Inside the water‑spray arms race making continuous casting cleaner, faster, and safer
Steelmakers are turning to mechatronic nozzles and digital‑twin controllers to keep strand temperatures on a tight leash — and the payoffs include fewer cracks, higher speeds, and less scrap.
In continuous casting, the quiet heroes are the secondary‑cooling sprays — the banks of nozzles that douse a red‑hot strand after it exits the mold. Get the water wrong and you court disaster: overcool corners and you seed surface cracks; undercool and the liquid pool runs long, flirting with breakout. The job now is surgical precision, tuned by algorithm.
Modern casters chase uniform cooling across the strand width and modulate spray intensities by grade and speed so the surface stays above about 850 °C until the shell is safely built, according to IspatGuru (note: secondary cooling refers to spray chambers that extract heat after the mold). Non‑uniform coverage — from roll collisions or inactive nozzles — wastes water and creates “hot spots” IspatGuru. So mills target predefined surface‑temperature curves for every steel grade and casting speed, then tweak flows zone‑by‑zone to follow that profile IntechOpen IspatGuru.
Thermal profile targets and risks
Product quality hinges on extracting heat uniformly so the solidification front advances smoothly. The surface temperature profile and the timing of solidification respond directly to spray intensity and distribution IspatGuru. Over‑cooling near corners elevates thermal stress and causes surface cracks; under‑cooling lengthens the liquid core and risks a breakout — a catastrophic shell failure IspatGuru.
To avoid cracks, many control strategies keep the surface above ~850 °C until solidification completes IspatGuru. Dynamic systems then adjust sprays in real time to stay within a few degrees of the setpoint even during speed changes or ladle‑temperature swings IspatGuru.
Related plant utilities can include water‑treatment ancillaries for the cooling circuit, such as supporting equipment for water treatment, which sit alongside control hardware without changing the core spray‑control problem.
From lookup tables to effective speed
Early approaches relied on physics‑based models and look‑up tables. Parametric controllers compute each zone’s water flow as a quadratic of casting speed, Qi = AiV² + BiV + Ci, tuned from steady‑state heat‑transfer models (conduction, latent heat, convection) to match grade‑specific cooling curves IntechOpen IntechOpen.
An “effective speed” model refines that by accounting for residence time — blending instantaneous and average speed into Ve — and then using Qi = AiVe² + BiVe + Ci. Simulations show effective‑speed control smooths temperatures better than simple parametrics during transients IntechOpen IntechOpen IntechOpen.
As mills modernized, they added online thermal models — “digital twins” (a live numerical simulator of the strand) — for closed‑loop control that continuously computes the temperature profile from operating data and adjusts flows accordingly. This approach can keep surface temperatures within a few degrees of target under speed or ladle‑temperature changes and has displaced fixed tables where speeds vary significantly IspatGuru.
Upstream of the sprays, facilities sometimes consider pretreatment steps for make‑up water; examples on the market include ultrafiltration used as pretreatment in industrial settings. Such utilities are separate from, but adjacent to, the spray‑control story described here.
Adaptive control: fuzzy logic and MPC
Adaptive and intelligent controllers are gaining ground. Fuzzy‑logic systems, for example, continuously adjust spray flow distribution based on instant feedback and, in one study, were shown by simulation to eliminate surface flaws and rejects by adaptively modulating flows under varying conditions ResearchGate. Compared with traditional PID loops, the same work reported “fuzzy logic offers better control performance, smaller overshoots, [and a] faster response” for spray‑flow control ResearchGate. In practice, inputs can include steel temperature, casting speed, and slab width, with real‑time valve commands as outputs.
Elsewhere, model‑predictive control (MPC, an optimization‑based method that predicts future behavior) and particle‑swarm algorithms are paired with a 3D digital twin calibrated online by temperature sensors to optimize water flows (and even final‑stirring parameters) MDPI MDPI. One case raised stable casting speed by about 6% — from 0.51 to 0.54 m/min — while holding the solid fraction at target and keeping surface‑temperature error within ±4 °C during transients. High‑magnitude centerline macrosegregation (grade >1.5; macrosegregation refers to composition variation) dropped from 11% of samples to 3.3% MDPI MDPI.
GPU‑accelerated MPC and neural networks are being explored to compute zone flows in real time, using inputs such as instantaneous casting speed and effective speed, within an Industry 4.0 framework ResearchGate MDPI.
Alongside controls, some plants manage cooling‑circuit chemistry with dosing equipment; category examples include a dosing pump for accurate chemical dosing. Such utilities are separate from the control algorithms themselves.
Mechatronic sprays and fine‑grained zoning
Hardware matters. Nozzle layouts are engineered to overlap without colliding with support rolls, avoiding dead zones and water wastage IspatGuru. Spray panels are typically split into zones — often center versus margins — so edges can be watered independently. Industry leaders now fit individually actuated, mechatronic nozzles.
One commercial approach uses pulse‑width‑modulated nozzles (PWM; high‑frequency on/off switching) with pneumatic drives. Mean cooling intensity is set by duty cycle while water pressure stays constant — decoupling turn‑down ratio from pattern uniformity. Results include a turn‑down of 1:15 (max/min flow) with a single‑cone nozzle versus about 1:4 in older fixed‑flow designs, and a constant spray pattern even at very low flows Primetals Primetals Primetals.
Finer sub‑zoning extends control. In one implementation, each cooling zone was split into four sub‑zones (one center plus three margins) with independent valves, enabling width‑adaptive cooling for 800–1650 mm slabs. After a 4‑step margin‑control system was installed at a Korean mill, there were no corner cracks on advanced high‑strength steel (AHSS) slabs; the products could be rolled “in non‑scarfing condition” — even without upstream inspection Primetals Primetals.
These nozzle systems are orchestrated by PLC/DCS‑level controllers (PLC/DCS: plant‑floor automation systems) running the algorithms above. Primetals’ Dynacs 3D, for example, continuously computes a 3D temperature profile and adjusts zone setpoints based on casting speed, slab dimensions, and steel grade, so the final solidification point (metallurgical length) hits target for each product Primetals. Cooling models can integrate with electromagnetic stirrers; after the mold, though, the cooling system itself exerts the largest effect on shell growth.
Protecting spray hardware from debris is a separate plant‑utility concern; options in industrial settings include an automatic screen for continuous debris removal or a steel filter housing suited to high‑pressure service. These utilities are adjacent to, not part of, the control strategies cited above.
Measured gains and water‑efficiency focus
Across case studies and simulations, the results are consistent. Tight control — keeping surface temperatures within ±4 °C of setpoints — reduces surface cracks and centerline segregations, while optimization can lift speed by ~6% without new defects MDPI MDPI. In one digital‑twin project, high‑magnitude centerline macrosegregation (grade >1.5) fell from 11% to 3.3% while speed rose from 0.51 to 0.54 m/min MDPI.
Adaptive control also trims coolant usage: narrower, duty‑cycled sprays can deliver the same average cooling with less water than constant high‑flow jets. Hard data on water savings are scarce, but industry reports emphasize water‑ and energy‑efficiency. An Indonesian steelmaker, PT Ispat Indo, was recognized for setting targets and continuously upgrading its cooling system to reduce waste and improve efficiency UNEP.
Some mills expand utility systems around cooling water as part of broader upgrades; for instance, facilities with challenging make‑up sources may deploy RO, NF, and UF systems in industrial water treatment, or brackish‑water RO where total dissolved solids are elevated. In closed‑loop utilities, operators also evaluate chemistries such as a scale inhibitor for cooling systems. These utilities are distinct from the spray controls themselves.
Key takeaways for capital decisions
Modern secondary‑cooling control merges two levers: advanced nozzles and advanced algorithms. Pulse‑modulated nozzles and multi‑zone valves deliver mechanical precision Primetals Primetals, while model‑based or adaptive controllers tune flows to strip width, speed, and grade IntechOpen Primetals. The payoff is measurable: fewer defects, higher casting rates, and lower scrap. One digital‑twin control project cut high‑grade macrosegregation from 11% to 3.3% while boosting speed MDPI.
Sourcing and upkeep of spray hardware sit alongside utilities such as strainers and housings; examples include a strainer in front of fine circuits or a 316L stainless cartridge housing where hygiene‑grade construction is specified by plant standards. These are utilities adjacent to, not a substitute for, the control technologies and nozzle designs described in the cited literature.
Documentation and sources
Sources: Peer‑reviewed articles and industry reports (IspatGuru, IntechOpen, Metals, Primetals, MATEC, UNEP) underpin the data and case examples. Each finding above is supported by the cited literature — see IspatGuru, IntechOpen, IntechOpen, IspatGuru, ResearchGate, Primetals, and MDPI.