Bottle Checkweigher for Glass Filling Lines | Fill Weight

A bottle checkweigher on a glass filling line weighs each filled, closed bottle in motion and rejects the ones outside a set limit, but it reads gross weight, so its usefulness depends on how well you know the empty bottle. Size it from the fill tolerance in grams on your smallest container, and let the instrument use only about a fifth to a third of that tolerance. Before comparing machines, weigh at least thirty empty bottles from the same mould: if their range is close to the tolerance you want to police, no checkweigher can tell a light bottle from a short fill.

Gross, net and tare weight in the specification

The platform sees the glass, the contents and the closure as one figure. The label, the customer and the regulator are interested in the net contents only. Tare is the bridge between the two, and an enquiry that asks a supplier to "weigh the bottle" without naming which weight is controlled and where the tare comes from will draw different accuracy figures from different suppliers, each of them defensible.

Tare can be handled in three ways:

• Fixed tare. A sample from the mould is weighed, averaged over a stated number of bottles and keyed in as a constant. This is the normal choice on a high-speed glass line.
• Per-container tare. Each empty bottle is weighed just ahead of the filler. It is accurate, but it needs a second weighing station, which takes rate and floor space and doubles the handling.
• Learned tare. The machine derives the tare from a run and then tracks it.

With a fixed tare, every gram of variation in the empty glass lands in the measurement. That variation comes from how wall and base thickness are distributed in the mould, and it is usually bigger than the error of the machine itself. So measure it first. Pull thirty or more pieces from one mould, weigh them on a calibrated laboratory balance, and note the mean and the range. Ask the bottle supplier to confirm the spread for your mould instead of working from an assumed figure.

When the spread turns out to be the real problem, a finer instrument rarely helps. The fix is normally a tighter dimensional requirement on the glass, because wall, base and finish all vary inside their permitted range; our page on dimensional tolerance standards for glass bottles explains what a mould is allowed to produce.

Turning a fill tolerance into an accuracy requirement

Two limits apply: what the instrument can hold, and what the product and the regulation demand. The smaller one governs. Fill tolerance is usually stated as a percentage, and the same percentage means very different things in grams. One and a half percent of a two hundred and fifty millilitre fill is about three and a half grams; on a thirty millilitre cosmetic bottle it is under half a gram. A small bottle therefore needs finer absolute accuracy, and a line for small cosmetic containers cannot share a specification with a line for large food jars.

The error budget

The checkweigher owns only one slice of the total error on the line. The others are:

• tare spread of the empty bottle;
• bottle-to-bottle repeatability of the filler, on most lines the biggest contributor;
• density of the contents shifting with temperature;
• evaporation of a volatile product between filler and scale;
• condensation on a cold bottle, or a water film left in a washed one.

A sensible target is for the machine to take a minority share, roughly a fifth to a third of the permitted tolerance, leaving the rest to the filling process. If the whole budget is larger than the tolerance, deal with that upstream: tighten the filler, stabilise the bottle weight, or widen the declared quantity. Choose the accuracy grade only afterwards. A checkweigher reports what everything before it has done and cannot improve the filler. Buying a finer grade first tends to end with sound bottles being thrown out at a rate the line cannot live with, after which someone loosens the limit until the machine no longer does anything.

Terms to use when writing the requirement

The verification scale interval, written e, is the unit in which an instrument's accuracy and permitted error are defined. The actual scale interval, d, is the display resolution; it is normally finer than e and says nothing about accuracy on its own. The maximum permissible error, MPE, is the largest error allowed in service. MPE is stated separately for static and dynamic operation, and a machine can hold one figure at rest and a noticeably larger one with bottles crossing the platform, so a quotation with a single unlabelled number is incomplete.

For an internationally recognised way to express accuracy, the usual reference for automatic catchweighing instruments is OIML R 51, which sets accuracy classes and permissible errors for equipment weighing items in motion. Treat it as a framework for wording the requirement. Any accuracy claim for a specific machine has to come from its supplier in writing.

Weight bands, typical accuracy and rejector by container size

Load cell, platform size and settling time all scale with the gross weight of the filled bottle, so that weight is the starting point for choosing a band. The ranges below are indicative market figures that overlap at the edges. Use them to frame the enquiry, and expect the accuracy figure to be demonstrated on your bottle, fill level and speed.

Where the station sits and how it matches the line

Rate and accumulation

Put three figures in the enquiry together: filler rate, checkweigher throughput, and the accumulation on each side of the scale. A checkweigher slower than the filler throttles the line. If a short stop downstream can back bottles up onto the platform, the readings mean nothing until buffer space is added. The usual sizing is a modest margin over line rate with accumulation before and after, and it should be agreed with the line builder, not left to the instrument supplier alone. Buyers still putting the sequence together will find washing, filling, capping, weighing, coding and packing, with buffer sizes and control architecture, on our glass bottle filling line overview; a weighing specification only makes sense once filler and capper are fixed.

Before or after the capper

Downstream of the capper, the scale sees the closure too. A missing cap or insert then shows as a weight deviation, as long as the absent part weighs clearly more than the machine's dynamic error. Caps, liners, dropper assemblies and pump heads generally weigh several grams or more, which a fine band detects comfortably on most small bottles, but check the real component weight. Upstream of the capper, the scale reads fill alone and tare is simpler, while closure faults need another detection method.

In both positions the reject has to leave the line ahead of packing. A bad bottle that reaches the case packer at the packing end is already in a case, and digging it out costs more than the bottle did. That is why the checkweigher normally sits upstream of packing.

Mechanical interface

The weighing conveyor must be isolated from filler vibration and from the drives of the conveyors beside it, otherwise the reading includes load that is not the bottle. Guide rails must stay clear of the bottle while it is being weighed. Two bottles in contact give a doubled reading that no limit can interpret, so the infeed and outfeed transfer, including any timing screw or star wheel, is specified with the instrument and not bolted on later. Settle at the same time the air quality and pressure stability for a contact-free rejector, and who is allowed to clear a jammed bottle from the platform without upsetting calibration.

Line conditions the checkweigher depends on

Most accuracy complaints about installed machines trace back to the surroundings. State the following in the enquiry as line conditions, with an owner for each; they are easy to design in and awkward to retrofit.

• Structure. A rigid, level frame and floor. A scale fixed to a structure that also carries a filling carousel reports the carousel's vibration as weight.
• Air movement. Draughts push on a light bottle. Look at doors opening onto the line, cooling fans and air-conditioning outlets pointed at it. A reject rate that climbs whenever one door opens is a draught problem.
• Bottle surface. A water film, condensation from a cold fill or product spilled down the outside changes the weight, sometimes by more than the tolerance.
• Hot fill. Straight after the filler the bottle carries vapour and a warm mass whose density is still moving, so the reading drifts with temperature. Either place the station where the bottle has stabilised or work with a known, documented offset.
• Debris. A loose sleeve, label or glass shard on the belt is weighed with the following bottle.
• Air supply. Air-blast rejection needs dry, clean air at steady pressure. When pressure sags, the blast weakens and the bottle stays put.
• Electrical supply. A stable feed, separation from variable-frequency drives and other noise sources, and a proper earth protect the signal.

Verifying accuracy at rest and in motion

Verification is two separate jobs, and treating one as the other is the commonest reason a machine is trusted when it should not be.

The static check uses certified test weights at the point of use: zero, a weight near mid-range and weights near the ends of the range, all at working temperature. It proves the load cell, amplifier and electronics are sound and correctly scaled. It tells you nothing about moving bottles, and a machine can pass it perfectly and still fail in production.

The dynamic check runs at production speed and bottle pitch. One filled bottle crosses the platform at least ten times, preferably twenty, with every reading logged. The mean exposes systematic offset; the standard deviation is the repeatability on that product. A further run with bottles filled to different levels checks linearity. Results do not carry over to another speed, another bottle or a replacement mould, so the record must state the conditions.

When to re-check

The plant decides the interval and writes it down. Certain events should always trigger a check:

• start of shift;
• a change of bottle or contents;
• mechanical work on the weighing conveyor or the transfer into it;
• a new tare value;
• an unexplained move in the reject rate.

A sealed reference bottle links the certified weights to the daily routine. Weighed on a laboratory balance and kept only for this job, it has the size and shape of the product and runs through the machine like any other bottle. Store it away from production stock and re-weigh it periodically so its own drift is known.

Setting the reject limit and tracing errors

Any limit produces two kinds of mistake. A false reject discards a sound bottle; an escape lets a bad one through. Tightening the limit trades escapes for false rejects, and no setting removes both. A false reject costs the glass, the contents and the labour at the bin. An escape costs whatever follows downstream: a complaint, a refused delivery or a regulatory finding. The choice is a commercial one, so make it deliberately and write the reasoning into the settings record instead of leaving it to the shift.

Common causes of false rejects

• Bottles touching, or arriving in pairs after infeed spacing has drifted, which reads as twice the true weight.
• Vibration from the filler or an adjacent conveyor.
• A draught from a door or cooling duct acting on a light bottle.
• Product or glass dust on the belt shifting the zero.
• A tare left over from the previous mould.
• Condensation on a cold bottle.
• A limit tightened during changeover and never put back, by a wide margin the most frequent reason for a sudden jump in rejects.

Common causes of escapes

• A rejector that fails to fire because of low air pressure, a blocked nozzle or a dead solenoid.
• A full or obstructed reject bin.
• A bottle that was struck but stayed standing on the belt.
• A dirty or misaligned confirmation sensor reporting a rejection that never happened. This is the most dangerous single fault because it conceals itself.
• A limit loosened to keep the line moving, or a standard weight entered wrongly.
• A bottle that crossed unweighed after transfer timing slipped.

Configure the machine so that a missing reject confirmation raises an alarm, not merely a log line. Remember also that weight cannot reveal a defect that leaves the mass unchanged; cracks, chips and foreign bodies in empty containers are a job for a glass bottle inspection machine further upstream.

Troubleshooting sequence

Working in this order saves hours, and the limit is the last thing to touch:

1. Check zero and span with certified weights at room temperature.
2. Run the sealed reference bottle at production speed and compare against its recorded value.
3. Confirm the tare matches the mould now running.
4. Watch the transfer for touching bottles and poor spacing.
5. Inspect the surroundings for draughts, vibration and belt debris.
6. Fire the rejector on purpose with a bottle known to be out of limits, and see that the confirmation sensor registers it.
7. Only then adjust the limit, and record the change with its reason.

A limit that moves without a record is how a plant loses the ability to stand behind its own data.

Records for batch traceability and declared quantity

Weighing becomes evidence only when it is logged against a batch. A useful record holds:

• batch or lot identity and time window;
• bottles weighed, bottles rejected and a reason code for each reject;
• the limit and tare in force;
• the reference bottle result at the start of the run;
• the operator, and every changeover or setting change.

A bare reject count cannot be analysed. A shift report stating that two hundred bottles were rejected does not tell a quality manager whether the glass, the filler or the setting was responsible. With reasons attached, a later discussion with another department or a supplier rests on facts.

Declared quantity is the second reason to keep records. The two regimes met most in export work are the European Union's average quantity system, under which the batch average must not fall below the declared quantity and individual packs are limited in how far they may fall short, and the sampling procedures of NIST Handbook 133 in the United States. Both are written rules, not equipment standards. Both put the duty on the packer and expect records that can be produced later. An end-of-line checkweigher makes those records continuous, and the destination market sets how long they are kept. We cite both as public references and make no certification claim for any plant or product.

Weighing every bottle also gives feedback on the filler. On a multi-head machine, sorting readings by filling head shows one valve drifting while the overall average still looks fine, and trending the mean through the run shows when it started. This requires each reading to be tied to the head that filled the bottle, which has to be arranged in the line control system at specification stage. Ask for it when the line is built.

Inline checkweighing compared with manual sampling

The manual route is a sample weighed at a set interval on a calibrated laboratory balance and entered in a log. It needs no floor space and no specialist, and on a slow line with a tolerant product it is fully defensible. Its weaknesses are plain. It describes the sample, so a valve failing two minutes after the draw goes unseen until the next one. It leaves no continuous record for a customer or market that wants proof of control. On a fast line the sample grows statistically thin, because an affordable interval is long compared with the number of bottles made in it.

An inline machine brings its own obligations: capital outlay, calibration and a reference-bottle routine, floor space, training, and reject handling with bins and records. It is justified by several drivers in combination:

• contents valuable enough that giveaway is a visible cost;
• a declared fill actively enforced in the destination market;
• a line rate at which sampling no longer represents the batch;
• complaints or refused deliveries already traced to fill;
• a customer contract calling for a hundred percent weight record.

Where two or more apply, the sums usually favour the machine. Where none applies, it is likely to generate a number nobody reads.

To decide, set the value at risk in a batch against the yearly cost of ownership, counting calibration, spares, air and the labour of clearing rejects. Value at risk is the product lost while one valve drifts for a whole run, plus a refused delivery if the drift leaves the plant. If it comfortably exceeds ownership cost, buy the machine as a process instrument. If not, a documented sampling routine with a calibrated balance, a written limit and a signed log is proportionate. Establish the actual costs and rates through an enquiry on your own bottle and line, since they differ from plant to plant.

What to send with a checkweigher enquiry

Three inputs turn a general machine description into an accuracy grade and a rejection proposal:

1. The bottle. Mould, filled gross weight, and tare spread measured from a sample.
2. The fill. Target and tolerance to be policed in both grams and millilitres, plus declared quantity and destination market.
3. The speed. Bottles per minute and the pitch at which they arrive, because accuracy depends on time spent on the platform.

Add the contents, as density, filling temperature and volatility all alter the error budget. Add the closure weight if the machine is expected to confirm a cap or insert.

A reply built on those inputs can name the band, the tare method and the sample size behind it, the station position relative to capper and packing, and a rejector suited to the bottle's mass and stability. Treat quoted accuracy and rate as ranges until they have been shown on your own filled bottles at your own production speed.

Questions buyers ask about bottle checkweighers

How accurate does the checkweigher have to be?

Work back from the tolerance. Express it in grams for the smallest container, deduct tare spread, filler repeatability and expected density or temperature effects, and what remains is the instrument's allowance. Going finer than about a fifth to a third of the total tolerance is paid for twice: once in the machine and again in false rejects the line must absorb.

How do the e value and the maximum permissible error differ?

The e value belongs to the instrument class, not to any single weighing: it is the unit in which accuracy is expressed. The maximum permissible error is the in-service limit and is normally given as a multiple of e. Each is stated for static and for dynamic operation, and a figure without that label cannot go into a specification.

Can it be relied on to catch a missing cap or dropper insert?

Yes, where the component clearly outweighs the dynamic error on that bottle. If the margin against your actual closure weight is thin, do not depend on the scale for closure presence; use a dedicated inspection.

Does the choice between pre-capper and post-capper weighing come down to accuracy?

Usually not. Weighing ahead of the capper gives a cleaner reading of fill; weighing after it adds closure detection. Since rejects must be out before packing in both layouts, the deciding constraint tends to be the space available between capper and case packer.

Which rejector suits glass: air blast, pusher or flap?

Match it to the bottle's mass and stability. Air blast is quick and contact-free and suits small, light containers, but it leaves the bottle lying down, so the bin must catch it without breakage. A pusher is positive and dependable on medium and large bottles; it needs room and must not strike the neighbouring bottle. A flap or drop-section diverter steers the bottle into a reject lane without lifting it. Whichever is used, a confirmation sensor proving the bottle left the line matters more than the mechanism.

Does higher line speed always cost accuracy?

Not always, though the two work against each other: a faster belt and tighter pitch leave less time for the load to settle before the reading. Compare machines on accuracy at your speed and pitch, not at a demonstration speed with wide spacing. If accuracy and rate cannot both be met, the options are a longer weighing section, better vibration isolation, or a coarser tolerance on the declared quantity.

Is an inline machine worth having for small batches?

Only when the value at risk in a batch comfortably exceeds the yearly cost of owning the machine, floor space included. Otherwise documented sampling on a calibrated balance is the proportionate control.