AI-Powered Predictive OEE for Aerospace Heat Treatment
By Grace on June 17, 2026
A vacuum furnace at an NADCAP-accredited aerospace heat treat operation ran for six months hitting every internal production target -- 97% schedule adherence, zero customer escapes, audit-ready paperwork. Then a routine capability study revealed something the OEE score had been hiding: the quench rate process was running at Cpk 0.94 on a key hardening characteristic for 4340 steel landing gear components. Mathematically, that meant approximately 2,700 parts per million were outside the acceptable hardness range -- roughly 80 suspect parts per month -- and the final hardness testing was simply too coarse to catch the marginal ones. The plant manager was winning the OEE game while losing the process capability game. Two weeks after switching from end-of-shift OEE reporting to predictive OEE with continuous Cpk tracking, the team identified a quench media temperature-correlated drift, adjusted the agitation setpoint, and pushed Cpk to 1.58. Defect rate dropped from 2,700 PPM to roughly 3 PPM -- a 900x improvement -- without a single new piece of furnace equipment. That is the difference predictive OEE makes. Not a better OEE score. A fundamentally smarter understanding of what the score is hiding.
1.67+
Sustained Cpk on critical characteristics when predictive OEE replaces end-of-shift reporting with real-time capability tracking per load
10-20
OEE points recovered within 90 days when predictive SPC and adaptive limits eliminate the gap between reported and actual process capability
45%
Rework and scrap cost reduction when predictive alerts catch nonconformances during the furnace cycle rather than at post-process inspection
55-70%
False alarm reduction when dynamic control limits replace static thresholds -- ensuring Cpk calculations reflect real process capability, not outdated baselines
Cp/Cpk + OEE + AS9100 Compliance
Your OEE Score Is Hiding a Capability Gap. Predictive OEE Surfaces It Before the Auditor Does.
iFactory's predictive OEE platform unifies live Cpk, real-time SPC, and automated AS9100 compliance evidence into a single plant manager dashboard. See it configured for your furnace types and alloy portfolio.
The OEE-Cpk Disconnect: Why Traditional OEE Cannot Surface Capability Loss
OEE is calculated as the product of three components: availability, performance, and quality. In most aerospace heat treat operations, the quality component is measured as first-pass yield -- the percentage of loads that pass inspection on the first attempt. This creates a structural blind spot. A process running at 91% OEE with 97% availability, 96% performance, and 98% quality yield appears to be performing well. But that 98% quality yield at final inspection can coexist with a Cpk of 0.94 -- because final inspection catches gross nonconformances while marginal drift goes undetected. The OEE score reports success. The process capability index reveals risk. Predictive OEE closes this gap by integrating live Cpk into the OEE calculation, so the plant manager sees not just how much time is being lost, but how much capability is being consumed.
Availability Blind Spot
Availability measures uptime. It does not measure whether the furnace was running at the correct parameters during that uptime. A furnace that completes a full 8-hour soak at 12 degrees Fahrenheit above setpoint due to zone drift is counted as available. It produced output. The OEE availability component shows no loss. But the Cpk for hardness on that load drops from 1.67 to 1.12, and the plant manager discovers the capability loss not during the cycle, but at the next TUS survey or customer complaint.
Predictive OEE fix: Live zone uniformity trend data feeds both availability and Cpk, surfacing parameter drift that traditional OEE treats as productive time.
Performance Blind Spot
Performance measures cycle time against ideal. It rewards shorter cycles. But when static control limits force supervisors to pad cycle times to avoid false alarms, the performance component already reflects a compromised baseline. The plant manager sees 89% performance and assumes the furnace is running near its capability. In reality, the cycle time has been inflated 20% above what the process actually needs. The Cpk at the inflated cycle time is 1.67. The Cpk at the true capable cycle time is unknown -- because nobody has run the furnace at real speed since the last capability study on a different alloy batch.
Predictive OEE fix: Adaptive limits enable true cycle time optimisation, and the Cpk at the new cycle time is tracked continuously to validate that capability is sustained.
Quality Blind Spot
Quality in conventional OEE is measured as first-pass yield. A load that passes inspection scores 100% quality. A load that fails and is reworked scores 0%. This binary measurement cannot distinguish between a process running at Cpk 1.67 and one running at Cpk 0.94 -- both produce loads that pass inspection until the marginal drift crosses the specification limit. The OEE quality component is a lagging indicator that only confirms what already happened. It provides no warning that capability is degrading. By the time the quality score drops, the nonconformances have already occurred and the audit record already shows the defect.
Predictive OEE fix: Live Cpk replaces binary pass-fail as the quality input to OEE, so capability degradation surfaces as a trend before it produces a nonconformance.
What Traditional OEE Shows
91%
Combined OEE score at shift end
97% availability x 96% performance x 98% quality yield = 91% OEE. The score says the plant is performing well. No signal of capability loss.
What Predictive OEE Reveals
0.94
Live Cpk on a critical hardening characteristic
2,700 PPM defect rate. The process is heading toward the specification limit. The OEE score did not flag it. The Cpk trend surfaced it 3 shifts before the first nonconformance would have occurred.
How Predictive OEE Lifts Process Capability to 1.67 and Above
Predictive OEE transforms OEE from a lagging score reported at shift end into a leading indicator that drives capability improvement during production. Three integrated capabilities -- real-time SPC with adaptive limits, ML-driven predictive alerts, and automated compliance documentation -- work together to lift Cpk from the 1.0 to 1.33 range into the sustained 1.67-plus territory that AS9100 and customer quality clauses demand.
Capability Builder 01
Real-Time SPC with Adaptive Limits
Real-time SPC generates control charts that update continuously during every furnace cycle -- not periodic charts that are reviewed after the fact. Adaptive control limits recalibrate UCL and LCL against a rolling baseline of current furnace data, so the Cpk calculation reflects the process as it is running, not the process as it was during a study conducted on a different alloy months ago. When Cpk drops below the 1.67 target during a production run, the system flags it immediately and links the event to the corrective action workflow. The plant manager sees the Cpk trend by alloy, recipe, and furnace -- live, per load, with the contributing parameter analysis that shows what is driving the capability change.
Cpk impact: Eliminates the structural gap between reported OEE quality and actual process capability by replacing binary first-pass yield with live Cpk as the quality metric.
Capability Builder 02
ML-Driven Predictive Alerts
The predictive ML model analyses current process parameters against historical patterns and generates alerts when a multivariate combination signals an impending nonconformance -- before the Cpk has crossed below the 1.33 minimum threshold. For the plant manager, this is the difference between discovering a capability problem during the next customer audit and intervening while the Cpk is still above minimum. The model detects drift trajectories that univariate SPC charts miss: the interaction between quench media age, zone temperature gradient, and load geometry that produces a hardness deviation pattern invisible in any single parameter chart. Predictive alerts include the specific parameter combination driving the risk and the recommended corrective action with expected Cpk improvement.
Cpk impact: Provides intervention lead time measured in hours, not inspection cycles, sustaining Cpk above 1.67 by catching drift before it reaches the specification limit.
Capability Builder 03
Automated Compliance Documentation
Every Cpk recalculation, every predictive alert, every corrective action, and every inspection result is logged automatically with the alloy code, AMS specification reference, furnace ID, and operator ID. The Cpk trend history is segmented by alloy and furnace automatically and exportable for any date range the auditor specifies. For the IA9100 transition audit, the system generates a gap analysis report showing how each new predictive quality management requirement is addressed by the existing deployment. The documentation trail demonstrates that process capability was monitored continuously, not reconstructed from paper logs when the auditor arrived. Audit preparation time drops from weeks of manual data compilation to a single export.
Cpk impact: Ensures that sustained Cpk above 1.67 is documented, auditable, and clause-mapped -- satisfying AS9100 Rev D and IA9100 requirements automatically.
The Plant Manager's Unified OEE-Cpk Dashboard
The unified dashboard combines OEE and Cpk on a single screen, giving the plant manager a complete performance-and-capability view that eliminates the blind spots in each metric when viewed alone. Three views address the questions that matter most: Is capability improving or declining? Where is OEE loss hiding capability risk? And is the audit evidence ready?
A
OEE + Cpk Composite View by Furnace
Each furnace displays its OEE score alongside the current Cpk for each monitored quality characteristic -- hardness range, case depth, quench rate uniformity. The composite view uses colour coding to flag four states: OEE above target and Cpk above 1.67 (green), OEE above target but Cpk below 1.67 (amber -- capability risk hidden by good OEE), OEE below target but Cpk above 1.67 (amber -- throughput opportunity), and both below target (red). Plant managers see instantly which furnaces are running well but building risk.
Action: Furnaces in amber with high OEE but low Cpk receive immediate capability investigation.
B
Cpk Trend by Alloy, Recipe, and Furnace
Process capability is calculated continuously for every key quality characteristic, segmented by alloy, recipe, and furnace. The Cpk trend line shows the current value, the 1.67 target line, and the 1.33 minimum threshold. When Cpk drops below 1.67, the dashboard flags it and provides the contributing parameter analysis. Plant managers investigate the root cause while the Cpk is still above the 1.33 minimum threshold rather than discovering the capability loss during the next customer audit. Trend data is available for any date range.
Action: Investigate any Cpk trend crossing below 1.67. Identify the parameter driver and intervene before it reaches 1.33.
C
Predictive Alert Log with Cpk Impact Projection
Every predictive alert includes the projected Cpk impact if the current drift trajectory continues unchecked. The plant manager sees not just that a parameter is drifting, but what the drift will cost in capability if not corrected. An alert that flags a quench media temperature rise of 8 degrees Fahrenheit includes the projection: Cpk on hardness will decline from 1.58 to 1.22 within 6 loads if no action is taken. This quantifies the intervention priority in terms of capability risk, not just process deviation. Alerts are ranked by Cpk impact severity.
Your OEE Score Says 91%. Your Cpk Says 0.94. Predictive OEE Closes the Gap Before the Next Audit.
iFactory's predictive OEE platform unifies live Cpk, real-time SPC with adaptive limits, ML-driven predictive alerts, and IA9100-compliant audit documentation -- so plant managers see capability and performance in one view, on every furnace, every load.
Our OEE was running at 86 percent. The quality yield component was 97 percent, so everyone assumed the process was under control. Then a customer requested Cpk data for a flight-critical hardening characteristic, and we discovered we were running at 1.08. The OEE score had been hiding the capability gap for six months because the quality yield only caught gross failures -- not the marginal drift that was consuming our capability margin. Predictive OEE changed our entire approach. Now we track Cpk per load, per furnace, per alloy. Within 90 days, every critical characteristic was above 1.67. The IA9100 transition audit was the first audit in five years where we had zero nonconformances. The difference was not working harder. It was seeing what the OEE score was hiding.
Process capability is the quality metric that actually determines whether an aerospace heat treat operation can sustain AS9100 and NADCAP compliance. OEE measures productivity. Cpk measures whether the output of that productivity meets specification. Running them separately -- as most operations still do -- creates a structural blind spot where a plant can report 91% OEE while a critical hardening characteristic drifts toward a Cpk of 0.94. Predictive OEE closes this gap by integrating live Cpk into the OEE calculation, replacing binary first-pass yield with continuous capability measurement that surfaces risk before it produces nonconformances.
The evidence from aerospace heat treatment operations that have deployed predictive OEE with real-time SPC, adaptive control limits, and ML-driven predictive alerts is consistent: 10 to 20 OEE points recovered within 90 days, 45 percent reduction in rework and scrap costs, 55 to 70 percent fewer false alarms, and sustained Cpk above 1.67 across alloy changes, recipe transitions, and furnace condition shifts. The incoming IA9100 revision, which mandates real-time SPC, predictive quality management, and comprehensive control plans as expected practices, makes predictive OEE not just a capability improvement strategy but a certification requirement in the making. Plant managers who deploy predictive OEE now will enter the IA9100 transition audit with documented evidence of continuous capability monitoring that exceeds the new standard's baseline.
iFactory's predictive OEE platform is designed for plant managers in aerospace heat treatment who need to lift and sustain Cpk above 1.67, eliminate the blind spots in traditional OEE, and arrive at every audit with complete, clause-mapped capability evidence. Book a Demo to see predictive OEE configured for your furnace types, alloy portfolio, and certification baseline, or talk to an expert about a free Cpk and OEE assessment for your heat treat operation.
Frequently Asked Questions
The incoming IA9100 revision, expected in late 2026, specifically mandates real-time statistical process control, measurement system analysis, design of experiments, and comprehensive control plans integrated into operational workflows. iFactory's predictive OEE platform addresses these requirements through real-time SPC with dynamic control limits that generate control charts updating continuously during every furnace cycle -- meeting the real-time SPC mandate. The predictive ML model analyses current process parameters against historical patterns and generates alerts when a multivariate combination signals an impending nonconformance, satisfying the predictive quality management expectation. Every data point, alert, corrective action, and control limit adjustment is logged with full audit trail and data integrity controls. For the IA9100 transition audit, the platform generates a gap analysis report showing how each new requirement is addressed by the existing deployment -- providing auditors with documented evidence of compliance before the formal transition deadline. Talk to an expert about configuring the IA9100 readiness report for your certification timeline.
Yes. The platform deploys in parallel with existing OEE systems during a 2-week shadow mode period. During shadow mode, predictive OEE generates live Cpk data, adaptive control limits, and predictive alerts alongside the existing OEE reporting without affecting any production decisions. The plant manager team compares the two data streams -- existing OEE versus predictive OEE with Cpk integration -- and validates that the predictive system surfaces capability gaps that the conventional OEE score misses. After the validation period, transition to primary operation is a configuration change, not a system replacement. The existing OEE data can still be retained for historical comparison if required. Book a Demo to see a side-by-side comparison of conventional OEE reporting versus predictive OEE with live Cpk integration.
For aerospace heat treatment, the minimum acceptable threshold is Cpk greater than or equal to 1.33 for general characteristics. Flight-critical and safety-of-flight characteristics demand Cpk greater than or equal to 1.67. World-class operations target Cpk of 2.0 or higher on key characteristics. These thresholds translate directly to defect rates: Cpk equals 1.0 yields approximately 2,700 PPM defects, Cpk equals 1.33 yields 63 PPM, Cpk equals 1.67 yields 0.57 PPM (near six-sigma), and Cpk equals 2.0 yields 0.002 PPM. The predictive OEE platform continuously calculates Cpk for every monitored characteristic against these targets and displays the current value, the trend direction, and the projected Cpk at current trajectory. When Cpk drops below the 1.67 target, the dashboard flags the contributing parameter analysis -- showing whether the cause is increased variation (Cp below target), process centering shift (Cpk below Cp), or both. The plant manager investigates and corrects while the Cpk is still above the 1.33 minimum. Talk to an expert about configuring Cpk targets for your specific alloy portfolio and customer requirements.
Your OEE Score Is Not Telling You the Full Story. Predictive OEE Surfaces What Is Hidden. Get a Free Cpk and OEE Assessment.
iFactory's predictive OEE platform for aerospace heat treatment plant managers -- live Cpk integrated into OEE, real-time SPC with adaptive limits, ML-driven predictive alerts with Cpk impact projection, and AS9100 and IA9100-compliant audit documentation generated automatically from every furnace cycle.
