Annealing Schedule Calculator: Soak and Cool Without Cracks

Glass that cools too fast keeps the stress inside it, and stress inside glass is a crack that has not happened yet. This calculator takes the thickest section of your piece and the anneal soak temperature your glass calls for, then returns three numbers: how long to hold at the soak, the fastest you can drop after it, and how many hours that drop will take to reach the strain point.

Thickness is the variable that matters, and it matters more than most people expect, because the safe cooling rate falls with the square of the thickness. Doubling the wall of a piece from 6 mm to 12 mm does not halve the cooling rate. It quarters it, and the cool from 516 °C down to the strain point goes from under an hour to three and a half.

Two fields, three outputs, and a red warning under them that is there on purpose. Annealing schedules belong to the glass, not to the calculator, and a kiln-formed piece in Bullseye and a torched piece in borosilicate hold at different temperatures for different reasons.

✂️ How to use the Annealing Schedule calculator

Start with Glass thickness. This is the thickest cross-section in the piece, measured in millimetres, stepping by 1, default 12. Not the average. Not the wall of the thin part. The thickest place, because that is where the stress lives and that is the section that has to release it.

Measure it honestly. A blown vessel with a 3 mm wall and a 14 mm foot is a 14 mm piece as far as the kiln is concerned, and entering 3 because most of the glass is thin will give you a schedule that anneals the body beautifully and leaves the foot under load.

The thickest section wins. Kiln-formed work with a thin sheet and a fat rim, a paperweight with a bubble in the centre, a bead with a heavy core: all of them are governed by the fattest measurement in the piece.

Next comes Anneal soak temp, in Celsius, stepping by 1, default 516. This number comes from your glass manufacturer and nowhere else. Bullseye and Spectrum COE 90 sheet glass anneal around 482 °C. Borosilicate COE 33 sits far higher, around 560 °C. Soda-lime studio glass in the 96 COE range typically lands near 516, which is why that is the default.

Lisa Mandel

The three results appear immediately. Soak at anneal tells you how long to hold at that temperature so the whole thickness reaches it and the internal stress relaxes. Max cool rate is the fastest safe drop in degrees per hour. Cool to strain point converts that rate into a duration, ending at 370 °C, which the calculator treats as the strain point for every glass.

The large tile below the table repeats the soak in minutes, which is the number you actually type into a kiln controller as the hold, and the one people most often forget to set.

Read the cool time before you commit to firing. A 12 mm piece at 516 °C needs 3.5 hours of controlled cooling after the soak, and if the kiln has to be opened in two hours because you have somewhere to be, do not start the firing.

📋 Calculator fields explained

Glass thickness (mm) – the thickest cross-section of the piece, in millimetres, stepping by 1, default 12. This value drives both the soak length and the cooling rate. The rate falls with the square of it, so the effect is dramatic rather than proportional.

Anneal soak temp (°C) – the temperature at which the glass is held to relax internal stress, in Celsius, stepping by 1, default 516. Take this from the manufacturer’s published schedule for your specific glass. It only affects the cooling duration, since the calculator measures the drop from this temperature down to 370 °C.

📊 Understanding the results

The soak is 5 minutes per millimetre of thickness, with a floor of 30 minutes. A 12 mm piece soaks for 60 minutes. A 6 mm piece would calculate to 30 and the floor keeps it there, so everything from 1 mm up to 6 mm gets the same half-hour hold.

The cooling rate is the number that punishes thick work. It comes from 6000 divided by the thickness squared, clamped between 20 and 300 °C per hour. At 6 mm that gives 167 °C/hr. At 12 mm it collapses to 42 °C/hr. Doubling the thickness cuts the safe cooling rate to a quarter, which is the single most useful fact in kiln work.

Cool to strain point is the drop from your soak temperature down to 370 °C, divided by that rate. With the defaults, 516 minus 370 is 146 degrees, and at 42 °C/hr that is 3.5 hours of controlled descent after the soak has finished.

The 370 °C strain point is hard-coded and is a soda-lime figure. Borosilicate strains nearer 510 °C and a real boro schedule cools over a much smaller window, so this tool will substantially overestimate the cooling time for COE 33 glass.

Both clamps are worth knowing about. Anything thinner than about 4.5 mm reports 300 °C/hr, because the formula would go higher and the ceiling stops it. Anything thicker than roughly 17.3 mm reports 20 °C/hr, because the floor catches it. A 20 mm block and a 40 mm block both show 20 °C/hr and a 7.3 hour cool, which is the point at which you should be reading a published slab schedule rather than a calculator.

The soak keeps rising past that floor, though, and that is where the calculator still earns its keep on thick work. A 30 mm piece soaks for 150 minutes, five times the minimum, and the cooling rate has already bottomed out.

Nothing here covers the ramp up, the process hold, or the final drop from 370 °C to room temperature. This is the second half of a firing schedule only, and the half where the damage is done.

🧮 Calculation formulas

Three lines. Let t be the thickness in millimetres and T the anneal soak temperature in Celsius.

soak (minutes) = max(30, t x 5)
cool rate (°C/hr) = clamp(6000 / t², 20, 300)
cool time (hours) = (T - 370) / cool rate

The inverse-square in the rate formula is the physical heart of it. Heat leaves glass through its surface, and the stress that builds during cooling depends on the temperature difference between the surface and the core. That difference grows with the square of the distance from surface to core, so the tolerable cooling rate falls with the square of the thickness.

Working the defaults through: t = 12, T = 516. Soak is max(30, 60) = 60 minutes. Cool rate is 6000 / 144 = 41.7, which sits inside the clamp and displays as 42 °C/hr. Cool time is (516 – 370) / 41.67 = 3.5 hours.

Change nothing but the thickness. At 6 mm: soak is max(30, 30) = 30 minutes, rate is 6000 / 36 = 167 °C/hr, cool time is 146 / 166.67 = 0.9 hours. The same 146-degree window that took three and a half hours at 12 mm now takes fifty-five minutes.

The soak and the cool rate move in opposite directions with thickness. The hold gets longer and the descent gets slower at the same time, which is why a thick piece costs so much more kiln time than its volume suggests.

The full curve, at the default 516 °C soak, looks like this, and the clamps are visible at both ends:

The soak temperature only touches the last line of the maths, and it does so linearly. A lower anneal temperature means a shorter window down to 370 °C and a shorter cool at the same rate:

That last row is the one to treat with suspicion. Boro really does soak around 560 °C, but its strain point is nowhere near 370 °C, so the 4.6 hour figure describes a cool far longer than any published boro schedule asks for.

🎨 Practical examples

1. Fused coaster, two layers of 3 mm. Thickness 6 mm, soak temp 482 °C. Soak is 30 minutes, cool rate is 167 °C/hr, cool to 370 is (482 – 370) / 166.67 = 0.7 hours. Forty-two minutes of controlled descent after a half-hour hold, which fits comfortably inside an evening firing.

2. Fused plate, three layers plus a rim. Thickness 12 mm, soak 482 °C. Soak 60 minutes, rate 42 °C/hr, cool 2.7 hours. The rim is the reason: the sheet itself is 9 mm but the rolled edge doubles up locally, and entering 9 instead of 12 would give a 74 °C/hr descent that is nearly twice too fast for that rim.

3. Blown tumbler with a thick base. Wall is 3 mm, base is 10 mm. Enter 10, soak temp 516 °C. Soak is 50 minutes, rate 60 °C/hr, cool 2.4 hours. The wall would have been happy at 300 °C/hr; the base decides the program.

4. Solid paperweight. Thickness 50 mm, soak 516 °C. Soak is 250 minutes, over four hours. The rate reads 20 °C/hr because it is pinned at the floor, and the cool shows 7.3 hours. A 50 mm solid needs a four-hour soak before cooling even begins, and the real published schedules for solids that size run considerably longer than the 7.3 hours the clamp allows.

Once the rate tile reads 20 °C/hr, the calculator has stopped telling you anything about the cooling. Treat that as a signal to open the manufacturer’s slab chart, not as an answer.

5. Lampworked beads. Thickness 8 mm at the widest, soak 516 °C. Soak 40 minutes, rate 94 °C/hr, cool 1.6 hours. Beads go into a kiln that is already at the anneal temperature straight off the mandrel, so the ramp up does not apply and this schedule is essentially the whole firing.

6. Slumped bowl in COE 90. Thickness 6 mm, soak 482 °C. Soak 30 minutes, rate 167 °C/hr, cool 0.7 hours. Slumping introduces almost no thickness change, so the schedule is the same as the flat blank it started as, and the total anneal segment runs about 72 minutes.

7. Studio production run of 20 tumblers. All at 10 mm base thickness, 516 °C. Fifty-minute soak, 2.4 hour cool, so 3.2 hours of anneal per kiln load, plus the ramp and the final descent below 370 °C. Loading all 20 in a single firing rather than four firings of five saves nearly ten hours of controlled cooling.

8. Thick cast panel. Thickness 25 mm, soak 516 °C. Soak is 125 minutes, rate is pinned at 20 °C/hr, cool 7.3 hours. Published casting schedules for a 25 mm panel typically call for something closer to 15 °C/hr and a longer soak, so the calculator is optimistic here and the honest move is to slow it down by hand.

9. Borosilicate sculpture. Thickness 15 mm, soak 560 °C. Soak is 75 minutes, rate 27 °C/hr, cool 7.0 hours. The soak and the rate are believable; the cool duration is not, because boro’s strain point is far above 370 °C and the real controlled window is much narrower. Use the soak and the rate, and end the descent where your boro supplier says to.

10. Thin sheet with a fat inclusion. A 4 mm fused panel with a 14 mm glass cabochon set into it. Enter 14, not 4. Soak 70 minutes, rate 31 °C/hr, cool 4.7 hours. The panel would happily cool at the 300 ceiling, and the cabochon would crack straight through if you let it.

💡 Tips and best practices

Take the anneal temperature from the manufacturer and the soak from the calculator. Those are two different kinds of number: the first is a chemical property of the glass, the second is a function of geometry.

Measure the thickest part with a caliper, not by eye. The difference between 10 mm and 14 mm is a cooling rate of 60 °C/hr against 31 °C/hr, and a two-hour difference in the firing.

Program the cool rate as the ramp for the segment below the soak, and set the segment to end at 370 °C. Below that temperature the glass has become rigid enough that stress no longer builds, and most controllers can be allowed to free-fall from there with the kiln closed.

Round the rate down, never up. If the calculator says 42 °C/hr, program 40. The kiln does not mind the extra six minutes and the glass gets a margin it will never use against you.

Do not open the kiln at 370 °C. The controlled segment ends there, but the piece is still hot enough to thermal shock in a cold room, and the last hundred degrees should still happen with the lid down.

Run a polariscope check on a test piece when you introduce a new glass or a new thickness. Two crossed polarising filters and a light source will show residual stress as coloured fringes, and it is the only way to know whether the schedule actually worked.

Soak longer than the calculator says if the piece has a large surface area as well as a thick section. A 12 mm slab that is 40 cm across takes longer to reach a uniform temperature than a 12 mm cube, and the 5-minute-per-millimetre rule only tracks the thickness.

Keep a firing log with the thickness, the soak, the rate and whether the piece survived. Three entries in and you have a house schedule for your own kiln, which drifts from the published one because kilns are not identical.

Treat the 20 °C/hr floor as a warning light. Anything above 17 mm has left the range where this formula is doing real work, and the glass manufacturers publish tables specifically for that territory.

Where a piece contains two different glasses, anneal at the lower of the two soak temperatures and use the thickness of the whole assembly. Compatibility is a separate problem, and it is not one a schedule can fix.

⚠️ Common mistakes to avoid

Entering the average thickness

The formula assumes the number you type is the thickest section, because that is the part that holds the most stress and cools the slowest. A blown vessel with a 3 mm wall and a 12 mm foot is not a 6 mm piece.

Entering 6 mm for a piece that is 12 mm at the base gives a cooling rate of 167 °C/hr instead of 42, four times too fast, and the base will crack days or weeks after the firing.

Caliper the thickest point and use that number, even if 90 percent of the piece is thin. The thin part will simply anneal comfortably.

Using the default temperature without checking the glass

516 °C is a reasonable soda-lime figure and a poor fit for anything else. Bullseye COE 90 wants 482 °C, and holding it 34 degrees too high risks devitrification on the surface and softening of fine detail. The number belongs to your glass, not to the calculator.

Skipping the soak because the piece is thin

The 30-minute floor exists for a reason. Even a 3 mm coaster needs the whole thickness to equalise at the anneal temperature before the descent begins, and a five-minute hold does not achieve that in any kiln with real thermal mass.

Trusting the cool time for borosilicate

The strain point in this tool is fixed at 370 °C, which is a soda-lime number. Borosilicate strains near 510 °C, not 370 °C, so the calculator describes a controlled descent almost three times longer than a boro schedule needs. The soak and the rate are usable; the duration is not.

Applying a soda-lime strain point to boro wastes kiln hours at best, and at worst gives a false sense that the schedule has been validated for a glass it was never derived from.

For COE 33, take the schedule from the tubing manufacturer and use the calculator only as a sanity check on the soak length.

Ignoring the 20 °C/hr floor on thick work

The rate stops falling at 20 °C/hr, so a 40 mm cast piece is given the same descent as an 18 mm one. That is not physics, it is a clamp in the code, and thick casting schedules routinely go below 10 °C/hr.

Cracking the lid to speed things up

Opening the kiln during the controlled cool destroys the whole point of the exercise. The surface drops fast, the core does not, and the temperature gradient you spent hours avoiding appears in a few seconds.

Annealing at the process temperature

The anneal soak is far below the fusing or working temperature. Holding a fused piece at 800 °C for an hour does not anneal it; it fuses it further, rounds the edges away, and leaves all the stress to be dealt with on the way down anyway.

🎯 When to use this calculator

Open it when the thickness changes. A schedule that has worked for a dozen 6 mm coasters says nothing about the 12 mm bowl you are about to fire, because the rate has fallen by a factor of four and the cool has gone from under an hour to nearly four.

Open it before you plan the day. The cool to strain point is a real duration in real hours, and a 15 mm piece at 516 °C commits you to five and a half hours of controlled descent that you cannot shorten by turning the kiln off.

Open it when you are quoting a piece. Kiln time is electricity and it is the studio’s biggest fixed cost per firing, and a 25 mm cast panel occupies the kiln for the best part of a working day.

The glass does not crack in the kiln. It cracks three weeks later, on a shelf, in a warm room, for reasons that were decided during a firing nobody remembers.

Leave it closed when you have a published schedule for the exact glass and the exact thickness in front of you. Bullseye, Spectrum and the boro tubing makers all publish annealing charts, and those are derived from measurements the calculator can only approximate.

🔗 Related calculators

Kiln Firing Schedule

Coe Compatibility

Kiln Electricity Cost

Glass Weight

Fusing Volume Control

Gather To Piece Size

Glass Studio Pricing

📖 Glossary

Annealing – holding glass at a temperature where it is rigid but still relaxes, then cooling it slowly enough that no permanent stress is locked in.

Anneal point – the temperature at which internal stress in glass relaxes in a few minutes. This is what the soak temp field asks for.

Strain point – the temperature below which glass is rigid enough that no further stress can be introduced or released. Fixed at 370 °C in this calculator.

Soak – a hold at a constant temperature, in minutes, giving the whole thickness time to reach that temperature.

Cooling rate – the controlled descent speed, in degrees Celsius per hour, programmed as a ramp in the kiln controller.

Why is the cooling rate an inverse square of thickness? Because the temperature difference between the surface and the core is what creates stress, and that difference grows with the square of the distance heat has to travel.

COE – coefficient of expansion, the number that says how much a glass grows when heated. Glasses of different COE cannot safely be fused together.

Borosilicate – COE 33 glass, low expansion, high working temperature, used for lampworked pipes and scientific ware.

Soda-lime – the common studio and window glass family, typically COE 90 to 104.

Thermal shock – cracking caused by a sudden temperature change across a piece, distinct from the slow stress annealing addresses.

Devitrification – a crystalline scum forming on the glass surface when it is held too long in a certain temperature band, usually appearing as a dull grey haze.

Polariscope – two crossed polarising filters and a light, used to see residual stress in finished glass as coloured fringes.

Ramp – a controlled change of temperature, up or down, at a set rate.

Slumping – reheating a flat fused piece so it settles into a mould under its own weight.

Kiln controller – the programmable unit that runs the ramps and holds. The soak in minutes and the rate in degrees per hour are the two figures you type into it.

❓ Frequently asked questions

What thickness should I enter for a blown piece?

The thickest cross-section, which for most blown work is the foot or the base rather than the wall. A tumbler with a 3 mm body and a 10 mm base is a 10 mm piece.

The difference is large. Ten millimetres gives 60 °C/hr, while three would give the clamped 300 °C/hr, and that base will not survive a descent five times faster than it needed.

Where does 370 °C come from?

It is the strain point the calculator uses for every glass, and it is a reasonable soda-lime figure. Below it, the glass is rigid and stress can no longer form.

The number is fixed and cannot be changed from the interface. That makes the tool accurate for COE 90 to 104 studio glasses and misleading for borosilicate, whose strain point sits near 510 °C.

Every published annealing schedule ends the slow section at the strain point and lets the kiln free-fall below it. The strain point is the door out of the dangerous part of the firing.

Why does the cooling rate stop changing on thick pieces?

Because it is clamped at a floor of 20 °C/hr. The formula would give 15 °C/hr at 20 mm and 6.7 °C/hr at 30 mm, but the code will not go below 20.

That means any piece thicker than about 17.3 mm reports the same rate and the same 7.3 hour cool at the default temperature. The soak keeps growing, but the cooling figure is no longer meaningful and you should be reading a published slab chart.

Can I just cool slower than the calculator says?

Yes, and it is always safe. The rate is a maximum, not a target, and a slower descent introduces no risk beyond electricity and time.

Programming 40 °C/hr where the calculator says 42 adds roughly ten minutes to a 12 mm firing. That is a good trade against a crack you find in November.

Does the calculator handle the ramp up and the process hold?

No. It covers the anneal soak and the controlled descent below it, which is the second half of a firing schedule.

The heat-up rate, the fusing or slumping temperature, the process hold and the final drop from 370 °C to room temperature are all outside this tool. A full program for a 12 mm fused piece has five or six segments, and this calculator gives you two of them.

What soak temperature should I use?

Whatever your glass maker publishes. Bullseye and Spectrum COE 90 fusing sheet anneal at 482 °C. Studio soda-lime in the COE 96 range typically sits near 516, which is the default here.

Borosilicate COE 33 anneals near 560 °C. Boro anneals 78 degrees above the default in this calculator, so leaving the field alone for a boro piece is not a small error.

How do I know if the annealing worked?

Look at the piece through two crossed polarising filters with a light behind it. Well-annealed glass shows almost nothing. Stressed glass shows coloured fringes concentrated at the thick sections.

The other test is time. Badly annealed glass often survives the firing and cracks weeks later when the ambient temperature changes, which is why a polariscope check on a test piece is worth the twenty pounds the filters cost.

Why does a thin piece show the same soak as a slightly thicker one?

Because the soak has a 30-minute floor. Anything from 1 mm to 6 mm calculates to 30 minutes or less and gets held at 30.

Above 6 mm the 5-minutes-per-millimetre rule takes over. A 9 mm piece gets 45 minutes, a 12 mm piece gets 60, and a 30 mm piece gets 150.

Can I anneal two different glasses together?

Only if they are compatible in COE, which is a separate question this calculator does not touch. Incompatible glasses will crack apart regardless of how well you anneal them.

If they are compatible but have different published anneal temperatures, use the lower of the two and the thickness of the whole assembly. The lower temperature keeps the softer glass from moving, and the longer soak accommodates both.

Do I need to anneal small beads?

Yes, and the fact that they look fine off the mandrel is not evidence otherwise. An 8 mm bead at 516 °C needs a 40-minute soak and a 1.6 hour cool by this schedule.

Unannealed beads survive the bench and fail in a customer’s hand, often after being knocked or after a change in room temperature. Bead kilns are cheap for exactly this reason.

⚖️ Disclaimer

This calculator provides educational craft-planning estimates for glass annealing. The soak, cooling rate and duration are produced by a general formula based on thickness and the temperature you enter, and they are not a substitute for the published schedule supplied with your glass.

Annealing behaviour is specific to each glass composition. Anneal points, strain points and safe cooling rates differ between manufacturers and between product lines from the same manufacturer, and the 370 °C strain point used here is a soda-lime figure that does not apply to borosilicate or to many casting glasses.

Kilns differ in thermal mass, element placement and controller accuracy, so the same program can behave differently in two studios. Verify results with a polariscope on a test piece before committing valuable work, and log what actually happened rather than what was programmed.

Hot glass and kiln firing involve real risk of burns, fire and glass failure. Follow the safety guidance supplied with your kiln and your glass, use appropriate eye protection, and never open a kiln during a controlled cooling segment.