Why varnish removal is rarely built into turbomachinery bearing and lube-oil-system design

Executive thesis

OEMs do consider varnish, but usually not as a built-in, full-flow, base-design removal function in the bearing lubrication console. They tend to treat varnish as a lubricant chemistry, operating-duty, oil-analysis, and condition-monitoring problem, not the same kind of design contaminant as hard particles or water.

That design philosophy exists because particulates and water are mechanically separable contaminants with relatively direct design controls, while varnish is a dynamic chemical degradation phenomenon: soluble oxidation products, submicron soft contaminants, polar degradation species, additive byproducts, and deposited lacquer-like films move in and out of solution depending on oil chemistry, temperature, residence time, flow regime, electrical charge, contamination, and machine duty cycle.

In other words, a normal lube-oil console is designed to keep the machine alive by delivering the correct oil pressure, flow, temperature, cleanliness, emergency rundown capacity, cooling, and sometimes water control. Varnish control, by contrast, is usually handled through oil selection, sampling, laboratory testing, periodic flushing, side-stream conditioning, and aftermarket mitigation.

---

1. What the OEM lube-oil system is primarily designed to do

For turbomachinery, the lube-oil system is fundamentally a hydraulic-mechanical reliability system. Its base design normally includes:

• main and auxiliary oil pumps;
• emergency rundown or DC backup pump logic;
• reservoir with retention volume;
• coolers;
• duplex particulate filters;
• pressure and temperature control;
• instrumentation and shutdown protection;
• sometimes oil mist eliminators, heaters, centrifuge connections, coalescers, or vacuum dehydrator connections.

The design target is to guarantee that journal bearings, thrust bearings, gears, couplings, seals, and hydraulic/control components receive oil at the correct flow, pressure, temperature, and particulate cleanliness during start-up, operation, trips, rundown, and standby.

This design philosophy is visible in API-style lube-oil-system standards. The IOGP/API 614 supplement describes its scope as covering minimum requirements for lubrication systems, oil-control systems, and auxiliaries for special-purpose critical-service equipment; it also treats system configurations and purchaser data sheets as part of procurement definition rather than assuming every possible conditioning technology is standard. (IOGP)

That distinction matters. The OEM base package is normally a minimum safe and reliable lubrication platform. Anything beyond that—such as varnish mitigation, electrostatic oil cleaning, ion-exchange resin, depth-media adsorption, or offline chemical cleaning—is often treated as an optional purchaser-specified or site-specific reliability enhancement.

---

2. Why particle filtration is built in, but varnish removal usually is not

2.1 Hard particles are a classical design contaminant

Particulate contamination is comparatively easy to define, measure, and design against. Particle cleanliness is commonly monitored using cleanliness codes such as ISO 4406, which counts particles at sizes such as ≥4 µm, ≥6 µm, and ≥14 µm. Turbine and hydraulic oils are among the systems that require high cleanliness because of tight component clearances, high shaft speed, oil-film sensitivity, and servo/control-valve tolerances. (STLE)

This makes particulate filtration an obvious OEM design item. A filter can be sized for:

• rated flow;
• differential pressure;
• collapse pressure;
• beta ratio;
• micron rating;
• dirt-holding capacity;
• bypass setting;
• duplex transfer operation;
• maintainability.

Industrial lube-oil filters are therefore largely specified as mechanical particle-control devices. API 614-type systems commonly use filters to protect bearings, compressors, turbines, and hydraulic systems, with particle-removal ratings in the 10 µm range and below often used in rotating equipment. (Bollfilter)

2.2 Varnish precursors are not equivalent to hard particles

Varnish does not begin as a simple particle problem. It usually begins as oil degradation chemistry: oxidation, thermal cracking, additive depletion, insoluble soft-contaminant formation, polar degradation products, and low-solubility byproducts.

Chevron describes varnish-related symptoms as including valve sticking, bearing overheating, reduced heat transfer, filter problems, erratic operation, and elevated membrane patch colorimetry values. But it also emphasizes a crucial limitation: oil-analysis tests may indicate varnish potential or current oil performance, yet no single test directly proves the amount of deposit actually sitting on internal machine surfaces. Samples can also be unrepresentative because varnish precursors can be oil-soluble, very small, and sensitive to temperature and sample handling. (Texaco Lubricants)

That is the core reason why a normal inline particulate filter cannot be assumed to solve varnish. Varnish precursors may be:

• dissolved in the oil;
• submicron and deformable;
• polar and sticky rather than hard;
• below the effective capture range of normal inline filters;
• thermally reversible in solubility;
• formed downstream of the filter;
• deposited preferentially in low-flow, cooler, or high-residence-time areas.

So, from a design standpoint, varnish is not just “smaller dirt.” It is a chemically generated, condition-dependent deposit mechanism.

---

3. Why steam turbines often receive water-control provisions, while varnish still remains optional

Water is also different from varnish. Water contamination can exist as free water, emulsified water, or dissolved water, and steam turbines are especially exposed to water-ingress mechanisms through gland sealing systems, condensation, cooler leakage, and operating transients. Water is harmful, but it is mechanically and physically separable using known technologies.

API-style specifications explicitly recognize oil-conditioning options such as filter-coalescers or vacuum dehydratorswhen a permanent or mobile oil conditioner is specified. The IOGP/API 614 supplement states that such oil conditioners, when specified, should be of filter-coalescer or vacuum-dehydrator type, with the justification that these are proven water-removal methods. (IOGP)

That is why water removal is easier to integrate into design logic:

• water has measurable concentration targets, commonly ppm by Karl Fischer or crackle-type screening;
• free and emulsified water can be separated by residence, coalescence, centrifugation, or vacuum dehydration;
• water ingress risk is strongly connected to machine type and plant layout;
• water removal does not normally require changing the chemical balance of the oil’s additive system.

Varnish, however, is not just an externally introduced contaminant. It is generated by the oil itself under stress. Therefore, the OEM cannot simply size a “varnish separator” the same way it sizes a water coalescer or particulate filter.

---

4. The chemistry problem: varnish is generated by oxidation, thermal stress, and additive behavior

GE’s technical information letter on lube-oil varnishing describes varnish formation as the result of a complex chain of chemical, mechanical, and thermal events. It identifies oxidation, elevated temperature, aeration, iron and copper catalytic effects, oil contamination, micro-dieseling, pressure-induced thermal degradation, and electrostatic discharge as contributors. (FocusLab)

This is why varnish is difficult to lock into the original mechanical design envelope. The same machine model can have very different varnish behavior depending on:

• base-oil group and solvency;
• antioxidant chemistry;
• top-up oil compatibility;
• oil age;
• reservoir residence time;
• filter type;
• oil conductivity;
• turbine firing temperature;
• peaking versus base-load operation;
• start-stop frequency;
• common versus separate hydraulic reservoirs;
• ambient environment;
• water level;
• particle catalytic metals;
• cooler performance;
• dead legs and low-flow zones.

GE also notes that Group II base stocks may have improved oxidation stability but lower ability to keep degradation products dissolved, making varnish formation possible once degradation begins. (FocusLab)

This is a major design-stage complication. A base lube-oil console designed around the machine’s bearing flow demand cannot know, with high certainty, the future oil chemistry, additive balance, duty cycle, contamination history, and thermal history over years of operation.

---

5. Temperature dependence makes varnish a moving target

Varnish potential is strongly affected by temperature. Oxidation products can remain dissolved at higher temperature, then come out of solution as the oil cools or sits in storage. A Tribology Online study on membrane patch colorimetry found that patch color changes depend on mixing solvent, oil temperature, and incubation period; it also reported that oxidation products become more soluble at higher temperature and less soluble after cooling or storage. (J-STAGE)

This creates a design and diagnostic problem. A sample taken hot, cold, immediately after shutdown, after incubation, after solvent mixing, or after storage may show different varnish potential. Likewise, a turbine operating at high temperature may keep more degradation products dissolved, while deposits may form in cooler bearing drains, tank walls, servo-valve clearances, filter housings, or low-flow regions.

That is why varnish cannot be treated like a fixed particulate burden. Its apparent concentration depends on the thermodynamic solubility state of the oil.

---

6. The measurement problem: no single varnish number is enough

ASTM D7843, the common membrane patch colorimetry method, extracts insoluble contaminants from in-service turbine oil onto a membrane and reports color using a spectrophotometer. ASTM describes the result as a condition-monitoring and trending tool, not a standalone absolute design criterion. (ASTM International | ASTM)

ASTM D4378, the broader turbine-oil monitoring practice, similarly treats in-service oil monitoring as necessary for reliable turbine operation, but stresses that test values are indicative and require interpretation according to equipment, workload, lube-circuit design, and top-up history. (iTeh Standards)

This is exactly why OEMs hesitate to make varnish removal a universal built-in design basis. There is no universally accepted equivalent of:

“Design the lube-oil system to remove X grams per hour of varnish precursor at Y microns.”

Instead, varnish is usually managed through a trend-based oil-analysis program, using tests such as:

• MPC / ASTM D7843;
• ultracentrifuge rating;
• RULER antioxidant depletion;
• RPVOT;
• FTIR oxidation;
• acid number;
• viscosity;
• Karl Fischer water;
• ISO 4406 particle count;
• ferrous and elemental analysis;
• visual inspection;
• filter debris analysis;
• servo-valve response data;
• bearing metal temperature trend;
• cooler approach-temperature trend.

Mobil’s turbine-oil testing guidance also cautions that MPC interpretation should be application- and oil-specific, confirmed by visual inspection, and selected at a frequency based on risk. It further notes that some antioxidant chemistries can bias MPC results. (Mobil)

So varnish risk cannot be reduced to a simple “filter size” or “micron rating” decision.

---

7. Soluble varnish versus insoluble varnish: why ordinary monitoring can miss risk

A major technical weakness in many varnish programs is that they focus only on insoluble varnish potential. However, varnish often begins as soluble degradation material. A Machinery Lubrication study on turbine oils from three power plants emphasized that monitoring only insoluble MPC can be misleading and argued for looking at both soluble and insoluble varnish potential. It also notes that the ASTM D4378/D7843 warning level of MPC ΔE 30 is used, but that varnish formation is a more complex process than one fixed threshold. (Machinery Lubrication)

This is highly relevant to your OEM-design question. A full-flow particulate filter is aimed at removing solid particles. A water-removal system is aimed at removing water. But soluble varnish precursors are still part of the oil phase. A standard filter cannot remove what is chemically dissolved.

That means a lube-oil system can meet its design cleanliness target and still be moving toward varnish deposition.

---

8. Varnish affects bearings, but the highest operational sensitivity is often in control systems

Varnish can affect bearings by reducing heat transfer, restricting oil flow, insulating surfaces, increasing bearing metal temperature, or contributing to oil starvation in small passages. It can also foul coolers and produce deposits in drains, tank surfaces, and filter housings.

However, in many gas and steam turbine incidents, the most sensitive components are hydraulic/control components, especially servo valves, trip valves, actuator valves, and small-clearance control-oil passages. GE states that varnish often accumulates in small, low-flow hydraulic passages and servo valves, potentially making valves sluggish or causing failure that can lead to a turbine trip. (FocusLab)

Mobil’s varnish article also highlights that gas turbines with a common hydraulic and bearing reservoir are more susceptible to varnish issues, while mild varnish on journal and thrust bearings may accumulate with little or no operational impact. It reports a survey of 192 gas-turbine power plants and 626 gas turbines in which about 40% of operators reported current or historical varnish issues within the first six years of operation. (Mobil)

This contributes to the OEM approach: varnish is recognized as serious, but its most damaging manifestation is often site-specific and duty-specific, not a universal bearing-design load case.

---

9. Why full-flow varnish removal is not the normal OEM solution

Most varnish-removal technologies are better suited to side-stream treatment than full-flow bearing lubrication. GE’s varnish bulletin discusses technologies such as electrostatic precipitation/collection and balanced charge agglomeration, noting that they differ from mechanical filters because they use induced electrical charge to make particles agglomerate or collect on plates. GE also describes these systems as side-stream systems connected to the existing lube-oil system and able to run online or offline. (FocusLab)

That side-stream placement is not accidental. Full-flow bearing-oil systems are safety-critical. Any treatment installed in the full-flow path must not jeopardize:

• bearing oil flow;
• trip and rundown lubrication;
• control-oil pressure;
• differential pressure margins;
• cold-start viscosity limits;
• filter bypass behavior;
• additive stability;
• anti-foam performance;
• demulsibility;
• air release;
• electrostatic behavior;
• maintainability during operation.

Some varnish-mitigation media can also create compatibility questions. Plant Engineering notes that OEM built-in filters may not be sufficient and that additional oil-treatment technologies may be needed, but also warns that fine filtration can remove foam inhibitors and that ion-exchange resins can remove soluble species, raising additive-depletion concerns. (Plant Engineering)

Therefore, OEMs often avoid embedding a specific varnish-removal technology into the base design because the “wrong” technology can create new risks.

---

10. Electrostatic discharge: a design paradox caused by modern clean oils and filters

A further complication is electrostatic discharge, or ESD. Highly refined Group II and Group III turbine oils can have lower electrical conductivity than older Group I oils. Fast-flowing dry oil through fine filter media can generate charge, and discharge events can create localized thermal degradation, filter damage, carbonaceous deposits, and varnish precursors.

GE notes that static discharge can occur in filters even though filter designs attempt to dissipate charge, and that conditions normally considered desirable—clean, dry oil with low moisture and low air—can still increase electrostatic sensitivity under some circumstances. (FocusLab)

Turbomachinery Magazine similarly explains that ESD risk is associated with fast-flowing dry oil through small areas, including lube-oil and last-chance filters, and that Group II and Group III oils may have conductivity below common risk thresholds. (turbomachinerymag.com)

This is another reason why “just add finer filtration” is not a complete answer. In some systems, aggressive filtration can reduce particles while worsening charge generation or removing additives. Varnish design is therefore not simply a matter of installing finer inline filters.

---

11. Why OEMs treat varnish as optional, even though they know it exists

GE’s lube-oil varnishing technical information letter is especially revealing. It applies to heavy-duty gas turbines and discusses varnish formation, effects, monitoring, and mitigation technologies, but it classifies the guidance as optional: beneficial to some operators but not necessarily all, with timing at customer discretion. (FocusLab)

That does not mean the OEM is unaware of varnish. It means the OEM is treating varnish control as a risk-based reliability enhancement, not a universal base-design requirement.

The reasons are mainly engineering and commercial:

1. Varnish is oil-chemistry dependent, not just equipment-geometry dependent.
   The same turbine model can perform differently with different oils, additive packages, top-up practices, operating temperatures, and cycling regimes.
2. Varnish is not directly measurable as an installed deposit load.
   MPC, ultracentrifuge, RULER, and oxidation tests are valuable, but they infer risk. They do not directly quantify deposit thickness on every bearing, servo valve, tank wall, or cooler surface. (Texaco Lubricants)
3. The technology choice is not universal.
   Electrostatic separation, balanced charge agglomeration, ion-exchange resin, depth adsorption, chemical cleaning, and flushing all have different strengths and side effects.
4. Side-stream treatment is safer than full-flow treatment.
   Bearing and control-oil flow should not depend on a chemistry-removal device that may load, saturate, create pressure drop, interact with additives, or require isolation.
5. The purchaser’s operation determines the risk.
   Base-load steam turbine duty, cycling combined-cycle duty, black-start standby duty, peaking gas turbine duty, and high-start-count operation create different varnish risks.
6. Standards define minimum lube-system reliability, not all degradation-control strategies.
   API 614-type packages are designed around lubrication and control-system integrity. Optional conditioning devices can be specified, but they are not automatically part of every base system. (IOGP)

---

12. Why this design gap matters

The absence of built-in varnish management can become expensive because the lube-oil system may remain “healthy” by traditional indicators while varnish risk rises.

A machine can show:

• acceptable ISO particle cleanliness;
• acceptable water content;
• acceptable viscosity;
• normal pump discharge pressure;
• normal filter differential pressure;

while simultaneously developing:

• servo-valve stiction;
• trip-valve sluggishness;
• increased bearing metal temperature;
• cooler fouling;
• tank-wall deposits;
• filter plugging by soft insolubles;
• reduced control-system repeatability;
• high MPC or ultracentrifuge values;
• antioxidant depletion;
• high varnish precursor loading.

Turbomachinery Magazine warns that deposits can reduce oil flow and cooling and may cause oil starvation, but also that MPC can be low even when deposits exist—for example after oil changes without flushing or after deposits have already left the oil phase. Conversely, high MPC can occur where deposits have not yet formed under operating conditions. (turbomachinerymag.com)

This is exactly why varnish is a reliability-management problem rather than a simple filter-design problem.

---

13. What a better design-stage approach should look like

The more defensible approach is not necessarily to make every turbine lube-oil system include a fixed varnish-removal skid. A better approach is to make the system varnish-ready at the design stage.

That means the OEM, EPC, and end user should define varnish provisions in the design basis, even if the removal equipment is installed later.

13.1 Mechanical provisions

A varnish-ready lube-oil system should include:

• side-stream supply and return nozzles on the reservoir;
• isolation valves for online connection of varnish-removal equipment;
• low-velocity reservoir return location to avoid aeration;
• sample points upstream and downstream of filters, coolers, bearings, hydraulic/control-oil headers, and side-stream conditioner;
• drain and flush connections at low points;
• adequate manways and inspection ports;
• provision for offline kidney-loop filtration;
• space, electrical supply, and controls integration for future varnish-mitigation skids;
• avoidance of dead legs and stagnant control-oil branches;
• filter housings designed with ESD risk in mind;
• oil-return layout that minimizes air entrainment;
• cooler design that avoids excessive oil hot spots.

13.2 Oil-analysis provisions

The condition-monitoring program should include, at minimum:

• MPC / ASTM D7843;
• soluble and insoluble varnish-potential testing where available;
• ultracentrifuge rating;
• RULER antioxidant remaining useful life;
• RPVOT;
• FTIR oxidation;
• acid number;
• viscosity at 40°C;
• Karl Fischer water;
• ISO 4406 particle count;
• elemental spectroscopy;
• ferrous density or particle quantifier;
• demulsibility;
• air release;
• foam tendency;
• conductivity where ESD risk exists.

ASTM D7843 should be treated as a trend tool, not an isolated pass/fail value. ASTM D4378 likewise frames turbine-oil monitoring as dependent on equipment, workload, circuit design, and top-up history. (ASTM International | ASTM)

13.3 Operating-data provisions

The varnish program should correlate oil analysis with machine data:

• bearing metal temperature;
• oil supply and drain temperature;
• cooler approach temperature;
• filter differential pressure;
• servo-valve current and hysteresis;
• trip-valve stroke time;
• actuator response;
• turbine start count;
• fired hours;
• trips and failed starts;
• reservoir temperature;
• oil top-up volume;
• offline duration;
• water excursions;
• oil changes or partial drain-and-fill events.

Varnish risk is best understood by combining laboratory chemistry with operating behavior.

---

14. Recommended specification language for new projects

For a new turbomachinery project, a stronger specification could say:

The lube-oil system shall be designed with permanent provisions for online or offline varnish mitigation by side-stream oil conditioning. The base system shall include valved supply and return connections, sample points, space allocation, electrical and controls allowance, and reservoir return design suitable for future installation of electrostatic, agglomeration, adsorption, or resin-based varnish-control equipment. The selected technology shall be validated for compatibility with the approved turbine oil, including antioxidant retention, anti-foam retention, demulsibility, air release, conductivity, and seal-material compatibility.

And:

The oil-analysis program shall include ASTM D7843 membrane patch colorimetry, soluble/insoluble varnish-potential assessment where available, RULER antioxidant monitoring, RPVOT, FTIR oxidation, acid number, viscosity, Karl Fischer water, ISO 4406 particle count, ultracentrifuge or equivalent insoluble-deposit assessment, and periodic visual inspection of filters, tank surfaces, servo components, and bearing drains.

This kind of language avoids forcing one technology into every machine while preventing the common problem of discovering, years later, that there is no good place to connect varnish-control equipment.

---

15. Final technical conclusion

Varnish removal is not usually built into turbomachinery bearing and lube-oil-system design because OEM lube systems are primarily engineered around lubricant delivery, cooling, pressure reliability, particulate cleanliness, and sometimes water removal. Those are physical design problems with measurable design parameters.

Varnish is different. It is a chemical degradation and deposition mechanism that depends on oil formulation, operating temperature, duty cycle, additive depletion, contamination, electrostatic behavior, sample handling, and machine-specific low-flow zones. It cannot be fully controlled by normal inline filters, and its best detection methods are trend-based rather than direct deposit measurements.

The correct criticism is not simply that OEMs “ignore” varnish. The more accurate criticism is that many OEM base lube-oil packages do not include varnish-ready design provisions unless the purchaser specifies them. Varnish is often pushed into the aftermarket reliability domain: oil analysis, technical bulletins, optional side-stream conditioning, and corrective flushing after symptoms appear.

For critical turbomachinery—especially gas turbines, combined-cycle units, high-start-count machines, common bearing/control-oil reservoirs, and steam turbines with aging oil systems—the design-stage philosophy should change from:

“Install particulate filters and water removal if required”

to:

“Design the lube-oil system for particulate control, water control, and future varnish-risk management.”

That does not mean every machine needs a permanent varnish-removal skid from day one. It means every critical machine should have the mechanical, sampling, monitoring, and operating-data provisions to detect and mitigate varnish before it becomes a bearing-temperature, servo-valve, cooler-fouling, or trip-reliability problem.