On this page7 sections
What a sample system has to protect
A sampling system extracts material from the process, transports it to the analyzer, conditions it into a form the analyzer can accept, and returns or disposes of it safely. Every one of those steps can change the answer. A good design protects three properties:
- Representative — the sample has the same composition as the process at the point of interest.
- Timely — it arrives quickly enough for the purpose of the measurement, whether control, quality or protection.
- Unaltered — nothing is added, lost, reacted or adsorbed between the tap and the analyzer.
Choosing the sample point
Take the sample where the process is well mixed and where its composition answers the question being asked — downstream of mixing points, and away from dead legs, injection points and anything that stratifies the flow. A probe inserted into the flow, rather than a tap flush with the wall, avoids the slower-moving and often unrepresentative material at the pipe wall.
- Gas service — tap from the top or side of a horizontal pipe, so that liquids and debris along the bottom are not drawn in.
- Liquid service — tap from the side of a horizontal pipe, avoiding gas pockets at the top and sediment at the bottom.
- Phase changes — know the sample's dew point or bubble point at the tap; a sample that changes phase in the probe is no longer representative.
- Access — provide isolation at the tap, and make the probe removable for maintenance — retractable where the process cannot be shut down.
Transport lag: a worked calculation
The delay between a change at the tap and its arrival at the analyzer is the internal volume of the sample path divided by the volumetric flow through it: lag = V ÷ Q. For tubing, V = π × d² ÷ 4 × L, using the internal diameter. The flow must be at line conditions: for gases, a flow set in normal litres has to be corrected to the actual pressure and temperature in the line.
Step 1: Tube volume
A 40 m run of 6 mm OD tube with a 1 mm wall has an internal diameter of 4 mm. Volume: π × 0.004² ÷ 4 × 40 ≈ 0.000503 m³, or about 0.50 L.
Step 2: Lag at the analyzer's own flow
If the line carries only the analyzer's flow of 0.5 L/min, the lag is 0.50 ÷ 0.5 ≈ 1.0 min — a full minute before the analyzer sees a process change.
Step 3: Lag with a fast loop
Circulating 5 L/min through the same tube at close to atmospheric pressure, and taking the analyzer's flow from near the analyzer, cuts the transport lag to 0.50 ÷ 5 ≈ 0.1 min — about 6 s.
Step 4: The effect of line pressure
If the tube instead runs at about 5 bar absolute and the 5 L/min is set in normal litres, the actual flow in the tube is only about 1 L/min, ignoring temperature. The lag rises to about 0.5 min — five times longer. Reducing pressure at the tap, not at the analyzer, avoids this.
Transport lag is only part of the total response. Filter housings, coolers, separators and other conditioning components add their own hold-up volume — often more than the tubing — and mixing inside them turns a sharp step into a slower ramp. Add the analyzer's own response time, and agree the acceptable total with the people who will use the measurement before the design is fixed.
Fast loops and bypass flow
A fast loop moves sample from the tap to a point near the analyzer at a flow much higher than the analyzer needs, then returns it to the process or sends it to a safe disposal point. The analyzer draws its small flow from the loop through a short branch. The loop needs a driving force: a pressure difference available in the process — across a control valve, orifice plate or pump, for example — or a sample pump or eductor.
A related technique is the bypass filter, in which the main sample flow sweeps across the filter element and only the analyzer's flow passes through it; the sweeping flow keeps the element surface clean and the lag short. In both cases, measure and indicate the loop flow. A fast loop that has quietly stopped flowing produces a stale reading that still looks plausible.
Conditioning the sample
Filtration
Remove particulates to protect valves, flow controllers and the analyzer, choosing the element for the particle load and for compatibility with the component being measured. Coalescing filters remove liquid aerosols; membrane filters can stop liquid carry-through. Every filter adds volume, so size it for the job rather than generously.
Pressure reduction
Reduce gas pressure close to the tap to shorten lag and reduce the energy stored in the sample line. A large pressure drop cools most process gases through the Joule–Thomson effect, which can cause condensation or hydrate formation; heated regulators or heat input at the reduction point are the usual remedies.
Temperature and dew point
A gas sample must either stay above its dew point — including any acid dew point — all the way to the analyzer, using heated or insulated lines and heated components, or be deliberately cooled and dried at a controlled point with the condensate removed. The choice matters: condensed water absorbs soluble components such as sulfur dioxide, ammonia and hydrogen chloride. A dry-basis system is designed to minimize contact between gas and condensate; a hot-wet system is designed so that no cold spot exists anywhere in the path.
Materials and adsorption
Wetted materials must resist the sample chemically, and must not adsorb or release the component being measured. Reactive and trace components such as hydrogen sulfide, ammonia and low-level moisture adsorb on untreated metal surfaces, which slows response and creates memory effects; inert-coated or passivated tubing and minimal wetted surface area help. Plastic tubing is permeable to moisture and oxygen, which makes it unsuitable for trace measurements of either. Check elastomer seals for compatibility too.
Validation, calibration and the analyzer environment
Distinguish validation — checking the analyzer against a known reference and recording the result without adjusting it — from calibration, in which the as-found response is recorded, zero and span are adjusted only if needed, and the as-left response is recorded. Wherever practical, introduce the reference gas at or near the probe so the check exercises the whole sample path; a check at the analyzer inlet proves the analyzer but not the probe, lines and conditioning. Use certified reference gases with traceable composition, and design stream switching so that calibration gas cannot leak into the sample, commonly with double block and bleed arrangements.
The analyzer's surroundings are part of the system. Many analyzers are sensitive to ambient temperature, so cabinets and shelters are usually temperature-controlled. Shelters that receive flammable or toxic samples need ventilation and gas detection, and electrical equipment must suit the area classification — by pressurization, for example. Plan the utilities — instrument air, power, carrier or zero gases, drains and vents routed to a safe location — and leave room to work on every component.
Troubleshooting from symptoms
| Symptom | Likely sample-system cause | What to check |
|---|---|---|
| Slow response to known process changes | Low sample or loop flow, oversized components, dead volume | Actual flows against design; hold-up volume of filters and pots |
| Soluble components read low while insoluble ones read correctly | Condensation in a line, or prolonged contact with condensate | Heat-tracing temperatures, cold spots, cooler and drain operation |
| Oxygen reads high and other components read low | Air leaking into a section under vacuum, such as the suction side of a sample pump | Leak-test the suction side; compare with an independent measurement |
| Noisy or spiking readings | Liquid slugs, particulate, or unstable pressure or flow | Filter condition, regulator stability, drains and liquid knock-out |
| Passes calibration at the analyzer but disagrees with the process | A fault in the probe, lines or conditioning that the check bypasses | Repeat the check with reference gas introduced at the probe |
| Trace moisture or a reactive component is slow to settle | Adsorption on wetted surfaces or permeation through tubing | Tubing material and surface treatment, purge history, dead legs |
Engineering questions
- Why does an analyzer pass calibration but read wrongly on the process?
- Usually because the calibration gas was introduced at the analyzer, bypassing the probe, sample line and conditioning system where the fault sits — a leak, condensation, a blocked filter or adsorption. Introducing the reference at or near the probe tests the whole path.
- What is an acceptable sample transport lag?
- It depends on what the measurement is used for. A measurement used for closed-loop control or protection needs a much shorter total response than one used for quality records. Agree the requirement first, then design the tap location, line size, flows and any fast loop to meet it.
- When should sample lines be heated?
- When the sample would otherwise cool below its dew point on the way to the analyzer — including acid dew points in combustion gases — or when components could condense or deposit. Either the whole path is held above the dew point, or the sample is cooled and dried at a controlled point with the line upstream of that point heated.
- Do liquid analyzers need sample conditioning?
- Often. Liquid samples may need filtration, pressure reduction, temperature control or degassing, and some measurements must be protected from air ingress. Transport lag follows the same volume-over-flow rule as for gases, although liquid volumetric flow does not change appreciably with line pressure.