For the manufacturing background, see how a glass bottle is formed; what follows assumes it.
Most glass bottle defects fall into three groups: surface flaws such as chips and stick marks, internal flaws such as bubbles, stones and locked-in stress, and dimensional drift in height, diameter, finish or wall thickness. Each group traces back to a different stage of production, so the fix is different too: handling alignment for chips, mould temperature and lubrication for stick marks, melting practice for inclusions, annealing for stress, and mould upkeep for dimensions. A buyer who can name the defect and its likely origin gets a far more useful answer from a plant than one who reports "bad bottles".
Where in production each defect starts
A bottle goes through batching of raw materials, melting, refining, forming by either press-and-blow or blow-and-blow, annealing and final inspection. Glass leaves the furnace above 1,500 degrees Celsius and is shaped quickly, then has to cool through its annealing range, the temperature window in which internal stress must be managed. A parameter that strays at any of these stages can leave a mark on the finished container.
The awkward part is the delay. A problem that begins in the batch mixer may only show once the bottle has been formed and cooled, which is why tracing a flaw backwards needs a method and not a guess. The table below links the defects buyers meet most often to the stage and the cause behind them.
| Defect | What you see | Usual origin | Control at the plant |
|---|---|---|---|
| Drops | Irregular, jagged area where glass has been knocked away | Contact with another bottle, a machine part or the conveyor | Calibrated transfer mechanisms, conveyor speed matched to output |
| Dings | Small dimples or indentations in the surface | Same impacts at lower force | Alignment checks, replacement of worn mould parts |
| Seizure (stick) marks | Raised ridges or streaks | Glass adhering to a mould that is too hot or poorly lubricated | Mould temperature monitoring, daily lubrication checks, mould cleaning |
| Bubbles | Round voids, from pinholes to large cavities | Melting and refining; moisture in the batch | Adequate furnace residence time, pre-dried raw materials |
| Stones | Solid specks embedded in the wall | Unmelted batch or fragments of furnace lining | Thorough batch mixing, refractory inspection |
| Residual stress and cracks | Often nothing, until the bottle fails | Cooling that is too fast or uneven; an off-schedule lehr | Controlled annealing schedule |
| Dimensional drift | Height, diameter, finish or wall thickness out of tolerance | Mould wear, clamping pressure, parison weight, blowing pressure | Preventive mould maintenance, in-line measurement |
Surface defects
Drops and dings
These are impact damage. A bottle hits its neighbour, a metal component or the conveyor hard enough to chip the glass or dent it. The cause is nearly always mechanical: transfer fingers out of line, worn mould parts, or conveyor belt tension set wrongly. On a line turning out hundreds of bottles a minute, a misalignment of a millimetre or so is enough to spoil thousands of pieces in one shift.
The remedy is routine maintenance: recalibrating the transfer equipment, changing worn components on time, and keeping conveyor speed in step with the production rate. Some plants also listen for the problem, using acoustic monitoring that picks out the sound of bottles colliding so that operators can correct alignment before scrap builds up.
Seizure marks
Seizure marks, also called stick marks, are ridges or streaks left when hot glass clings to the mould during pressing or blowing and is dragged as the mould opens. They matter most on clear bottles for high-end products, because the mark scatters light and reads as a distortion.
Two conditions cause most of them. One is a mould surface running too hot, which makes glass stick instead of releasing. The other is mould lubricant, usually an oil-based solution that forms a thin barrier between glass and metal, applied too thinly or unevenly. Residue left on the mould from an earlier run can produce the same sticking in one spot.
Moulds fitted with thermocouples report their temperature in real time, so an operator can raise the cooling water flow or idle the press before marks appear. Lubrication nozzles should be looked at every day for clogging and for a steady flow rate, and moulds cleaned on a set routine.
Internal defects
Bubbles and stones
A bubble is a void inside the glass; a stone is a solid inclusion. Both are more than cosmetic, because each one concentrates stress and can be the starting point of a crack when the bottle is pressurised or knocked.
Bubbles come from the melting and refining stages. Damp raw materials are a known contributor, since the water turns to steam in the melt and leaves bubbles on its way out. Stones have other sources: sand grains that never reacted, clumps of soda ash, or pieces of the refractory lining that have spalled off the furnace wall. Poorly mixed batch makes this worse, because zones of different composition melt at different temperatures and some resist fusing completely. An ageing furnace lining sheds particles into the glass stream as it deteriorates.
Control sits in the melting process: enough residence time in the furnace, the right temperature profile along the melting tank, and raw materials dried before use. Better-equipped plants model the furnace with computational fluid dynamics (CFD) to find dead zones where glass stagnates and stones develop, and run scheduled refractory inspections to head off spalling.
Stress and cracks
Stress builds during cooling, when one part of the bottle sets before another. Cool too quickly or unevenly and the thermal gradient stays locked into the glass. A bottle in that state may crack on its own during handling, give way early under internal pressure, or shatter at a change of temperature.
Annealing exists to remove this. Bottles travel through a lehr, a long oven divided into temperature zones, where they are held at the glass transition temperature long enough for the structure to relax and are then brought down slowly to room temperature. If the schedule slips through equipment failure or operator error, the result is the hardest defect for a buyer to catch: bottles that look perfect and still fail performance tests.
Cracks themselves point to their cause by their shape, and an experienced inspector can usually tell them apart on sight:
- Radial cracks spreading from the point of highest stress indicate thermal shock from rapid cooling.
- Circumferential or spiral cracks indicate mechanical stress in handling.
- Stress corrosion cracking develops over time under chemical attack from aggressive cleaning solutions or from the product inside.
Stress and wall distribution become more important as bottles get lighter, which is one reason we treat lightweighting a glass bottle as a specification exercise and not only a weight target. Handling cracks also arise after the bottle has left the plant, so packing and loading belong in the same conversation as freight planning that keeps breakage down.
Dimensional defects
Dimensions decide whether a bottle runs on an automated filling and capping line. One that is a little too tall, too narrow at the finish or uneven in shape can jam the filler, fail to seal, or fall short in stack strength testing. The usual faults are variation in height, deviation in diameter, inaccurate finish dimensions and inconsistent wall thickness.
Four variables account for most of it: mould wear, mould clamping pressure, the weight of the parison (the pre-formed gob of glass before final blowing), and blowing pressure. A small change in any one can push a bottle outside its tolerance band.
Plants check dimensions with precision gauges, optical comparators and, increasingly, automated vision systems that take hundreds of measurements a second. An in-line measuring station at the exit of the annealing lehr can reject out-of-specification bottles automatically. The more useful output is the trend: plotted over time, the measurements show when a mould will need replacing or a setting adjusting before any bottle goes out of tolerance.
How plants detect defects in 2026
Inspection used to mean people watching bottles pass on a fast conveyor, a method limited by fatigue, attention span and personal judgement. Automated systems now photograph every bottle with high-resolution cameras, multi-spectral imaging and 3D profilometry.
Machine learning models trained on thousands of labelled defect images classify flaws with accuracy above 99.5% and hold that level through a whole shift. They also grade severity, separating a cosmetic blemish that some applications can accept from a structural defect that calls for stopping the line at once.
Pooling inspection results across lines and shifts exposes patterns no single inspector would see. A jump in bubbles on Line 3 at 2 AM may line up with a shift change at the furnace; slow dimensional drift on Line 7 may be progressive mould wear that should be scheduled for replacement. Underneath this sits statistical process control (SPC): temperature profiles, furnace pressure and mould temperatures are tracked continuously, and an alert fires when a reading leaves its control limits.
A few leading manufacturers are trialling digital twins, virtual copies of a production line used to simulate how a parameter change affects defect rates. Engineers can test an adjustment in the model first, which cuts trial and error and lowers the chance of a process change creating a new defect.
What to look for in a plant's defect prevention programme
With demanding retail specifications, even a 0.5% defect rate can mean a significant financial loss, so prevention has to be a system and not a string of one-off fixes. When we match a project to a plant, these are the elements we look for, and they are fair questions for any buyer to put to a supplier.
- A shared defect classification. Every known defect type has a visual definition and a severity rating, so furnace operators, inspectors and customer service all use the same words for the same flaw.
- Real-time monitoring. Temperature, pressure, flow rates and dimensions are logged continuously at the critical process points, with automatic alerts for out-of-range readings.
- Root cause analysis. Defects are investigated with a structured method such as a fishbone diagram, five-whys or fault tree analysis, and the findings are recorded in a searchable knowledge base for the next time the flaw appears.
- Corrective action protocols. Each root cause has a standard response with a named owner and a deadline, followed by an inspection to confirm that the fix worked and introduced nothing new.
- Trained operators with authority. Detection equipment does not replace people who understand how process settings produce defects. Operators should be trained across production roles and allowed to stop the line when they see conditions that lead to flaws.
Defect control of this kind is continuous work. Inspection technology and quality expectations both move, and a programme that is reviewed against them keeps scrap and rework down as they do.