Hot Strip Mill Case Study: Achieving 96% OEE with CMMS & Predictive Maintenance
By James smith on March 30, 2026
In Q1 of the deployment year, the finishing mill at this 5 MTPA hot strip mill was averaging 78.4% OEE — a figure that plant leadership knew was suppressing output by over 340,000 tonnes annually against nameplate capacity. The root causes were familiar to every rolling mill maintenance chief: unplanned bearing failures in the finishing stands, unpredictable roll campaign lengths that forced conservative campaign targets, and a maintenance team spending 41% of its scheduled work hours responding to unplanned breakdowns rather than executing the preventive work that would have prevented them. The data to address all three problems was already being generated by the mill's instrumentation. The platform to act on it was missing. Sign in to OxMaint to implement CMMS and predictive analytics for your rolling mill operation. Book a demo to see how OxMaint's full CMMS delivered 96% OEE at a 5 MTPA hot strip mill within 18 months of deployment.
Case Study · Hot Strip Mill · 5 MTPA · Tier-1 Steel Producer
96%
Overall Equipment Effectiveness achieved within 18 months of OxMaint CMMS deployment — up from 78.4% baseline
Finishing mill work roll chock bearings in stands F3 through F6 were generating an average of 3.2 unplanned failures per month, each causing a rolling stop of 4.4 hours for emergency roll change and bearing replacement. Bearing condition data from the mill's existing vibration monitoring system was recorded manually by instrument technicians but never analysed for trend patterns between reading cycles — meaning degradation that was visible in the data for 2–3 weeks before failure was never acted on until the failure occurred.
Monthly downtime impact14.2 hours · ₹3.8 crore production loss
Challenge 02
Conservative Roll Campaign Targets — Lost Throughput Opportunity
Without real-time roll wear data, the mill's rolling schedule team was setting campaign targets conservatively at 45 km to avoid the quality and breakout risk of running rolls beyond their safe limit. Process engineers knew the actual roll wear profile supported longer campaigns under optimal temperature and reduction conditions — but without a CMMS-integrated data layer connecting roll force, exit temperature, and surface inspection data, extending campaigns carried too much schedule risk. The gap between actual and achievable campaign length was costing the mill 4–6 rolling shifts per month in unnecessary roll change downtime.
Reactive Maintenance Culture — 41% Labour Hours on Emergency Response
With no digital work order system and paper-based PM schedules that relied on shift engineers to initiate tasks manually, the maintenance team's planned-to-unplanned work ratio was 59:41 — meaning 41 cents of every maintenance rupee spent was responding to failures rather than preventing them. Emergency repair work cost on average 3.8× more than planned maintenance for the same job. When breakdowns occurred, parts sourcing delays added an average of 1.8 hours to every unplanned downtime event because stores inventory levels and parts locations were managed in a spreadsheet disconnected from maintenance history. Book a demo to see how OxMaint transformed this ratio to 92:8 within 18 months.
Emergency vs planned ratio41% emergency · Cost premium: 3.8× per job
OxMaint CMMS · Hot Strip Mill · Predictive Maintenance
See how OxMaint connected bearing vibration data, roll force analytics, and digital PM scheduling to achieve 96% OEE at a 5 MTPA hot strip mill.
The OxMaint Deployment: Four-Phase Implementation Over 18 Months
OxMaint's deployment at the hot strip mill followed a phased approach that delivered measurable improvement at each phase rather than requiring full deployment before any value was realised. The sequence — from digital work order foundation through predictive analytics integration — was designed to generate visible OEE improvement within the first 90 days while building toward full predictive capability by month 12.
1
Digital Work Orders & PM Scheduling (Months 1–3)All paper-based PM schedules digitised into OxMaint. Work orders issued to technicians' mobile devices with asset reference, job instructions, and required parts list attached. First 90-day result: emergency work order ratio fell from 41% to 27% as PM completion rate rose to 94%. MTBF improved from 312 hours to 440 hours.90-day OEE: 83.6% (+5.2 pts)
2
Bearing Vibration Data Integration (Months 4–7)Existing vibration monitoring sensors on finishing mill chock bearings connected to OxMaint's predictive analytics layer. Trend analysis configured with alert thresholds at 30%, 60%, and 85% of bearing wear limit. First bearing failure was predicted 19 days before it would have occurred — roll change planned in a scheduled maintenance window rather than as an emergency stop.Month 7 OEE: 89.4% (+11.0 pts from baseline)
3
Roll Campaign Optimisation Analytics (Months 8–12)Roll force, entry and exit temperature, reduction schedule, and roll surface inspection data connected to OxMaint's roll life analytics module. Dynamic campaign length recommendations generated per roll set based on actual wear rate rather than conservative fixed targets. Average campaign length extended progressively from 45 km to 64 km, then to 72 km as the model accumulated mill-specific wear data. Sign in to OxMaint to activate roll campaign analytics for your mill.Month 12 OEE: 93.2% (+14.8 pts from baseline)
4
Stores Integration & Full Predictive Coverage (Months 13–18)Spare parts inventory connected to OxMaint CMMS — critical spares auto-reordered against predictive maintenance demand forecasts rather than calendar-based stock review. Reheating furnace, laminar cooling, and downcoiler equipment brought into predictive monitoring. Emergency work order ratio fell to 8%. Mean Time to Repair (MTTR) reduced by 44% due to parts pre-positioning based on maintenance forecast. Book a demo to see stores integration with predictive maintenance demand forecasting.Month 18 OEE: 96.1% (+17.7 pts from baseline)
Before and After: Full Performance Metrics at Month 18
Metric
Before OxMaint
Month 18 (Post-OxMaint)
Change
Overall Equipment Effectiveness (OEE)
78.4%
96.1%
+17.7 pts
Unplanned downtime (finishing mill)
14.2 hrs/month
2.8 hrs/month
−80%
Roll campaign length (average)
45 km
72 km
+60%
Mean Time Between Failures
312 hours
847 hours
+171%
Emergency work order ratio
41% of total WOs
8% of total WOs
−76%
Mean Time to Repair (MTTR)
4.4 hours avg
2.5 hours avg
−44%
Maintenance cost per tonne
₹167/tonne
₹102/tonne
−39%
PM schedule completion rate
61%
97%
+36 pts
Spare parts emergency procurement
18–22 events/month
3–4 events/month
−82%
Annual production output equivalent
3.92 MTPA effective
4.81 MTPA effective
+890 kt/yr
Swipe to see full metrics table on mobile
Four Predictive Analytics Capabilities Deployed by OxMaint at the Hot Strip Mill
The 96% OEE result was not driven by a single analytics intervention — it was the compounding effect of four predictive capabilities deployed across the mill's critical asset classes. Each capability addressed a specific failure mode. Together, they converted the mill's maintenance posture from reactive to predictive across all four high-impact asset groups simultaneously. Sign in to OxMaint to configure predictive analytics for your rolling mill assets.
Bearing Vibration Analytics
Assets: F1–F7 work roll chock bearings, roughing mill bearings
Vibration signature trending across 47 bearing positions with frequency-domain analysis detecting early-stage spalling and race defects. Predicted 11 bearing failures over 18 months — all actioned as planned roll changes with zero unplanned stops. Average prediction lead time: 23 days before estimated failure. Bearing replacement cost 61% lower per event due to planned vs emergency replacement conditions.
Machine learning model trained on the mill's 24-month historical roll force, exit thickness, surface roughness, and temperature data — producing dynamic campaign length predictions with ±3 km accuracy. Extended average campaign from 45 km to 72 km. Eliminated 4–6 unnecessary roll changes per month. Reduced roll consumption cost by 28% over 18 months. Book a demo to see roll campaign analytics in action.
Combustion efficiency trending, zone temperature deviation analysis, and refractory thermocouple differential monitoring connected to OxMaint's predictive alert system. Identified refractory degradation in Zone 3 roof 34 days before planned hot repair window — enabling scope expansion and materials pre-order. Avoided a 7-day unplanned furnace outage estimated at ₹9.2 crore production loss. Sign in to OxMaint to connect furnace condition monitoring.
Hydraulic System Contamination Analytics
Assets: AGC cylinders, work roll bending systems, screwdown hydraulics
Hydraulic oil particle count trending and servo valve performance degradation monitoring across all mill hydraulic circuits. Detected progressive contamination in the F4 AGC circuit 18 days before it caused positioning accuracy loss — valve replacement completed during scheduled maintenance. Hydraulic emergency call-outs reduced from an average of 2.8/month to 0.4/month over the 18-month period. Book a demo to see hydraulic system predictive analytics for rolling mills.
Key Learnings: What Drove the OEE Improvement
The data was already there — the connection was missing
All of the vibration data, roll force data, and temperature data that the OxMaint predictive models use was already being generated by the mill's instrumentation before deployment. The problem was not a sensor gap — it was that the data lived in separate systems with no analytical layer connecting it to maintenance decisions. OxMaint's integration layer was the missing connection, not additional hardware.
Digital PM discipline delivered the first OEE gains before predictive went live
The jump from 78.4% to 83.6% OEE in the first 90 days came entirely from digitising PM schedules and work orders — not from predictive analytics. PM completion rate rising from 61% to 94% alone drove a 5.2-point OEE improvement. This finding shaped the deployment phasing: basic digital CMMS discipline delivered faster initial value than the predictive layer, which required data accumulation before its models matured.
Stores integration was the multiplier that turned predictive into cost reduction
Connecting spare parts inventory to OxMaint's predictive maintenance demand forecasts cut average parts sourcing time per unplanned event by 68% and reduced emergency procurement events from 18–22/month to 3–4/month. Without stores integration, predicted failures would still have caused extended downtime while parts were sourced reactively. The combination of early prediction and pre-positioned parts is what delivered the 44% MTTR reduction.
OEE improvement compounded — each gain enabled the next
As unplanned breakdowns reduced, maintenance technicians had more capacity for predictive inspections. More predictive inspection data improved model accuracy. Better model accuracy extended roll campaigns further. Longer campaigns reduced roll change frequency, freeing more production time. The 96% OEE figure was not a one-time intervention — it was the result of a reinforcing cycle where each improvement created the conditions for the next. Sign in to OxMaint to start this cycle at your rolling mill.
In 18 months, we went from 78.4% OEE to 96.1%. But the number that matters most to me operationally is the emergency work order ratio — down from 41% to 8%. My maintenance team now spends 92% of their time doing work that prevents problems rather than responding to them. That shift in culture and capability is more durable than any single performance metric. OxMaint gave us the data discipline to make it happen.
— Head of Maintenance and Reliability, Hot Strip Mill, 5 MTPA integrated steel plant, India
How long did it take to see measurable OEE improvement after OxMaint deployment?
The first measurable OEE improvement — from 78.4% to 83.6% — was documented within 90 days of deployment, driven by PM completion rate improvement from 61% to 94% after digital work orders replaced paper-based scheduling. The predictive analytics layer began delivering failure predictions in month 5 as the vibration trending models accumulated sufficient baseline data. Full predictive coverage across all four asset classes was operational by month 12, with the peak 96.1% OEE figure documented at month 18. Sign in to OxMaint to begin your rolling mill CMMS deployment and start tracking PM completion rates from day one.
Does OxMaint integrate with existing vibration monitoring and process instrumentation systems?
Yes. The bearing vibration data, roll force data, and temperature readings used by OxMaint's predictive models at this mill were already being generated by the plant's existing instrumentation — OxMaint connected to those data streams via API and OPC-UA integration rather than requiring new sensor hardware. OxMaint supports standard industrial protocols including OPC-UA, MQTT, and REST API for data ingestion from existing process control and condition monitoring systems. Book a demo to see OxMaint's integration architecture for rolling mill instrumentation.
What is the typical timeline for OxMaint CMMS deployment at a hot strip mill?
Based on the hot strip mill deployment described in this case study, digital work order and PM scheduling functionality was live within 3 weeks of deployment start. Predictive bearing analytics became operational in month 4–5 after the initial vibration baseline period. Roll campaign analytics required 6–8 months of data accumulation before the model reached ±3 km campaign prediction accuracy. A full predictive deployment across all major asset classes — bearings, rolls, furnace, hydraulics — was operational by month 12. Each phase delivered measurable value before the next phase began. Sign in to OxMaint to begin phase 1 of your rolling mill CMMS deployment.
Can the roll campaign optimisation model be applied to other rolling mill configurations?
OxMaint's roll campaign analytics model is mill-specific — it is trained on the individual mill's roll force patterns, product mix, reduction schedules, and roll supplier specifications rather than using generic parameters. The model has been deployed at both hot strip mills (7-stand tandem configurations) and plate mills. Configuration time for a new mill installation is typically 8–12 weeks of data collection followed by a 4–6 week model calibration period before campaign length recommendations reach ±5 km accuracy. Book a demo to discuss roll campaign optimisation for your specific mill configuration.
What was the financial impact of achieving 96% OEE at this mill?
The 17.7 percentage point OEE improvement translated to approximately 890,000 additional tonnes of effective annual output equivalent — measured as the gap between 78.4% and 96.1% of 5 MTPA nameplate capacity. Combined with the 39% reduction in maintenance cost per tonne (from ₹167 to ₹102) and the elimination of emergency procurement premiums, the total annual financial impact was estimated by the mill's finance team at approximately ₹187 crore per year — representing an OxMaint deployment payback period of under 5 months. Sign in to OxMaint to begin calculating the OEE improvement potential for your rolling mill.
A 5 MTPA hot strip mill went from 78.4% to 96.1% OEE in 18 months. The data to do it was already there. OxMaint was the platform that connected it to maintenance decisions.
Digital PM scheduling. Bearing vibration analytics. Roll campaign optimisation. Furnace condition monitoring. Hydraulic system analytics. Stores integration with predictive demand forecasting. Active from your first deployment phase.
