Ladle slag detection is not simply a matter of identifying a colour change in a stream of molten steel. It is a process-control decision made at a difficult point in steel production: near the end of ladle draining, when the steel level falls, flow conditions change and the risk of slag entering the tundish or the next treatment stage increases.
The purpose of detection is not to remove all slag from the process. Ladle slag has a role in secondary metallurgy. The operational objective is to limit the transfer of unwanted primary-steelmaking slag or excessive slag during steel transfer, while preserving a stable and safe pouring operation.
This distinction matters because there is no single best detection method for every steel plant. Optical, infrared, electromagnetic and vibration-based methods observe different physical effects. Manual observation can still be useful, but it depends heavily on visibility and operator judgement. The correct choice depends on the transfer point, ladle and nozzle arrangement, desired response, existing gate-control system, maintenance conditions and the plant’s ability to generate reliable process data.
Why slag carry-over matters
During primary steelmaking, part of the furnace slag can be transferred with the steel into the ladle. Later, during ladle treatment or casting, this carry-over slag can affect the steelmaking sequence. Its impact depends on slag chemistry, quantity, steel grade and the operating practice used at the plant.
Published technical work has associated excessive slag carry-over with increased deoxidizer consumption, phosphorus reversion, oxide inclusion formation and ladle-refractory wear. These effects are linked to the oxidizing components and composition of the transferred slag, not merely to the visible presence of a second phase.1
In continuous casting, the end of ladle draining is also a hydraulic problem. As the steel level falls, vortexing or unstable flow can develop around the outlet. Slag may then enter the outlet stream, and the transition may not be easy to identify from a safe distance. A detection system is therefore most useful when it provides a timely signal that can support an operator decision or a controlled gate-closing action.
The objective should be defined in plant terms. A steelmaker may want to reduce slag carry-over into the tundish, protect steel cleanliness, reduce nozzle-related instability, improve heat-to-heat consistency or create a traceable record of the end of the cast. These are related goals, but they are not the same performance requirement.
What exactly must be detected?
Before comparing technologies, define the event. “Slag detection” can refer to several different situations:
- slag entering the tapping or pouring stream;
- the transition from predominantly steel flow to mixed steel-and-slag flow;
- the onset of a slag vortex near the ladle outlet;
- the start or stop of steel flow after opening a slide gate;
- slag carry-over during ladle-to-tundish transfer;
- steel penetration or an abnormal condition around a nozzle or shroud.
A system designed to observe an exposed tapping stream is not automatically suitable for monitoring flow through a submerged ladle shroud. A system that detects a transition event is not necessarily measuring the mass or thickness of the slag layer. The plant should state which event is required, where it occurs and what action should follow.
This is also why a detection signal should not be treated as a laboratory analysis. Most online systems identify a change in physical behaviour and provide an operational indication. Converting that indication into an estimate of slag quantity requires plant-specific calibration and a defined reference method.
Method 1: Manual visual observation
The simplest method is for an operator to observe the stream, the ladle outlet or the visual condition around the shroud and close the gate when slag is judged to be entering the flow.
Manual observation has one obvious advantage: it uses the operator’s direct experience and does not require a new sensor. It may also remain useful as a reference during commissioning and as a backup when an automatic system is unavailable.
Its limitations are equally clear. Visibility can be affected by glare, fumes, dust, low light, splashing and the geometry of the ladle and shroud. The decision is also influenced by fatigue, workload, viewing position and differences between operators. Manual observation is difficult to standardize and produces limited historical data.
Manual practice should therefore be treated as an operating method that needs defined rules, not as a reliable measurement standard. If it is used as the reference during system commissioning, the plant should record who made the judgement, under what conditions and with what supporting evidence.
Method 2: Optical and infrared detection
Optical systems observe the visible or thermal characteristics of the stream. Depending on the design, the camera may look at the exposed flow between the ladle and tundish, the outlet region or the pouring stream during tapping. Algorithms then distinguish changes associated with steel and slag.
Longer-wavelength infrared imaging can be useful because the system does not rely only on visible colour. Published technical work describes infrared imaging as a non-contact approach for identifying slag in a tapping stream.1 Industrial optical systems may also combine a camera, protected housing, lighting or infrared sensing, image processing and an alarm or control output.2
Optical and infrared methods are attractive when there is a clear line of sight to the event. They can provide a visual record and may be easier for operators to understand because the signal is related to what is happening in the stream.
The plant must nevertheless address several practical conditions. Smoke, dust, steam, glare, scale, changing ambient light, lens contamination and camera vibration can affect the signal. The field of view must remain stable, and the system needs a cleaning and inspection routine. The test procedure should include the actual range of steel grades, flow rates and environmental conditions rather than relying only on a clean demonstration sample.
Method 3: Electromagnetic detection
Electromagnetic systems infer the change from the electrical response of the material passing through or near the sensing assembly. A commonly described arrangement places a sensor assembly around the inner nozzle or slide-gate outlet. An alternating magnetic field induces a response in the conducting molten material, and the change in the measured signal is analysed when slag enters the flow.3
The advantage of this approach is that it does not require a clear optical view of the stream. It can be integrated with a slide-gate control system and can provide a direct trigger for an alarm or gate-closing sequence, subject to the plant’s safety and control design.
Its suitability depends strongly on the nozzle and gate arrangement, sensor position, material properties, mechanical fit and the control system available at the plant. Installation may require a defined mechanical interface and careful protection of the sensor assembly. The signal should also be checked against changes in flow condition, equipment configuration and start-of-casting behaviour.
An electromagnetic system should not be judged only by its nominal detection sensitivity. The plant should review installation constraints, compatibility with the existing slide gate, maintenance access, calibration, fault indication and the consequences of a missed or false signal.
Method 4: Vibration and vibro-acoustic detection
Vibration-based systems monitor the mechanical response generated as molten material flows through the ladle outlet, shroud or associated manipulation structure. The physical behaviour changes when the flow changes from predominantly molten steel to a mixture containing slag. Sensors such as accelerometers can be mounted on a ladle shroud, a shroud manipulator or another mechanically coupled point away from the hottest part of the stream.
The principle is not that “slag makes more vibration” in every situation. The usable signal is the change in vibration characteristics caused by differences in density, viscosity, flow pattern and interaction with the outlet or shroud. The signal can be analysed in the time domain, frequency domain or through extracted features. Research on vibration-based methods has used acceleration signals, displacement or other process signals to distinguish flow states.4 5
DBM’s VSD2000 is described as a vibrational ladle slag detection system that analyses vibration signals from ladle manipulation and obtains information associated with slag vortex formation.6 In practical terms, this places the system in the vibration-based category. It should be evaluated as an online event-detection method, not assumed to be a direct measurement of slag mass.
Vibration detection can be useful when optical access is poor or when the plant wants to sense the event from a mechanically coupled structure. It can also be developed with signals already related to the ladle manipulation arrangement. However, vibration is not generated by slag alone. Mechanical contact, manipulator movement, nozzle condition, drive action, ladle position, flow rate and other disturbances can influence the signal.
This is the main engineering requirement for a vibration system: the algorithm and installation must separate the slag-related change from the normal vibration of the equipment. A single threshold based only on amplitude may be insufficient in a complex casting environment. Multi-axis sensing, signal features, gate or flow signals, ladle weight and timing information can improve the interpretation when they are available and properly synchronized.5
Method 5: Process-signal and hybrid detection
Some plants combine a physical detector with process signals. Examples include slide-gate position, opening command, ladle weight, casting status, shroud position, flow start and the timing of the heat. These signals do not replace the primary detector, but they provide context.
A hybrid system may reduce false alarms by distinguishing the start of casting from the end of casting, separating a normal manipulator movement from a flow-related vibration or recognizing that a gate command was issued without a corresponding steel-flow response.
The plant should be careful not to confuse correlation with detection. A change in ladle weight or gate position may indicate that the process is changing, but it does not by itself prove that slag has entered the stream. The value of process signals lies in combining them with a physical observation and a known operating sequence.
Comparing the methods in a real plant
The following questions are more useful than asking which technology is “the most advanced”:
Where is the event visible or physically transmitted?
If there is a stable line of sight, optical or infrared observation may be appropriate. If the stream is enclosed or the view is unreliable, electromagnetic or vibration-based sensing may be more practical.
What action must follow the detection?
A warning for an operator, a recommendation to begin gate closing and an automatic gate-closing command have different requirements for response time, reliability, interlock and risk assessment.
How much modification is acceptable?
A plant may prefer a non-contact camera installation, while another may already have a suitable slide-gate sensor interface or a manipulator structure for vibration measurement. Mechanical access, hot-zone protection and maintenance space can determine the feasible method before accuracy is compared.
Can the signal be validated and maintained?
The system should provide a way to review the event, the operating state and the reason for the alarm. It should also identify sensor failure, poor signal quality and abnormal installation conditions. A method that cannot be checked during maintenance is difficult to trust in production.
Is the plant trying to detect an event or quantify slag?
Online detection usually identifies a transition or abnormal flow condition. If the plant needs a mass estimate or a quantitative slag-carry-over model, it must define a separate calibration method and reference measurement.
How to commission a detection system
Commissioning should be treated as an application study, not simply as a sensor installation.
First, define the reference event. The plant should agree on how “slag detected” will be recorded and how the reference judgement will be made. Depending on the process, this may use operator observation, video review, process records, downstream samples or a controlled operating test. No reference method is perfect, so its uncertainty should be acknowledged.
Next, collect data across the real operating range. Include different ladles, nozzles, flow rates, steel grades, temperatures, operators and equipment conditions. For vibration systems, record normal manipulator movements and known non-slag disturbances. For optical systems, include changes in light, smoke and contamination. For electromagnetic systems, verify the response with the actual nozzle and gate configuration.
Then, tune the system around the plant’s decision. A high-sensitivity warning may be suitable when the priority is to limit slag carry-over. A different setting may be used when a premature gate closure would create a larger operational problem. The chosen logic, delay, alarm level and automatic action should be documented and approved.
Finally, review the first period of operation. Track false alarms, missed events, late detections, manual overrides, gate actions, sensor faults and the conditions associated with each event. A detection system becomes valuable when its records are used to improve the process and not merely to display an alarm.