Oil-in-water measurement is one of those jobs where the number on the screen has to stand up in front of a regulator as well as a process engineer. That is what makes technology selection awkward: the reference methods and the online instruments do not measure the same thing.
This guide walks through the main detection principles, what each one actually responds to, and how to choose between them.
Table of Contents
What “Oil” Means in a Water Sample
Oil is not a single analyte. Produced water, refinery effluent and bilge water all contain a mixture:
- Light hydrocarbons — largely dissolved or easily stripped
- Medium-weight hydrocarbons — usually the fraction of interest for compliance
- Heavy hydrocarbons — often emulsified and slow to separate
- Polar compounds — phenols and organic acids that carry some water solubility
Different methods respond to different fractions of that mixture, which is why two laboratories can report different numbers for the same sample and both be technically correct. Any comparison between methods has to start from that.
Sample handling
What arrives at the analyser is rarely a clean, stable sample:
- Stable emulsions resist extraction and bias low
- Temperature changes both solubility and viscosity
- Production chemicals (demulsifiers, corrosion inhibitors) fluoresce or absorb
- Suspended solids scatter light and inflate optical readings
UV Fluorescence
Principle
Aromatic hydrocarbons — BTEX and larger PAHs — absorb UV light (roughly 250–400 nm) and re-emit at longer wavelengths (roughly 300–500 nm). The emitted intensity tracks the aromatic concentration, so the measurement is effectively an aromatic-hydrocarbon measurement rather than a total-oil measurement.
Performance
- Range: roughly a few tenths of a ppm up to a few hundred ppm, set by optical path length and gain
- Response: seconds to minutes, continuous
- Selectivity: aromatic fraction only; aliphatic hydrocarbons are largely invisible to it
Advantages
- Sensitive enough to see trace levels in treated water
- Continuous, consumable-free, no solvent handling
- Well suited to trend monitoring and alarm/diversion logic
Limitations
- Response depends on the oil’s aromatic content, so calibration with the wrong oil gives the wrong number
- Temperature compensation is required
- Background fluorescence from some production chemicals and dyes
- UV source and optical window age; lamp or LED replacement is part of maintenance
The point that is usually got wrong
ISO 9377-2 is a laboratory method — solvent extraction followed by gas chromatography with flame ionisation detection — and ASTM D7066 is a laboratory method based on infrared absorption after S-316 extraction. An online fluorescence analyser does not perform either of them. What makes it useful is a site-specific correlation: calibrate the analyser against the laboratory reference on the site’s own produced water, and re-check that correlation periodically. Marketing copy that claims a fluorescence sensor “meets” or “complies with” a GC-FID or IR method is simply wrong, and it is the kind of claim that falls apart the first time a regulator asks to see the method.
ChiMay inline oil-in-water sensors use UV fluorescence for continuous measurement, with the calibration performed against the site’s reference-method results.
Infrared Absorption
Principle
Infrared methods measure absorption by the C-H bond, around 2,800–3,100 cm⁻¹ (roughly 3.2–3.6 µm). Two standards dominate:
- ASTM D7066 — extraction into S-316 followed by infrared determination
- ASTM D7678 — solvent extraction followed by mid-IR laser spectroscopy, reporting total oil and grease and total petroleum hydrocarbons
Both are laboratory methods built around an extraction step. Both give a broader hydrocarbon response than fluorescence, because aliphatic hydrocarbons also carry C-H bonds.
Performance
- Range: from sub-mg/L up to high concentrations, depending on extraction ratio and path length
- Method detection limit: low mg/L range, better with larger sample volumes
- Analysis time: minutes to hours, batch
Advantages
- Wide hydrocarbon coverage, including aliphatic fractions
- Long history of regulatory acceptance
- Available in laboratory and field-portable formats
Limitations
- Solvent consumption, storage and disposal
- Extraction efficiency varies between operators and samples
- Not continuous, so it cannot follow a process upset
Note on method families: EPA Method 1664 is the US oil-and-grease method and is gravimetric (n-hexane extractable material), not infrared. IR methods such as ASTM D7066 and D7678 have their own approvals and their own reference oils; they are not substitutes for Method 1664 by definition.
GC-FID as the Reference Method
ISO 9377-2 defines the hydrocarbon oil index by solvent extraction, chromatographic separation and flame ionisation detection. The FID responds to any carbon-hydrogen compound, so the method covers a broad set of hydrocarbons and can resolve individual peaks.
Performance
- Method detection limit: around 0.1 mg/L for the standardised procedure, lower with optimised extraction
- Analysis time: tens of minutes per sample, batch
- Selectivity: all combustible organic compounds
Advantages
- The reference method most discharge limits are written against
- Best detection limits of the standardised methods
- Chromatographic detail supports source identification and method comparison
Limitations
- Laboratory instrumentation and trained analysts
- Batch turnaround only, no real-time capability
- Highest operating cost of the options here
Because it is the method permits are written around, GC-FID is usually the anchor: online analysers are calibrated to it, not the other way round.
Online and Inline Systems
A working online installation normally has four parts:
Sensor assembly — optical or extractive, in probe form for inline mounting
Sample conditioning — filtration and temperature control, matched to the sensor’s tolerance
Transmitter/controller — signal processing, compensation, local display
Data layer — historian, alarms, and reporting for compliance records
ChiMay inline oil-in-water sensors include self-cleaning (ultrasonic or mechanical) interfaces, multiple measurement ranges, temperature compensation, and Modbus, HART and Profibus outputs.
Choosing Between Them
Work back from the requirement:
For a discharge permit, determine which method the permit names, then choose an online instrument whose correlation with that method you can demonstrate. Continuous data is what lets you show an excursion was brief and controlled, and average-based standards (which is what most produced-water limits are) are much easier to demonstrate with continuous data.
For treatment optimisation, an inline fluorescence analyser gives the fastest feedback for separator optimisation, produced-water polishing and diversion control.
For method disputes or baseline characterisation, GC-FID or an IR method in the laboratory remains the arbiter.
Practically, most operators run both: continuous analysers on the control side, reference-method analysis on the compliance side, with the correlation between them documented and re-verified.
Cost shape rather than a price list
Equipment prices move with configuration, so treat published ranges with suspicion. The structural differences are stable: laboratory methods carry recurring per-sample cost plus instrument capital; infrared field instruments sit in the middle; inline fluorescence has the highest capital cost and the lowest running cost per measurement, with consumables limited to cleaning, calibration standards and, eventually, source replacement. For a site measuring continuously, the per-measurement cost of the inline option is an order of magnitude below laboratory analysis — that is the argument that usually decides procurement, not the sticker price.
Calibration and Quality Assurance
Primary calibration uses certified standards traceable to a national metrology institute. For oil-in-water work, hexadecane, a mineral oil, or a standard matched to the site’s own crude is common. Whichever is chosen, the calibration oil has to match the oil actually being measured as closely as possible.
Frequency: inline sensors are typically verified frequently (daily to weekly, by a two-point check) and fully recalibrated at longer intervals; laboratory methods calibrate with each batch.
Quality control that actually catches problems:
- Blank checks for contamination between samples
- Duplicates for precision
- Spike recoveries to confirm extraction efficiency
- Calibration verification at the working range, not just mid-scale
- Regular correlation against the reference method — this is the one that catches slow drift in an online analyser
Where the Technology Is Going
Multi-wavelength fluorescence improves discrimination between oil types and reduces interference from chemicals that fluoresce in the same band. Machine-learning calibration builds correction models from operating history rather than a single reference oil, which helps where the oil composition changes. Smaller optical assemblies and lower-cost sources widen the deployment envelope. Sensor fusion — fluorescence plus light scattering plus image analysis — is starting to appear in commercial instruments, mainly to distinguish oil droplets from suspended solids.
Wrapping up
The three families are complementary rather than competing. UV fluorescence gives continuous sensitivity to the aromatic fraction, which is what process control needs. Infrared methods give broader hydrocarbon coverage with a long record of regulatory use. GC-FID is the reference against which the others are calibrated. Choose on the basis of which one the permit accepts, what the operator needs to see in real time, and how the correlation between the two will be maintained over years of operation.
