The new frontier in steel casting: teaching water to hit ±4 °C

In continuous casting, secondary cooling sprays are the only actively controllable way to pull heat out of a moving strand—everything else is passive. Industry and academic sources call this stage “vital” to solidification rates and shell quality (Lechler; MDPI Metals). The catch: even small variations in spray intensity move the needle on quality. One analysis of high‑carbon billets found an optimal cooling intensity of ≈0.9 L/kg minimized center carbon segregation (ResearchGate).

That’s why suppliers emphasize that different steel grades and section sizes need different cooling profiles to avoid thermal gradients—too much at the corners invites cracks; too little leaves a long liquid core (Lechler; MDPI Processes). As Lechler puts it, “different steel grades require individual cooling processes” with separate spray profiles (Lechler).

Speed, grade, and thermal load

Faster casting speeds raise the thermal load per meter, demanding proportionally higher spray intensity to remove extra heat in the same machine length (MDPI Processes). And grade matters: high‑alloy or high‑carbon steels often have narrower ductility temperature ranges (the temperature band where steel can deform without cracking), so they need more gradated cooling; ultra‑low carbon grades can tolerate more aggressive cooling.

In practice, advanced systems store grade‑specific target surface temperatures—using steel‑grade lookup tables or models—and then monitor actual casting speed to modulate water flow by zone in real time (IntechOpen; MDPI Processes; MDPI Metals). Position encoders (to track strand position) and pyrometers (non‑contact temperature sensors) feed a PLC (programmable logic controller) that solves the cooling model and updates each nozzle’s flow on the fly (MDPI Processes).

On the utilities side, pretreatment options such as ultrafiltration can support the industrial water management that sits behind stable spray circuits.

Nozzles, turn‑down, and pattern control

Hardware has caught up with the software. Traditional air‑mist nozzles have limited turn‑down (≈4:1), restricting how far operators can throttle without sacrificing uniformity (Primetals Magazine). New pulse‑width‑modulated (PWM) water‑only nozzles—Primetals’ “DynaJet Flex” is one—switch water on/off at high frequency; cooling intensity scales with duty cycle (fraction of time “on”) while the nozzle runs at a fixed pressure (Primetals Magazine). That boosts turn‑down to ≥15:1 and keeps the spray pattern nearly constant across flows—critical for narrow slabs and low‑flow conditions (Primetals Magazine).

Pattern resolution is improving, too. Plants divide each spray zone into multiple sub‑zones across the width and individually pulse nozzle clusters to shape the footprint—say, one center strip plus multiple edge strips to manage corners (Primetals Magazine; Lechler). One cited arrangement uses a four‑subzone layout (one center + three edge strips) over a caster segment for precise corner control (Primetals Magazine). For manifolds and headers, selecting industrial pressure–rated steel filter housings is a practical consideration in spray‑water service.

Model‑based control and optimization

Coordinating dozens of nozzles demands more than set‑and‑forget logic. Modern architectures embed strand heat‑transfer models and use optimization or model‑predictive control to compute flows in real time. One recent framework routes flow‑meter and temperature signals into a steel‑surface temperature model and sends optimal setpoints back to PLC‑driven valves (MDPI Processes).

Digital‑twin studies go further, coordinating sprays with electromagnetic stirring; a particle‑swarm optimization (PSO, an AI optimization technique) controller kept strand temperatures within ±4 °C of targets across all zones (MDPI Metals). Genetic‑algorithm and neural‑network methods are in the mix as well: using an improved genetic algorithm in a billet caster, simulation on a T91 steel bloom showed a 2% reduction in total cooling water use and smoother peak cooling rates, yielding a higher fraction of equiaxed grains (grains of similar dimensions) (University of Wollongong). In coordinated control trials (cooling + stirring), billet tests achieved a central carbon segregation index ≈1.06 (near ideal) after optimization (Wiley Online Library).

Utilities programs can complement control projects; for example, accurate chemical additions from a dosing pump can support stable spray‑water circuits as operators scale up automation.

Measured outcomes and operating impact

The case for precision control is quantifiable. Custom cooling solutions enable increased casting speed and wider steel grades, boosting productivity and quality, according to Lechler. In the R&D examples above, optimization smoothed cooling curves—reducing scrap indicators like hot tears or cold laps—and cut water usage by ~2% in the T91 bloom case (University of Wollongong).

On the plant floor, precision systems can hold surface temperatures within a few degrees and maintain the molten core length exactly at the mold exit. Coordinated digital‑twin control confined temperature fluctuations to ±4 °C (MDPI Metals). Reduced thermal gradients also minimize downtime from breakout or rework. For sites upgrading water management alongside casting controls, integrated membrane systems are an option to support industrial water quality without changing the control logic.

What the evidence shows

Across vendors and journals, the throughline is clear: high‑turn‑down nozzles plus real‑time, model‑based adjustment tied to casting speed and grade deliver uniform, stable cooling. The evidence spans tight temperature control (±4 °C), improved segregation indices, liters saved, and faster casting (MDPI Metals; University of Wollongong). That combination is translating into higher yield and efficiency on modern casters.

Sources: Scholarly and industry publications on continuous‑casting cooling (e.g., IntechOpen, MDPI, ISIJ Int.); equipment vendors (Primetals, Lechler) and case studies (Primetals Magazine; University of Wollongong; Lechler; MDPI Metals). Each claim above is supported by experiments, models, or trials reported therein.