How to Know Turbine Oil Sampling Was Wrong, the Lab Made Mistakes, or the Diagnostic Conclusion Is Not Trustworthy

A practical guide for turbine oil reliability engineers

In turbine oil analysis, a wrong sample can create a wrong diagnosis, and a wrong diagnosis can create unnecessary shutdowns, wrong filtration decisions, wrong oil change decisions, or false confidence.

A turbine oil report should never be accepted only because it has numbers, a lab logo, and color-coded alarms.

A professional must ask:

“Do the results make physical, chemical, operational, and historical sense?”

If the answer is no, then either:

The sample was not representative.
The sampling method was wrong.
The bottle or sampling hardware was contaminated.
The lab made an analytical or reporting error.
The test method was not suitable for the fluid.
The diagnostic interpretation was weak.
The oil condition really changed, but the report needs confirmation.

This article explains how to detect these issues by looking carefully at turbine oil reports.

1. The Most Important Rule: One Report Is Not a Diagnosis

A single oil analysis report is only a snapshot.

A trustworthy diagnosis needs:

Historical trend
Correct sampling point
Operating condition at sampling
Machine status: running, standby, shutdown, recently topped up, recently filtered
Oil temperature
Reservoir level
Filter changes
Recent maintenance
Oil type and age
Same lab or different lab?
Same method or different method?
Same bottle type and cleanliness?

For turbine oil, especially varnish diagnostics, trend is more powerful than one number.

A wrong sample may look like a serious oil failure.
A real failure may look normal if the sample was taken from the wrong place.

2. Signs That a Turbine Oil Report May Not Be Trustworthy

Before going test by test, look for general red flags:

A. Sudden impossible change

Example:

TAN last month: 0.08 mg KOH/g
TAN this month: 0.38 mg KOH/g
RULER phenol still 92%
RPVOT still 1,650 minutes
MPC still 8
No overheating, no contamination, no oil mixing

This combination is suspicious. TAN jumped severely, but oxidation indicators did not support it.

Possible causes:

Wrong sample
Wrong unit or sample ID
Lab titration issue
Contaminated bottle
Acidic cleaning residue
Mixed sample from another system

B. Results contradict each other

Example:

Water by Karl Fischer: 1,800 ppm
Appearance: clear and bright
Demulsibility: 40/40/0 in 10 minutes
No rust, no water drain, no cloudy oil

High water may be real, but the rest of the report does not support it. Confirmation is required.

C. Data does not match the oil type

Example:

Phosphate ester EHC fluid report shows zinc, calcium, magnesium as if it is engine oil.
Mineral turbine oil report shows resistivity alarm interpreted like phosphate ester FRF.
Turbine oil report includes TBN, soot, nitration, fuel dilution.

This suggests wrong test slate, wrong template, or weak diagnostic knowledge.

D. Missing critical information

A turbine oil report without sampling point, oil hours, oil type, operating condition, or previous trend is incomplete.

Without context, the report may be analytically correct but diagnostically weak.

3. Sampling Problems That Commonly Create False Results

Wrong sampling can affect almost every turbine oil test.

Common sampling mistakes

1. Sampling from the drain valve

Drain points often contain settled water, sludge, rust, sediment, and dead-zone oil.

This can falsely increase:

Water
Particle count
Iron
Copper
MPC
Insolubles
Acidic degradation products

2. Sampling after shutdown

When the turbine is stopped, oil cools down. Soluble varnish precursors can become less soluble and start depositing or forming soft insoluble material.

This can distort:

MPC
Particle count
Appearance
Water distribution
Patch color

3. Sampling from stagnant lines

A sample from a dead leg, gauge line, low-flow line, or external cooler bypass may not represent the reservoir or active oil.

4. Not flushing the sample valve

The first oil from a sample valve can contain old oil, debris, water, rust, or residue from previous sampling.

5. Using dirty bottles

Dirty or unsuitable bottles can affect:

Particle count
Water
TAN
Metals
MPC
FTIR

For MPC, transparent bottles exposed to light may also influence varnish-related interpretation. Amber or non-transparent bottles are preferred for varnish-sensitive sampling.

6. Sampling after filtration, not from the main system

Sampling immediately downstream of a filter can make the oil look cleaner than the reservoir and control system.

7. Sampling immediately after oil top-up

Fresh oil dilution can temporarily improve:

TAN
RULER
RPVOT
MPC
Color
Water

But this may hide the actual degradation state of the bulk system.

4. Lab Problems That Can Create False Diagnostics

Even with a good sample, lab errors can happen.

Common lab-related issues

1. Sample mix-up

The most dangerous issue. The report may belong to another machine.

2. Wrong test method

Example:

Using ASTM D974 instead of ASTM D664 for low-TAN turbine oil.
Reporting water by crackle test instead of Karl Fischer.
Using particle count method not suitable for dark, wet, or aerated samples.

3. Wrong dilution or preparation

Some tests need precise preparation. Errors affect FTIR, RULER, ICP, MPC, and viscosity.

4. Instrument calibration issue

A lab instrument out of calibration can create a trend shift across many customer samples.

5. Reporting unit error

Example:

Water reported as % instead of ppm
Viscosity reported at 40°C but interpreted as 100°C
Resistivity reported in ohm-cm but interpreted incorrectly
MPC reported as patch color but not ΔE or ΔL

6. Poor diagnostic comments

The numbers may be correct, but the interpretation may be wrong.

For example:

“MPC is high. Change oil immediately.”

This is incomplete. High MPC needs correlation with RULER, TAN, RPVOT, operating temperature, varnish symptoms, servo valve history, filtration condition, and deposit history.

5. How to Capture Sampling or Lab Mistakes by Looking at the Report

You need to use five checks.

Check 1: Trend logic

Ask:

Does this result follow the previous trend?

A slow-moving property should not jump without a reason.

Slow-moving properties:

Viscosity
TAN
RULER
RPVOT
MPC trend
FTIR oxidation
Additive metals
Base oil fingerprint

Fast-changing properties:

Water
Particle count
Appearance
Foam
Air release
Some wear metals after abnormal event

Check 2: Internal consistency

Ask:

Do the results support each other?

Example: If oxidation is severe, you may expect some combination of TAN increase, RULER decrease, RPVOT reduction, FTIR oxidation increase, MPC increase, color darkening, or deposit symptoms.

Not all must move together, but the story should make technical sense.

Check 3: Machine reality

Ask:

Does the report match what happened in the machine?

If the report says severe contamination but:

Filters are normal
No alarms
No water drain
No bearing temperature rise
No control valve issues
No oil color change
No maintenance event

Then confirm before acting.

Check 4: Method suitability

Ask:

Was the correct method used for this oil and this question?

For turbine oils:

TAN: ASTM D664 preferred for critical low-TAN oils.
MPC: ASTM D7843 for varnish potential.
RULER: ASTM D6971 for antioxidant monitoring.
RPVOT: ASTM D2272 for oxidation stability.
Water: Karl Fischer preferred for low-level water.
Particle count: ISO 4406 with proper sample preparation.
Demulsibility: ASTM D1401.
Foam: ASTM D892.
Air release: ASTM D3427.
Resistivity: especially critical for phosphate ester EHC fluids, often linked to ASTM D8323-style condition monitoring.

Check 5: Cross-test contradiction

Ask:

Which result is the outlier?

If nine results are stable and one result is abnormal, do not immediately blame the machine. Investigate the abnormal result.

6. Ten Practical Cases With Numbers

Case 1: False High Water Due to Sampling From Drain Point

Report result

Test	Previous	Current
Water, Karl Fischer	85 ppm	1,250 ppm
Appearance	Clear	Slightly hazy
ISO particle count	17/15/12	23/21/18
TAN	0.07	0.08
MPC	12	13
RULER phenol	78%	77%

Why suspicious?

Water and particles jumped together, but TAN, MPC, and RULER remained stable.

This suggests the bulk oil did not suddenly oxidize. The sample likely captured local water and sediment.

Possible cause

Sample taken from:

Reservoir drain
Bottom dead zone
Low-point drain
Poorly flushed valve

How to capture it from report

Look for:

Water jump + particle jump
No matching oxidation change
No change in TAN/RULER/MPC
No site report of water ingress

Correct action

Take confirmation samples:

Live zone sample while turbine is running
Upstream of filters
Reservoir mid-level sample
Drain sample separately for comparison

If drain sample is high but live sample is normal, the issue is localized reservoir bottom contamination, not full system water contamination.

Case 2: False High MPC Due to Cold Shutdown Sampling

Report result

Test	Previous hot-running sample	Current shutdown sample
Oil temperature at sampling	62°C	28°C
MPC ΔE	18	49
TAN	0.09	0.10
RULER amine	64%	63%
RPVOT	1,120 min	1,090 min
ISO code	18/16/13	21/19/16

Why suspicious?

MPC jumped from 18 to 49, but TAN, RULER, and RPVOT barely changed.

This is a classic warning that the sample condition changed, not necessarily the oil chemistry.

Possible cause

The sample was taken after shutdown and cooling. Varnish precursors that were soluble at operating temperature may have become less soluble during cooling and appeared as higher MPC/particles.

How to capture it from report

Check:

Oil temperature at sampling
Machine running or stopped?
Time since shutdown
MPC jump without oxidation trend support
Particle count increase at the same time

Correct action

Repeat MPC:

From same point
Same bottle type
Turbine running
Oil at normal operating temperature
Sample handled quickly and protected from light

Diagnosis should mention:

“MPC may be influenced by sampling temperature. Repeat hot-running sample required before making oil-change or filtration conclusions.”

Case 3: False Low MPC Because Sample Was Taken After Fine Filtration

Report result

Test	Main reservoir previous	Current downstream filter
MPC ΔE	36	9
RULER amine	42%	41%
TAN	0.13	0.13
RPVOT	680 min	670 min
Servo valve sticking	Yes	Still yes

Why suspicious?

MPC suddenly improved dramatically, but the machine still has varnish symptoms, and antioxidant/oxidation indicators did not improve.

Possible cause

Sample was taken downstream of a varnish removal unit, fine filter, or temporary kidney loop return point.

This sample represents cleaned oil at that location, not the total system condition.

How to capture it from report

Look for:

MPC improvement without corresponding system recovery
Same TAN/RULER/RPVOT
Sampling point changed
Report says “after filter,” “return line,” or “polishing unit outlet”

Correct action

Use multiple sample points:

Reservoir
Main lube supply header
Control oil/hydraulic supply
Filter inlet
Filter outlet

For reliability diagnosis, do not rely only on outlet samples. Outlet samples show equipment performance, not full system health.

Case 4: TAN Jump Not Supported by RULER, RPVOT, or FTIR

Report result

Test	Previous	Current
TAN by ASTM D664	0.08 mg KOH/g	0.31 mg KOH/g
RULER phenol	81%	80%
RPVOT	1,520 min	1,500 min
FTIR oxidation	4 Abs/cm	5 Abs/cm
MPC ΔE	11	12
Viscosity at 40°C	32.1 cSt	32.2 cSt

Why suspicious?

TAN increased almost four times, but oxidation, antioxidant depletion, viscosity, and MPC did not support that severe acid formation.

Possible causes

Lab titration error
Contaminated bottle
Wrong sample ID
Residual cleaning chemical
Wrong endpoint detection
Sample mixed with acidic fluid
Reporting decimal error: 0.031 reported as 0.31

How to capture it from report

A true TAN increase usually has supporting evidence, especially in aged turbine oils.

You should question it when:

TAN jumps sharply
RULER remains unchanged
RPVOT remains unchanged
FTIR oxidation remains stable
Viscosity remains stable
No varnish or deposit trend change

Correct action

Ask lab to:

Re-run TAN from same bottle
Confirm method used
Confirm sample ID
Confirm duplicate titration agreement
Test retained sample if available

Then send a fresh confirmation sample to same lab or second lab.

Case 5: RULER Result Impossible After No Oil Top-Up

Report result

Test	Previous	Current
RULER phenol	38%	86%
RULER amine	22%	79%
RPVOT	520 min	1,460 min
TAN	0.16	0.07
MPC ΔE	44	13
Oil top-up	None	None

Why suspicious?

This looks like the oil became almost new again.

Without major oil replacement, purification, antioxidant replenishment, or large top-up, this is not realistic.

Possible causes

Wrong sample from fresh oil drum
Wrong asset ID
Lab sample mix-up
Report copied from new oil reference
Large top-up not recorded
Sampling from a newly filled side reservoir

How to capture it from report

Look for:

RULER increases dramatically
RPVOT increases dramatically
TAN decreases sharply
MPC decreases sharply
No maintenance record explaining it

Correct action

Immediately verify:

Was oil changed?
Was oil topped up?
Was antioxidant additive added?
Was the sample taken from the correct machine?
Was the lab sample ID correct?

If no physical explanation exists, reject the report as diagnostically unreliable.

Case 6: Particle Count Very High but Metals and Patch Do Not Support It

Report result

Test	Previous	Current
ISO 4406	17/15/12	25/23/20
Iron by ICP	0.8 ppm	0.9 ppm
Copper by ICP	0.4 ppm	0.5 ppm
Silicon	1.5 ppm	1.6 ppm
Water	70 ppm	75 ppm
MPC ΔE	10	11
Membrane patch photo	Clean	Clean

Why suspicious?

The particle count is extremely high, but wear metals, dirt indicators, water, MPC, and patch appearance do not support massive contamination.

Possible causes

Air bubbles counted as particles
Poor sample agitation or excessive agitation
Dirty particle count bottle
Automatic particle counter interference
Sample not degassed
Wrong dilution
Lab instrument contamination

How to capture it from report

Question the result when:

ISO code jumps severely
ICP metals do not move
Silicon does not move
Water does not move
Patch photo does not show contamination
No filter differential pressure issue

Correct action

Request:

Repeat particle count after proper degassing
Microscopic patch examination
Pore blockage method if optical particle count is unreliable
Fresh sample in certified clean bottle

For turbine oil, particle count should be interpreted with patch morphology, filter history, and machine condition.

Case 7: Viscosity Shift Suggests Wrong Oil or Contamination

Report result

Test	Normal turbine oil	Current
Viscosity at 40°C	32.0 cSt	46.5 cSt
Viscosity at 100°C	5.4 cSt	6.8 cSt
TAN	0.08	0.09
FTIR oxidation	5	6
RULER	75%	74%
Water	90 ppm	95 ppm

Why suspicious?

Viscosity increased by about 45%, but oxidation did not increase. TAN and FTIR are stable.

This is not normal oxidation thickening. It suggests contamination or wrong oil.

Possible causes

Gear oil contamination
Hydraulic oil contamination
Wrong oil top-up
Sample bottle previously contained higher viscosity oil
Sample taken from wrong asset
Lab viscosity bath/reporting error

How to capture it from report

Use the logic:

Viscosity changed strongly
Oxidation did not change
TAN did not change
RULER did not change
No thermal degradation evidence

This points to mixing or identification issue.

Correct action

Ask for:

FTIR fingerprint comparison with new oil
Elemental additive comparison
Viscosity repeat
Check top-up records
Check oil transfer containers
Check whether same sample bottle was used for other oils

A turbine oil viscosity shift above ±5% should be taken seriously. Above ±10%, immediate investigation is required.

Case 8: High Silicon but No Particles, No Wear, No Water

Report result

Test	Previous	Current
Silicon	2 ppm	38 ppm
ISO 4406	17/15/12	17/15/12
Iron	0.8 ppm	0.9 ppm
Aluminum	0.2 ppm	0.2 ppm
Water	80 ppm	82 ppm
Appearance	Clear	Clear

Why suspicious?

If silicon came from dirt ingress, you would often expect particle count increase and possibly aluminum or other dust-related indicators.

Silicon increased alone.

Possible causes

Silicone sealant contamination
Anti-foam additive carryover
Bottle contamination
Lab ICP interference
Wrong sample
Maintenance activity using silicone-based product

How to capture it from report

Question dirt diagnosis if:

Silicon high alone
Particle count stable
Aluminum stable
Iron stable
No filter issue
No air intake or breather failure evidence

Correct action

Do not immediately diagnose “dust ingress.”

Ask:

Was silicone sealant used?
Was anti-foam additive added?
Was maintenance recently done?
Was the bottle clean?
Did the lab re-run ICP?

Silicon interpretation must be contextual. Silicon is not always dirt.

Case 9: Demulsibility Failure Not Matching Water and Field Condition

Report result

Test	Previous	Current
Demulsibility ASTM D1401	40/40/0 in 20 min	35/35/10 after 60 min
Water	65 ppm	70 ppm
TAN	0.09	0.09
MPC ΔE	14	15
Foam tendency	Normal	Normal
Appearance	Clear	Clear

Why suspicious?

Demulsibility suddenly failed, but water is low, oil is clear, and other degradation indicators are stable.

Demulsibility can fail due to real polar contamination, oxidation, additive depletion, or contamination, but isolated failure should be confirmed.

Possible causes

Lab glassware not clean
Wrong water purity
Poor temperature control
Emulsion created by sample handling
Wrong oil tested
Contamination in test cylinder
Sample from stagnant area with detergent contamination

How to capture it from report

Look for:

Demulsibility failure without water increase
No TAN increase
No MPC increase
No foam issue
No appearance issue
No operational water separation complaint

Correct action

Repeat D1401 and add:

FTIR for polar contamination
Water by Karl Fischer
MPC
Visual bottle inspection
Check recent chemical cleaning, flushing, steam leak, cooler leak, or detergent contamination

For steam turbines, do not ignore demulsibility failure, but do not make a major oil change decision from one isolated abnormal result.

Case 10: Resistivity Result in EHC Fluid Contradicts Acid and Water Data

Report result for phosphate ester EHC fluid

Test	Previous	Current
Resistivity	8.0 GΩ-cm	0.4 GΩ-cm
Acid number	0.06 mg KOH/g	0.07 mg KOH/g
Water	350 ppm	360 ppm
Chloride	Low	Low
Particle count	Stable	Stable
Servo valve issue	No	No

Why suspicious?

Resistivity collapsed dramatically, but acid number, water, chloride, particles, and servo valve performance did not support a severe fluid condition change.

In EHC systems, resistivity is critical because low resistivity can increase electrokinetic charging, current leakage tendency, servo valve sensitivity, and fluid degradation risk. But the result must still be technically consistent.

Possible causes

Wrong temperature correction
Contaminated test cell
Poor electrode cleaning
Sample contamination
Wrong fluid sample
Lab reporting unit error
Testing delay or exposure issue

How to capture it from report

Question the result when:

Resistivity changes by one order of magnitude
Acid number is stable
Water is stable
Ionic contamination indicators are stable
No operational symptoms exist
No maintenance/top-up event explains it

Correct action

Repeat resistivity with:

Correct temperature control
Clean test cell
Same method
Fresh sample
Confirmation of fluid type

For EHC phosphate ester fluids, resistivity should never be interpreted alone. It must be linked with acid number, water, chloride, particulate, varnish/deposit tendency, and servo valve behavior.

7. More Examples of Report Contradictions

A. High wear metals but normal particle count

Example:

Iron: 45 ppm
ISO code: 16/14/11
Patch: clean

Possible reasons:

Dissolved/very fine metals below optical particle range
Lab ICP contamination
Wrong sample
Recent corrosion products
Particle counter missed dark/soft particles

Action: ferrography or analytical patch.

B. High copper with no machine reason

Example:

Copper increased from 0.4 ppm to 28 ppm
Iron unchanged
TAN unchanged
No cooler leak
No bearing issue

Possible reasons:

Copper-containing sampling valve corrosion
Brass fitting contamination
Lab contamination
Wrong asset
Copper cooler issue, if applicable

Action: check sample valve metallurgy and cooler condition.

C. FTIR oxidation high but TAN/RULER/RPVOT normal

Example:

FTIR oxidation: 35 Abs/cm
TAN: 0.08
RULER: 88%
RPVOT: 1,700 min

Possible reasons:

Wrong FTIR baseline
Wrong new oil reference
Ester-containing oil misinterpreted
Additive peak misread as oxidation
Contamination

Action: compare with correct new oil reference.

D. RPVOT drops but RULER remains high

Example:

RPVOT: 1,600 min to 480 min
RULER phenol: 82%
RULER amine: 78%
TAN stable
MPC stable

Possible reasons:

RPVOT test variability
Wrong sample
Catalytic contamination not captured
Lab issue
Different test lab/method practice

Action: repeat RPVOT and correlate with RULER and trend.

8. A Practical Checklist for Reviewing Turbine Oil Reports

Use this before accepting any diagnosis.

Step 1: Confirm identity

Correct plant?
Correct turbine?
Correct reservoir?
Correct oil type?
Correct sample date?
Correct sample point?
Correct running condition?

Step 2: Confirm sampling condition

Running or shutdown?
Oil temperature?
Before or after filter?
Before or after varnish removal unit?
Drain or live zone?
Was valve flushed?
Bottle type?
Time from sampling to lab?

Step 3: Compare with history

Ask:

Is the change gradual or sudden?
Is the direction logical?
Did all related tests move together?
Did the alarm appear only after lab change?

Step 4: Compare related tests

Oxidation health

TAN
RULER
RPVOT
FTIR oxidation
Viscosity
MPC
Color

Water health

Karl Fischer water
Appearance
Demulsibility
Rust
Particle count
Reservoir drain condition

Contamination

ISO 4406
Silicon
Aluminum
Sodium
Potassium
Water
Patch photo

Varnish risk

MPC
RULER
TAN
RPVOT
FTIR
Operating temperature
Servo/control valve history
Bearing temperature
Filter changes

Wear

Iron
Copper
Tin
Lead
Chromium
Nickel
Particle count
Ferrography
Patch morphology

Step 5: Challenge isolated abnormalities

One abnormal result does not always mean one real failure.

Ask:

“Which other result confirms this?”

If no result confirms it, repeat the test.

9. The Report Review Matrix

When to trust the report

You can have higher confidence when:

Trend is consistent
Related tests support each other
Sampling point is correct
Sample was taken during normal operation
Same lab and same methods were used
Field symptoms match the report
No sudden unexplained jump exists
Lab comments are technically specific

When to question the report

Question the report when:

One result changes dramatically alone
Report comments are generic
Sampling point changed
Machine was shutdown during sampling
Sample was from drain
Lab changed
Method changed
Units are unclear
Oil type is wrong
Report includes irrelevant tests
Diagnosis ignores turbine oil chemistry

10. What a Good Lab Comment Should Look Like

Weak comment:

“MPC high. Oil is bad. Replace oil.”

Better comment:

“MPC increased from 18 to 42. TAN and RULER remained stable, but sample was taken after shutdown at 30°C compared with previous hot-running samples at 62°C. Result may be influenced by reduced varnish solubility during cooling. Repeat hot-running sample from the main supply header is recommended before deciding on oil change or varnish removal strategy.”

Weak comment:

“Water high. Use dehydration.”

Better comment:

“Water increased from 90 ppm to 1,250 ppm together with ISO code increase from 17/15/12 to 23/21/18. TAN, RULER, and MPC were stable. Since the sample was taken from the reservoir drain, localized bottom water/sediment is possible. Confirm with live-zone sample during operation before diagnosing bulk oil water contamination.”

11. How to Handle Suspicious Reports Professionally

Do not accuse the lab immediately.

Use this wording:

“The result is not fully consistent with the historical trend and related parameters. Before making a maintenance decision, we recommend confirmation by repeat testing and fresh sampling from the correct live-zone sampling point under normal operating temperature.”

Or:

“This may be either a real abnormal condition or a sampling/testing artifact. The current data is insufficient for a reliable diagnosis.”

This is professional and technically defensible.

12. Recommended Confirmation Strategy

When a turbine oil report looks suspicious, do this:

A. Re-test from the same bottle

Useful for checking lab repeatability.

B. Take a fresh sample from the same point

Useful for checking sample handling or bottle contamination.

C. Take a sample from a second point

Useful for checking whether contamination is local or system-wide.

D. Send split samples to two labs

Useful when the result may drive major cost or shutdown decisions.

E. Add confirmatory tests

Examples:

High MPC → repeat MPC + patch photo + RULER + TAN
High water → Karl Fischer + appearance + demulsibility
High particles → repeat ISO + patch microscopy
High TAN → repeat D664 + FTIR + RULER
High metals → ICP repeat + ferrography + patch analysis
Low resistivity → repeat with temperature control + acid number + water + chloride

13. Practical Rules of Thumb

Rule 1: One bad number is a question, not a conclusion.

Rule 2: A trend break needs a reason.

Rule 3: Drain samples are not representative of circulating oil.

Rule 4: Shutdown samples can distort varnish and particle interpretation.

Rule 5: Sampling point changes can look like oil condition changes.

Rule 6: New oil top-up can temporarily hide degradation.

Rule 7: Outlet samples show filter performance, not total system health.

Rule 8: High MPC without context can be misdiagnosed.

Rule 9: TAN, RULER, RPVOT, FTIR, and MPC should be interpreted together.

Rule 10: The lab gives data; the reliability engineer must test the story.

14. Final Message

Turbine oil analysis is not only about reading numbers. It is about checking whether the numbers tell a technically possible story.

A good turbine oil diagnostician behaves like an investigator:

Is the sample representative?
Is the lab method correct?
Is the trend logical?
Are related tests confirming each other?
Does the machine condition match the report?
Is the diagnostic conclusion proportional to the evidence?

The most dangerous reports are not always the ones with bad results.

The most dangerous reports are the ones with bad results that are accepted without challenge.