A finishing mill customer spec rarely gives much room to work with, and a strip that drifts even a few microns outside the agreed tolerance band becomes a reject, a downgrade, or a customer claim, none of which are cheap once coil weight and shipping are factored in. Automatic gauge control systems have existed for decades, but most still rely on a fixed model of roll force and mill stretch that doesn't fully account for how temperature variation across the strip length changes deformation behavior stand to stand. iFactory adds that missing context back into the loop, and the difference is easiest to see through Book a Demo.
Your AGC Model Assumes The Strip Behaves The Same Stand To Stand. It Doesn't.
Temperature drop along the strip length, roll wear between reground cycles, and entry thickness variation from the roughing mill all change how the strip actually deforms at each finishing stand. iFactory layers real-time correction on top of your existing AGC model so thickness deviation stays inside the tight tolerance your customers expect.
The Gap Between Model Prediction And Actual Strip Behavior
Traditional automatic gauge control relies on a mill stretch model that relates roll force, roll gap, and expected strip thickness through a physical formula calibrated during commissioning. That formula assumes relatively stable inputs, but a hot strip cools continuously as it moves through the finishing train, meaning the tail end of a coil enters each stand at a different temperature than the head end did moments earlier. Since flow stress and therefore roll force requirements change with temperature, a fixed stretch model applied uniformly across the coil length inevitably produces small thickness deviations that accumulate especially toward coil tail, exactly where many customer specifications are strictest.
Head-To-Tail Temperature Drop
A coil's tail end can run measurably cooler than its head by the time it reaches the finishing stands, changing flow stress and producing a thickness trend across the coil length that a static model doesn't fully correct for.
Roll Wear Between Regrinding Cycles
Work roll surface condition changes gradually between scheduled regrinding, altering friction and roll force response in ways the original stretch model calibration doesn't capture until the next recalibration cycle.
Entry Thickness Variation From Upstream
Small thickness variations carried over from the roughing mill or transfer bar propagate into the finishing train, and a gauge control loop that only reacts at the last stand has limited room left to fully correct them.
Grade-To-Grade Model Mismatch
Different steel grades have different flow stress behavior at the same temperature, and a stretch model tuned as an average across grades performs worse on any grade whose actual behavior sits far from that average.
Adding Real-Time Context Back Into The Gauge Control Loop
iFactory doesn't replace your existing AGC system, it feeds it better information. The model tracks strip temperature profile, roll force trend, and measured thickness feedback from the x-ray gauge in real time, and continuously refines the stretch model's parameters to reflect what the mill is actually doing right now rather than what it was calibrated to do at commissioning.
Give Your Existing AGC The Real-Time Context It's Missing
iFactory refines your stretch model against live temperature and roll force data to tighten thickness deviation stand to stand.
What Changes When The Stretch Model Adapts In Real Time
| Aspect | Static Stretch Model | Adaptive Real-Time Model |
|---|---|---|
| Head-to-tail thickness consistency | Drifts as strip cools | Corrected against live temperature trend |
| Roll wear compensation | Only updated at scheduled recalibration | Continuously refined from force feedback |
| Grade change response | Uses averaged behavior across grades | Adjusts to grade-specific flow stress |
| Reject rate from off-spec gauge | Higher, especially near coil tail | Reduced through tighter stand-level control |
We had a persistent pattern of slightly thin gauge toward the tail end of longer coils that our existing AGC never fully corrected, and it took a while to trace it back to temperature drop along the strip. Since iFactory started refining our stretch model against live pyrometer data between stands, that tail-end thickness trend has largely disappeared, and our gauge-related reject rate has come down noticeably.
What Mills Report After Adopting Adaptive Gauge Correction
What Undermines Adaptive Gauge Control Rollouts
Poorly Calibrated Pyrometers
If temperature readings between stands are inaccurate or inconsistently calibrated, the model's correction is built on a flawed input, which can make thickness control worse rather than better until the instrumentation issue is fixed.
X-Ray Gauge Drift Uncorrected
The model relies on measured thickness feedback to confirm its corrections are working, so an uncalibrated x-ray gauge undermines the entire feedback loop the correction depends on.
Ignoring Roll Change Timing
A fresh roll change resets the wear-related parameters the model has been refining, so plants that don't flag roll changes to the system see a temporary dip in correction accuracy until it recalibrates against the new roll surface.
Underestimating Grade Transition Effects
Switching between grades with significantly different flow stress behavior mid-campaign requires the model to adapt quickly, and mills that don't log grade changes clearly see slower correction accuracy during the transition coils.
Frequently Asked Questions
Q: Does this replace our existing AGC system?
No, iFactory works alongside your existing automatic gauge control system rather than replacing it, refining the stretch model's parameters in real time and passing corrected roll gap targets to your current AGC hardware and control logic. This means the safety interlocks, mechanical actuators, and control architecture you already have in place continue operating exactly as they do today, with the improvement happening at the model accuracy level rather than the control mechanism itself. Reach out through Support Contact to review compatibility with your current AGC vendor and setup.
Q: How does the model handle a new grade with limited rolling history?
For a grade with limited historical data, the model initially relies more heavily on physical flow stress relationships and conservative correction margins until enough coils have been rolled to build a grade-specific calibration. Accuracy for that grade improves progressively as more coils are processed and logged, similar to how an experienced mill operator builds intuition for a new grade over the first several campaigns.
Q: Can this help identify when work rolls need to be changed sooner than the scheduled interval?
Yes, because the model continuously tracks the gap between predicted and actual roll force response, a work roll wearing faster than expected shows up as an increasing correction magnitude before it would otherwise be noticed at the next scheduled inspection. This can give mill setup teams an earlier signal to consider adjusting the regrinding schedule for a specific stand rather than waiting for a fixed interval to elapse.
Q: How long does it take to see a measurable improvement in gauge consistency?
Some improvement is visible within the first few days as the model begins refining stretch model parameters against live data, but the full benefit, particularly on less common grades or thickness ranges, builds over several weeks as more coils are processed and logged across the mill's typical product mix. A Book a Demo session can walk through a realistic timeline based on your specific product portfolio.
Q: Does this work for both thin gauge and heavier plate-range products on the same mill?
Yes, the model's correction is built around the underlying temperature and flow stress relationships rather than a single thickness range, so it adapts to both thin gauge coil products and heavier plate-range products as long as sufficient rolling history exists for each product category on that mill.
Tighten Thickness Deviation To What Your Customers Actually Specify
iFactory refines your finishing mill's gauge control against live temperature and force data, stand by stand.
