A gas turbine driving a compressor at a midstream station or a refinery tells you it's in trouble long before it trips — if you're reading the right signals against the right baseline. The hot section runs above 1,400°C, where thermal-barrier coatings degrade, blades creep, and first-stage nozzles oxidize; the axial compressor fouls a fraction of a percent at a time until surge margin erodes; and the bearings that carry the rotor spall quietly until vibration spikes. None of it is invisible. A widening exhaust-gas-temperature spread flags combustion or hot-gas-path distress weeks before turbine inlet temperature or vibration responds, a rising inlet-to-discharge pressure ratio at corrected flow reveals fouling, and bearing defect frequencies emerge long before seizure. Yet a single forced outage on one of these machines runs $500K to $2.5 million, and hot-section and combustion components alone account for 60 to 70 percent of gas turbine maintenance cost and most forced outages. The plants that stay ahead don't run to a calendar — they trend these signals continuously and let a thermodynamic model separate real degradation from ambient noise, so hot-gas-path intervals are extended on evidence, not guesswork. To see it on your fleet, book a demo.
OIL & GAS · GAS TURBINE & TURBOCOMPRESSOR RELIABILITY
Read the Hot Section, the Compressor, and the Bearings — Before the Trip.
EGT spread, compressor fouling, surge margin, and axial vibration each carry a machine's warning weeks ahead — but only against a corrected baseline that separates degradation from ambient swings. iFactory pairs predictive maintenance with a thermodynamic model to trend all three signal domains at once, so you catch hot-gas-path distress early and extend inspection intervals on evidence instead of a calendar.
4–12 wks
EGT-spread warning ahead of blade failure
$500K–2.5M
Cost of a single gas turbine forced outage
3–8%
Efficiency lost to undetected compressor fouling
60–70%
Of GT maintenance cost in hot-section and combustion parts
Why Calendar Maintenance Costs You Both Ways
Most gas turbine and turbocompressor maintenance still runs on fired-hours and start-count schedules designed decades ago — and that fixed cadence is wrong in both directions. It pulls a healthy machine offline for a hot-gas-path inspection it didn't need yet, consuming an outage slot and burning parts life that was still good; or it misses a developing fault that a signal was flagging weeks earlier, and the machine trips between intervals at the worst possible cost. For a reliability engineer, the frustration is that the data to do better already streams off every unit — thousands of points a second — while the maintenance plan ignores it in favor of the calendar.
The Calendar Doesn't Know the Duty Cycle
A turbine driving a refrigerant or pipeline compressor at a high continuous load factor accelerates hot-section degradation relative to the duty cycle the OEM interval assumed. A fixed schedule set for average service either over-inspects a lightly loaded unit or under-protects a hard-run one — the interval and the actual consumption of component life drift apart.
Absolute Alarms Miss the Subtle Shift
A machine can sit inside every absolute alarm limit while trending steadily toward failure. The shifts that precede most gas turbine failures are subtle drifts against a rolling baseline, not threshold breaches — so a program built only on fixed alarm limits stays silent until the degradation is advanced and the intervention is urgent.
Ambient Noise Hides the Signal
Raw EGT, power, and pressure readings swing with ambient temperature, inlet conditions, and load, so a genuine degradation trend is buried under operational noise. Without correcting performance to reference conditions, a real efficiency loss looks like a hot afternoon — and the fouling or hot-section distress underneath it goes unread.
One Technique Can't See It All
Vibration sees the rotor and bearings but never EGT spread; performance analytics see the hot gas path but never a bearing defect frequency; oil analysis sees the lube system but not blade coating loss. A single-technique program has structural blind spots, and the fault it can't see is the one that takes the machine down.
Recognized condition-monitoring frameworks treat these techniques as a layered set, not a menu — vibration for the rotor, oil analysis for the bearings and lube system, and performance analytics for the hot gas path, each specified per asset. Trending the combination against a corrected baseline is what catches the 70 to 90 percent of failures that a calendar and an alarm limit miss.
The Three Signal Domains a Turbine Speaks In
A gas turbine and its driven compressor announce trouble across three distinct signal domains, and each catches faults the others are blind to. A credible program reads all three continuously — which is exactly what makes the difference between a caught degradation and a forced outage.
THERMAL
EGT spread & hot-gas-path distress
Exhaust gas temperature spread — the variation between thermocouples around the exhaust annulus — is the most sensitive early indicator of combustion and hot-section trouble, often producing a detectable increase weeks before turbine inlet temperature or vibration respond. A widening spread points to combustion-can imbalance, a fuel nozzle starting to coke, or first-stage blade and nozzle distress; a roughly 15°C spread between thermocouples commonly precedes a combustor-can or nozzle issue, and AI trends the pattern to flag blade degradation 4 to 12 weeks before failure. Since the hot section drives most forced outages, this is the highest-value domain to read early.
PERFORMANCE
Fouling, surge margin & efficiency
Particulate, salt, and oil-mist deposits foul the axial compressor blades, disrupting airflow and raising compressor discharge temperature — a 3 to 8 percent efficiency loss that translates straight to wasted fuel. Trending the inlet-to-discharge pressure ratio at corrected flow reveals both fouling severity and eroding stage surge margin, giving a direct read on wash timing. Because a fouled compressor loses the head needed to overcome system resistance, tracking this margin is also the first line of defense against a surge event, and correlated output-and-efficiency loss at fixed firing temperature is the signature of hot-gas-path degradation.
MECHANICAL
Vibration, thrust & bearing health
Proximity probes measure shaft orbit, eccentricity, and thrust position, catching rub, oil whirl, and thrust-bearing overload on large rotors, while spectral analysis reads the fault frequencies directly: a rising 1X running-speed component points to rotor imbalance, often from blade fouling, and a 2X component to misalignment or a coupling problem. Envelope-detected high-frequency energy at a bearing defect frequency confirms an incipient spall, and rising wear-metal in oil corroborates it — with a bearing-related forced outage averaging over $800K, this domain protects the most expensive failure mode from turning catastrophic.
See All Three Domains on Your Turbines
Bring a unit that's near a hot-gas-path interval or showing a nagging EGT or vibration trend. iFactory engineers will show the corrected-performance model, the EGT-spread trending, and the vibration analytics working together — and how the composite health picture supports a defensible interval decision.
The Degradation Cascade — and Why It Compounds
The reason early detection matters so much on these machines is that the degradation mechanisms feed each other. Left unread, a small compressor fouling problem doesn't stay small — it drives a chain that accelerates hot-section wear and erodes the very margin that protects against catastrophic surge. Understanding the cascade is understanding why a corrected, continuous view pays for itself.
1
Fouling Raises Exhaust Temperature
Deposits on compressor blades disrupt airflow and raise compressor discharge temperature. A new turbine loses 2 to 3 percent of baseline output within the first 1,000 operating hours and up to 5 percent before the first major overhaul — much of it fouling that a corrected-performance trend catches while an online or offline wash can still restore 1 to 3 percent per cycle.
2
Higher EGT Forces More Fuel
A fouled compressor raises exhaust temperature, and the control system compensates by burning more fuel to hold firing temperature — which pushes the hot section harder. What began as a cleanable aerodynamic deposit becomes a thermal load on the most expensive components in the machine.
3
Hot-Section Wear Accelerates
The added thermal load speeds thermal-barrier-coating loss, creep, and tip-clearance growth on blades and nozzles running above 1,400°C. Combustion hardware distortion shifts the flame profile and forces the dry-low-NOx system into richer, less efficient operation — so combustion wear and hot-section wear compound together.
4
Surge Margin Erodes Toward Risk
Fouling, erosion, and growing tip clearances reduce the compressor's ability to overcome system head, shrinking surge margin. If margin erodes far enough, a surge event can drive hot combustion gas backward through the compressor, damaging seals, thrust bearings, and blades — the catastrophic endpoint the whole cascade builds toward.
Each mechanism feeds the next, which is why catching the first link — fouling — early is worth so much more than its own small efficiency number suggests. A wash scheduled on a corrected-performance trend doesn't just recover a few points of output; it breaks the cascade before it reaches the hot section and the surge line.
Thermodynamic AI: Correcting the Signal Before Trending It
The hardest part of turbine diagnostics isn't collecting data — it's knowing whether a change is real degradation or just a hot day and a heavy load. iFactory pairs predictive maintenance with a thermodynamic model, a hybrid of physics-based and data-driven analytics, so every signal is corrected to reference conditions before it's trended. That correction is what turns noisy raw data into an early, trustworthy warning.
01
Correct Performance to Reference Conditions
The thermodynamic model normalizes power, heat rate, EGT, and pressures for ambient temperature, inlet conditions, and load, so a genuine efficiency or fouling trend is separated from operational swings. A degradation that would hide under a hot afternoon in raw data stands out clearly once performance is corrected — the physics does the de-noising.
02
Trend Against a Rolling Baseline
Rather than waiting for absolute alarm limits, the model trends each corrected parameter against a rolling baseline for that unit, surfacing the subtle drifts that precede the large majority of failures. The combination of physics and data means the baseline updates reliably with minimal field tuning, staying accurate as the machine and conditions evolve.
03
Fuse the Domains Into One Health Picture
EGT spread, compressor performance, fuel-system parameters, and vibration are fused into a composite driver-health picture, so a developing combustion problem, fuel-nozzle fouling, or first-stage blade degradation is identified from the pattern across domains — not from any single sensor reading in isolation. The whole is more diagnostic than the parts.
04
Predict Component Life and Lead Time
Using operational history and real-time thermal profiles, the model predicts blade life and names the lead time to intervention — flagging hot-section distress weeks ahead so refurbishment happens during a planned outage rather than as an emergency replacement. The output is a date and a scope, not just an alarm.
This physics-plus-data approach is what makes the warnings both early and trustworthy. A purely data-driven model struggles to tell degradation from a change in operating point; a purely physics-based one is slow to adapt. The hybrid gives you the sensitivity to catch a subtle shift and the grounding to trust that the shift is real.
Extending Hot-Gas-Path Intervals — Safely
The payoff of continuous, corrected condition monitoring is that it lets you extend inspection intervals on evidence rather than run blindly to a fixed schedule. Hot-gas-path and combustion inspections are among the most expensive events in a turbine's life, and moving from time-based to condition-informed timing is where the largest reliability and cost gains live — provided it's done on a defensible basis.
THE BASIS
Condition Evidence, Not a Guess
Fired-hours and start-count triggers stay as the backstop, but a corrected-performance and EGT-spread history gives a real read on how much component life the machine has actually consumed. When the hot section is demonstrably healthy on the data, an interval can be extended with evidence behind the decision; when the data shows accelerated consumption, it's pulled in before a trip — the schedule follows the machine's true condition.
THE GUARDRAILS
Events and Starts Still Count
Interval extension is bounded by the events that consume margin: each emergency start, trip, or surge consumes a portion of the inspection interval, and those counts are tracked against the asset so the projected due date adjusts automatically. A known event like a compressor surge or fuel-nozzle alarm triggers an interim borescope regardless of the trend — the condition view extends intervals without ever overriding a hard safety trigger.
Done this way, interval extension isn't a gamble — it's the disciplined result of knowing the machine's real condition. The turbine that's genuinely healthy earns more time between outages, the one that's degrading gets caught before it trips, and every decision carries the corrected data trail to defend it to the OEM and the reliability committee.
Turbocompressor-Specific: Surge, Thrust, and Seals
When the gas turbine drives a process or pipeline compressor, the driven end brings its own failure modes that the same monitoring platform has to cover — because on a compressor string, the driver and the driven machine are one reliability problem, not two.
01
Surge Margin & Anti-Surge
A compressor operates a set margin away from its surge boundary, and fouling or erosion erodes that margin toward instability. Trending pressure ratio at corrected flow against the surge line gives an early read, so a wash or an anti-surge tuning happens before an event where hot gas could flow backward and damage the machine.
02
Axial Thrust Position
Proximity probes track thrust-bearing position and shaft axial float, catching thrust-bearing overload before it becomes a rub or seizure. Because a surge event imposes large transient axial loads, watching thrust position is both a routine health check and a defense against the damage a surge would otherwise do to the bearing.
03
Dry Gas Seal Health
Dry gas seals permit a planned, limited leakage, and monitoring seal-gas differential pressure and vent flow detects a developing seal problem from a change in that leakage pattern. A drifting seal signature flags degradation before a seal failure risks a release or an unplanned shutdown of the compressor.
04
One Composite String Health View
Driver and driven machine are trended together, so a vibration, thermal, or performance shift is read across the whole string. A problem that originates in the turbine and shows up at the compressor — or the reverse — is diagnosed as one connected picture rather than two disconnected monitoring systems missing the link.
What Changes for the Reliability Engineer
Pairing predictive maintenance with a thermodynamic model changes the reliability engineer's role from running a fixed schedule and reacting to trips, to managing each machine on its true, corrected condition — with the data trail to defend every call.
01
Weeks of Warning, Not a Trip
EGT-spread and vibration trends flag hot-section and bearing distress weeks ahead, so a refurbishment lands in a planned outage instead of erupting as a forced outage costing $500K to $2.5M. The worst events become scheduled work with parts and crew arranged.
02
Intervals Extended on Evidence
A demonstrably healthy hot section earns more time between expensive inspections, backed by a corrected-performance history that defends the decision to the OEM and the reliability committee — while a degrading unit gets pulled in before it fails. The schedule finally matches the machine.
03
Washes Timed to Real Fouling
Corrected-flow pressure-ratio trending schedules compressor washes when fouling actually warrants them, recovering efficiency and breaking the degradation cascade — instead of washing on a calendar that's either too early to matter or too late to help.
04
The Whole String, One Picture
Driver and driven compressor are managed as one health picture across thermal, performance, and mechanical domains, so no fault hides in the gap between two monitoring systems. The engineer sees the string the way it actually behaves — as a single connected machine.
Frequently Asked Questions
The questions reliability and rotating-equipment engineers ask most often when evaluating AI-driven gas turbine and turbocompressor monitoring.
Why is EGT spread such a valuable early signal?
Because it responds before the other indicators do. Exhaust gas temperature spread is the variation between thermocouples around the exhaust annulus, and it's among the most sensitive early indicators of combustion and hot-section trouble — often producing a detectable increase weeks before turbine inlet temperature or vibration signals respond. A widening spread points to a specific set of causes: combustion-can imbalance, a fuel nozzle starting to coke, or first-stage blade and nozzle distress, with a roughly 15°C spread commonly preceding a combustor-can or nozzle issue. Because the hot section drives most forced outages and 60 to 70 percent of maintenance cost, catching its distress this early — trended by AI to flag blade degradation weeks before failure — is one of the highest-value things a monitoring program can do. To see spread trending on your units,
book a demo.
What does the thermodynamic model add over standard vibration monitoring?
Vibration monitoring is essential but blind to a whole class of faults — it sees the rotor and bearings but never EGT spread, compressor fouling, or hot-gas-path efficiency loss. The thermodynamic model adds the performance domain and, crucially, corrects it. Raw EGT, power, and pressure readings swing with ambient temperature, inlet conditions, and load, so a real degradation trend hides under operational noise; the model normalizes performance to reference conditions so genuine fouling or hot-section distress stands out from a hot afternoon. It's a hybrid of physics and data, which gives it both the sensitivity to catch a subtle drift and the grounding to trust it's real. Combined with vibration and oil analysis, it completes the layered coverage that recognized condition-monitoring frameworks call for — each technique covering the others' blind spots.
Can we really extend hot-gas-path intervals without adding risk?
Yes, when it's done on condition evidence with the right guardrails. The fired-hours and start-count triggers stay as a backstop, but a corrected-performance and EGT-spread history shows how much component life the machine has actually consumed, so a demonstrably healthy hot section can earn an extended interval with real data behind the decision — while a unit showing accelerated consumption gets pulled in early. The guardrails matter: each emergency start, trip, or surge consumes a portion of the interval and is counted against the asset so the projected due date adjusts automatically, and a known event like a compressor surge or fuel-nozzle alarm triggers an interim borescope regardless of the trend. Done this way, extension isn't a gamble — it's the disciplined result of knowing the machine's true condition, with a data trail to defend the decision to the OEM.
How does this handle the driven compressor, not just the turbine?
On a compressor string the driver and driven machine are one reliability problem, so the platform trends them together. On the turbocompressor side that means watching surge margin by trending pressure ratio at corrected flow against the surge line, tracking axial thrust position through proximity probes to catch thrust-bearing overload, and monitoring dry gas seal health through seal-gas differential pressure and vent-flow patterns to detect a developing seal problem before it risks a release. These matter because a surge event can drive hot gas backward through the compressor and damage seals, thrust bearings, and blades, and eroding surge margin is often the upstream cause. By fusing the turbine and compressor signals into one composite string-health picture, a fault that originates in the driver and manifests at the driven end — or the reverse — is diagnosed as a single connected problem rather than missed in the gap between two separate systems.
Does this work on both heavy-duty industrial and aeroderivative units?
Yes. The three signal domains — thermal EGT spread, corrected performance, and mechanical vibration and thrust — apply across heavy-duty industrial and aeroderivative gas turbines alike, because the underlying degradation mechanisms are common to both: hot-section thermal distress, compressor fouling and surge-margin erosion, and bearing and rotor faults. The thermodynamic model is configured to the specific machine's design and reference performance, and the baselines are learned per unit from its own operational history, so the analytics fit the actual asset rather than a generic template. Aeroderivative units in particular respond well to the corrected-performance approach given their higher power density and sensitivity to hot-section condition. Whether the fleet is compressor-station aeroderivatives, refinery heavy-duty frames, or LNG-train drivers, the same layered, corrected methodology applies, tuned per asset class. Contact
iFactory support to discuss your specific fleet.
READ THE MACHINE · CORRECT THE SIGNAL · EXTEND ON EVIDENCE
Catch Hot-Section Distress Weeks Early — and Extend Intervals Safely.
EGT-spread trending, corrected-performance fouling and surge-margin analytics, and axial vibration and thrust monitoring fused into one composite health picture by a thermodynamic model — so you catch degradation before the trip, time washes to real fouling, and extend hot-gas-path intervals on evidence, not a calendar. Across heavy-duty and aeroderivative fleets, driver and driven compressor as one string.
