Quench Oil Selection and Testing for Indian Heat-Treatment Shops

Viscosity is the wrong number to buy on

Quenching oils are usually quoted the way every other industrial oil is quoted: kinematic viscosity at 40°C, flash point, density. Those numbers are real and they matter — viscosity governs how much oil leaves the shop on the parts, how well the load drains, and how hard the pumps have to work. What they do not describe is the only thing the steel responds to, which is how quickly heat comes out of the section and at what temperature.

Two oils that sit within a couple of centistokes of each other can produce different hardness and very different distortion on the same part. If a supplier can only tell you the viscosity of a quenchant, they have not told you what it does.

The three stages of a quench, and where parts actually go wrong

Drop a hot part into oil and three things happen in sequence, each with its own heat-transfer physics.

• Stage A — the vapour blanketThe part is hot enough to boil the oil around it instantly, and it wraps itself in a stable film of vapour. Vapour is an insulator, so cooling in this stage is slow. The film does not collapse evenly: it clings to recesses, to the underside of horizontal surfaces, and to anywhere the flow is dead. Wherever it holds on longest, that area of the part cools slowly enough to miss the transformation it was supposed to make. That is where soft spots come from.
• Stage B — nucleate boilingThe film collapses and the oil boils violently against the metal. This is the fast stage, and it is where the heat actually comes out. A fast-quench oil is one formulated to reach this stage sooner and hold it longer.
• Stage C — convectionBelow the boiling range of the oil, cooling is by ordinary convection and it is the slowest stage of the three. For most steels this is the range in which martensite forms, and the cooling rate here governs residual stress, distortion and cracking.

Put plainly: you want the oil to be fast where the steel needs speed, high in the range, and unhurried where it does not, low in the range. An oil that is fast everywhere hardens beautifully and cracks parts. An oil that is slow everywhere leaves them soft. Every quenchant on the market is a position taken between those two errors.

What a cooling-curve test actually measures

The standard test is the same in substance under both ISO 9950 and ASTM D6200. A cylindrical probe of Inconel 600, 12.5 mm in diameter and 60 mm long, with a type K thermocouple at its geometric centre, is heated to 850°C and quenched into a fixed volume of the oil at a stated bath temperature, without agitation. The instrument records temperature against time, then differentiates it to give cooling rate against temperature.

That derivative curve is the document worth reading. Four things come off it:

• Maximum cooling rateThe peak of the curve, in degrees per second. On its own this is the number people quote and the number that misleads them.
• The temperature at which the maximum occursRead together with the peak, this tells you whether the oil is fast in the range where speed prevents soft structures, or fast lower down where speed cracks parts. This is the pair that decides the job.
• Cooling rate at 300°CA proxy for behaviour through the martensite range on most steels. High here means distortion and cracking risk.
• Time to reach 600, 400 and 200°CThe practical summary, and the numbers easiest to trend on a used bath against the new-oil certificate.

The test is run unagitated for a reason: it characterises the oil, not your tank. What happens in the tank is a separate question, and one most shops answer badly.

The Quenchometer, and what it cannot tell you

ASTM D3520 — the General Motors Quenchometer test — heats a nickel ball to 885°C, drops it into 200 millilitres of quenchant, and times how long it takes to fall to the Curie point of nickel at around 355°C, the temperature at which the ball becomes magnetic and a magnet below the beaker picks it up. The answer is one number in seconds. A faster oil gives a shorter time.

It is quick, cheap and repeatable, which makes it genuinely useful in two places: checking an incoming batch against the supplier’s figure, and watching a working bath for drift. What it cannot do is separate an oil that is fast high in the range from one that is fast low down, because it collapses the whole cooling history into a single interval. Two oils with the same Quenchometer number can behave completely differently on a real part.

Treat it as a control chart. Treat the cooling curve as the specification.

The Castrol Iloquench range, as Castrol publishes it

Four grades cover most of what an Indian heat-treatment shop needs, and they are separated by cooling behaviour and bath temperature rather than by viscosity.

Two rows in that table carry most of the decision. The Wolfson cooling time separates Iloquench 1 from Iloquench 21 — roughly half the time over the same range, which is the whole difference between a hardness-led and a distortion-led job. And the vapour-pressure row is why Iloquench 25 exists at all: an order of magnitude below the standard grades, which is what a vacuum furnace and its pumps require. Take per-batch figures from the certificate of analysis rather than the catalogue minimum, particularly the flash point.

Castrol also lists specialty grades for austempering, isothermal quenching and specified water-emulsion quenchants. If an OEM heat-treatment specification names something outside the four above, send us the specification rather than the nearest-looking grade — the full Iloquench range is wider than the four rows that cover most work.

Fast or delayed: making the decision in practice

The honest version of this decision is short. Choose Iloquench 1 when hardness is the constraint: tough alloy steels, through-hardening and case-hardening work, sections uniform enough to take the thermal gradient without moving. Choose Iloquench 21 when distortion is the constraint: thin sections, complex geometry, parts where the scrap is dimensional rather than metallurgical, and lower-alloy steels that still reach target hardness on a slower curve.

When those two descriptions both fit the part — and on real work they often do — the argument cannot be settled from data sheets. Run the actual part through both, and measure two things: a hardness traverse through the section, and the distortion against the drawing tolerance. One trial answers a question that a year of specification meetings will not.

Martempering is a different operation, not a hotter bath

In a conventional quench the part goes from austenitising temperature straight through the martensite range in one movement, and the surface transforms well before the core does. That difference in timing across the section is where distortion comes from.

Martempering interrupts it. The part is quenched into a bath held above the martensite start temperature — 150–200°C for Iloquench M — and held there long enough for the section to equalise. Only then is it removed and allowed to cool through the martensite range in air, with the whole section at much the same temperature. The transformation still happens; it happens more evenly. On complex tooling and thin-walled forgings that is the difference between a part that needs straightening and one that does not.

One safety point deserves stating plainly. On a cold bath the gap between working temperature and flash point is well over a hundred degrees. On a martempering bath at 150–200°C against a minimum flash point of 230°C it is much narrower. That is not an argument against martempering — it is why hot baths are run with covers and level control, why water ingress is treated as an emergency rather than a maintenance item, and why the flash point that matters is the one on the batch certificate rather than the catalogue minimum.

Vacuum-furnace duty

A vacuum furnace changes the problem. Anything that evaporates from the oil ends up in the chamber and in the pumps, so the property that matters most is vapour pressure. Iloquench 25 sits at or below 0.1 mbar at 50°C where the standard grades sit between 0.5 and 1.0 — the reason it is specified for clean-furnace work and for components where surface condition after hardening is part of the acceptance criteria.

Its cooling behaviour sits between the fast and delayed grades, which is usually workable. If a vacuum job needs genuinely fast cooling, that is a furnace and gas-quench conversation before it is an oil conversation.

Agitation beats grade more often than anyone admits

The severity of a quench is not a property of the oil alone. It is a property of the oil, the bath temperature and the flow across the part, and the third of those is the one most shops leave to whatever the tank was built with twenty years ago.

A bath with dead zones will produce soft spots on parts sitting in them no matter which Iloquench grade is in the tank, because the vapour blanket in Stage A survives where nothing is moving. Load density, fixture design, basket orientation and the direction of flow relative to the load all move the outcome further than a grade change usually does. Before changing oil to fix a soft-spot problem, map where on the load the soft parts are sitting. If they cluster, the problem is flow.

What ages a bath, and the tests that see it

A quench bath does not wear out so much as drift, and it drifts along several axes at once.

• Viscosity at 40°C, ASTM D445Rises as the oil oxidises and polymerises. A rising trend against the new-oil figure is the earliest cheap warning.
• Total acid number, ASTM D664Tracks oxidation directly. Judge it against the new-oil baseline on the certificate of analysis, not against a universal number — different grades start in different places.
• Water content, Karl FischerThe one that causes sudden trouble rather than gradual trouble. See below.
• Flash point, ASTM D92Matters most on hot baths, where the operating margin is narrowest.
• Sludge and insolublesOxidation products and carried-over scale. They foul the tank, coat the parts and change how the vapour blanket forms.
• A repeat cooling curve, ASTM D6200The definitive test, because it measures the property you actually bought. Everything above is a proxy for this.

Bath life varies enormously between shops running the same oil, and the largest single reason is drag-out. Oil that leaves on the parts has to be replaced, and a bath that is topped up frequently is partly refreshing itself; a well-draining load in a tight shop is not. That is why a drain interval quoted in months from someone else’s plant is close to meaningless. Set yours from your own trend data.

Water is the contaminant that cracks parts

Water reaches a quench bath from a small number of predictable places: a leaking cooling coil or heat exchanger, carry-over from a washer, parts that go in wet, and straightforward condensation during the monsoon months. Storage practice upstream matters here too, and the same discipline that protects drums in the yard protects a quench bath — our guide to lubricant storage in Indian conditions covers the sealing and handling side of it.

What water does is disproportionate to how much of it there is. It destabilises the vapour blanket, so cooling becomes uneven across the part: soft where the film held, highly stressed where it broke early, and at higher levels cracked outright. It causes foaming. And on a bath above the boiling point of water it flashes, which turns a metallurgical problem into a fire one.

Take the action limit from the product data sheet rather than a rule of thumb, test by Karl Fischer on a schedule, and treat a rising trend as a leak to be located rather than a reading to be recorded.

A changeover checklist

• Start from the part, not the tankSteel grade, section thickness, the hardness specification and the distortion tolerance. Those four decide fast against delayed before any product name is mentioned.
• Get the new-oil cooling curve and keep itIt is the baseline every later test is judged against. Without it, trending is guesswork.
• Check the tank before you change the oilAgitation, dead zones, heater and cooler capacity, filtration, and whether the bath can actually hold the temperature the new grade needs.
• Clean out the old bath properlySludge left in the tank contaminates the new charge immediately and makes the first cooling curve meaningless.
• Confirm the fire and level systems for a hot bathCovers, level control, temperature interlocks and the batch flash point. A martempering bath is a different safety case from a cold one.
• Run a trial load with measurementHardness traverse and distortion against the drawing, on the real part, before the line runs on it.
• Set the sampling schedule on day oneViscosity, acid number, water and a periodic cooling curve. A bath with no baseline and no trend cannot be managed, only replaced.

Frequently asked questions

How do I choose between Iloquench 1 and Iloquench 21?

Decide which failure you are managing. Iloquench 1 is the fast-quench grade: it gets into the nucleate-boiling stage sooner and pulls heat out faster in the range where soft transformation products form, so it is the choice when a tough alloy steel needs the hardness and the part is simple enough in section to take the thermal gradient. Iloquench 21 is the delayed-quench grade: Castrol characterises it at a longer cooling time over the same range, which lowers the thermal gradient through the section and with it the distortion. Choose it for thin sections, complex geometry, and any job where the distortion budget rather than the hardness number is what scraps parts. When the answer is not obvious, run the same part through both and measure a hardness traverse and the distortion, rather than arguing from the data sheets.

What does a cooling curve tell me that viscosity does not?

Almost everything that matters. Viscosity governs drag-out, drainage and pumping. It says nothing about how fast the oil extracts heat, or at which temperature it does so. Two quenching oils with the same viscosity at 40 degrees Celsius can produce very different hardness and very different distortion on the same part. A cooling curve to ISO 9950 or ASTM D6200 gives you the cooling rate against temperature: the maximum rate, the temperature at which it occurs, and the rate low down where martensite forms. Those are the numbers the steel responds to.

Is the GM Quenchometer test enough on its own?

No, and it is not meant to be. ASTM D3520 heats a nickel ball to 885 degrees Celsius, drops it into 200 millilitres of quenchant and times how long it takes to fall to the Curie point of nickel, around 355 degrees Celsius, where a magnet picks it up. It is fast, cheap and repeatable, which makes it a good incoming-batch check and a good drift alarm on a working bath. What it cannot do is tell you where in the temperature range the oil is fast, and that is the thing that decides whether a part comes out hard or cracked. Use it as a control chart alongside a cooling curve, not instead of one.

Why does a small amount of water in a quench bath matter so much?

Because it attacks the stage of the quench that is already the least stable. Water changes how the vapour blanket forms and collapses, so cooling becomes uneven across the part: soft spots where the film held, high local stress where it broke early, and in the worst cases quench cracking. It also causes foaming, and on a martempering bath running above the boiling point of water it flashes, which is a fire and safety problem rather than a metallurgy one. Take the action limit from the product data sheet, test by Karl Fischer, and treat a rising trend as a leak to be found rather than a number to be logged.

How do I know when a quench bath is finished?

From trend data, not from a date. Track viscosity at 40 degrees Celsius by ASTM D445 and total acid number by ASTM D664 against the new-oil baseline on the certificate of analysis, water by Karl Fischer, flash point, and sludge or insolubles. Then re-run the cooling curve, because that is the only test that measures the property you actually bought. Bath life varies enormously between shops on the same oil, mostly because of drag-out: a bath that loses oil on every load and is topped up is partly refreshing itself, and one that does not is not.

Quench-oil selection is one of the few lubrication decisions where the wrong answer shows up as scrap rather than as wear, which is why it is worth settling with a cooling curve and a trial load rather than with a catalogue. We supply the Castrol quenching-oil range across Andhra Pradesh and Telangana and can arrange bath sample analysis as part of the supply relationship.