A thermal shock test tells you how big a sudden temperature change a glass container survives without cracking, and the endurance version tells you how many times it survives a fixed change. The result is a temperature difference in degrees Celsius, tied to a medium and a method; it is never a maximum working temperature. For hot fill, pasteurising and washing, the step that breaks glass is nearly always fast cooling of a hot container, so the requirement you write should come from the coldest thing your line does to a container that is still hot.
Specify a temperature difference, a medium and a cycle count
Asking a supplier for "heat resistant glass" or for a bottle "good to" some temperature gives the plant nothing to build to and the laboratory nothing to verify. An empty jar can sit in a hot oven and come out intact, yet crack at a far lower temperature when cold water lands on it. What damages glass is the gap between the temperature of the wall surface and the temperature inside the wall.
When a hot container touches something cold, its outside skin shrinks first while the glass underneath is still expanded. The skin is stretched, and glass tolerates stretching badly although it resists squeezing well. A crack therefore opens at the outer surface and travels inward and round the container. Heating a cold container quickly does the opposite, squeezing the surface and stretching the core, which glass forgives much more readily. That is why the standard method moves a hot sample into cold water and not the reverse.
Four things decide whether a container survives, and a requirement is incomplete without all of them:
- The size of the worst temperature difference in the process.
- The medium and speed. Air carries heat away slowly, so glass tolerates a very large difference in air. Moving water strips heat fast and creates a steep gradient through the wall, so a difference that is harmless in air can break the same container in water. A spray sits in between, and how harsh it is depends on the pattern, the flow and whether the jet hits a single spot.
- The number of repeats. Glass fails by slow crack growth: a tiny surface flaw lengthens slightly with each stress cycle until it is long enough to let go. A container that shrugs off one large shock can still fail after many smaller ones.
- The state of the glass surface at the moment the shock arrives.
Because of the second point, two laboratories can test identical containers at the same nominal differential in different media, report different results and both be right. Because of the third, thermal shock resistance and thermal shock endurance are separate tests that give separate numbers for one container. A supplier who quotes a figure without asking about these four items has answered some other question.
Which products and channels make thermal shock a hard requirement
The need is settled by the contents, the sales channel and the food safety plan, well before anyone draws the container.
Contents
Acidic foods (sauces, jams, preserves, fruit preparations, pickles, many juices) are often filled hot so the product's own heat handles the microbiology and no separate sterilising step is needed before sealing. Syrups and concentrates are hot filled too, because heat helps them dissolve and keeps them stable. For these, a hot fill thermal shock test is the normal basis for the container requirement, since it treats the fill and the cooling after it as two linked events.
Low-acid and protein-containing products need a heavier treatment, usually pasteurisation after filling or a retort cycle, and the container then follows a defined temperature profile instead of meeting one fill temperature. Spirits, edible oils, dry goods and most cosmetic and pharmaceutical liquids go in at or close to ambient, so their thermal demand is slight or nil. Sparkling and other pressurised products carry a pressure requirement in addition to any thermal one, and each must be specified and tested on its own.
Channel
A one-way pack for a shelf-stable product is filled once and cooled once, and a single-shock requirement covers it. A bottle in a deposit return scheme goes through a washer every trip, so the trips in its working life equal the thermal cycles it must endure, and a single-shock value says very little about whether it will last. A "dishwasher safe" claim on a label works the same way: it reads like a feature, but it commits the jar to surviving a domestic cycle again and again, which is an endurance requirement.
Market rules and your own plant
If the market expects a defined pasteurisation regime for an ambient-stable product, that step cannot be dropped to suit the glass; the glass has to be specified against it. Food contact frameworks such as FDA 21 CFR, EU 10/2011, EU 1935/2004 and LFGB control what the glass and its decoration may contain. They sit in the same specification document but say nothing about how much heat a container takes.
Your equipment adds its own constraint. The filler, pasteuriser tunnel, bottle washer and cooling conveyor all run at settings chosen for a container already on the line. A new container normally has to fit that existing profile, so its requirement comes from the plant, not the recipe. A measured profile of the line, as opposed to an assumed one, is the most useful thing you can give a glass supplier, and it is what enquiries most often lack.
Typical differentials for hot fill, pasteurising and washing
The ranges below show the order of magnitude each process family produces. They describe processes, not container ratings, and each should be replaced with figures measured on your own line.
| Process | Temperatures involved | Wall thickness need | Annealing need | How failure shows | Agree with the plant |
|---|---|---|---|---|---|
| Hot fill, then cooling | Product at roughly 82 to 95 degrees Celsius entering an ambient container: an inward heating step of about 60 to 70 degrees. Cooling by cold spray or immersion can apply a similar gap in the harmful direction. | Thin and even, which keeps the gradient small. Heel, base knuckle and a heavy base cool slowest and are the weak spots. | Low, uniform residual stress. Stress left by an incomplete annealing curve stacks on the thermal stress and uses up the margin. | A crack round the body or up from the heel soon after cooling, often with no bruise or impact origin. Breaks minutes to hours later are common when inspection is too early. | Wall thickness distribution, the heating and cooling profile expected, the annealing evidence available, and the surface quality level accepted on finish and body. |
| Tunnel pasteurisation | Product around 60 to 85 degrees Celsius, raised and lowered through successive spray zones. Each zone step is only a few degrees, but the swing from cold entry to coldest exit can be large. | Evenness matters more than any minimum. A thick base under a thin body cools at a very different rate, and that internal restraint causes the crack. | Very uniform, since every container gets the same stress and any lehr variation appears as a spread of failures. | Body cracks under the label or a crack from the heel, partway down the tunnel. Breaks grouped at one position point to the zone profile. | Spray temperature by zone, dwell per zone, the largest step between neighbouring zones, and the verified differential of the container. |
| Returnable bottle washer | Caustic bath around 60 to 80 degrees Celsius, then rinses that bring the bottle down. The biggest step is normally from the hot section into the first cooler rinse, repeated every trip. | Consistency from bottle to bottle. Thin sections are not the issue; variation is, because it creates a tail of early failures. | Endurance outranks single-shock strength. Residual stress that one test would tolerate turns into a failure mechanism across many trips. | A crack at the finish or through the neck during or right after washing, or a body crack found later in the filling hall. Breakage rising with trip count indicates slow crack growth. | Stage temperatures, the largest step between stages, expected trips, and an endurance result given as cycles at a stated differential. |
| Domestic dishwasher | Wash phase around 55 to 70 degrees Celsius, then a cooler rinse (the damaging step) and drying. Moderate gap, very many repeats over the life of a household item. | Even and moderately thick, because the jar must also cope with handling and stacking in the basket. A very thin wall trades the thermal problem for a mechanical one. | Low residual stress. Nothing inspects the jar between cycles, and a crack in a consumer's machine is a claim, not a rework. | A crack at finish, shoulder or body on a later cycle signals cumulative damage. A crack on first use more likely comes from a surface defect. | The cycle profile typical of the target market, the number of cycles to be claimed, and whether a cycle test or only a single-shock result will back the label. |
From a line profile to a container requirement in six steps
Leaving out any of the middle steps produces a requirement nobody can verify.
- Record the process as a profile. For each station list the container temperature, the temperature of what it meets, whether that is air, spray or immersion, and the dwell. A hot fill, an invert and hold, a cooling tunnel with three spray zones, labelling and a cold rinse are each their own thermal event. A spreadsheet with one column for container temperature and one for the medium is the right working document.
- Find the largest step and its direction. Cooling steps are the ones to worry about. In most plants the worst one is not the fill: it is a cold rinse on a freshly filled container, the first cold zone of the tunnel, or the move out of a hot washer bath into the first rinse.
- Count the cycles. Take how often the worst step occurs in one use and multiply by the uses planned. A returnable bottle making twenty trips meets it twenty times; a jar with a dishwasher claim may meet it hundreds of times if the household keeps it.
- Add a stated margin. In service a container is scratched by handling, stands on cold metal, is gripped by machinery and is not at an even temperature when the shock comes. Each of these lowers its effective resistance, so the verified value needs to sit above the plant requirement by a margin written down as a number.
- Convert the requirement into container features. Wall thickness and how it is distributed, annealing quality, surface condition and, to a smaller degree, glass composition carry the resistance. They belong in the specification next to capacity and finish, because the plant can adjust them and a laboratory can measure them.
- Fix verification and change control. Name the test, the differential, the sample size, the acceptance rule and the triggers for repeating it. A new mould, a different wall weight, a changed furnace load or altered annealing settings all shift thermal behaviour and void an earlier result, so the document should require a re-test after any of them.
Container features that set the limit
Wall thickness
For a given rate of heat removal, thickness sets the gradient. A thin wall evens out fast, so surface and core never differ much and the stress stays low. A thick wall lags, the surface gets far ahead of the core, and the stress climbs. A light, evenly blown container often beats a heavier one in thermal shock for this reason, contrary to the instinct that more glass means more strength.
It also explains why thermal cracks in hot filled jars so often begin at the heel and base knuckle, where the glass is unavoidably thickest. Stacking strength and impact resistance generally want more glass in the base, so the thermal requirement has to be balanced against them; the stacking side of that balance is set out in our guide to the vertical load test for glass bottles.
Annealing quality
Forming locks stress into a container. The annealing lehr, a long controlled-temperature tunnel, holds the glass in a band where that stress can relax and then cools it no faster than the glass can follow. Whatever stress remains is a permanent internal load, and a temperature difference adds to it, so the container cracks at a lower differential. A good shape can therefore fail simply because the lehr curve was off on the day of production.
A thermal shock failure may be caused by poor annealing, but a thermal shock pass does not show how well annealed the glass is. Residual stress is measured with a different instrument and method, described under the annealing test and how to specify it.
Surface condition
Glass under tension breaks from a flaw, and any flaw on a stretched surface can become the starting point. Plant defects such as checks, stones, blisters, mould seams and finish chips qualify. So do scratches from handling, rub marks from conveyor guides and damage from rough bulk packing. A lot that passed at the plant can fail the identical test after a scuffing trip through distribution, which is why the accepted surface quality level for finish and body goes into the specification and why packing and handling are more than a logistics matter.
Glass composition
Ordinary soda-lime container glass, well made, has enough resistance for hot filling, pasteurising and washing. Borosilicate glass resists considerably more and is the right choice when a process truly needs a very large differential, but it is a different product from a different supply chain. Before switching glass, check whether the process can be made gentler with slower cooling stages, a warmer rinse, staged spray zones or a longer dwell. Lowering the differential is a change the buyer controls.
How the test is run
ISO 7459 is the reference standard for both thermal shock resistance and thermal shock endurance of glass containers. Samples soak in a hot water bath until their temperature has evened out, move to a cold bath within a set time, stay there for a set dwell, and are then taken out and inspected. The differential is simply the gap between the two bath temperatures.
The resistance test finds the difference the samples survive, or the difference at which a stated share of them fail. The endurance test keeps the difference fixed, repeats the cycle and reports how many cycles were completed before failure.
Four test conditions decide whether a result can be carried over to a real process, so each must be recorded:
- Dwell in each bath. The wall has to reach bath temperature all the way through before transfer, or the shock applied is smaller than the bath gap suggests. Thick-based containers need longer than thin ones; using a thin-container dwell on a thick container flatters the result.
- Transfer time. The sample starts cooling in the air between baths. A transfer lasting several seconds softens the shock compared with one lasting a second, so the method should state the time. Hand-transferred bench tests cannot be compared with mechanised ones.
- Immersion or spray. A circulated bath is the repeatable laboratory case. A plant rinse is a spray: glass under the jet chills fastest while the shadowed side lags, concentrating the gradient in one area. A container that passes immersion at one differential may fail under spray at a smaller nominal one.
- Inspection method and hold. Thermal cracks can stay hidden at first, because glass keeps relieving stress and may let go minutes or hours afterwards. A polarised viewer or a dye finds more cracks than the naked eye, and a pass called one minute after removal is weaker than a pass called after a hold.
Writing the pass criterion
A clause that both sides will read identically names the process step being imitated, the hot and cold temperatures (and so the differential), the medium, dwell and transfer times, the cycle count when endurance is tested, the sample size, the acceptance rule, the inspection method and the hold before inspection. Without the last two, a batch can be released as sound and then break in the warehouse.
Express the acceptance rule through a recognised sampling scheme such as ISO 2859-1, with an AQL (acceptable quality limit) chosen to match how serious a failure would be. An informal understanding about how many breaks are tolerable does not hold up in a dispute.
For continuous production, separate the original qualification test from the routine batch check. A full endurance run is a qualification exercise; demanding it on every shipment is slow and costly for little gain.
Keep three figures apart in the document. The process requirement is what the line does to the container. The tested value is what the container showed in the laboratory. The margin between the two is the allowance for surface damage, mechanical handling and process variation, and having it in writing is what makes the specification defensible when a shipment breaks.
Reading a crack to separate thermal from mechanical causes
A typical thermal crack travels round the container, not along it. It may ring the body, cut through the wall at a shallow angle, or climb from the heel in a curve. There is usually no bruise, no contact mark and no origin with radiating lines of the kind an impact leaves, and the two broken faces often mate cleanly. It can happen at the instant of immersion or some minutes after, sometimes with an audible sound. A crack at the finish or through the neck after a hot wash or cold rinse is thermal too, though it is often blamed on handling because the finish is the most-touched part.
Delayed breaks
A container can leave a temperature event carrying a crack too small to see. That crack lengthens over the next minutes and hours until the container fails with nothing new acting on it. On the floor this appears as bottles bursting on a conveyor for no visible reason, when the real cause lies upstream. Match failures to each container's process history, not just to where it broke.
What the origin shows
If the origin can be located, it reveals which surface was stretched and so which way the heat flowed. An outer-surface origin means something cold on the outside: a rinse, a spray, a cold conveyor, a cold case, water on the plant floor. An inner-surface origin usually means hot product into a cold container or a cold internal rinse. A scratch, check or chipped finish at the origin says the flaw decided the outcome, and the fix lies in surface quality or handling, not in the annealing curve.
Ruling out other mechanisms
A vertical crack from the heel that appears under load is a compression failure. A pressure failure, whether from carbonation or from vacuum pulled under a lug lid, tends to branch and run through the body wall from the inside; the internal pressure test covers that mechanism and how to specify against it. Neither needs a temperature event, so a breakage with no thermal event in its record should be treated as mechanical until shown otherwise. Transport and handling damage normally leaves evidence such as a bruise, a chipped heel, a scrape or a visible contact origin. Thermal cracks seldom do, which is why they get misattributed.
Line practices that keep the process inside the limit
The verified differential is a ceiling. How the line is run decides how much of it gets used.
Cooling after hot fill
Cooling method is the biggest lever. From harshest to mildest: plunging into a cold bath, a directed cold spray, staged spray zones on a controlled profile, then air. When the verified value sits close to the process requirement, softening the cooling is usually simpler than changing the container. A staged spray ramp, a warmer first zone, a longer dwell or a slower tunnel conveyor are all changes your own plant can make, and none needs a new mould.
Cold water and cold surfaces
Do not put cold water straight onto a hot container. A large share of hot fill breakage comes from exactly this, since the container is at its peak temperature leaving the filler and the rinse hits one side as a jet. Where rinsing is necessary, stage the water temperature and let the container cool first. Surfaces count as well: a hot jar set on a cold metal conveyor, a stainless table or a wash bay floor gets a local chill that can start a crack even when the room feels normal.
Capping
A closure applied hot draws a vacuum as the contents cool. The vacuum pulls the container inward and loads the finish and shoulder in tension while thermal stress is still there, and the capping head presses on the finish, the area least suited to a concentrated load when hot. For a hot filled container closed with a lug lid or press-on cap, put the capping sequence and temperature into both the process description and the container requirement.
Washers and pasteuriser tunnels
Washing is the step most likely to overshoot the limit unnoticed. The first rinse after the caustic bath sets the biggest differential in the cycle, and if it runs off cold mains water the step can exceed the fill and cooling steps the specification was based on. Set the rinse stages so every step stays within the verified differential and neighbouring sections differ only slightly. In a tunnel, likewise, the gap between adjacent spray zones is what counts, not the absolute temperature of any one zone.
Late heat and cold storage
Heat added later is still a thermal event. A shrink sleeve tunnel on a cool container, a hot melt glue applicator touching the glass or an ink drying step can each create a local differential on a container specified only for fill and wash. Storage matters as well: a pallet left overnight in an unheated winter warehouse is a cold container, and hot filling it produces a bigger gap than filling at room temperature. If you fill through winter, add the seasonal temperature of the container store to the profile.
What to send for a thermal shock assessment
We need the process and the container. Send the process as a temperature profile giving container temperature, medium and dwell at each step. Send the container as a shape or drawing with capacity, finish and whatever wall thickness data exists. Tell us whether it is used once or many times, and whether the true requirement is single shock or endurance.
From that we can identify the governing process step and the differential the container must meet, frame a test both sides will accept, propose a margin for handling and surface damage, and word the clause so a change of mould or furnace settings triggers a re-check. If you have not yet settled on a container family, the glass bottle and jar ranges show the shapes and selection variables to start from.
Questions buyers ask about thermal shock testing
What temperature difference can a glass container survive?
No single figure applies. It depends on that container's wall thickness and distribution, annealing and surface condition, and on how the difference is applied. Well made hot fill containers generally cope with hot filling and washing when the process is staged, but the only number that belongs in a specification is one measured on the container you will buy, at your differential and in your medium.
Does the hot fill or the cooling crack the container?
Mostly the cooling. Fast cooling stretches the outer surface while the core is still hot; heating compresses the surface, which glass handles far better. On real lines the usual culprit is a cold rinse or spray on a container that has not yet cooled.
Is the test needed for a product filled at ambient temperature?
Not for the fill. Check the other stations, though: a caustic wash on returnables, a hot rinse before filling, a shrink sleeve tunnel or hot glue. If one of them puts a temperature difference across the glass, specify for it. If none does, thermal shock is not the governing requirement and the testing budget is better spent on the characteristics that are.
Why are a heavy base and thick heel a weakness here?
Thick glass evens out slowly, so its surface gets ahead of its core and the stress is higher than in a thin wall. The same mass that helps with stacking and impact is where thermal cracks tend to begin, so neither requirement should be pushed to an extreme.
How do I recognise a thermal shock crack on the line?
Look for a crack circling the container with no sign of contact, appearing soon after a wash, rinse or cooling step. Treat it as thermal until something proves otherwise.