Analysers and sampling systems — where the real problem usually is

Every other instrument on this site measures a physical property — pressure, temperature, flow, level. An analyser measures what the stream is made of.

That difference is why analysers behave so differently from everything else in the instrumentation section.

What they measure

Analyser	Measures	Response	Typical use	Watch out for
Gas chromatograph	Full composition, component by component	Minutes per cycle	Product quality, custody transfer, column control	Carrier gas supply, column ageing, long cycle time
Infrared	Specific gases — CO, CO₂, hydrocarbons	Seconds	Combustion products, emissions, leak detection	Cross-sensitivity to other gases present
Paramagnetic or zirconia oxygen	Oxygen concentration	Seconds	Combustion control on fired heaters, inerting checks	Zirconia cells need high temperature and age steadily
pH	Acidity or alkalinity	Seconds	Effluent neutralisation, water treatment	Electrodes foul and drift; frequent calibration needed
Conductivity	Dissolved ionic content	Immediate	Boiler feed water, condensate contamination	Temperature compensation must be right
Moisture	Water content in gas or liquid	Seconds to minutes	Gas dehydration, dry product specification	The sample line itself holds moisture and slows response
Density and viscosity	Physical properties used to infer composition	Immediate	Blending, product grading	Temperature correction is essential
Flammable gas detector	Gas in air, as a percentage of the lower explosive limit	Seconds	Area gas detection, building intake protection	A safety device, not a process analyser — tested on a set cycle

An ordinary transmitter measures a physical property — pressure, temperature, flow. An analyser measures what the stream is made of. That difference is why analysers need a sample taken, conditioned and delivered, and why they are the least reliable instruments on most plants.

Analysers are used where composition is what matters: product on specification, combustion efficiency, emissions compliance, custody transfer, environmental discharge, and safety detection.

They are also, on most plants, the least available instruments. Understanding why is the whole of this topic.

The analyser is rarely the problem

Fault	How it shows	Cause	Addressed by
Sample line plugged	Reading frozen, or drifting very slowly	Wax, hydrate, solids or polymer in the line	Heat tracing, filtration, larger bore, fast loop
Transport lag too long	Analyser correct but always late; control hunts	Long sample line at low flow	A fast loop returning most of the sample to the process
Phase change in the line	Erratic or nonsensical readings	Sample cooling and condensing, or flashing on pressure drop	Tracing, insulation, and reducing pressure at the right point
Unrepresentative take-off	Analyser disagrees with the laboratory, consistently	Probe in a dead leg, at the wall, or downstream of a mixing point	Relocating the probe — a design fix, not a maintenance one
Calibration gas exhausted	Validation fails, or drift goes uncorrected	Cylinder not replaced, or regulator leaking	Cylinder pressure monitoring and a stocking routine
Carrier gas lost	A chromatograph stops producing results entirely	Cylinder empty or supply interrupted	Duplicate cylinders with automatic changeover
Sample disposal blocked	No sample flow, and possibly back pressure on the analyser	Return line or vent plugged, or return pressure risen	Checking the disposal route, which is often forgotten entirely

Industry experience consistently puts the large majority of analyser unavailability in the sample system rather than the instrument. The analyser is generally a sound piece of laboratory equipment being asked to work on whatever arrives down a tube from a process line.

A pressure transmitter sits on the pipe and measures it. An analyser sits in a shelter some distance away and measures a sample that has been taken, transported, conditioned and delivered to it.

Every one of those steps can go wrong, and most of them do.

The take-off point decides everything

If the sample is not representative, nothing downstream can fix it.

A probe should sit in well-mixed flowing stream — not in a dead leg, not at the pipe wall where the boundary layer differs from the bulk, not immediately downstream of a mixing point before the streams have combined, and not where liquid can collect around it.

Conditioning and lag

Between the tap and the analyser the sample usually has to be filtered, have its pressure and temperature reduced, kept in a single phase, and delivered at a controlled flow.

Two things dominate the design.

Phase. A gas sample that cools and condenses, or a liquid that flashes across a pressure reduction, is no longer the thing you meant to measure. Heat tracing, insulation, and taking the pressure drop at the right point are what prevent it.

Transport lag. Sample trickling slowly down thirty metres of small-bore tubing may take several minutes to arrive. The answer is a fast loop — a high flow kept moving through the line and returned to the process, with only a small slipstream taken off for the analyser.

That lag matters beyond patience:

Where they live

Analysers are often grouped in an analyser house, which is the awkward building described in buildings and shelters: a small shelter that deliberately contains process samples, and is therefore a hazard rather than a refuge from one. Ventilated rather than sealed, gas detected, and often a classified area inside.

Field-mounted analysers avoid the sample transport problem entirely but then have to survive the weather and be maintained at the point of measurement, in whatever hazardous area classification applies.

Calibration and validation

Analysers drift, so they are checked against known reference gases or solutions.

• Validation — passing a known sample through and confirming the reading, without changing anything.
• Calibration — adjusting the response to match.

Both need calibration gas in date and in stock, and the routine that keeps it there. An analyser whose calibration cylinder has been empty for two months is still producing a number on the operator’s screen, and it is the sort of failure nobody notices because the display never goes blank.

What to take away

• An analyser measures composition; everything else measures a physical property.
• Most analyser downtime is the sample system, not the analyser.
• If it disagrees with the lab consistently, move the probe — do not recalibrate.
• Keep the sample in one phase. Condensing or flashing destroys the measurement.
• Use a fast loop to cut transport lag; add cycle time to it to get the real age of the reading.
• A slow analyser trims a faster loop in cascade rather than controlling directly.
• An analyser with expired calibration gas still shows a number. That is what makes it dangerous.