The Heavy-Duty Restorer: Rail Milling Explained

The stretch of East Coast Main Line between Peterborough and Grantham was, by 2003, consuming rail at a rate that alarmed even experienced track engineers. Gauge corner cracking — the signature RCF defect of heavily loaded, moderately curved track — had developed on the high rail of almost every curve in the 60 km section. Ultrasonic inspection was revealing crack depths of 3–5 mm on multiple sites, well beyond the 0.5 mm threshold at which conventional grinding becomes effective. The infrastructure manager faced a choice that, in that era, was binary: either grind repeatedly at shallow depth and re-inspect every six weeks, or renew the rail at a cost of approximately £800,000 per track-kilometre. Then a third option was trialled: a Linsinger MG31 rail milling train, brought over from Austria at the request of Network Rail’s research team. The machine worked at walking pace — 2.2 km/h — but in a single overnight possession it removed 2.5 mm from the gauge corner of the high rail through a 3 km curve at Stoke Bank, eliminating crack depths that had been accumulating for four years. Post-milling ultrasonic inspection showed no residual RCF defects. The rail surface hardness profile was measured at 320–340 HBW — slightly above the virgin rail value of 300 HBW for grade R260, consistent with a cold-worked cutting process that hardens the surface without creating a brittle white etching layer. The Peterborough–Grantham trial did not immediately transform Network Rail’s maintenance strategy, but it established, for the first time in the UK, a validated evidence base that milling could reproduce RCF life extension comparable to rail renewal at approximately one-third of the cost. That evidence base is now embedded in Network Rail’s Rail Milling Strategy, which authorises milling as the primary corrective treatment for RCF defects exceeding 1.5 mm depth on passenger routes carrying more than 20 million gross tonnes per annum.

What Is Rail Milling?

Rail milling is a track maintenance process in which the surface of the rail head is machined — using rotating multi-tooth cutting tools with hardened carbide inserts — to remove a controlled depth of steel and restore the rail’s design cross-sectional profile. It is classified as a corrective maintenance technique: where rail grinding is applied preventively to control surface defects before they propagate, milling is applied when defects have already penetrated beyond the reach of grinding, or when the volume of metal removal required would demand so many grinding passes as to be operationally or economically impractical.

The governing European standards for rail milling are EN 13231-3 (Railway applications — Track — Acceptance of works — Part 3: Acceptance of reprofiling rails in track by milling) for surface quality acceptance, and EN 13674-1 (Railway applications — Track — Rail — Part 1: Vignole railway rails 46 kg/m and above) for the design profiles that milling must restore. In the UK, Network Rail’s NR/SP/TRK/3090 standard governs rail milling operations and acceptance criteria on its managed infrastructure.

The Cutting Mechanism: How Milling Differs from Grinding

The fundamental distinction between milling and grinding is the mechanism of material removal. Grinding uses abrasion: a bonded abrasive wheel rotates at high speed and removes material by the action of hard abrasive particles (aluminium oxide or silicon carbide) scratching across the steel surface. Each individual abrasive particle removes a tiny chip of material, but the aggregate action of billions of particles produces a rapid temperature rise at the contact zone — typically 600–900 °C at the wheel–rail interface during aggressive grinding. This thermal input can cause metallurgical changes in the rail steel: the white etching layer (WEL), a thin (5–50 μm) re-hardened martensitic layer that forms when rail steel is rapidly heated above its austenitising temperature (~720 °C) and then quenched by the surrounding cold metal. WEL is brittle, has hardness values of 800–1,100 HV (compared to 260–320 HV for normal rail steel), and is itself a crack initiation site if not subsequently removed by further grinding passes.

Milling uses chip-forming cutting: each carbide tooth of the rotating cutter engages the rail steel at a controlled depth of cut (typically 0.1–0.5 mm per tooth per revolution), shearing a discrete chip that carries the heat of cutting away from the rail surface within the chip itself. The rail surface temperature remains below 150 °C. No WEL forms. The cut surface has defined chip marks (the “milling texture”) from the discrete tooth engagements — which is why all modern milling trains include a finishing stage (polishing stones or light grinding passes) that smooths this texture to the roughness values required by EN 13231-3 before the machine leaves the treated section.

Chip Formation and Metal Removal Rate

In practice, the total swarf generated per kilometre of double track during a corrective milling pass (removing 1.5–2.5 mm) is significantly higher than this simplified calculation — typically 1.5–4.0 tonnes/km double track when accounting for the full cutter head engagement geometry across the rail head profile. This material is captured in real time by an integral vacuum extraction system mounted directly behind the cutter heads on all modern milling machines.

Rail Milling Machines: The Leading Systems

The global market for purpose-built rail milling trains is dominated by two manufacturers: Linsinger Maschinenbau GmbH of Austria and Stahlberg Roensch GmbH of Germany. A third significant player, Schweerbau International, has developed a hybrid grinding-milling concept (the RGM — Rail Grinding Machine with Milling capability) that offers both process types in a single consist.

The Cutter Head: Tungsten Carbide Inserts

Each cutter head on a rail milling machine is a circular disc — typically 300–600 mm in diameter — carrying 12–24 tungsten carbide (WC-Co) cutting inserts arranged in a helical pattern around its circumference. The inserts are indexable: when one cutting edge wears, the insert is rotated to present a fresh edge, then replaced entirely when all edges are consumed. A standard insert geometry for rail milling uses a cutting edge angle of 0–5° positive rake (aggressive cut for soft-to-medium hardness rail, grade R260 to R320) or up to 10° negative rake (conservative cut for premium-grade rail R350HT and R370CrHT, which have surface hardness of 350–380 HBW). Insert material is typically ISO K20–K30 cemented carbide — 78–82% WC, 8–12% Co binder, with TiN or TiAlN PVD coating — with a hardness of 1,400–1,600 HV and fracture toughness sufficient for the interrupted-cut conditions of rail milling. Insert life is typically 8–15 km of rail treated per insert edge, depending on rail grade and depth of cut.

Target Defects: Rolling Contact Fatigue and the Milling Decision Matrix

Rail milling’s primary application is the removal of Rolling Contact Fatigue (RCF) defects — the family of surface and near-surface cracks that develop in rail steel under repeated wheel loading. Understanding which RCF defects are treatable by milling (versus those requiring rail renewal) is central to optimising maintenance strategy.

The White Etching Layer and Milling

One of the most technically significant advantages of milling over grinding in heavily RCF-affected rail is its treatment of the white etching layer (WEL). WEL forms in rail steel when the surface is rapidly heated above ~720 °C and quench-cooled — a process that occurs naturally during wheel slip events (wheelset spin under traction or lock-up under braking), and paradoxically can also be induced by aggressive grinding operations. WEL has a martensitic microstructure with hardness of 800–1,100 HV, is brittle, does not deform plastically under load, and cracks at the WEL–bulk steel interface under repeated loading. The cracking initiates RCF defects or propagates existing ones.

Milling, operating below 150 °C, does not generate WEL. More critically, because milling removes 1.5–3.0 mm per pass, it removes pre-existing WEL — whether formed by wheel slip or by previous grinding operations — along with the RCF cracks that initiated at the WEL–bulk interface. Post-milling EBSD (electron backscatter diffraction) analysis of milled rail surfaces consistently shows a fine-grained, work-hardened ferrite–pearlite microstructure with no martensitic phase — the expected state of rail steel that has been cold-cut rather than thermally processed. This microstructure is more fatigue-resistant than the coarse pearlite of the original hot-rolled rail, because the cold-working from the cutting process introduces compressive residual stresses in the surface zone to a depth of approximately 0.3–0.5 mm.

Operational Parameters: Planning a Milling Campaign

Rail milling is an expensive intervention — typically £40–80 per track-metre in the UK and €35–75/m in continental Europe, compared to £8–20/m for conventional grinding. A successful milling strategy requires careful pre-treatment measurement, machine parameter selection, and post-treatment verification to justify the cost against the alternative of rail renewal or repeated grinding.

Pre-Milling Survey Requirements

Before a milling campaign, every section scheduled for treatment must be assessed by:

• Rail profile measurement: Laser profilometer or contact gauge measurement of the as-worn cross-section at intervals of 5–25 m. This defines the depth of metal removal required at each location along the section and allows the milling machine operator to programme cutter head depth continuously along the run.
• Ultrasonic inspection: To confirm no transverse fatigue cracks are present (which would mandate renewal rather than milling) and to establish the depth of existing RCF cracks, confirming they are within the treatable range (<4 mm for squats, <5 mm for GCC).
• Rail head hardness measurement: To select appropriate carbide insert geometry and cutting speed. Premium-grade rail (R350HT, R370CrHT) requires different insert geometry than standard R260 to avoid insert chipping on the harder surface.
• Remaining rail head height: EN 13674-1 specifies a minimum residual rail head height for continued service. A worn rail that is already at or near this minimum cannot accept the additional metal removal that milling requires without triggering premature renewal. The pre-milling survey must confirm adequate remaining head height (typically ≥ 5 mm above the minimum permitted) to allow the planned removal depth.

Possession Length and Output Rate

At a working speed of 2.0–3.0 km/h and allowing for set-up, travel to the worksite, and post-treatment verification, a milling machine typically treats 3–6 km of single track per 8-hour overnight possession on a deep corrective pass. A lighter preventive pass (0.5–1.0 mm removal) at higher speed (4–5 km/h) can achieve 6–10 km per possession. Network Rail’s operational experience on the ECML confirms that a Linsinger MG31 operating on a corrective campaign for GCC-affected high rail on 600 m radius curves averages 4.2 km per possession at 2.5 km/h working speed — producing approximately 2.8 tonnes of swarf per possession recovered and returned to the steel supply chain.

Rail Milling vs. Rail Grinding: Full Technical Comparison

Rail Milling in Service: Global Deployments and Results

Frequently Asked Questions

1. Why can’t conventional rail grinding simply be applied more aggressively to remove RCF cracks as deep as milling can reach?

The theoretical answer is that it can — a grinding train making repeated passes can remove 3 mm of rail steel if given enough time. The practical answer is that aggressive grinding to achieve deep metal removal creates its own set of defects that undermine the purpose of the operation. The core problem is thermal: grinding stones working at high material removal rates generate surface temperatures of 700–900 °C at the stone–rail interface, well above the austenitising temperature of rail steel (~720 °C for pearlitic grades). At this temperature, a thin layer of rail steel transforms to austenite, and when the wheel passes and the grinding zone is rapidly cooled by the surrounding rail mass and ambient air, the austenite transforms to martensite — the white etching layer. WEL has hardness of 800–1,100 HV and is brittle; it is itself a crack initiation site, not a crack cure. Attempting to remove 2–3 mm of RCF-cracked rail by aggressive grinding therefore risks replacing one defect population with another. The correct protocol for grinding at any depth is to work at low material removal rates per pass and use multiple light passes — but this multiplies the number of possession hours required by a factor of 5–10 and consumes grinding stones rapidly. On a 10 km section of GCC-affected high rail needing 2.5 mm removal, a grinding strategy would require 8–12 overnight possessions to achieve the same result as a single milling pass. In a network operating 500+ trains per day, those possession opportunities are simply not available at reasonable notice. Milling achieves the depth in one pass, at the cost of slower working speed, without the WEL risk.

2. How does a milling machine maintain profile accuracy when the cutter head is flexibly mounted on a vehicle moving at 2–3 km/h over track that may have vertical irregularities?

Profile accuracy is the central engineering challenge of any rail milling machine, and the solution involves a combination of mechanical isolation, active referencing, and CNC control. The cutter head assembly on a machine such as the Linsinger MG31 is mounted on a longitudinal reference beam — a stiff steel girder supported on the rail at points 3–5 m ahead of and behind the cutting zone. This beam averages out short-wavelength track geometry irregularities (wavelengths below ~3 m) and provides a stable datum from which the cutter head height is controlled. The lateral position of each cutter head is set relative to the rail gauge face using a gauge face follower roller that rides against the rail’s inner face — this ensures that the profile cut by the cutter heads is registered to the actual rail position, not to the machine’s centreline, which may wander due to track curvature and cant variation. The depth of cut at each cutter head is controlled by a CNC servo system that reads the as-measured rail profile data from the pre-milling survey (uploaded to the machine’s control computer before the campaign) and adjusts cutter depth continuously to achieve the target final profile at each point along the rail. The result is a reprofile accurate to ±0.2 mm in cross-section and ±0.5 mm in longitudinal level — tighter than the EN 13231-3 acceptance tolerances of ±0.3 mm and ±1.0 mm respectively. On tight curves (R < 400 m), additional cant compensation is applied to the cutter head assembly to account for the cant-induced lateral shift of the wheel–rail contact band.

3. What happens to the swarf — is it genuinely recyclable or does carbide insert contamination make it difficult to process?

Rail milling swarf is, from a metallurgical standpoint, an excellent feedstock for electric arc furnace (EAF) steelmaking. It consists of nearly pure carbon steel (UIC Grade R260 or similar: 0.62–0.80% C, 0.80–1.30% Mn, 0.13–0.60% Si, negligible alloy additions) with no coatings, no surface treatments, and no hazardous materials. The chips are 2–8 mm in length with a bulk density of approximately 1.5–2.5 t/m³ — dense enough to handle without briquetting, which is required for grinding dust before it can be charged to a furnace. The small concern about carbide insert contamination (tungsten carbide chips that may fall from a worn insert before it is detected and changed) is managed by the fact that the machine’s chip extraction system passes swarf over a magnetic separator that removes ferrous chips from non-ferrous carbide fragments. The separated swarf is certified as clean steel scrap to EN 10131 tolerances by the milling contractor before delivery to the EAF operator. In Switzerland, where SBB has operated a closed-loop swarf recycling agreement with Swiss Steel AG (formerly von Roll) since 2009, the EAF operator pays approximately €80–120 per tonne of sorted swarf — meaning the milling contractor receives a material credit that partially offsets the machine operating cost. Network Rail has investigated a similar arrangement with UK steelmakers but has not yet formalised a closed-loop agreement, and current swarf is sold to general metal recycling brokers at spot market rates of €40–80/tonne.

4. Can rail milling be used on turnouts (switches and crossings), and how does the machine handle the variable geometry?

Turnout milling is technically possible but operationally complex. The variable geometry of switches and crossings — rails converging and diverging, check rails at different heights, crossing noses with acute profile changes, switch blades with tapered sections — requires the milling cutter head to adjust its depth, angle, and lateral position continuously and rapidly. Standard line-milling machines are not designed for this: their reference beam requires a consistent rail cross-section and gauge width to maintain accurate profile registration. The solution, where turnout milling has been attempted (notably on the German high-speed network and on the Swiss Bahn 2000 upgrade programme), is a purpose-designed turnout milling attachment — a smaller, more manoeuvrable cutter head assembly mounted on a road-rail vehicle or a compact rail-mounted platform, manually guided and CNC-controlled for depth but lacking the automated profile-following system of a full milling train. Working speeds in turnout milling are typically 0.3–0.8 km/h — effectively a crawl — and the total treatment time for a single medium-speed turnout (e.g., a 1:12 design speed 130 km/h turnout) is 4–8 hours. This makes turnout milling very expensive per unit treated, but the alternative — renewal of a heavily RCF-affected turnout — costs €80,000–200,000 per unit including installation, compared to €15,000–25,000 for a corrective milling treatment. Where crossing nose RCF in R350HT manganese steel has reduced the crossing to below its minimum section, milling can restore the profile without removing the manganese insert, deferring renewal by a further 2–4 years on medium-traffic routes.

5. How does the milling process interact with track circuits and axle counters — does it affect the electrical continuity of the rail and the signalling system during or after treatment?

Rail milling does not break the rail’s electrical continuity — unlike rail cutting or joint renewal operations — because the cutter heads remove surface material without interrupting the rail section. Track circuits, which rely on the electrical conductivity of the rail to detect train presence, are therefore unaffected during milling operations in the sense that the rail remains a continuous conductor throughout. However, there are two more subtle interactions. First, the milling process cleans the rail surface to bare metal, removing the oxide layer and any contamination (oil, moisture, ballast dust) that was present on the as-worn surface. A freshly milled rail is a better electrical conductor than an oxidised or contaminated one, which means that track circuit tuning — particularly for audio-frequency track circuits that are calibrated for a specific rail conductance — may need to be re-verified after milling on circuits whose tuning was adjusted to compensate for degraded rail surface conditions. Operators such as DB Netz include a standard track circuit impedance check in the post-milling acceptance procedure for this reason. Second, the chip extraction system on the milling machine uses strong vacuum flow, and if a vacuum hose seal fails adjacent to a track circuit rail bond, metallic chips could potentially bridge the bond connection and the adjacent rail surface. This is managed by a standard post-milling visual inspection of all rail bond and impedance bond connections within the treated section, which forms part of the EN 13231-3 acceptance procedure. No case of milling-induced track circuit failure due to swarf accumulation on a bond has been formally documented in published Railway Accident Investigation Branch (RAIB) or BEU reports, though several precautionary bond re-checks have been noted in infrastructure manager post-maintenance reports.