower GCR values, typically 0.28–0.35, because wider spacing increases ground-reflected rear irradiance. The backtracking algorithm should be reviewed with the bifacial model: a strategy optimised for monofacial front-side yield may not maximise total bifacial gain.
Controller Calibration
The backtracking algorithm lives in the tracker controller, not in the design model. At commissioning, the controller must be loaded with the correct site coordinates, GCR, module width, hub height, and slope data. A mismatch between the controller and the PVsyst input is one of the most common causes of tracker underperformance.
For EPCs executing ground-mount projects in India or elsewhere, Heaven Designs’ tracker engineering guidance recommends obtaining the controller configuration report at handover. It should be checked against the approved single-line diagram and shading model.
Field Observation
On a 250 MW site in Rajasthan, the operations team noticed that morning production consistently trailed the PVsyst yield by 4–5% for the first two hours after sunrise. The root cause was a controller GCR setting that matched an early design iteration, not the as-built spacing. Updating the GCR input recovered most of the gap within one week.
The Hidden Cost of Suboptimal Backtracking
Suboptimal backtracking is not the same as no backtracking. It occurs when backtracking is enabled but the algorithm does not match the real site geometry. The tracker rows still avoid the worst shading, but they over-flatten and sacrifice sun-facing orientation.
The loss is hard to spot with standard performance ratio metrics. PR compares actual energy to in-plane irradiance, but it assumes the tracker is following the modelled angles. If the tracker is operating at a flatter angle than the model assumes, the in-plane irradiance estimate is wrong and PR looks normal.
Signs of suboptimal backtracking include:
- Morning and evening production curves that are lower than expected but not obviously clipped
- Visible ground illumination patterns during backtracking hours
- Tracker angles from SCADA that are systematically flatter than the design schedule
- Yield gaps that persist after cleaning, inverter checks, and soiling corrections
The Madrid study found that annual losses attributable to tracking-related effects can exceed 5% when compared to ideal simulations without operational constraints. The difference between ideal flat-ground backtracking and real controller behaviour is now recognised as a distinct loss category.
To avoid it, use terrain-corrected backtracking algorithms where available. Grade the site to tighter tolerances than a fixed-tilt project would require. Then validate controller inputs against as-built survey data.
How to Verify Backtracking in Your Solar Design Software
Modern solar design platforms model backtracking in different ways. The verification process depends on the tool, but the principles are the same.
In PVsyst
PVsyst implements backtracking for horizontal north-south trackers, tilted axis trackers, horizontal east-west trackers, and two-axis frames. It uses a common tracking angle per orientation group based on a reference GCR. Key steps:
- Define the tracker geometry: width, pitch, hub height, and axis azimuth
- Enable backtracking in the tracking parameters
- Confirm the GCR reference matches the dominant spacing in the scene
- Run the shading simulation with electrical shading enabled
- Review the near-shading loss, which PVsyst states typically remains around 2–3% even with backtracking
If the site has variable pitch or sloped terrain, PVsyst warns that backtracking becomes inefficient. For those cases, run a sensitivity with backtracking disabled. Compare the energy results, but remember that the no-backtracking case uses linear shading rather than electrical mismatch.
In NREL SAM and pvlib
NREL’s System Advisor Model and the open-source pvlib Python library both include single-axis tracking functions with a backtrack flag. The pvlib singleaxis function takes GCR, max_angle, and backtrack as inputs and returns the surface tilt and azimuth for each time step. This is useful for scripting custom sensitivity studies or validating commercial controller outputs.
In SurgePV
For integrated layout, electrical design, and energy modelling, solar design software like SurgePV models single-axis trackers with backtracking and supports GCR sensitivity studies. The platform flags spacing configurations where backtracking losses exceed user-defined thresholds and exports the tracker schedule for SCADA comparison. For project owners, the solar proposal software module surfaces the backtracking-adjusted yield curve and LCOE impact.
Field Verification
The final check happens in SCADA. Export the commanded tracker angle from the controller for a clear-sky day. Compare it to the angle predicted by the design model at the same timestamp. A persistent offset during backtracking hours points to a GCR, slope, or module-width mismatch.
When Backtracking Is Not Enough
Backtracking is necessary but not sufficient for maximum tracker yield. There are situations where the algorithm cannot recover all losses.
Very high GCR. Above GCR 0.50, backtracking dominates the daily operation. The energy shoulders shrink so much that the tracker behaves almost like a fixed-tilt system at low sun angles. The incremental cost of tracking may no longer justify the yield.
Severe terrain. On steep or highly irregular terrain, no common-angle backtracking strategy can keep all rows unshaded. Terrain-following trackers or row-specific controllers may be required, but they add cost and complexity.
Diffuse-dominated climates. In regions where diffuse irradiance makes up most of the available light, the benefit of tracking — and therefore backtracking — is smaller. A tracker adds only 8–15% yield over fixed tilt in low-DNI climates, and backtracking recovers only a fraction of that.
String-level shading from objects. Backtracking handles row-to-row shading. It does not handle shading from external objects such as trees, buildings, or equipment. Those must be addressed with layout changes, stringing segregation, or optimisers.
FAQ
What is solar backtracking?
Solar backtracking is a tracker control algorithm that deliberately rotates single-axis tracker rows away from the sun-facing position at low sun angles to prevent adjacent rows from shading each other. It accepts a small irradiance loss from the off-angle orientation in exchange for eliminating the much larger electrical mismatch loss caused by partial row shading.
How much energy does backtracking save?
Backtracking typically recovers 1–3% of annual energy on tracker arrays with GCR values of 0.35–0.45, according to NREL analysis. On tighter layouts or complex terrain, advanced slope-aware or 3D backtracking can add another 1–4% compared to standard flat-ground backtracking.
What is GCR in solar backtracking?
GCR, or Ground Coverage Ratio, is the ratio of collector width to row pitch in a tracker array. It determines how densely the trackers are packed. Higher GCR means tighter spacing, more inter-row shading risk, and more frequent backtracking. Lower GCR means wider spacing, less shading, and less need for backtracking.
Does backtracking work on sloped terrain?
Standard backtracking assumes perfectly flat ground. On sloped or undulating terrain, the shadow geometry changes and a single common tracking angle can no longer eliminate all inter-row shading. Slope-aware or 3D backtracking algorithms calculate per-row or terrain-corrected angles to reduce these losses.
What is the difference between standard and slope-aware backtracking?
Standard backtracking uses a single tracking angle for all rows based on a flat-ground width-to-pitch ratio. Slope-aware backtracking, described in NREL research, adjusts the correction angle for cross-axis terrain slope so each row avoids shading its neighbor on real terrain.
Do all single-axis trackers use backtracking?
Essentially all commercial single-axis tracker controllers offer backtracking. It is standard on utility-scale horizontal single-axis trackers. The implementation quality varies: basic controllers use flat-ground astronomical backtracking, while advanced systems use slope-aware, 3D, or diffuse-optimised algorithms.
Can backtracking be modelled in PVsyst?
Yes. PVsyst includes a backtracking strategy for horizontal north-south trackers, tilted axis trackers, horizontal east-west trackers, and two-axis frames. It uses a common tracking angle per orientation group and requires uniform pitch and flat terrain for ideal results. Electrical shading calculations should still be enabled because diffuse and albedo components remain partially shaded.
What causes suboptimal backtracking losses?
Suboptimal backtracking losses occur when real tracker controllers deviate from ideal flat-terrain angles to avoid shading on slightly uneven ground. A 2026 study by the Polytechnic University of Madrid found these losses can reach 2% of tracking-related irradiance gains. Controllers over-flatten tracker angles during morning and afternoon backtracking periods.
Should backtracking be enabled for bifacial trackers?
Yes, but with care. Backtracking changes the panel orientation and ground illumination pattern, which affects rear-side irradiance on bifacial modules. Wider spacing and lower GCR generally improve bifacial gain because more reflected light reaches the rear side.
How do I verify backtracking settings in the field?
Obtain the tracker controller configuration report at commissioning and confirm it matches the design GCR, module width, row pitch, and GPS coordinates. Then compare actual tracker angles against the PVsyst or design-model backtracking schedule during morning and evening hours. Mismatched GCR inputs are a leading cause of unexplained early-morning and late-afternoon underperformance.
