Hot Gas Path Inspection & AI Borescope Analytics for Gas Turbines
By Johnson on July 2, 2026
A borescope inspection puts an experienced technician's eyes inside a hot gas path that would otherwise require a partial teardown to see, but the value of that inspection has always depended entirely on one person's judgment in the moment — sizing a pit by eye, deciding whether a coating loss area is within limits, remembering how this same blade looked three inspections ago. Two inspectors looking at the same footage can reach different conclusions, and a defect that was genuinely borderline at the last inspection is easy to miss as "unchanged" when there is no structured record to compare against. Maintenance managers running hot gas path programs on a fixed inspection calendar are increasingly turning to AI-assisted borescope image analysis to remove that variability, and many begin by choosing to book a demo to see how automated defect measurement compares against their last three inspection reports.
HOT GAS PATH INSPECTION & AI BORESCOPE ANALYTICS
Extend Inspection Intervals Without Extending Your Risk
iFactory analyzes borescope images and video for coating loss, oxidation, creep deformation, and foreign object damage — building a trended condition record between inspections instead of a one-time snapshot.
Defect detection accuracy demonstrated by deep learning models on borescope blade imagery
90%+
Confidence level achievable in remaining useful life estimates from trended condition data
10–21 days
Typical forced outage length when a single blade failure is caught too late
0
New field hardware required — analysis runs on existing borescope footage
How AI Borescope Analysis Works
From Raw Footage to a Trended Condition Record
A borescope inspection generates far more image and video data than any technician can manually annotate in detail during a limited outage window. AI analysis doesn't replace the inspector — it gives every frame the same consistent measurement standard, and keeps a structured record that survives long after the inspection report is filed.
1
Video and Image Ingestion
Borescope video and still images from the inspection are uploaded directly, tagged automatically by stage, blade position, and inspection date using the access point the footage was captured from.
2
Component Segmentation
A vision model isolates each individual blade or vane in the frame, distinguishing the component under inspection from background structure, lighting artifacts, and reflections that commonly confuse manual review.
3
Defect Detection and Sizing
Coating loss area, crack length, pitting, and erosion depth are measured against a consistent standard for every frame, removing the inspector-to-inspector variability that comes from sizing defects by eye under time pressure.
4
Trend Comparison Against Prior Inspections
Each finding is compared automatically against the same component's measurements from previous inspection cycles, turning "looks about the same" into a quantified progression rate per operating hour.
5
Remaining Useful Life Estimate
Physics-informed models combine the measured degradation trend with operating hour projections to produce a remaining useful life estimate with a confidence interval, supporting a data-backed replacement window decision.
What Gets Detected
The Defect Categories That Drive Most Hot Gas Path Decisions
Not every mark on a blade surface matters equally. These four categories account for the overwhelming majority of findings that actually change an inspection interval, repair scope, or replacement decision.
01
Thermal Barrier Coating Loss
Coating spallation exposes base metal to temperatures it was never designed to withstand directly, accelerating oxidation at a rate that compounds quickly once it starts. Measured coating loss area trended over successive inspections is one of the strongest predictors of remaining blade life.
02
Oxidation and Base Metal Distress
Surface oxidation patterns and discoloration indicate how much thermal stress a component has actually absorbed, often diverging significantly from what equivalent operating hours alone would predict for units running hotter duty cycles.
03
Creep Deformation
Elongation and shape distortion under sustained high temperature and centrifugal load develop slowly and are difficult to judge visually without a precise geometric reference from a prior inspection to compare against.
04
Cracking and Foreign Object Damage
Crack initiation and FOD nicks or dents require immediate disposition against OEM manual limits, and consistent length and depth measurement is what separates an acceptable finding from one requiring urgent removal.
RUN YOUR LAST INSPECTION THROUGH IT
Upload Your Last Borescope Footage and See the Measured Trend
Most maintenance teams are surprised by how much progression the trended data shows once prior inspection footage is measured against a consistent standard.
The gap between manual and AI-assisted borescope review isn't about replacing the inspector's expertise — it's about giving that expertise a consistent, comparable record to work from every single time.
Aspect
Manual Review Only
AI-Assisted Review
Defect sizing
Visual estimate, inspector-dependent
Consistent measurement standard every frame
Inspection-to-inspection comparison
Relies on inspector memory or notes
Automatic trend against prior measurements
Documentation
Written report, limited image reference
Structured record with quantified findings
Replacement decision confidence
Judgment call against generic limits
Remaining useful life with confidence interval
Maintenance Manager Perspective
Field Perspective
D
David O.
Maintenance Manager, 350 MW Simple-Cycle Facility
Our old process was three inspectors over three outages, each writing their own report with their own sense of what "acceptable wear" looked like. Once we had measured coating loss area tracked across those same three inspections, we realized one row of blades had been degrading faster than the others the whole time — nobody had connected the dots because the reports lived in three separate binders.
David O.Maintenance Manager, Simple-Cycle Facility
Impact on Planning
What a Trended Condition Record Changes Downstream
A trended, quantified borescope record doesn't just inform the current inspection — it feeds directly into overhaul scope planning, spare parts procurement timing, and OEM warranty documentation.
4–8 wks
Earlier Warning Before a Component Reaches Limit
90%+
RUL Confidence for Replacement Planning
Zero
New Field Hardware Required
Full History
Every Inspection Retained and Comparable
FAQ
Hot Gas Path Inspection & AI Borescope Analytics — Frequently Asked Questions
It works with footage from your existing borescope equipment. The platform analyzes video and still images after they are captured, rather than requiring a specific camera model or new field hardware, which means any inspection footage your team or contractor already produces can be uploaded and measured. Higher resolution footage generally improves measurement precision, but the core defect detection and trending functions work across the image quality most modern borescope inspection equipment already produces.
No, and it isn't designed to. Borescope inspection still requires a trained technician to physically access the hot gas path, position the probe correctly, and make judgment calls about hardware condition that go beyond image analysis, including catching findings that fall outside typical defect categories. What AI analysis adds is consistent measurement and automatic trending on top of that inspector's footage, reducing the variability between different inspectors and different inspection cycles rather than replacing the inspection itself.
A structured, quantified degradation trend across multiple inspection cycles gives you a data-backed argument for interval adjustment that a single snapshot inspection cannot provide, since it shows the actual measured progression rate for your specific unit rather than relying on a generic OEM assumption built for a worst-case operating profile. Any interval change still needs to stay within OEM-approved limits and your own risk tolerance, but the trend data makes that conversation with engineering and the OEM a much more concrete one.
Historical borescope footage from prior inspections can be uploaded and processed retroactively, which is one of the most valuable steps teams take when first adopting the platform, since it immediately establishes a multi-cycle trend instead of starting from a single data point. Many maintenance managers find it useful to run their last two or three inspection cycles through the analysis before their next scheduled outage, giving the upcoming inspection real trend context from day one rather than waiting years to build up history.
Remaining useful life estimates are presented with a confidence interval rather than a single number, reflecting the genuine uncertainty in projecting future degradation from historical trend data, and confidence typically improves as more inspection cycles are added to a given component's history. The estimate is meant to support your replacement planning decision alongside your own engineering judgment and OEM guidance, not substitute for it. If you want to see how this would look applied to your specific fleet's inspection history, the team can walk through it during a scheduled demo, or you can raise implementation questions through support.
COATING LOSS · OXIDATION · CREEP · FOD · REMAINING USEFUL LIFE
Turn Every Borescope Inspection Into Data You Can Trend, Not Just a Report You File
iFactory measures every inspection against a consistent standard and compares it to everything that came before — so hot gas path decisions are backed by a trend, not a memory of what the blade looked like last time.
