Why hard milling runs dry: flood coolant, air blast and thermal cracking

Every machinist learns early that coolant is good. More coolant, longer tool life, better finish. Hard milling inverts that, and the first time someone sees a CBN insert destroyed by flood coolant it is genuinely surprising.

Here is the mechanism, and where the exceptions are.

Heat is doing useful work

In hard machining, the cutting zone reaches temperatures in the range of 800-1000 °C. That heat is not a side effect to be suppressed — it locally softens the material immediately ahead of the cutting edge, which is a large part of how a CBN tip can shear 62 HRC steel at all.

Cool that zone aggressively and you are asking the edge to cut fully hardened material at room temperature. Cutting forces rise sharply and the edge chips.

Thermal shock is the real killer

The more acute problem is cycling, not temperature.

In milling, each tooth is in the cut for a fraction of a revolution and out of it for the rest. With flood coolant on, that tooth goes: hot in the cut, quenched out of the cut, hot again, quenched again — thousands of times per minute.

CBN and ceramic are hard and brittle with limited thermal shock resistance. That cycling propagates comb cracks — a series of fine cracks perpendicular to the cutting edge — and the edge fails by chipping away between them. The insert does not wear out; it fractures.

This is why the standard practice in hard milling is dry cutting with compressed air, and why a quote that mentions flood coolant on a CBN finishing pass is worth a question.

What air blast is actually for

Running dry does not mean running with nothing. Compressed air through the spindle or a nozzle does two jobs:

Chip evacuation. In hard milling the chips come off small, hot and hard. Recutting them damages the edge and marks the surface. Getting them out of the cut is the main purpose of the air.

Bulk temperature control. Not cooling the cutting zone — cooling the part. A die block that gains 15 °C during a finishing pass grows measurably, and you will measure it later when it has cooled back down and the dimension has changed. Air keeps the bulk temperature closer to stable without quenching the edge.

Cold air guns and vortex tubes take this further, delivering air at well below ambient. They help on long finishing passes where thermal growth is the limiting factor.

Where MQL fits

Minimum quantity lubrication — a fine oil mist, typically a few tens of millilitres per hour — sits between dry and flood.

It works because the volume is far too small to quench the edge. It provides lubricity at the tool-chip interface and helps with chip evacuation, without the thermal cycling of flood.

MQL is genuinely useful for:

• Drilling and tapping in hardened steel, where chip evacuation from a deep hole is the problem and there is no interrupted-cut cycling.
• Hard milling with coated carbide rather than CBN, where the substrate tolerates the thermal variation better.
• Long continuous finishing passes where built-up edge is the failure mode.

It is not the default for CBN milling. If a shop proposes it there, ask why.

When flood coolant is correct

There are real cases:

Grinding. Completely different mechanism — the wheel needs flood both to flush swarf and to prevent thermal damage to the workpiece surface. Skimping on grinding coolant causes burn.

Wire EDM. The deionised water is part of the process, not a coolant.

Hard turning, sometimes. Turning is a continuous cut with no thermal cycling, so the argument against flood is weaker. Some shops run hard turning wet, particularly on long cuts where bulk part temperature drifts. Others run dry and control temperature by other means. Both are defensible; dry is more common because it avoids the coolant management entirely.

Pre-hardened steel at 30-35 HRC with carbide tooling. This is ordinary machining, not hard machining, and normal coolant practice applies.

What it means for your part

You are not specifying coolant strategy — that is the shop’s call. But two things follow for the drawing:

Thermal growth affects your tolerances. A part measured hot reads differently from the same part measured at 20 °C. If you are holding ±0.005 mm, say so, because it changes whether the shop lets parts normalise before inspection. See inspecting hardened parts.

Surface integrity is affected by heat management. Too much heat with a worn edge produces a re-hardened white layer with tensile stress beneath it. On a fatigue-loaded surface that matters — see white layer and surface integrity. Say on the drawing if a surface carries cyclic load.

If a shop describes their hard milling setup and flood coolant is in the description, that is a reasonable signal they are newer to this than their machine list suggests. Send the drawing and we will route it to shops where it is routine.