By Harris | Technical Team, FUSHUN METAL
Semi-Steel Roll Heat Treatment: Why a 1000 mm Roll Neck Bends at 940 °C — and the Four Fixes That Stopped It
A semi-steel roll — known as adamite in Western mill practice — is a casting that deliberately sits between cast steel and cast iron. With carbon in the 1.4%–2.3% range it keeps the small hardness drop and high wear resistance that mills want in roughing and intermediate stands, while retaining enough strength and toughness to survive heavy section duty. These rolls serve section, rail-and-beam, bar and billet mills, hot-strip roughing stands and the front positions of finishing trains, plus vertical rolls, universal-mill roll rings and sleeves. The manufacturing route is short: cast, rough machine, heat treat, finish machine. The heat treatment is high-temperature diffusion followed by normalize and temper. And this is where a surprising share of large rolls quietly lose money.
A note on designations before we begin: the grade in the case below, ZUB180CrNiMo, is a Chinese casting-roll designation under GB/T 1503. Internationally, this family is called alloy adamite or semi-steel rolls — “adamite” being the term used in US and European mill practice — and comparable export grades such as AD140 to AD190 carry the same C 1.3%–2.3% with Cr–Ni–Mo chemistry. Hardness is the universal acceptance unit across all systems, typically 45–65 HSD after normalizing and tempering.
Our view, after studying these failures: at 940 °C a large roll is not a metal part, it is a piece of hot plastic. Every millimeter of uneven support inside the furnace becomes a permanent bend after cooling. Deformation control is therefore a gravity problem before it is a metallurgy problem.
The case: one roll, seventeen failures
A roll maker supplied 18 semi-steel work rolls, 1000 × 1100 mm in body size, grade ZUB180CrNiMo (an AD180-type alloy adamite), for the intermediate stand of a section mill. The drive-side neck carries a length-to-diameter ratio of about 5.2 — a long, slender neck, exactly the geometry that a hot furnace punishes. The result after heat treatment was a recurring quality accident: the drive-side neck bent, the center hole shifted off true, machining allowance ran short, and the only recovery was a corrective heat treatment. Of the 18 rolls delivered, 17 showed some deformation; six were severe enough to fail — a 33% rejection rate — and the remaining batch consumed extra furnace time, rework and schedule.
Root cause one: the metal itself expands unevenly
Thermal expansion follows a simple rule: δ = α · d · (T − T₀), where α is the linear expansion coefficient, d the local diameter, T the furnace temperature and T₀ room temperature. The roll body is roughly twice the diameter of the neck, so at the 940 °C diffusion soak the body grows about 11.73 mm while the neck grows only about 4.5 mm — a mismatch of about 7 mm. That temperature is far above the alloy’s plastic-deformation temperature. With the roll resting on supports, gravity simply settles the neck downward by that mismatch, and the bend becomes permanent on cooling.
Root cause two: the bogie hearth itself bends (the main driver)
The rolls were heat treated in a resistance-heated bogie hearth furnace built from, bottom to top, wheel beams, I-beam frame girders, a floor plate, refractory concrete and refractory brick. Furnace bogies of this type have a chronic weakness: the outer and inner layers of the girder frame run at very different temperatures, the top of the girders expands more than the bottom, and the middle of the hearth bows upward. Measured deformation on this furnace showed the hearth crown lifting enough that the roll neck ends, sitting near the hearth edges, dropped 36–40 mm relative to the body resting near the hearth center. This was the dominant contributor — roughly five times the effect of the metal’s own expansion.
Root causes three and four: how the roll is supported and lifted
Two further contributors were procedural. First, support placement: where the hearth carried resistance strips, support points were set on the strip cover plates, which themselves deform at temperature; support spacing was sometimes too wide for a slender neck; and supports were not always verified as firm — loose insulating bricks and gaps under the neck let it sag. Second, lifting: during transfer from the hearth to the air-cooling/quench station, the roll was lifted at the neck end while still above its plastic temperature, so the weight of the body bent the neck further. Neither fault is metallurgical. Both are controllable by discipline.
| Driver | Measured effect | Nature |
|---|---|---|
| Differential thermal expansion of body vs neck at 940 °C | Body 11.73 mm, neck 4.5 mm; 7 mm mismatch | Physics of the material |
| Bogie hearth bowing under thermal gradient | Neck ends drop 36–40 mm vs body at hearth center | Main cause; equipment |
| Support placement errors | Local sag where supports shift or loosen | Procedural; controllable |
| Lifting at the neck end at plastic temperature | Gravity adds to existing bend during transfer | Procedural; controllable |
The four fixes, and what they achieved
Fix one targeted the hearth, the main offender. An asbestos felt insulation layer was laid between the bogie floor plate and the refractory concrete to cut the temperature gradient through the metal structure, and expansion joints were gas-cut into the longitudinal I-beam girders every 3 m — 100 mm deep by 3 mm wide — to relieve the difference between inner and outer expansion. Hearth relative deflection fell from the 36–40 mm range to 5–10 mm.
Fix two handled the physics: pre-calculated counter-deformation. Before charging, 5–7 mm thick wood planks were placed under the roll body. During heating the planks burn away, leaving a controlled gap that absorbs the body-versus-neck expansion difference, so the neck no longer sags into the mismatch.
Fix three was procedural: such rolls are no longer scheduled on hearths fitted with resistance strips (support covers that deform at heat); a furnace-charging inspection record was introduced; and every support point gets a tap test before the charge is accepted. Fix four changed the lifting plan: hot rolls are now hooked at the body–neck junction, never at the neck end, so gravity no longer bends the slender neck during transfer.
The outcome on the next 20 rolls was decisive: neck bending stayed within 8 mm across the batch, comfortably inside the machining allowance, with no corrective heat treatment and no rejected roll.
| Measure | Action | Result |
|---|---|---|
| Hearth stabilization | Asbestos felt between floor plate and refractory concrete; expansion joints in girders every 3 m | Hearth deflection reduced from 36–40 mm to 5–10 mm |
| Counter-deformation | 5–7 mm wood planks under the body; planks burn off during heating | Gap compensates the 7 mm body–neck expansion difference |
| Support discipline | No resistance-strip hearths for such rolls; charging inspection record; tap test on every support point | No loose support, no local sag |
| Lifting plan | Hook at the body–neck junction, never the neck end | No gravity bending during transfer |
| Batch outcome | Next 20 rolls produced under the revised procedure | Neck bending within 8 mm; zero rejects; no corrective heat treatment |
Why this case is a lesson for every heavy-furnace shop
The four fixes here are not exotic. They match the rules that heavy-forging shops apply to long, heavy workpieces everywhere: multi-point equidistant supports so no section hangs unsupported; staged heating with controlled ramp rates to hold temperature gradients in check; furnace cooling or controlled-rate cooling instead of free air cooling; multi-point, symmetrically loaded lifting; and deliberate pre-deformation compensation. The roll maker in this case simply applied the same logic to its own bottleneck — the slender neck on a hot hearth — and removed a 33% rejection rate from one product line. That is the point worth remembering: heat treatment distortion of large rolls is rarely a metallurgy puzzle that needs a new alloy. It is usually a mechanical problem — a hearth that bows, a support that shifts, a hook placed in the wrong spot — and mechanical problems have mechanical solutions that cost far less than the scrap they prevent.
