Home / Insights / Conveyor belt tracking
Manufacturing & Maintenance 8 min read

Conveyor belt tracking: the belt runs to one side because something upstream is telling it to.

Mistracking is a symptom that almost everyone treats as a problem. That is why it keeps coming back.

By Frank Guo · Technology & Product Leadership, addanode

TL;DR — A conveyor belt moves toward whichever side it contacts first, so mistracking is always a message about geometry or loading somewhere before the point where you noticed it. The five real causes are off-centre loading, material build-up on idlers and pulleys, frame out of square or level, idler and pulley misalignment, and belt condition (a stretched edge, a bad splice, camber). The most common mistake is reaching for tracking-roller adjustment first — it can mask a cause for weeks while the belt edge is being destroyed. Diagnose by walking the belt backwards from the drift, running it empty and then loaded, and fixing causes in order: clean, then square, then align, then adjust. Continuous drift monitoring earns its place because mistracking is progressive: the sensor sees the trend long before the edge starts fraying against the structure.

The one principle that explains almost every case

A belt steers toward the side of a roller or pulley it touches first. That is the whole mechanism. Anything that makes one side of the belt arrive earlier — a pulley face that is not perpendicular to the belt path, an idler rotated slightly in plan, a lump of carry-back building up on one end of a roller, a load dumped off-centre — will walk the belt in that direction, continuously, until something stops it.

Two consequences follow, and they are the reason so much tracking work fails. First, the cause is upstream of where you see the drift, sometimes many metres upstream. Second, the belt is obeying, not misbehaving: if you correct the symptom without removing the instruction, the belt will simply go on trying, now fighting your correction, usually against its own edge.

A belt steers toward the side of a roller it touches first A belt walks because something makes one side arrive earlier: a pulley face out of perpendicular, an idler rotated in plan, carry-back built up on one end of a roller, a load dumped off-centre. The cause sits upstream of the drift, so correcting at the point of drift leaves the instruction in place and the belt keeps obeying it. Treating the symptom Removing the instruction Where you look At the place the belt is drifting Upstream of it, sometimes many metres upstream What you change The nearest adjustable idler, in the opposite direction Whatever is making one side of the belt arrive first What the belt does next Goes on trying, now against your correction Stops steering, because the instruction is gone What wears The belt edge, against whatever is now fighting it Nothing that was not wearing before How long it holds Until the next shift notices it again Until something else changes What the record shows A tracking adjustment, repeatedly, at the same frame One adjustment, at a different frame
A belt steers toward the side it touches first The cause is upstream of the drift, so the nearest adjustment fights the belt instead of releasing it. Where you look Symptom At the place the belt is drifting Instruction Upstream of it, sometimes many metres upstream What you change Symptom The nearest adjustable idler, in the opposite direction Instruction Whatever is making one side of the belt arrive first What the belt does next Symptom Goes on trying, now against your correction Instruction Stops steering, because the instruction is gone What wears Symptom The belt edge, against whatever is now fighting it Instruction Nothing that was not wearing before How long it holds Symptom Until the next shift notices it again Instruction Until something else changes What the record shows Symptom A tracking adjustment, repeatedly, at the same frame Instruction One adjustment, at a different frame
The two consequences of the principle above, side by side. The belt is obeying, not misbehaving — so an adjustment made where the drift is visible leaves the instruction in place and simply gives the belt something to fight.

The five real causes

Cause How it presents Tell-tale
Off-centre loading Belt tracks true empty, drifts as soon as it is loaded — always toward the light side. The single most useful test on any conveyor: run it empty, then loaded. If drift only appears loaded, stop looking at idlers and go to the chute.
Material build-up Carry-back caking on return idlers or the tail pulley creates a bigger effective diameter on one side. Drift that gets worse over days and improves after a clean-up. Usually a scraper or cleaner problem in disguise.
Frame not square or level Persistent drift in one region of the conveyor that no amount of idler adjustment fixes for long. Measure diagonals and check level across the stringers. Common after a structure has been bumped, resited or built in a hurry.
Idler / pulley misalignment Drift begins at a specific point along the belt and continues from there. Walk the belt backwards from where the drift starts; the offending idler is usually within a few frames upstream.
Belt condition Drift that travels with the belt — the same section wanders each time it passes a fixed point. Camber, a stretched edge, or a splice not square to the belt centreline. Mark the belt and watch: if the wander follows the mark, it is the belt.

The diagnostic sequence that actually finds it

  1. Watch it empty, then loaded. This single comparison splits the problem space in half — loading and chute issues on one side, geometry on the other.
  2. Establish whether the drift is fixed to the structure or to the belt. Chalk-mark a belt section. If the wander travels with the mark, the belt or the splice is the cause; if it happens at the same physical location regardless, it is the structure.
  3. Walk backwards from where drift begins. The instruction always comes before the symptom. Look for build-up, a rotated idler, a shifted bracket, a bent frame.
  4. Fix in this order: clean, square, align, then adjust. Cleaning and squaring are free and remove most causes. Alignment of idlers and pulleys comes next. Tracking rollers are last, and they are for holding a properly aligned belt in place — not for compensating a fault you have not found.
  5. Change one thing at a time and let the belt make several full passes. A conveyor responds slowly; adjusting three idlers at once produces a result nobody can attribute.

Why the tracking-roller shortcut is expensive: forcing an incorrectly loaded or misaligned belt back into position with training idlers keeps the belt on the structure — while the edge grinds against whatever is holding it there. Belt edges are not repairable in the way a splice is; by the time the fraying is obvious, you are pricing a belt rather than an afternoon's alignment. The shortcut does not avoid the cost, it converts a maintenance job into a capital one.

What monitoring adds — and what it doesn't

Mistracking is progressive and mostly invisible between inspections. A belt that has begun to wander does not announce it; it simply moves a little further each week until an operator notices spillage, a rubbing noise, or a torn edge. That is precisely the shape of problem continuous monitoring is good at: drift sensing at the points where the belt is most likely to move gives you a trend rather than a discovery, and the trend can be acted on during planned time.

Being honest about the limits: a drift sensor tells you the belt has moved and when it started. It cannot tell you whether the cause is a chute, a caked idler or a splice — that is still a walk along the belt with the sequence above. What it changes is when that walk happens, and whether it happens before or after the edge is damaged. On a plant where a conveyor stoppage halts everything behind it, that timing is the whole value — the wider scope of what to watch on a conveyor system (idler and pulley bearings, motor and gearbox condition, pull-cord status) is covered in conveyor belt monitoring, and the failure-detection physics behind the bearing side in bearing fault detection.

The South African footnote

Two local realities change the calculus. Long overland and underground conveyors mean the inspection walk is expensive in time and, on some sites, in access permits — which raises the value of knowing which section to walk. And repeated stop-start cycles from load shedding are hard on tracking: a belt that starts under an uneven load behaves differently from one already running, so plants that shed frequently see tracking problems appear faster than their inspection intervals were designed for. Both argue for trending drift continuously rather than sampling it on a route.

FAQ

Conveyor belt tracking — common questions

Why does a conveyor belt run to one side?

Because a belt steers toward whichever side of a roller or pulley it contacts first. Anything that makes one side arrive earlier will walk it: off-centre loading, material build-up on one end of an idler, a frame out of square, a misaligned idler or pulley, or a belt fault such as camber or an out-of-square splice. The cause is always upstream of where you notice the drift.

How do I adjust conveyor belt tracking?

Only after diagnosis, and in order: clean off build-up, check the frame is square and level, align idlers and pulleys, and use tracking rollers last to hold a properly aligned belt. Change one thing at a time and allow several full belt revolutions before judging the result. Adjusting training idlers to mask an unfound cause keeps the belt on the structure while destroying its edge.

How can I tell whether it's the belt or the conveyor?

Chalk-mark a section of belt and watch it pass a fixed point several times. If the wander travels with the mark, the cause is in the belt — camber, a stretched edge or a splice not square to the centreline. If the drift always happens at the same physical location regardless of which belt section is passing, the cause is in the structure or the idlers there.

Why does the belt only mistrack when loaded?

That pattern points almost exclusively to loading, not geometry. Material landing off-centre pushes the belt toward the lightly loaded side. Look at the chute, skirting and load point — including whether the material stream has shifted as a liner has worn — rather than at idler alignment.

What are the consequences of ignoring mistracking?

Belt edge damage first, which is not economically repairable and eventually forces a belt replacement; then spillage and the clean-up labour and housekeeping risk that comes with it; then accelerated wear on idlers, structure and skirting; and finally, on a badly mistracked belt, the risk of a rip. Almost all of it is preventable with a diagnosis that takes an afternoon.

Can sensors detect belt mistracking automatically?

Yes — drift sensing at the points where a belt is most likely to move gives you a continuous trend and an alert when the belt starts wandering, rather than a discovery when someone notices spillage. What it will not do is diagnose the cause; that still needs a walk along the belt. Its value is that the walk happens before the edge is damaged, and on the right section.

Primary sources

Know the belt is wandering before the edge tells you.

Drift, bearing condition, motor load and pull-cord status on the conveyors whose stoppage halts everything behind them.

Start a Conversation See the conveyor case study →