A glass bottle reject system takes a bottle off the conveyor after an upstream station has flagged it, using an air blast, a vacuum head, a mechanical pusher or flap, a rotary wheel, or a diversion of the whole lane. Which of these works is settled by the mass and stability of the bottle, not by taste. What makes the installation reliable is the delay between the detection signal and the reject action, which equals the conveyor distance from detector to reject point divided by the belt speed, plus a sensor that proves a bottle really left the line.
The scope here is narrow on purpose. Whether a bottle is acceptable is decided elsewhere: weight limits and weighing accuracy sit with the checkweigher that verifies fill weight, and cracks, chips, stones or foreign bodies in an empty container sit with the optical inspection machine. The trigger is taken as given, and only the removal that follows it is discussed. No rejection rate or accuracy figure is given for any machine or supplier, because the only numbers worth relying on are the ones demonstrated with your bottle, on your conveyor, at your production speed.
Four product facts that define the reject job
Poor reject stations are rarely poorly built. More often the buyer named a mechanism at purchase and never described the task it had to perform. Four facts fix that task, and every one of them comes from the product, not from an equipment catalogue.
- Mass. A thirty gram cosmetic dropper bottle and a two kilogram catering jar are different handling problems whatever the frame around them looks like. Weight decides whether a contact-free device can work at all and whether a mechanical push is safe.
- Stance. A tall, narrow bottle with a high centre of gravity tips in a way that a squat wide-mouth jar of equal weight does not. Straight off a fast filler, the liquid may still be moving inside and shifting the centre of gravity at the moment the bottle is touched.
- Arrival state. A bottle alone on a clear belt allows options that are closed to one arriving in a single-file queue at tight pitch, shoulder to shoulder with the bottles either side.
- Reason for rejection. The station that raises the trigger sets how precise the timing must be and what physical evidence the operator will want to see afterwards.
Settling these early is a commercial matter as much as a technical one. A reject station deliberately destroys or diverts product, so each false rejection is an immediate loss and each escape is a complaint waiting to happen. Both rates are largely locked in at specification and respond little to later adjustment. Plants that buy the hardware first and think about tolerances afterwards tend to relax the threshold until nothing important is rejected any more, and at that point the equipment has been paid for and does no work.
Reject mechanisms and the bottles each one suits
Reject actions on a glass line fall into three families: removal without contact, removal by a moving part, and diversion of a whole lane or belt section.
Non-contact rejection by air or vacuum
A burst of air blows the bottle off the belt, or a vacuum head lifts it clear. Nothing solid touches the container and nothing sits in the product path, so the method is quick and kind to light bottles. The drawbacks are just as definite. Force drops with distance and with the square of the nozzle offset, so the nozzle must stay exactly where it was set, and the air has to be dry, clean and steady in pressure. Of the three families this is the one a pressure dip at the compressor hurts most, and it is normally kept to small, light containers.
A bottle blown sideways leaves the belt lying down. The receiving bin therefore has to catch it without letting it roll back into the following bottle or shatter against a hard edge.
Mechanical rejection by pusher, arm, flap or wheel
Here a moving part does the work. A pusher strokes across the belt, a swing arm sweeps the bottle sideways, a flap or drop section opens beneath it so gravity takes over, or a rotating paddle wheel lifts it out of the main lane. These devices act positively, their operation is easy to watch, and medium and heavy bottles are within their reach because no pressure gradient is involved.
The cost is contact. A pusher must miss the bottles ahead and behind. A swing arm describes an arc that has to stay clear of the adjacent lane. A drop section must be shut again before the next bottle arrives, or the queue falls into the gap. In every case the stroke time has to fit inside the bottle pitch, and that single constraint determines whether a design can run at a given speed.
Diverting a lane or a belt section
A lane diverter, a traversing belt section or a gate does not pick out one bottle. It sends a whole lane, or a defined stretch of conveyor, down a second path where suspect bottles collect. On a multi-lane line, where five or six bottles reach the same point side by side at the same instant, this avoids duplicating a pusher on every lane.
The method is coarse. Good bottles in the diverted group go with the bad ones, and someone then sorts or reprocesses the group by hand. It tends to be chosen when rejections are infrequent and the product value cannot carry a mechanism per lane, or when one filling head is under suspicion and the owner wants everything from that head quarantined.
Mechanism and trigger compared
The table is a frame for selection, not a specification. Mechanism and trigger source are linked: a trigger that arrives early or late rules mechanisms out well before it shows up in the reject rate.
| Mechanism and usual trigger | Bottles it suits | What normally calls the reject | Timing demand | Typical failure | Confirmation |
|---|---|---|---|---|---|
| Air blast nozzle, from a checkweigher or vision head | Small, light containers of a few tens of grams up to a few hundred grams, standing alone with free space alongside | Weight check after capping, or a camera that has seen a missing closure, a wrong label or a fill-level fault | Brief and exact. The blast window lasts a fraction of a second and the nozzle offset cannot vary, which calls for a rigid bracket and stable air | Low pressure, a blocked nozzle or a stalled solenoid gives a weak or absent blast. The bottle stays upright on the belt and the system logs a successful reject | Photoelectric sensor across the reject aperture, tested at the start of each run with a bottle known to be outside the limit |
| Pusher or swing arm, from a checkweigher or fill-level probe | Medium and heavier bottles, roughly a few hundred grams to a few kilograms, steady enough to be pushed without tipping | Weight, fill height, or a closure torque check that has found a loose cap | Moderate. Stroke and retract must both finish in the gap between consecutive bottles, so pitch caps the speed, not belt speed alone | The stroke lands on a neighbour, or the retract is late. A sound bottle is knocked off or the lane jams | Actuator position sensor proving full stroke, paired with a pass sensor at the aperture so a stroke into empty space is not counted |
| Flap or drop-section diverter, from a checkweigher or vision head | Medium to heavy bottles and wide containers that cannot be pushed sideways without upsetting the row behind | Weight, or an inspection result flagging a container defect | Moderate. The section opens under the right bottle and must be closed and locked before the next one, so return time matters as much as opening time | An incompletely returned section leaves a gap in the track and a sound bottle tilts or drops through | Section-closed sensor interlocked so the conveyor cannot run with the section open |
| Rotary paddle wheel or star-wheel, from a checkweigher or metal detector | Regular cylindrical bottles in one format, generally at higher speed and already spaced by a timing screw | Weight, metal detection on filled product, or a camera result on a defined lane | High. The wheel pocket has to meet the bottle at an identical phase each cycle, so the drive is taken from the line and not left free-running | Worn pockets or phase drift lift bottles that should pass and miss bottles that should be lifted, seen as unexplained drift in the reject rate | Counts at the pass lane and the reject lane, with their sum compared against total production for the run |
| Vacuum or suction pickup head, from a vision head on empty containers | Small, light bottles, typically below a few hundred grams, which need lifting clear and not pushing into a bin | Camera or optical station that has flagged a chip, crack or foreign body in an empty bottle | High. Arrive, seal, lift and release all happen in the available gap, and the vacuum must hold on an irregular surface | A partial seal on a chipped or wet neck drops the bottle back on the conveyor, where it carries on as if sound | Vacuum switch confirming grip before the lift, plus a presence sensor in the discharge chute |
These five are not ranked. Lines often run two of them together, for example a non-contact device on a lane of light containers and a mechanical one on a lane carrying a heavier format. The mistake to avoid is picking the cheapest-looking mechanism and then feeding it a bottle that was never in the description of the line. That mismatch does not announce itself as a breakdown. It appears as a confirmation signal that fires although no bottle has moved.
Setting and holding the reject delay
The delay between detection and action is the one number on which the whole installation rests. It is a physical quantity, not a software preference: detector-to-reject distance divided by belt speed, expressed in whatever unit the controller counts. It changes when line speed changes, when a detector is relocated, and when a new format alters the spacing.
Measure the delay on the running line
A calculated value is only a starting point. Belt speed under load can differ from the nameplate, and a sensor does not detect at its front face but at some point within its field of view. The working method is to send through a bottle that is deliberately out of limit, see where it comes off, and trim the delay until that position matches the intended one.
Write the result down alongside the line speed it was set at. A delay carried over from an earlier speed setting or an earlier bottle format is among the commonest reasons a system looks healthy while removing the wrong bottle.
Leave conveyor length for it in the layout
Some delay cannot be avoided, and on a fast line the detector and the reject point may sit several metres apart. The reject point is therefore often placed downstream of anything it might upset, such as a transfer or a lane change, even when the operator would like it somewhere else. Agree the position with the line builder while a conveyor can still be moved on the floor plan. Recovering the distance later by slowing the line, or by bolting a device into a spot that does not suit it, seldom survives production speed. The spacing a reject point depends on is a property of the conveyor that carries bottles between stations, so the two are best specified together.
Two detectors or two lanes
When two detectors can both demand a rejection, each brings a delay and an alignment of its own, and the controller has to hold both values without one interfering with the other. When one reject point serves two lanes, the device must act on one lane and leave the other alone. In practice that means a dedicated device per lane or a lane diversion, not one shared mechanism. Decided at specification this is simple. Found after installation, it usually costs a new mechanism and not just a new setting.
Proving the bottle left the line
The word "reject" hides three separate questions. Did the system decide to reject? Did the mechanism remove a bottle? Was it the flagged bottle? An installation that answers only the first is the most hazardous arrangement on a line, since it reports a reject whenever it fires and stays silent whenever it fails to.
Confirmation answers the second question, and its worth depends on what it is wired to. A sensor that only adds a log entry helps little, because logs are not read mid-shift. A sensor that sounds an alarm or stops the line when a reject was called and nothing crossed the aperture is real protection. Treat a missing confirmation as a fault condition, never as a record. It is the single failure that hides itself: everything reads normal while rejected product keeps moving towards packing.
False rejects and escapes
A false reject is a sound bottle removed because the trigger judged it out of limit. An escape is an out-of-limit bottle that stayed on the line, because the trigger missed it or the mechanism failed. The two rates pull against each other. A tighter threshold cuts escapes and raises false rejects, and no setting brings both to zero.
What the reject station can influence is the mechanical share of escapes. Keep two separate records: how often a reject was called, and how often a call ended with a bottle leaving the line. The second figure is the one for the shift report.
Bottles put back on the line
A rejected bottle that goes back without a re-check has not been rejected. It has been deferred, and whether it is caught again depends on which station happens to look at it next. Where rejects are collected and reintroduced, write down what may return and what must be discarded, and tie that rule to the reason for rejection and not to what is convenient for the operator.
Where the rejector sits among neighbouring stations
The rejector closes a short chain that starts with a judgement, and each station in the chain knows something different. The weigher knows a number: whether the gross weight of a filled, closed bottle lies inside a band. The inspection head knows an image of an empty container. A metal detector knows a signature and a fill-level probe knows a height. The capper knows none of this, yet it produces the fault a weigher most often catches, a closure that is missing or badly seated. The rejector knows only that a signal arrived. Its entire skill is turning that signal, accurately, into a bottle that is no longer on the conveyor.
This division tells a buyer where to look when results go wrong:
- The reject rate has climbed and the rejected bottles really are out of tolerance: ask the detection station and review its threshold.
- Sound bottles are being removed: check the timing and alignment of the reject action before touching the threshold.
- Out-of-limit bottles are reaching packing: find out whether the mechanism fired, whether it hit the right bottle, and whether the confirmation is reporting truthfully.
One neighbour is often mistaken for a reject device. An accumulation table that buffers the line holds bottles so that a short stoppage elsewhere does not spread. It sorts nothing, and it disturbs spacing. A rejector directly after it would face irregular gaps, which is why the reject point normally goes on a stretch of conveyor with controlled pitch and not at a buffer exit. A buffer absorbs variability. A rejector acts on one bottle at a known instant. The two need opposite conditions.
For anyone still assembling the line and not yet specifying a single station, the sequence of stations, the control architecture and the balancing of rates are set out in our overview of the complete glass bottle filling line. Start with the weighing decision if that is not yet settled, since a reject mechanism can be no better than the judgement handed to it.
What to send when asking for a reject proposal
An enquiry that names only a mechanism gets a catalogue page in return. One that fixes the points below in writing removes the guesses a supplier would otherwise have to make.
- Bottle. Every format to be handled, with body diameter, total height, empty weight and filled weight. Mark the slowest or least stable format, because it sets the hard case.
- Line. Speed in bottles per minute and the pitch at which bottles reach the reject point, since a mechanism is rated by the time available per bottle. Also the number of lanes, and whether rejection must work lane by lane or a whole-lane diversion is acceptable.
- Triggers. Which stations can call a reject, what each one measures, whether calls can overlap, and the conveyor distance from each detection point to the intended reject point.
- Bottle condition at the reject point. Free-standing or in a touching queue, dry or wet, upright or already tilted by an upstream transfer.
- Destination. Counted and returned, counted and discarded, or quarantined by head or by lane for later inspection.
- Confirmation. Whether a missing confirmation must raise an alarm or stop the line, and how the confirmation is recorded.
- Control interface. Signal type, logging of each reject with a reason code, and whether the reject history has to be exported.
Two items are habitually omitted. Without the pitch, a supplier given a nominal rate assumes generous spacing, and the mechanism then falls behind when bottles arrive close together at full speed. Without the destination, the proposal may suit a bottle headed for recycling and be wrong for one that has to be inspected, counted and traced to a filling head.
A useful reply names a mechanism family and gives the reasoning, sets out the timing arrangement and the delay values the alignment needs, states what the confirmation device is wired to do, describes the reject path with a bin or chute matched to the bottle, and covers the interface to line control, including the reason-code logging that makes reject history analysable. If one mechanism cannot cover every format, the reply should say so and propose a split, not a single device that is marginal on half the range. Every rate, distance and delay in it remains a planning estimate until shown on your own bottle at production speed, and that demonstration is what the installation should be accepted against.
Questions buyers ask about bottle reject systems
How does a reject system differ from a checkweigher?
The checkweigher judges and the reject system acts. One weighs a filled, closed bottle against a limit and issues a signal. The other converts that signal into a physical removal, and its work is timing, force and confirmation. They are purchased separately and fail for different reasons, so they are diagnosed differently: correct weights with a rising reject rate point to the decision, while sound bottles leaving the line point to the action.
Which reject mechanism is best for glass bottles?
None is best in general. Light containers up to a few hundred grams usually go by air blast or vacuum head, since little force is needed and a heavy device risks disturbing the row. Medium and heavy bottles usually get a pusher, swing arm or flap, which is easy to verify. Regular cylindrical formats at high speed often suit a paddle or star-wheel driven from the line, and side-by-side lanes may leave lane diversion as the only practical choice.
How do I know a bottle was actually rejected?
A command to fire is not evidence. Proof comes from a sensor showing that a bottle crossed the reject aperture, wired so that a called reject with no bottle leaving raises an alarm or stops the line. Rejects counted at the mechanism and bottles counted out of the reject chute are different numbers, and the chute count is the one to record for the shift.
What false reject rate is acceptable on a glass line?
There is no universal figure, and one quoted without knowledge of the product would mislead. The threshold is a commercial choice. A false reject costs the container, its contents and the labour of handling the bin, while an escape costs whatever the defect causes downstream: a customer complaint, a refused delivery or a regulatory finding. Choose the threshold deliberately and note the reasoning. Then eliminate mechanical causes before changing it, because a bottle touching its neighbour, a draught across the belt or a stale delay will produce false rejects that no threshold cures.
Can rejected bottles go back on the line?
Only under a written rule, and only where a re-check can resolve the cause. A bottle short on fill may be topped up or reworked if the plant has a defined procedure and a re-inspection step. A bottle rejected for a glass defect, a damaged finish or a foreign body should stay out, since the cause is still present. Quietly returning bottles is how a defect the line detected still reaches the customer.
Is an automatic reject system worth having for small batches?
The cost of an escape matters more than batch size. A slow line filling a tolerant product for a domestic customer can be defended with manual inspection at a defined interval and a written record. An automatic rejector is easiest to justify where the contents are valuable enough that each rejected bottle is a visible loss, where the destination market enforces a declared quantity, where line speed is high enough that sampling no longer describes the batch, or where a customer contract demands a hundred percent rejection record. Two or more of these usually make the case. If none applies, a documented manual routine is proportionate, and the sums should be run on your own volumes.