Product
SIC™ BF Smart Agent X
Physics-constrained AI for the blast furnace
Agent X turns real blast furnace operating data into predictions of what cannot be measured inside the furnace, and recommends quantified low-carbon operating changes.
“SIC™ opens the black box of the blast furnace, so steelmakers can adopt new ores — from Australian ores to new blends — and low-carbon operations with confidence instead of trial and error.”
What it predicts
The states no sensor can reach.
Thermocouples and probes see the shell, the top and the tuyeres. Agent X fills in the rest — from top charging to hearth drainage.
- 01
Gas flow and velocity fields
Where the reducing gas actually goes, and how burden changes redirect it.
- 02
Temperature fields and thermal reserve zone
The internal heat distribution behind fuel rate and stability.
- 03
Cohesive zone shape and position
The permeability bottleneck that decides how far an ore blend can be pushed.
- 04
Raceway conditions
Combustion and flow in front of the tuyeres under different injection regimes.
- 05
3D transient evolution
How the furnace interior moves over time, not just a single snapshot.
Stability early warning
Warns before the furnace goes wrong.
A stability-monitoring module watches gas-dynamic, thermal and chemical indicators together, calibrates a baseline from each furnace's own data, and flags fluctuations before they grow into an upset.
- Gas-dynamic, thermal and chemical subsystems combined into one furnace status: Stable · Watch · Warning · Alarm
- Thresholds calibrated per indicator from the furnace's own baseline
- Early-warning tracking shows which variables moved first, and how far ahead
- warning raised before the plant found a tuyere failure
- ~1 daywarning raised before the plant found a tuyere failure
- alarm raised before the failure was found
- ~½ dayalarm raised before the failure was found
Replay of hourly operating data from a running blast furnace.
What it recommends
Low-carbon strategies, evaluated before they reach the furnace.
Burden and ore-blend optimisation
New ore sources and blend ratios, higher pellet ratios — including transitions to Australian ores.
Hydrogen-rich gas and COG injection
Injection rates and conditions compared side by side.
Low-carbon burden materials
Advanced materials and emerging technologies tested in the model first.
Fuel and coke rate
Operating moves ranked by their quantified effect on fuel rate and CO₂.
The agent
An expert you can question.
Agent X explains its results in plain language. Ask why the cohesive zone moved, or what a change will do to the raceway, and it answers from the model results and a knowledge base of ironmaking literature.
- Expert analysis and Q&A on in-furnace results
- Retrieval-augmented knowledge base that your team can extend
- Multilingual technical literature, including Chinese and English sources
- Every answer grounded in physics-constrained model output
Capabilities
Built for engineers who run real furnaces.
Mechanism-based results
In-furnace phenomena derived from CFD-based advanced models.
Operating-data visualisation
Plant measurements and imported tables, visualised instantly.
Instant contour rendering
Dynamic playback, multi-variable comparison, custom ranges and HD export.
Modular expansion
Add mathematical and physical model components as needs grow.
Data security
AES-256-GCM encryption protects operating data and model assets.
Access control
User authentication, permissions and structured data management.
How it compares
What other approaches each leave out.
| Sensor screens & 3D visualisation | Data-driven AI | Steady-state models | One-off CFD studies | SIC™ BF Smart Agent X | |
|---|---|---|---|---|---|
| Shows the in-furnace state | No | Partly | Yes | Yes | Yes |
| Follows how the furnace evolves over time | No | Partly | No | Partly | Yes |
| Evaluates operations with no history | No | No | Partly | Yes | Yes |
| Results in seconds, on site | Yes | Yes | Partly | No | Yes |
| Explains why | No | Partly | Yes | Yes | Yes |
Yes YesPartly PartlyNo NoTypical capabilities of each approach; individual products vary.
From site to system
Deployed on site today. Engineered for live integration next.
Agent X already runs on site as an offline application at a large steelmaker in North China, where engineers import operating data and see the in-furnace response in seconds. The next step connects it directly to the plant's data and control systems.
01
Plant data systems
DCS/PLC · OPC UA · historian
02
SIC™ BF Smart Agent X
Physics-constrained engine
03
Open interfaces
Predictions · warnings · recommendations
04
Plant users
Level 2 · control room · engineers
Today
Offline deployment on site
In offline trial at a large steelmaker in North China. Engineers import operating data and get in-furnace predictions in seconds.
Read moreNext
Plant data connectors
Read furnace operating data online — burden, blast, injection, top gas — from DCS/PLC through industrial standards such as OPC UA, or from the plant's historian (for example PI). No more manual exports.
Next
Open interfaces
A standard API lets the furnace's Level 2 system, control-room screens and enterprise data platforms call Agent X's in-furnace predictions, warnings and recommendations directly.
Next
On-premise deployment
Delivered as a container (for example Docker) or an installer and run on the plant's own servers inside its network — operating data never leaves the site, and no internet connection is needed.
Deployment
In the control room or the engineering office.
Agent X can sit beside the operators for fast in-furnace answers during shift decisions, or with process engineers for offline studies and technical discussion. Choose the model that fits your data policy.
| On-premise | Cloud | Managed service | |
|---|---|---|---|
| Installation | Your PC or server | Cloud-hosted | Deployed and run by SIC™ |
| Access | Offline, on site | Online, multi-user | Anywhere, secure login |
| Updates | Scheduled releases | Automatic | Managed by SIC™ |
| Support | Per agreement | Online support | Dedicated engineer |
On-premise suits sites with limited connectivity or strict data rules.
Contact
Bring one furnace question. We'll show you what the model sees.
Steel plant, research partner or investor — leave a short message and an engineer from our team will reply within one business day.