What grinding depth is safe before you over-grind a rail?

Question

Accepted Answer

What grinding depth is safe before you over-grind a rail?

There is no single safe number, because depth is two limits rather than one. The first is depth per pass: what one grinding run takes off the running surface. In Chinese practice that is planned at about 0.1 mm for preventive grinding and at least 0.2 mm for pre-grinding or corrective grinding at the rail top centre, while acceptance caps parent-metal removal at 0.5 mm. The second is depth over the life of the rail: how much the head can afford to lose before the profile leaves its envelope and too little section remains. A job is safe only when both hold — each pass inside its band, and the total ground since the rail was new still leaving the head inside envelope.

Rail grinding in progress on track — Depth is decided twice on every job: once per pass on the machine, and again over the life of the rail in the profile records.

The two limits that bound every grinding depth

Limit	What it actually is	Where it comes from	How it is checked
Depth per pass	The metal one run removes at a stated position on the head, normally rail top centre between −1° and +3°	The grinding strategy: pre-grinding and corrective work are planned at ≥0.2 mm, preventive work at ≥0.1 mm; acceptance caps parent-metal removal at 0.5 mm	Profile measured before and after the run at the same cross-section
Lifetime depth	Everything ground off since the rail was laid, added up	The profile envelope the infrastructure manager sets: the rail top centre may sit +0.2 / −0.4 mm from target and zones 2, 3 and 4 within ±0.2 mm	GQI or an equivalent profile index against the target template, section by section
Design wear allowance	The part of the head intended to be consumed by traffic over the rail's service life	The rail section and the operator’s renewal plan — grinding spends this allowance, it does not create it	Head loss recorded against the renewal forecast, not against the last grinding job
Defect depth window	How deep the defect still is, and therefore whether one pass can clear it	Rolling contact fatigue is best removed while the crack is under about 0.2 mm	Crack depth on a removed sample, or the defect pattern plus the depth the last job actually reached

The mistake behind most over-grinding is treating the 0.5 mm figure as a target. It is a ceiling for acceptance, not an allowance to be used up on every visit. The corrective trigger is separate again: on conventional lines in China, repair grinding is organised once corrugation depth or lipping width passes 0.3 mm — a threshold for acting, not a depth to grind to.

Why one number cannot describe the whole head

Position on the head	Removal measured	What it tells you
Inner side, R = 800 m curve, 6 passes	Up to 1.8 mm at the deepest point	Gauge-corner work moves far more metal than the running surface ever does
Outer side, same curve, same 6 passes	Up to 0.3 mm	The two sides of one curve are not the same job
Inner rail zone (+20° to +5°), train grinding	0.142-0.241 mm	Quoted depths usually mean the high rail or the field side — always ask which
Outer rail zone (−20° to −2°), train grinding	0.165-0.326 mm	Same job, same pass, roughly double the removal of the inner zone
Rail top zone (+5° to −2°), train grinding	0.193-0.222 mm	This is the figure that should be compared with the 0.5 mm acceptance cap

So when a specification says “grind 0.2 mm”, it is describing one location on a head whose cross-section is being changed by different amounts at different angles. The curve figures come from a 6-pass job at 12 km/h on 60 kg/m U71Mn and U75V rail; the zone figures come from instrumented train grinding. A depth claim without a position on the head cannot be verified, and cannot be wrong either — which is worse.

Rail grinding machine working on the running surface — Removal is not uniform across the head, so the depth that matters is the one recorded at a named zone against a named template.

Three measurable signs you have already gone too far

Sign	What the record shows	How to confirm it
The hardened surface is gone	Hardness falls instead of holding. A worked surface reaches roughly 870 HV 0.3 against 340 HV 0.3 for parent metal; once that layer is cut away the wheel is removing sound, softer steel and the profile stops improving while the head keeps shrinking	Portable hardness at the surface and at depth on a coupon, or a metallographic section from a sample
The profile is outside the envelope	GQI or the equivalent index drops below the pass mark even though the visible defects have gone, and the contact band grows wider than it should — it is intended to sit at 25-30 mm on high-speed turnout rail and 35-40 mm on conventional and heavy-haul rail	Profile measurement against the target template, zone by zone, before and after the job
Fatigue comes back faster	Corrugation, squat or spalling returns within one interval, or the defect pattern shifts to the field side — a symptom of ground geometry pushing contact away from where the wheel should be running	Defect mapping over two consecutive cycles at the same sections, not just the first inspection after grinding

None of the three needs a laboratory to suspect, and all three need a fixed cross-section to prove. That is why the sections you measure have to be marked and reused, otherwise the “before” and “after” are two different parts of the track. The general measurement route is set out in grinding acceptance standards.

Railway grinding machine on a track possession — A depth limit is only protective when the same cross-section is measured before and after, so the number can be defended later.

How to set the depth for your own line

Start from the defect, not from the machine. Corrugation peaks, lipping and a shallow fatigue layer each need a different depth. Removing the maximum the machine can take is not a strategy, and the entry point for that reasoning is what grinding actually removes from the head.
Take the last two cycles of profile records. If the head has already moved 0.3 mm from target in one zone, the next job there is a recovery job, not a repeat of the last depth.
Fix the measurement sections. Three to five per kilometre, marked, measured before and after, reused every cycle. Unmarked spot checks cannot separate a real change from natural variation.
Decide the fallback before the possession. If the first pass does not reach the planned removal, the answer is a parameter change — speed, pressure or angle coverage — not simply more passes at the same depth.
Hand the records forward. Depth per pass plus cumulative depth is what tells you whether this rail has years left or is now a renewal candidate; that judgement is covered in grinding on heavy-haul and mining railways.

Everything above concerns geometry, not consumables. Which Molaton wheel a machine head takes, and how the abrasive behaves at a given depth, is a separate question answered on the rail grinding solution page.

Why trust this answer — RailwayCare (product brand Molaton) has manufactured rail grinding wheels since 2004 — the first dedicated producer in China, born from the friction-materials laboratory of Wuhan University of Technology. We drafted the industry standard JB/T 11431 for rail grinding wheels, are certified to ISO 9001 / ISO 45001 / CRCC, and every claim below is backed by on-track tests on high-speed, heavy-haul and metro networks.

Set the depth from your own profile records

Send us the last two cycles of profile measurements, the defect pattern and the possession length you work with. We will tell you what depth the rail can still afford, and whether Molaton wheels and the grinding plan behind them fit that envelope before you commit a possession.

Ask About a Grinding Depth Plan