Why Billet Internal Quality Limits Rolled Wear Plate Yield
Wear-resistant plate is a through-hardened or low-alloy quenched and tempered product in which hardness, and therefore wear life, is carried by the whole section rather than by a surface layer. Grades such as NM400, NM450 and NM500 under GB/T 24186 are supplied in the 370-430 HBW, 420-480 HBW and 470-530 HBW hardness windows respectively, and that hardness is developed by austenitising, quenching and tempering a plate rolled from a continuously cast slab. Because the transformation behaviour of these steels is sensitive to local composition, any point in the slab where carbon, manganese, chromium or molybdenum has segregated will respond differently to the same quench. The result is banded hardness, soft spots, crack initiation sites at the plate centre line and, in the worst case, delayed cracking during flame cutting or bending.
For plate producers working from a 250 mm x 1600 mm continuous casting billet, the practical ceiling on rolled-plate yield is therefore set upstream of the rolling mill. Centre segregation, centre porosity, shrinkage cavities and internal cracks in the slab are not eliminated by rolling reduction; they are flattened, elongated and redistributed into the plate as centre-line stringers and laminations. A slab that carries severe class A or class B centre segregation will produce plate that fails ultrasonic inspection regardless of rolling schedule, cooling practice or level-2 automation. This is why improvements in billet internal quality translate almost linearly into improvements in rolled plate inspection pass rate.
The most cost-effective lever available on an existing casting machine is dynamic soft reduction, also described in plant practice as dynamic light pressing. It is a mechanical means of compensating solidification shrinkage and of suppressing the inter-dendritic flow that produces centre segregation, applied at exactly the right position and with exactly the right amount of reduction for the steel grade and casting speed in use. The engineering difficulty lies less in the withdrawal-and-reduction unit itself than in knowing where the solidification end point actually is, second by second, on a heat-by-heat basis.
The Metallurgy of Soft Reduction in the Mushy Zone
During solidification of a 250 mm thick billet, a columnar dendritic shell grows inward from the mould faces. Between the fully solid shell and the liquid pool there is a mushy zone in which a coherent dendrite network coexists with enriched interdendritic liquid. As the shell contracts, the residual liquid must feed the shrinkage of the last material to solidify. If feeding is blocked, or if the strand is bulged by ferrostatic pressure, the enriched liquid is squeezed toward the centre, producing positive segregation in the core. Solute elements with partition coefficients well below unity, principally carbon, sulphur, phosphorus and manganese, are redistributed preferentially.
Soft reduction works by applying a controlled thickness reduction to the strand while the core is still partially liquid. Three mechanisms act together:
1. Shrinkage compensation. The draft replaces the volume lost to solidification shrinkage, so the core is not starved of feed metal and cavity formation is suppressed.
2. Squeeze-out of enriched liquid. Reduction applied where the solid fraction is high but the core is still permeable pushes solute-enriched liquid back toward the pool, where it is diluted, instead of allowing it to concentrate at the centre line.
3. Closure of internal porosity. Pores and micro-shrinkage that have already formed can be partly welded shut if the reduction is applied while the surrounding metal is still hot and plastic.
The window is narrow. Too early, and the reduction acts on a thick liquid core: the shell buckles, internal cracks open, and the draft is simply absorbed by liquid displacement. Too late, and the centre is fully solid: the draft does nothing except load the machine, and may generate internal cracks in a low-ductility temperature range. In practice the useful window lies where the solid fraction at the centre line is roughly 0.3 to 0.8, corresponding to a liquid core thickness of the order of 20-60 mm on a 250 mm billet. Reduction efficiency, defined as the ratio of centre-line densification achieved to nominal draft applied, falls steeply once the core falls below about 15 mm.
Locating the Solidification End Point with the Nail Shooting Test
Every soft reduction model is only as good as its estimate of shell thickness. Empirical heat-transfer models calibrated on water-flow data will drift with changes in mould flux, secondary cooling nozzle condition, superheat and casting speed. The most reliable plant method for anchoring the model is the nail shooting test, sometimes called the pin shooting or pin shooting test: a small steel nail or pin of known composition and known length is fired into the strand at a defined position, and the strand is subsequently sampled to see how far the pin penetrated and how it was bonded.
The procedure typically runs as follows. Pins of a grade with a distinctly different sulphur or boron content from the cast steel are shot at several positions along the machine radius at a steady casting speed. The resulting slab section is examined by sulphur print, by macro-etch, or by direct measurement of the remaining liquid core. Where the pin is fully melted and diluted, the core was still liquid; where the pin remains intact and unbonded, the section was already solid; the transition zone brackets the solidification end point. Repeating the shots at three or four casting speeds generates the relationship between speed and liquid pool length that the dynamic model needs.
The measured data are used to fit the shell growth relationship, conventionally expressed as a power law of the form s = K t^n, where s is shell thickness, t is time below the meniscus and K is a solidification coefficient specific to the steel grade, section and cooling intensity. For plain carbon and low-alloy wear plate grades, n is usually taken as 0.5 and K is calibrated from the shot data rather than assumed. The liquid core half-thickness at any machine position is then the difference between the half-thickness of the strand and the shell thickness at that point, and the solidification end point is the position at which the core closes.
Two practical points matter for validity. First, the shots must be taken at the casting speed and superheat that production actually uses, not at a nominal condition, because a 10 degrees C change in superheat or a 0.1 m/min change in speed moves the end point by hundreds of millimetres. Second, the pin material must be selected so that it does not itself alter the local solidification; a pin whose melting point and carbon content are close to the cast steel gives a cleaner signal than a strongly dissimilar one.
Building the Dynamic Control Model
Once the pool profile is anchored by shot data, the dynamic model converts it into a per-segment reduction schedule in real time. The control architecture normally comprises four layers.
Measurement layer. Casting speed from the withdrawal drive, secondary cooling water flows and temperatures, tundish temperature and mould level are sampled at short intervals. Mould heat flux and thermocouple trends can be used to detect shell thinning or sticking events that shift the end point.
Thermal model layer. A one-dimensional or slice-based solidification model computes the liquid core thickness at each roll segment for the current speed and cooling setpoints. The model is re-anchored against shot-test data whenever the grade or section changes.
Reduction decision layer. The segments whose core thickness falls inside the soft reduction window are identified, and a draft is distributed across them according to a target total reduction and a target reduction gradient. Draft per segment is limited by machine capability and by the permissible strain in the shell.
Execution and feedback layer. Hydraulic cylinders close the gap to the setpoint, cylinder position is compared with the model, and the gap is trimmed. Because the model output changes continuously with casting speed, the segment setpoints are updated dynamically rather than set once per grade.
Two parameters govern the outcome. The total soft reduction amount is the arithmetic sum of the drafts applied while the core remains in the window; for a 250 mm billet, industrial practice typically falls in the range of 3 to 8 mm total, with the upper part of that range used for higher-carbon and higher-alloy wear grades. The reduction gradient is the rate at which draft accumulates per unit of strand length or per unit of decreasing core thickness; a gradient that is too steep concentrates strain in one segment and causes internal cracking, while a gradient that is too shallow fails to suppress segregation. A common industrial approach is to place the largest single draft where the core is around 30-50 mm and to taper the following drafts.
Typical Operating Windows for 250 mm x 1600 mm Wear Plate Billets
The table below summarises the ranges that are generally used in industrial practice for low-alloy wear-resistant plate grades on a 250 mm x 1600 mm billet caster. The values are process windows, not guarantees; each caster must confirm them against its own shot-test data and metallographic results.
| Parameter | NM400 / Q355-class base | NM450 / Cr-Mo added | NM500 / higher C and alloy |
|---|---|---|---|
| Typical carbon range, % | 0.14-0.20 | 0.18-0.24 | 0.22-0.30 |
| Superheat above liquidus, degrees C | 15-25 | 12-22 | 10-20 |
| Casting speed, m/min | 1.0-1.3 | 0.9-1.2 | 0.8-1.1 |
| Liquid core at first reduction segment, mm | 40-60 | 35-55 | 30-50 |
| Total soft reduction, mm | 3-5 | 4-6 | 5-8 |
| Reduction gradient, mm per metre of strand | 0.4-0.8 | 0.6-1.0 | 0.8-1.3 |
| Target centre segregation rating, GB/T 1979 | class 1.0 or better | class 1.0 or better | class 1.5 or better |
| Secondary cooling specific water flow, L/kg | 1.4-2.0 | 1.3-1.8 | 1.2-1.7 |
Note that as carbon and alloy content rise, the permissible strain before internal cracking falls, so the higher-alloy grades need larger total reduction applied more gently, at higher shell strength and at lower superheat. Treating NM500 like NM400 by copying the same setpoints is one of the most common reasons a soft reduction installation disappoints.
Coupling Soft Reduction with Secondary Cooling and Superheat
Soft reduction is not a stand-alone fix. Its effectiveness depends on the shape of the pool it is acting on, and that shape is set by cooling and superheat.
Superheat control. Low superheat produces a shorter, narrower liquid pool with a thicker shell at the reduction segments, which improves reduction efficiency and reduces the volume of enriched liquid available to segregate. Too low a superheat risks nozzle blockage and meniscus freezing. A working band of 10-25 degrees C above liquidus is normal, with tighter control on the higher-carbon grades.
Cooling intensity and uniformity. Uneven secondary cooling produces shell thickness variation around the perimeter, and soft reduction applied to an eccentric shell will overstrain the thin side. Nozzle clogging, spray overlap defects and distorted rolls all show up as asymmetry in the macro-structure. Specific water flow is therefore kept in the moderate band and monitored per zone, not just in total.
Mould powder and oscillation. A stable flux film and consistent oscillation parameters reduce shell thickness variation at the meniscus, which propagates all the way to the reduction zone. Deep oscillation marks act as stress raisers when reduction begins.
Electromagnetic stirring. Where a mould electromagnetic stirrer is available, it homogenises the superheat and breaks the dendrite tips, producing a more equiaxed core. An equiaxed core responds better to soft reduction because the permeable network permits liquid to be squeezed back rather than pushing a rigid columnar structure. When stirring is added, the reduction schedule generally needs re-optimising, since the pool profile changes.
What Changes in the Rolled Plate: Segregation, Hardness and Inspection
The production benefit shows up in three measurable places.
Macro-structure. Centre segregation in the billet is assessed against the macrostructure reference charts of GB/T 1979, with class A referring to the centre-line segregation pattern and class B to the associated shrinkage and porosity. Systematic application of a correctly tuned dynamic model, combined with disciplined casting practice, typically moves a billet that would have produced class 2 or worse centre segregation into the class 1 to 1.5 band. Centre porosity and centre cracks are reduced at the same time.
Hardness uniformity after quenching. Because banded segregation is reduced, the local hardenability scatter across the plate width narrows. A plate that previously showed a spread of 30-50 HBW between edge and centre after tempering will usually show a much tighter distribution once the core composition is uniform. This matters directly for NM400 and NM450 acceptance, where the specified hardness window is only 60 HBW wide.
Ultrasonic inspection yield. Centre-line laminations and clusters are the dominant rejection cause in heavy wear plate. Plate is normally inspected to GB/T 2970 for heavy steel plate, with acceptance levels agreed between producer and purchaser according to the end use. Reducing centre segregation removes the reflectors that cause rejection, so the pass rate of rolled plate inspection rises. In plant terms, the improvement in billet internal quality converts directly into a reduction in the downgrade and scrap rate of the rolled product, which is where the economic case for the model is made.
A secondary benefit appears in fabrication. Plates with clean centre lines are less prone to centre-line cracking during flame cutting of thick sections and during cold bending, because the weak plane that would open under through-thickness tensile strain is no longer there.
Commissioning, Verification and Common Failure Modes
A dynamic model should be commissioned in stages rather than switched on at full authority. A workable sequence is: verify the shot-test pool profile at three casting speeds; run the model in advisory mode against existing static setpoints and log the difference; apply soft reduction with a conservative total of 2-3 mm and confirm no internal cracking by sampling two heats per grade; then step the total reduction up in increments while monitoring macro-structure and ultrasonic yield; finally enable full dynamic updating with speed changes and grade transitions.
Verification should use three independent evidence streams: billet macro-sections from the transition heats (sulphur print or macro-etch, rated to GB/T 1979), rolled plate ultrasonic results, and hardness distribution maps across width and thickness on quenched and tempered plate. A model that improves macro-structure but degrades ultrasonic yield is almost always over-reducing or reducing too early.
The failure modes seen most often in practice are these. Reduction applied too early, with an over-thick liquid core, produces internal cracks and a "squeezed" oval section. Reduction applied too late does nothing but load the frame. A model anchored to a single casting speed drifts whenever the machine runs faster or slower, so the end point must be re-anchored. Ignoring superheat means the same setpoints produce different results on different heats of the same grade. Gap control hysteresis on hydraulic cylinders causes the actual draft to lag the setpoint during speed changes, which is why closed-loop position feedback is essential. Finally, roll wear and misalignment change the effective gap; a schedule that worked on a new segment set will over-reduce once the rolls are worn if the gap is not referenced to a calibrated datum.
Procurement and Acceptance Notes for Wear Plate Buyers
A buyer of NM400, NM450 or NM500 plate cannot see the casting machine, but the consequences of billet internal quality are visible in the delivered product and in the documents that accompany it. The following points are worth building into a purchase specification.
Heat and slab traceability. Require the plate heat number to be traceable to the casting heat and to the slab, so that any inspection finding can be traced back to casting parameters. This is the only way to confirm that a supplier's claimed soft reduction practice is real rather than descriptive.
Macro-structure evidence. Ask for the centre segregation rating of the corresponding billet, rated to GB/T 1979, and for the sampling position. A supplier who regularly achieves class 1 to 1.5 will have the data available; one who does not will offer general statements.
Ultrasonic acceptance. Define the inspection standard and acceptance level explicitly, typically GB/T 2970 with a level agreed for the application. For abrasion parts subject to impact, a stricter level is justified than for liners in low-stress sliding wear.
Hardness verification. Specify hardness testing to the requirements of GB/T 24186 for the ordered grade, with the number and location of test points agreed, and confirm whether surface or through-thickness hardness is required. Surface hardness alone will not reveal centre-line soft spots caused by segregation.
Mechanical and chemical confirmation. Require the mill test certificate to report carbon, manganese, chromium, molybdenum, boron where applicable, plus tensile properties and impact energy at the specified test temperature. Chemistry that sits at the top of the specification range with wide scatter between heats is a warning sign of process instability upstream.
FAQ
Q: What is dynamic soft reduction in continuous casting?
Dynamic soft reduction is the application of a controlled thickness reduction to a continuously cast strand while its core is still partially liquid, with the draft position and amount adjusted in real time to the actual solidification end point. It compensates solidification shrinkage and suppresses centre segregation.
Q: Why is the nail shooting test used to calibrate the model?
Because shell thickness calculated from heat-transfer models drifts with cooling, flux and superheat changes. Firing pins of known composition into the strand and examining where they melt or remain intact gives a direct physical bracket on the position of the liquid core at a known casting speed.
Q: How much total reduction is normally applied to a 250 mm billet?
Industrial practice generally falls between 3 and 8 mm total, distributed over the segments where the liquid core is roughly 20-60 mm. Higher-carbon and higher-alloy wear grades tend toward the upper part of the range, applied more gently to avoid internal cracking.
Q: What happens if reduction is applied too early?
The draft acts on a thick liquid core, the shell buckles, liquid is displaced rather than the core being densified, and internal cracks open. The section may also become oval, which complicates subsequent rolling.
Q: How does billet centre segregation affect quenched wear plate hardness?
Segregated carbon and alloy raise local hardenability, so the centre line transforms differently from the rest of the section. After tempering this appears as banded hardness and soft spots, and the spread across the plate can exceed the 60 HBW width of an NM400 or NM450 specification window.
Q: Which standards are used to judge the result?
Billet macro-structure is rated against the reference charts of GB/T 1979, plate ultrasonic inspection is commonly performed to GB/T 2970 at an agreed acceptance level, and the product grade itself, including hardness requirements, is specified in GB/T 24186 for NM-series wear-resistant plate.




