Xinysun Instrument (Shanghai) Co., Ltd.

Applications

Breadcrumb: Home > Applications > Fluorescent Leak Testing

How to Make Early Leak Detection Effective: Cost Curves, Interpretation Thresholds, and Inspection Protocols

Updated:2026-11-14Views:

Leak costs are not linear. A seepage gap widens as wear proceeds, so the later a leak is found the higher the cost of a single repair. This article separates early leak detection into three parts: how the cost curve of a leak is calculated, how small a leak fluorescent tracing can actually resolve, and how to put detection into the routine inspection plan with records that can be traced. It includes a tracer selection table for oil and water systems, the time windows for the five-step procedure, inspection intervals and the fields of a leak log, so a workshop can use it as a checklist.

Equipment does not start leaking on one particular day. It seeps, then drips, then sprays: the sealing lip wears a gap, the gap grows a little with every pressure and temperature cycle, and the leak rate rises non-linearly. The value of finding it early is therefore not the price of a drum of oil, but trading an unplanned stop for a planned one-hour job.

Fluorescent leak detection has a clear place in this. It turns finding the leak from a shutdown task - dismantling, pressure testing, brushing on soap solution - into a routine job that fits into the inspection plan and can be done while the machine runs. It depends on getting three things right: reading conditions, tracer selection, and the discipline of working and recording. The article takes them in that order.

1. The cost curve of a leak: why early beats precise

The same leak costs very different amounts depending on when it is found. Split by the moment of discovery:

Point of discoveryTypical evidenceAction and cost
Found in a planned inspectionSeepage, staining, a thin oil film; oil level almost unchangedChange the seal or re-torque; a one-hour job, no loss of output
Found on a routine roundOil level falling steadily, filters blocking often, local temperature risePlanned stop, parts replacement, check for consequential wear
Shutdown after an alarm or abnormal noiseOil mist spreading, bearing and shaft seal damage, system contaminationUnplanned stop, loss counted by the hour, including dismantling and cleaning
Found when the leak is already out of controlA whole batch contaminated, waste liquid escapingA quality incident and a compliance incident together, with cleanup and corrective action

The point of this table is the gap between the top two rows: a few months separate them physically, an order of magnitude separates them in cost. A leak accelerates itself - abrasive particles enter the gap with the oil, and a wider gap admits more particles. Early intervention moves the knee of the curve to the left.

Divide leaks into three grades by form and the response rhythm becomes clear:

FormTestRiskSuggested deadline
SeepOil film or stain on the surface, no dripping, level essentially unchangedParticles enter the gap and wear acceleratesPut on the plan, fix at the next stop
DripVisible dripping, clear accumulation in the drip trayFalling level, contamination, slippery floorSchedule a stop within two weeks
FlowContinuous flow, rapid change in pressure or levelSeized bearing, fire, spreading contaminationStop and deal with it on shift

Gas systems keep a different set of books, and the comparison is worth making. A compressed air leak leaves no oil trace: it is found by ear, by ultrasound or by flow accounting. Its cost follows a chain - leak flow (m³/min) × specific power of the compressor (about 6-7 kW/(m³/min) at 0.7 MPa) × annual running hours × electricity price. A 0.1 m³/min leak running 6000 hours at 0.7 CNY/kWh comes to 2500-3000 CNY a year in electricity; several dozen such points across a plant are a steady waste.

One point needs to be clear: fluorescence cannot find gas leaks. A fluorescent tracer has to be dissolved in a liquid and has to seep out with that liquid to show under a UV lamp; compressed air, steam and fuel gas belong to ultrasound and flow accounting instead. The right division of labour in a plant is gas to ultrasound, liquid to fluorescence: the two do not overlap.

2. The physical chain of fluorescence: from excitation to reading

The whole mechanism compresses into four stages:

  1. The tracer dissolves in the medium. The dye dissolves completely in the oil or water and travels with circulation to every corner of the system, including the leak path;
  2. The medium seeps out at the leak. Dyed liquid leaves an extremely thin residue on the leak surface, and the dye stays behind as the solvent evaporates;
  3. UV excites the dye. 365nm UV-A photons are absorbed by the dye molecules and the electrons move to an excited state;
  4. Visible fluorescence is emitted. As the molecules fall back to the ground state they emit visible light at a longer wavelength. That is the Stokes shift, and it is why invisible light can reveal visible light.

Take the oil-soluble LUYOR-6100: excitation peaks at about 494nm and emission at about 535nm. A shift of some 40nm looks small, but in engineering it is what makes the method work - emission and excitation have to be separated so that a filter can block the excitation and let the fluorescence through to the eye. The UV filter in a leak detection lamp does exactly that: it stops visible violet above 400nm, the background darkens, and the fluorescence stands out.

This also explains a common misunderstanding: a lamp that looks more violet and brighter is not necessarily better for leak detection. Lamps at 395nm and above carry more visible light: the eye finds them bright, but what is bright is the background, and the contrast of the fluorescence drops.

Colour matters too. Photopic vision is most sensitive to yellow-green near 555nm, so tracers that fluoresce yellow-green (the colour of 6200 and 6100 under UV) look brighter and carry further under the same excitation; the blue fluorescence of 6300 is inherently dimmer to the eye. That is not a defect, because the purpose of blue is zoning - where several water circuits run in parallel in one workshop, colour separates the loops far more reliably than remembering positions.

3. Reading threshold: how small a leak fluorescence can resolve

The conclusion up front: fluorescence does not measure an instantaneous leak rate but the amount of dye accumulated on the leak surface. A reading is possible when the seeping liquid has left enough dye molecules on the surface, so sensitivity depends on three variables at once: concentration, time and viewing conditions.

MethodTypical detectable levelConditions and limits
Visual (white light)Already dripping or flowing, several millilitres per hour or moreNeeds an oil trace or wet mark thick enough to see; nearly useless on dark or oily surfaces
Soap bubble or smear testGas systems, of the order of 10⁻³ Pa - m³/sNeeds a shutdown and access to the leak; fails on micro-leaks and hidden locations
Fluorescent tracingReadable once the leak surface has accumulated microlitres; refrigerant systems down to a few grams of refrigerant per yearNeeds dosing and waiting; blind spots that light cannot reach are missed
UltrasoundGas leaks, depending on differential pressure and orifice sizeStrong directional location, but poor at telling which medium is leaking
Helium mass spectrometryOf the order of 10⁻⁹ Pa - m³/sCostly, needs dedicated tooling, limited to laboratories and high-end manufacturing

Note how the fourth row is worded: its sensitivity is not a fixed number but a cumulative process that grows with waiting time. The same 0.01 ml/h seep looks completely different after 10 minutes and after 2 hours. Hence the rule for scheduling: leave enough time between dosing and reading, otherwise a "no leak" result is a false negative.

Ranked by influence, the variables that decide the detection rate usually run: 1. ambient light (background brightness) > 2. excitation irradiance > 3. dye concentration > 4. waiting time. That is why switching off the workstation lights and hanging a blackout cloth usually does more than buying a brighter lamp.

4. Tracer selection: start with the medium, then the colour

The rule has no exceptions: the tracer must be fully miscible with the medium. Oil systems take an oil-soluble tracer, water systems a water-based one; the wrong choice wastes the work, because oil-soluble dye poured into a water tank merely floats or sinks, the fluorescence goes with the sediment, and nothing shows under the lamp.

ModelSystem and fluorescence colourTypical doseSuitable systems
LUYOR-6100Oil-soluble, Solvent Red 175, yellow-white fluorescence under UV (excitation 494nm/emission 535nm); the oil looks orange-red in daylight0.5-2.5 ml per litre of oil (500-2500 ppm)Hydraulic oil, lubricating oil, fuel, transmissions, compressors
LUYOR-6200Water-based, green fluorescence, miscible with water in any ratioFrom 1000:1, dilutable to 5000:1 for pressure testingCooling water, circulating water, boilers, tanks, piping
LUYOR-6300Water-based, blue fluorescence, clearly distinct from 6200500ml treats about 3000 litres of waterAs above; used for zoning where several water circuits run in parallel
SPI-WGB/WGGSpectroline water-soluble tracer, blue or green, filtered to 2 microns, NSF certifiedAccording to the water/glycol system dilution ratioWater, water/glycol and coolant systems

Three hard limits apply when selecting a tracer:

  • Concentration has a window; more is not better. Too much dye causes self-absorption and quenching, and the fluorescence turns duller, whitish or yellowish, which at reading time looks like a severe leak. Dose to the recommended ratio rather than by feel;
  • Check what the system forbids. Industrial tracers must not be used in systems in direct contact with food, drinking water or pharmaceuticals; water-based grades are for industrial water systems and must not enter potable lines;
  • Assess material compatibility. The oil-soluble carrier is a solvent system and can swell some elastomers under long immersion at high concentration. On older equipment and with special seal materials (certain EPDM grades, silicone), soak a sample before a large dose.

Colour does not change the principle, but it changes the efficiency of reading: green (the 6200 family) is enough for a single system, and where several water or oil circuits run in parallel in one workshop, alternating blue and green lets one pass with the lamp show which loop belongs to which machine.

5. Reading conditions: lamp, distance and ambient light

Reading conditions can borrow the established wording of NDT practice: UV-A irradiance not below 1000 µW/cm² and ambient white light held below 20 lx (the viewing conditions for penetrant and magnetic particle testing in ISO 3059 and GB/T 5097, which leak detection can follow). Below that irradiance the fluorescence of a seep is lost in the noise.

ApplicationSuggested lampHow to read
Whole pipe runs, whole machinesLUYOR-365L (about 6500 µW/cm² at the centre at 38cm, useful area about 200mm)Sweep slowly at 30-40cm, overlapping the spot edge by 1/3
Final judgement on fittings, valve stems, shaft sealsLUYOR-3180 (30000-40000 µW/cm² at 38cm, spot about 30mm)Work at 10-20cm, dark-adapt before reading
Fixed stations, washdown areasIP68 fixed installation lamp (such as LUYOR-BBMA)Continuous illumination with a light shield

Three things that are often overlooked: 1. dark adaptation - coming from a bright area into darkness, the eye needs 5-10 minutes to recover its sensitivity to weak fluorescence, so switch off and wait before reading;2. lens cleanliness - oil on the head filter absorbs UV and produces the classic "bright lamp, no fluorescence";3. eye protection - 365nm is UV-A, and prolonged close viewing requires UV safety glasses (LUV-10 or equivalent); there is no "just a quick look" exception.

6. The five-step procedure: making leak detection routine

  1. Dose and record. Choose the tracer by medium, dose to the ratio, and hang a dosing tag (date, system, quantity); this step is the precondition for any later re-check;
  2. Run and let it penetrate. A running system mixes and circulates within 5-10 minutes; a static system needs 2-3 hours for the medium to soak the leak path. For a very slow seep, look again the next day rather than waiting once;
  3. Illuminate and read. Dim the ambient light and read in the order: wide scan for anything at all → narrow down for what it is → locate the exact point;
  4. Mark and grade. Mark the leak and grade it by rate: seep (can be planned)/drip (schedule soon)/flow (deal with it on shift);
  5. Re-inspect. After the repair, illuminate again under the same conditions, and clean the old trace beforehand, or residual fluorescence will be read as "still leaking".
StepTime window (running system)Time window (static system)Evidence to keep
Dose and penetrate5-10 minutes2-3 hours; overnight for a slow seepPhoto of the dosing tag, dose record
Illuminate and readAfter 5-10 minutes of dark adaptationAs abovePhoto of the leak location with a reference object
Repair and re-checkImmediately after repairAfter repair, hold under pressure for over an hourBefore and after photos, one file per leak

7. Inspection interval and log: letting the system find the leak

Do not copy a neighbour's interval; set it from three factors: medium pressure and temperature, the severity of the consequence, and whether the machine sits on the critical path. A workable starting point:

  • Critical equipment (high-pressure hydraulics, pressure vessels, continuous lines): every shift or daily, with fixed points illuminated each time;
  • General rotating equipment (pumps, fans, gearboxes, motors): weekly or monthly, dosing and illuminating around shutdown windows;
  • Utilities and auxiliary systems (cooling water, tanks, peripheral piping): quarterly, alongside water or oil change cycles;
  • Newly commissioned, after overhaul, after modification: regardless of the interval, run a full survey and establish a baseline photo set.

The log needs six fields at minimum; miss one and no trend is possible: machine number or system name, dosing date and quantity, tracer model and batch, leak location with photo, grade conclusion, repair and re-check dates. After six months it shows which machine keeps seeping and which location is the weak point in the design or the assembly, which is far more useful than assigning blame afterwards.

A simpler start is a kit: for oil systems LUYOR-6801 (LUV-100 leak detection lamp, 240ml of 6100 tracer, cleaner, safety glasses, dosing tags); for water systems LUYOR-6802 (green) or LUYOR-6803 (blue). Everything needed is in the case, so a crew can draw it and start work.

8. Why one machine leaks again and again: without the root cause, detection treats the symptom alone

If the log shows three or more leaks at the same location within a year, the seal itself is rarely the cause; it is usually one of four root causes:

Root causeSite evidenceDirection
Assembly and toleranceSeveral machines of the same model leak on the same sideCheck alignment, face flatness and tightening sequence
Seal selection or materialSeals hardened, cracked or extrudedRe-select the material for the medium and temperature (NBR/FKM/PTFE)
Pressure pulsation and temperature riseLeakage follows the duty cycle, with clear peaks and troughsCheck the accumulator, relief valve and cooling capacity
Particle contaminationOil cleanliness outside the target ISO 4406 classFilter, change the oil, keep the filler cap clean

Read leak results alongside oil analysis and a three-level warning appears: Level 1 (a seep under UV, cleanliness normal) goes on the watch list and is fixed at the next stop; level 2 (the leak growing, cleanliness one class worse) calls for a planned stop; level 3 (dripping, with a temperature rise or noise) is handled on shift. The logic matches the condition monitoring framework of ISO 17359: leak detection is not a one-off yes/no judgement but a continuous correction of the machine's condition picture.

9. False positives: six sources and how to rule them out

SourceAppearanceHow to rule it out
Residue from a previous checkFixed location, sharp boundary, still faintly visible after wipingClean with solvent and illuminate again; see whether it returns
Assembly or rust-preventive oilSpread in patches, unrelated to the leak pathClean, run under pressure and watch whether it reappears at the joint
Autofluorescence of rubber and plasticThe whole part glows, the boundary following its outlineChange angle and distance; autofluorescence does not extend along the leak path
Cleaner or degreaser residueBlue-white haze, widely spreadRinse with water or the specified cleaner and look again
Ambient reflectionMoves with the lampMove the lamp; real fluorescence stays put
Excess concentrationLarge areas glowing yellow-white, hiding the leakReduce the dose or change the medium and retest

One action decides the reading: wipe, illuminate, look again. It counts as a leak when it reappears in the same place after wiping and follows the direction of flow.

10. Working with other methods

CasePreferredIn support
Seepage in hydraulic or lube systemsFluorescent tracingOil analysis (ISO 4406 cleanliness, spectroscopy) for the wear trend
Cooling and circulating waterWater-based fluorescent tracingPressure decay for quantitative confirmation
Compressed air, steam, fuel gasUltrasonic leak detectionFlow accounting, soap bubble location
Vacuum and ultra-high vacuumHelium mass spectrometry-
Fast screening on a production lineFluorescent scanConfirm suspect points with a high-intensity narrow beam

11. References

Sources for the standards and methods (check the original by number; the current edition from the issuing body applies)

  • ISO 3059, Non-destructive testing - Penetrant testing and magnetic particle testing - Viewing conditions: UV-A irradiance and ambient white light for reading;
  • GB/T 5097, Non-destructive testing - Penetrant testing and magnetic particle testing - Viewing conditions: the national standard version of the same conditions;
  • ASTM E3022, Measurement of emission characteristics of LED UV-A lamps used in fluorescent penetrant and magnetic particle testing: how the 365nm band and visible light leakage are evaluated;
  • ASTM E2297, Guide for the use of black lights in penetrant and magnetic particle testing: use and maintenance of black lights;
  • ISO 4406, Hydraulic fluid power - Fluids - Method for coding the level of contamination by solid particles: the basis for assessing the contamination that comes with leakage and wear;
  • ISO 17359, Condition monitoring and diagnostics of machines - General guidelines: the framework for placing leak detection inside condition monitoring;
  • ISO 8573-1, Compressed air - Part 1: Contaminants and purity classes: the basis for accounting gas system leaks and quality.

12. Frequently asked questions

How small a leak can fluorescent detection find?

It depends on how much dye accumulates on the leak surface, so there is no fixed figure. With a 365nm high-intensity UV lamp and a sensible waiting time, a seep that has not yet formed a drop can be found (microlitres accumulated on the surface), and refrigerant systems can show a slow leak of a few grams per year. Below that level, laboratory methods such as helium mass spectrometry are needed.

The tracer is in, but nothing fluoresces. What should be checked before anything else?

In order of likelihood: 1. wrong tracer medium (oil-soluble dye in a water system, or the reverse); 2. not enough waiting time; 3. ambient light too strong; 4. oil on the head filter, or a violet torch with no filter at all; 5. dye concentration too low, or the product is out of date; 6. the leak sits behind insulation or a guard, where the light never reaches.

Does the tracer have to be drained out?

An oil-soluble dye such as LUYOR-6100 can stay in the system long term when dosed to specification; it does not change oil performance. After the leak is repaired, illuminate again directly, with no need to add more. If the system is sensitive to additives or an oil change is due, it leaves with the old oil.

How much goes into one system?

Oil systems take 0.5-2.5 ml/L, so a 200L tank takes 100-500ml. Water systems start at 1000:1, and 500ml of concentrate covers hundreds to thousands of litres. For exact figures, calculate from system volume and pipe diameter.

Can fluorescent detection find compressed air leaks?

No. A gas carries no liquid medium to hold the dye. Gas leaks are found with ultrasound, backed by flow accounting and soap bubble location; fluorescent tracing applies to systems with a closed circulating liquid.

Can it be done in daylight?

Yes, but the detection rate drops. Sunlight and workstation lighting are the main interference; prefer dusk, night or a blackout cloth. Where light cannot be excluded, wear yellow contrast glasses and work as close to the surface as possible.

Further reading: Fluorescent leak detection explained: wavelength, sensitivity and 18 reading questions, Finding industrial leaks: lighting up the leak with the LUYOR-365L, Preventive maintenance on hydraulic systems: fitting fluorescent detection into the inspection plan, Finding equipment leaks in a paper mill.

Related sections and models

This article belongs to the fluorescent leak detection section.

Other sections on the same product line include fluorescent leak detection (add a tracer, find the leak under UV, locate before dismantling), 365nm blacklights (UV area lamps for NDT, leak detection and fluorescence inspection, judged by irradiance rather than wattage), industrial test instruments (the entry point for the four lines: UV inspection, speed measurement, stroboscopic observation and surface inspection).

Models mentioned in this article: LUYOR-6100, LUYOR-365L, LUYOR-6200. Ranges, accuracy and accessory lists for each model are listed on the product pages.

  • Official WeChat QR code

    Email: info@xinysun.com

  • Tel: 131-2093-6518

  • Address: 1525 Minqiang Road, Bldg 26, Songjiang, Shanghai

Official WeChat QR code Follow

Official WeChat QR code

Contact

131-2093-6518

Top