Inside a Steam Turbine Overhaul: How NDT and Oil Discipline Cut Failures and Downtime

A steam turbine lives or dies by its inspection discipline. One plant cut an oil leak by ~97.3% — from 2,826 mL/hr to 75 mL/hr — simply by replacing worn oil deflectors and regapping steam seals during a major overhaul, which in turn lowered lube top‑ups and safety risk (jurnal.syntaxliterate.co.id). The same mindset — risk‑based inspections that zero in on high‑stress zones like blade roots and bearing journals — is showing up across the sector, where an annual or biennial teardown typically includes end‑to‑end blade, casing, seal, and bearing checks, oil flushing, and alignment verification (worldwidepower.cn) (worldwidepower.cn).

The payoff is material. One combined‑cycle site that pivoted to phased inspections and predictive monitoring cut annual downtime by ~20% (worldwidepower.cn). Targeted blade repairs have shortened outages by ~25% and trimmed maintenance costs by up to 60% (asian-power.com).

Major overhaul, step by step

Periodic major overhauls — generally after defined operating hours — strip the unit for cleaning and a full examination. All blades, rotors, and stationary diaphragms come out for deposit removal and visual checks for corrosion, erosion, and rubs; borescopes reach internal rows and casing pockets (worldwidepower.cn). Tip and seal clearances are measured; trailing edges get special scrutiny for erosion; any blade exceeding design wear limits is replaced or repaired. Allied Power warns that tiny surface cracks can escalate quickly — a “small” fix in the tens of thousands can otherwise become a $1,000,000 catastrophe (alliedpg.com).

Rotors and discs go through non‑destructive evaluation (NDE/NDT) — often phased‑array ultrasonic scans at shaft and fir‑tree attachments — plus shaft runout and balance checks. Gland packing seals, oil deflectors, thrust collars, and bearings are renewed or adjusted to OEM tolerances, with clearances reset. One maintenance guideline underscores the fundamentals: “the turbine should not be rotated without lubrication,” due to severe contact damage under dry operation (jurnal.untirta.ac.id). Lube pumps, filters, coolers, reservoirs, and system cleanliness are verified; teams flush or replace oil filters and strainers, then refill with turbine‑grade oil.

Reassembly brings alignment verification (including alpha‑beta stands), no‑load spin or pressure‑decay tests, match‑marking, and vibration/clearance confirmation. The Indonesian case that cut leaks by ~97.3% after replacing oil deflectors and regapping seals highlights how such corrective actions also reduce lube top‑ups and safety risk (jurnal.syntaxliterate.co.id).

NDT methods and defect detection thresholds

NDT (non‑destructive testing) finds cracks, fatigue, and corrosion without sacrificing parts. Fluorescent or dye penetrant inspection (FPI/DPI) is the low‑cost surface screen for blade airfoils, disc fillets, welds, and nozzle plates; with proper prep and lighting it can reveal cracks about 0.1 mm wide (alliedpg.com) (alliedpg.com). It is especially useful for tight fatigue cracks at platform edges or shrouds that a visual might miss.

Magnetic particle inspection (MPI) — for ferromagnetic parts like steel rotors, hubs, and journals — highlights hairline surface or near‑surface cracks at ~0.05 mm sensitivity via magnetization and iron particles (alliedpg.com). MPI is common on blade roots and rotor fillets; executed well, it can find defects other methods overlook (alliedpg.com). Demagnetization closes the loop.

Eddy‑current testing (ECT) induces currents in conductive alloys (nickel, stainless) to detect surface and slightly subsurface flaws, typically to ~6 mm depth, with sensitivity on the order of 0.1–0.3 mm crack length; phased‑array scanners can sweep blade surfaces quickly (alliedpg.com) (researchgate.net).

Ultrasonic testing (UT) — including phased‑array and TOFD (time‑of‑flight diffraction) — reaches deep internal flaws across blade and rotor cross‑sections; table data indicate around 0.5 mm through‑thickness crack sensitivity (alliedpg.com). Skilled technicians use UT to size internal defects in thick sections and decide repair versus replacement (alliedpg.com).

High‑resolution visual and borescopes accelerate discovery of obvious defects — chipped blades, broken bolts, loose vanes — in hard‑to‑reach zones. In practice, teams combine UT for roots and thick sections, ECT for surface/near‑surface cracks, and DPI for fine surface flaws to approximate complete defect coverage; early crack fixes cost “tens of thousands” versus “millions” for skipped failures (alliedpg.com) (alliedpg.com) (alliedpg.com). Allied Power cites analogous wind‑sector data: ~70% of severe blade damage originates from undetected cracks — a caution for steam turbine operators (alliedpg.com) (alliedpg.com).

Lubrication system design and monitoring

A turbine’s bearings ride on a continuous oil film that must remove heat and prevent metal‑to‑metal contact. Lubrication problems — contamination, degradation, or misapplication — account for about 50–80% of bearing failures (evolution.skf.com) (machinerylubrication.com). Oils are typically ISO VG32–68 (ISO viscosity grade) synthetic or premium mineral turbine oils, formulated for hot bearings (~100°C), with high thermal/oxidation resistance, rust and corrosion control, foam resistance, and easy water separation; high viscosity index helps preserve lubricity across operating temperatures (turbomachinerymag.com) (maintenanceandengineering.com).

Filtration and purification are continuous. Plants target ISO 4406 cleanliness codes (an oil contamination standard) and use online filters (Beta ≥200, a filter efficiency index) and offline centrifuges/dehydrators to remove particulates, varnish, and water. High‑pressure filters often sit in steel filter housings for industrial service. Portable purifiers handle episodic ingress. In one Indonesian case, engineers added a purifier after water entry and installed a “steam block” seal to stop high‑pressure steam leaking into VG46 turbine oil; together with proper mesh selection and checks, these changes prevented water/oil mixing (jurnal.untirta.ac.id).

Condition monitoring centers on TAN (total acid number) and water percent, with guidance to top up or change oil as TAN approaches ~0.3 mg KOH/g and water ~0.20% by volume; particle counts, wear metals, debris analysis, and dielectric strength round out the picture (turbomachinerymag.com). A well‑run program trends these data and acts before damage occurs; one plant (Banten‑3) explicitly monitors “tribology” data to assess purifier/filter performance under a condition‑based maintenance program (CBM) (jurnal.untirta.ac.id).

System design matters. Lube skids typically use main/standby/emergency pumps, a reservoir with degassing tray and level sensors, high‑pressure filtration, and a water‑cooled heat exchanger. Oil coolers are usually sized to keep oil near ~48°C under load, and the oil‑side pressure should always exceed water‑side pressure to prevent coolant ingress (plant.ca). Maintenance guidance is explicit: lubrication systems should be inspected weekly and overhauled annually, or sooner if contamination is detected (worldwidepower.cn).

The outcomes are hard to ignore. SKF estimates poor lubrication or contamination “accounts for around half of all premature rolling bearing failures,” and getting lubrication right can cut failures by roughly 50% (evolution.skf.com) (machinerylubrication.com). In the Sumsel‑5 case, oil leak reduction from 2.8 L/hr to 0.075 L/hr (~97%) also lowered fire hazard and oil wastage (jurnal.syntaxliterate.co.id).

Data points that shape maintenance plans

Inspection‑led overhauls that embed targeted repairs and NDT have reduced outages by ~20–25% and saved “tens of percent” in maintenance costs (asian-power.com) (worldwidepower.cn). NDT detection thresholds sit on the order of 0.05–0.5 mm — MPI around 0.05 mm; DPI/FPI around 0.1 mm; ECT around 0.1–0.3 mm; UT around 0.5 mm through‑thickness — per table and method data (alliedpg.com). Meanwhile, lubrication management addresses a failure mode responsible for ~50% of bearing problems (evolution.skf.com). The earlier overhaul example — cutting a 2.8 L/hr leak to 0.075 L/hr — underscores both reliability and cost gains (jurnal.syntaxliterate.co.id).

In short, a thorough overhaul backed by historical failure data, systematic NDT (FPI/DPI, MPI, ECT, UT/TOFD), and vigilant oil care (TAN, water %, particle counts) is the cornerstone of turbine reliability — the combination that turns “millions” into “tens of thousands” (alliedpg.com) (turbomachinerymag.com) (plant.ca).