Field note · 18 November 2026

Narrow belt sorters: how the divert actually works

A narrow belt sorter moves a parcel with a row of thin belts, then lifts and angles them to push it sideways. The transfer is short, the pitch is tight, and the reliability lives in the gap between the belts and the chute.

A narrow belt sorter moves a parcel with a row of thin belts, then lifts and angles them
A narrow belt sorter moves a parcel with a row of thin belts, then lifts and angles them to push it sideways. The transfer is short, the pitch is tight, and the reliability lives in the gap between the belts and the chute.

01What the belts do, and what they do not

A narrow belt sorter carries a parcel on a row of narrow belts, each maybe 20 to 30 mm wide, running in parallel along the frame. The belts carry; they do not grip. A parcel with a soft base, a polybag, or a loose corner rides on the belt surface and is only as controlled as its own footprint.

That is the first thing to understand about the technology: the transfer is friction-driven and short. There is no shoe, no cell, no carrier. The belt row lifts, the pulley axis rotates, and the parcel is pushed diagonally off the frame.

The consequence is that the sorter is gentle on rigid cartons and less predictable on anything that does not sit flat. A 300 mm carton with a clean base diverts well. A sealed bag of clothing with 40 mm of air in it behaves like a different object every time.

02Where narrow belt wins, and where it loses

It wins on footprint. The divert module is short - typically a metre and a half along the frame - and the belts can be angled from both sides of a single module. Two divert directions from one metre of line is a density that cross-belt and tilt-tray cannot match.

It wins on cost per divert point. There is no carrier fleet to buy and maintain, no cell to keep aligned, and the moving mass is a belt row rather than a tray with a parcel on it.

It loses on heavy or high-friction items. A 25 kg carton with a rough base needs more lateral force than a 20 mm belt can transmit without slipping, and once it slips the parcel arrives at the chute at an angle the chute was not designed for.

It loses on items that cannot be singulated at the pitch the belts require. A narrow belt line wants a gap between parcels, and a line that presents parcels shoulder to shoulder will either recirculate or divert into the wrong chute.

03Divert geometry: angle, speed and the parcel that turns

The divert angle is usually set between 30 and 45 degrees to the line of travel. A shallow angle is gentler and needs more length; a steep angle is compact and throws the parcel harder into the chute wall.

The lateral speed matters more than the angle. If the belt row moves the parcel sideways faster than the line speed, the parcel rotates as it leaves the belts; if it moves slower, the parcel arrives late and can catch the chute lip.

The design case is not the average parcel. It is the lightest parcel that still has to reach the end of the chute, and the heaviest parcel that must not overshoot it. Both cases have to be satisfied by the same belt speed setting, which is why the setting is a compromise rather than an optimum.

04Pitch, gaps and the throughput number

Throughput on a narrow belt line is set by pitch, and pitch is set by the longest parcel the line expects. If a 600 mm parcel is in the mix, the minimum gap has to accommodate it, and the items-per-hour figure falls accordingly.

The practical number for a single-side divert is often quoted at 2,000 to 3,500 items per hour, and it assumes a parcel mix that singulates cleanly. In a mixed operation with polybags, that number is optimistic by a third.

The gap budget is worth writing down: parcel length, minimum gap, belt response time, and the length of the divert module. Four numbers that decide whether the line's advertised rate is reachable on the actual mix.

The belts also need a gap to lift and settle. A row that lifts while a parcel is crossing the module boundary will tip the parcel, and a tipped parcel arriving at a chute is a jam waiting at the next induction.

05Belt tracking, and the pulley that drifts

Each belt in a narrow row runs on its own pair of pulleys, and every pulley has a tracking adjustment. A belt that has drifted will still carry a parcel, which is why drift is usually found by maintenance rather than by operations.

The symptom is subtle: parcels begin to arrive at the chute slightly rotated, and the divert point moves a few centimetres along the module. Operators compensate by nudging the parcel at induction, and the real cause stays untouched for months.

Tracking is checked with a straight edge across the module and a mark on each belt. A belt that is more than three millimetres off the mark is a belt that is loading its pulley unevenly, and the pulley bearing will be the next failure.

The check belongs in a weekly routine, not in a breakdown response. Ten minutes with a straight edge saves a bearing, a belt and a shift.

06The gap sensor and the parcels it cannot see

The gap sensor looks for the space between parcels, and it is usually a photoelectric pair across the line. Its job is to tell the control that the next parcel can be released into the divert zone.

Black cartons, shrink-wrapped trays and glossy bags do not always break the beam reliably. A sensor that misses a parcel will release the next one too early, and two parcels arrive at the divert together.

The usual fix is a sensor with a background suppression setting, or a second sensor at a different height. Both are cheap; the alternative is a line that jams at the divert every hour.

The sensor's alignment drifts with vibration, and a sensor that has moved five millimetres will read the belt surface as a parcel. A monthly alignment check is part of keeping the pitch honest.

07When a belt row is worn out

A belt row does not wear evenly. The belts nearest the divert point see the most lateral load, and they lose grip first, which is why a worn row tends to divert late rather than not at all.

The replacement decision is economic: a row costs a fraction of a module, and a late-diverting row costs a recirculation every twenty parcels. The row is usually replaced long after it should have been.

Keep one spare row per size on site. The lead time on a matched belt set is weeks, and a line that waits for a belt is a line that pays for itself in overtime.

The old row is worth keeping for the pulleys, which rarely fail and are the expensive part of the assembly.

08The two-minute check before the shift

Release a test parcel at the induction, watch it lift and leave the belts, and time how long the row takes to return. A slow return is a pneumatic or a cam problem, and it shows up as a fine on the next parcel's base.

Look along the belt row while it runs. A belt that is flapping has lost tension, and a flapping belt is a belt that will fail at the divert rather than at a convenient moment.

Confirm the chute behind the module is clear and the container is in position. A module that diverts into a full chute is a module that recirculates a parcel and pays for it twice.

09One number for the operations board

Track diverts made and parcels recirculated per module, weekly. The ratio is the module's accuracy, and a module whose ratio worsens over two weeks is a module that needs a belt check rather than a settings change.

A single ratio on a board beside the line is worth more than a monthly report in a folder, because the team that runs the module can act on it the same week.

The number also gives maintenance a way to argue for the belt row. A worn row shows up as a flat ratio, and a flat ratio is a business case in one line.

Where the parcel mix is mostly cartons at that pitch, the narrow belt sorter is the module to compare against - its divert length and belt count are the two numbers that decide the gap budget.

What to check

  • Belt width against the smallest and largest base footprint in the mix
  • Divert angle and lateral speed agreed against the chute length
  • Minimum gap written into the line control, not left to the induction operator
  • Polybag handling tested, not assumed