The Non-Linear Math Behind Every Cold Roll Reject
Five percent chromium forged cold work rolls look like a mature, solved problem. The defect-rate equation that governs them is not — and that equation quietly decides who is buying 5% scrap and who is buying 0.5% scrap.
A new study on forged 5%Cr steel cold work rolls — the grade behind most modern Sendzimir and tandem-mill work rolls in the 64–68 HRC class — ties the surface-wave defect rate directly to two process variables in a single equation:
DRS = 0.15 · [O]1.6 + 0.08 · Dmax2.3
The 1.6 and 2.3 exponents are not rounding artefacts. They are the difference between an improvement that looks impressive on a mill certificate and an improvement that actually drops scrap on a cold strip mill. Halving [O] cuts the DRS contribution from oxygen by 33%. Halving Dmax cuts the inclusion contribution by 80%. That is a 4× difference in marginal return — and it inverts the optimization order that most procurement specs assume.
A quick note on grade designation, because 5%Cr forged cold work roll steel is one of the most poorly-mapped grades in the international table. The 5%Cr forged cold work roll family used in modern mills (typically C 0.85–0.95, Cr 4.8–5.2, Mo 0.3–0.5, V 0.1–0.3) sits below AISI A8 / EN 1.2360 (X48CrMoV8-1-1, ~7.5% Cr) in chromium content and above the 2%Cr work-roll grade in alloy depth. The closest international references are W.Nr 1.2367 (X38CrMoV5.3, AISI H10 modified) and the EN 1.2360 / AISI A8 family used as a higher-Cr reference. The 5% Cr composition itself is largely a Chinese / Japanese mill-roll tradition (9Cr2Mo, MC3, MC5, and similar proprietary grades); when you ask for a cross-reference, the honest answer is “there is no single drop-in AISI/EN grade — here is the closest band.”
What the DRS Equation Actually Says
Plug in three realistic melt-shop scenarios for a 5%Cr roll, all of them already passing ASTM E45:
| Scenario | [O] (10−6) | Dmax (mm) | 0.15·[O]1.6 | 0.08·Dmax2.3 | DRS (%) |
|---|---|---|---|---|---|
| Conventional EAF + LF | 25 | 0.20 | 2.34 | 1.49 | 3.83 |
| + VD vacuum degas | 15 | 0.20 | 1.16 | 1.49 | 2.65 |
| + ESR + Ca treatment | 10 | 0.08 | 0.60 | 0.13 | 0.73 |
Computed from Equation (2) of the source paper; Dmax in the third row is reduced by 90% through Ca-treatment converting SiO2-MnO-Al2O3 aggregates >200 µm to dispersed 12CaO·7Al2O3 (m.p. 1450 °C).
The table makes the math tangible. Vacuum degassing alone — the most common “clean-steel” investment — knocks only 31% off DRS. Adding ESR with Ca-treatment knocks another 72%, because it crushes the Dmax2.3 term. Procurement specs that read “O ≤ 15 ppm” but ignore maximum inclusion size are paying for the cheaper 31% and missing the 4× one.
The Inclusion-Type Table That Decides Your Inspection Floor
The four inclusion families that drive cold roll surface-wave defects are not equally harmful. EDS work on real roll-body surface defects in the source study (Figure 1, scale bars 10–50 µm) finds complex multi-phase populations rather than single-phase particles, but the families still classify cleanly:
| Family | Chemistry | Morphology | Source | Failure mechanism |
|---|---|---|---|---|
| Silicate | SiO2-CaO-Na2O-K2O | Plastic flow → long spindle | Electrode ingot slag entrapment | Interfacial micro-void coalescence |
| Al2O3 | Al2O3-MgO | Brittle fracture → chain fragments | Deoxidation product | Radial micro-cracks |
| Nitrogen pore | N2 gas | Spherical, Φ10–50 µm | VD/ESR N-control failure | Stress concentrator |
| SiO2-MnO-Al2O3 | SiO2-MnO-Al2O3 | Large aggregates >200 µm | Incomplete refining removal | Crack initiation kernel |
Table 1 of the source paper; morphology and source verified by Figure 1 SEM-EDS work (20/10/50/20 µm scale bars).
The bottom row is the one that buys the equation its 2.3 exponent. SiO2-MnO-Al2O3 clusters are the Dmax term in physical form. They survive VD and standard ladle refining; they need Ca-treatment to convert to low-melting-point 12CaO·7Al2O3 that floats out, and they need ESR slag washing to fully disperse. Removing them is the single highest-leverage process change in the whole roll-making chain.
Why “Halving [O]” Is the Wrong Optimization Target
Cold roll producers tend to optimize the oxygen side of the equation because it is the easier number to publish on a certificate. The source paper’s full-process data is a useful reality check. The reported DRS contributions from the inclusion-distribution analysis (Table 2) at three defect-zone positions are:
| Defect zone | Density (pcs/mm2) | Dominant inclusion | Typical shape | Crack signature |
|---|---|---|---|---|
| Wave trough | 15 | Al2O3 fragments | Chain (length > 50 µm) | Radial micro-cracks |
| Wave crest | 8 | Plastic silicate | Spindle, aspect > 3.5 | Interfacial micro-void coalescence |
| Transition | 3 | H/N gas pore | Spherical Φ20–30 µm | Non-oriented micro-cracks |
Table 2 of the source paper; numbers reproduced from the published image.
Wave troughs concentrate 62% of the inclusion count, and the dominant particle is an Al2O3 fragment chain — the brittle, un-deformable, broken-up daughter of a single larger alumina that survived refining. This is the Dmax2.3 story in micrograph form. VD-only producers do not address it; only Ca-treatment that prevents the parent aggregate from forming, plus ESR that breaks up survivors, makes the troughs go away.
The Forging Red Line at 35% Reduction
Even with a clean melt, the forging step can manufacture defects out of nothing. The source paper’s forging-parameter matrix (Table 3) is one of the cleanest illustrations:
| Single-pass reduction | Carbide morphology | Ledeburite rating | Impact toughness | Structure uniformity |
|---|---|---|---|---|
| 18% | Fine, dispersed | < 1.0 | High | Uniform |
| 35% | Coarse, aggregated | > 1.5 | Significantly lower | Non-uniform |
Table 3 of the source paper.
A 35% single-pass reduction is the number that looks impressive on a press log but that the lab will not forgive: ledeburite carbide rating jumps from <1.0 to >1.5, and impact toughness falls off a cliff. The good practice from the same study is a multi-pass sequence at 0.3–0.5 s−1 strain rate (below the adiabatic shear band regime), 40% upset, a finish-forge from 1100 °C to 850 °C avoiding the blue-brittle range, and a 1150 °C × 4 h high-temperature hold that closes over 60% of the micro-cracks and reduces primary carbide size by 40%. The valid stress criterion to achieve both interface welding (for ≤ 50 µm inclusions) and recrystallization densification is σeff = (F/A0)·eε > 2σy (Equation 1), with the dynamic recrystallization critical strain ε = 0.25.
ASTM E45: The Floor, Not the Target
The Chinese GB/T 1299 thresholds cited in the source paper — sulfides and oxides at ≤ 1.5, banded carbides at ≤ 2.5, ledeburite at ≤ 2.0, network carbides at ≤ 2.5 — are necessary but not sufficient. The closest equivalent international rating is ASTM E45, where the usual cold roll acceptance is type B (alumina chains) at ≤ 1.5 and type D (globular oxides) at ≤ 2.0 by the JK chart. A roll that meets E45 with [O] = 15 ppm and Dmax = 0.20 mm still scores DRS ≈ 2.6%. The full-process optimized target from the same study:
| Process step | Key measure | Process parameters | Effect achieved |
|---|---|---|---|
| Steelmaking | Ca treatment + Al control | [Al] ≥ 0.023%, [O] ≤ 15 ppm | SiO2-MnO-Al2O3 clusters −90% |
| ESR | Atmosphere protection + water cooling | O2 ≤ 10 ppm, water 15–35 °C | Reject rate −5%, carbide segregation eliminated |
Table 4 of the source paper; full process chain ESR + Ca + controlled forging → surface-wave defect rate < 5% (versus the industry-baseline ~30% reject rate cited in the same paper).
That is the real benchmark. The source paper reports a drop from an industry baseline of ~30% surface-wave reject rate down to < 5% — an 80%+ reduction. Note that this is not the same as cutting DRS by 80% (the equation is calibrated against a different dataset), but the order of magnitude is the same: Dmax is the lever, [O] alone is not.
What a Cold Roll Buyer Should Ask the Mill
A purchasing spec that asks only for “[O] ≤ 15 ppm” will reward the wrong process. Five questions that re-aim the spec at the 2.3 exponent:
- What is your maximum inclusion size Dmax on the last 10 heats — not the average, the maximum? (Targets: < 50 µm for the Ca + ESR route, < 200 µm for VD-only.)
- Do you run Ca-treatment, and do you have a [Al] ≥ 0.023% verification on the heat analysis? (Without Ca, the SiO2-MnO-Al2O3 row of Table 1 stays alive.)
- Is the ESR atmosphere O2 controlled to ≤ 10 ppm and the cooling water to 15–35 °C? (Atmosphere leak undoes the slag washing.)
- What is your single-pass upset ratio and your 1150 °C × 4 h soak policy? (35% single-pass will wreck Table 3’s impact row even on a clean melt.)
- Show me the ASTM E45 JK chart ratings for the last 10 heats, not just the MTC. (B ≤ 1.5 and D ≤ 2.0 is the floor; ask for the trend.)
By Harris, FUSHUN METAL technical team. Source paper: Meng Yanjun, “Inclusion deformation and precision process control in forged cold roll steel manufacturing”, Shandong Metallurgy, 2026.2. Tables 1–4, Figure 1 and Equations (1)·(2) reproduced from the original publication. The 5%Cr forged cold work roll grade sits between EN 1.2360 (AISI A8) and EN 1.2367 (AISI H10 modified); the closest international references for cross-spec conversations are W.Nr 1.2360 X48CrMoV8-1-1 (8% Cr, higher alloy) and W.Nr 1.2367 X38CrMoV5.3 (5% Cr, lower alloy depth), with no exact drop-in equivalent in the AISI/EN system.
