2.3 Analytical Instruments and Sampling Systems

Analyzers are loops with chemistry failure modes

NCEES PE Control Systems Domain 1.B (analytical instruments and sampling systems) is where otherwise good control engineers lose points by treating a pH probe like a thermocouple. The sensor chemistry, the sample system, and the process time constant are one specification. Do not invent unpublished analyzer model numbers on the exam or in the field; specify principle, range, wetted materials, sample conditioning, and lag.

pH, ORP, and conductivity

pH is a Nernstian glass-electrode measurement: about 59 mV per pH unit at 25 °C, temperature-dependent (2.303 RT/F). Automatic temperature compensation is mandatory on any loop you care about. The reference (typically Ag/AgCl with a KCl junction) is the weak part. Reference poisoning occurs when sulfide, cyanide, or similar species convert silver salts or clog the junction; the reading drifts or freezes while the glass is still intact. Dehydration of the glass membrane (stored dry, left in a hot anhydrous solvent, or rinsed and left in air) kills slope; the glass must stay hydrated. Coating (oil, latex, biological film) adds lag and offset. High-pH sodium error and low-conductivity high-purity water (streaming potentials) are separate traps—pure-water pH is often a conductivity/resistivity problem in disguise.

Oxidation–reduction potential (ORP) is a noble-metal electrode in millivolts, used on bleach, chrome, and cyanide destruct loops. It is not a pH meter and does not replace a pH loop.

Conductivity (and resistivity in ultrapure water) is ionic. Two-electrode probes fit clean, low-fouling water. Toroidal (electrodeless) sensors belong in coating, slurry, or high-conductivity service where electrodes would polarize or foul. Ultrapure water resistivity is about 18.2 MΩ·cm at 25 °C; temperature compensation to a 25 °C reference is part of the spec. Hydrocarbons are not a conductivity application unless you are actually measuring a water phase.

Oxygen, density, moisture, and GC concepts

Zirconia oxygen cells are solid-electrolyte Nernst sensors that must run hot (typically on the order of 600–800 °C). They are the standard in-situ flue-gas O2 probe with instrument-air or a known reference. They are not a 25 °C dissolved-oxygen probe. Paramagnetic oxygen analyzers use O2's paramagnetism on an extractive process-gas sample; they need a conditioned, often dry, gas and a sample system. Electrochemical (galvanic or polarographic) and optical (luminescence) sensors measure dissolved oxygen in water and wastewater, or gas O2 in portable form—again, not a hot zirconia flue probe.

Density for process control comes from Coriolis (line density plus mass flow), a vibrating-fork point sensor (interface, high/low density, some tank duty), or nuclear gamma (through-pipe, severe service). Pick nuclear last: licensing and cost. Do not use a tank hydrostatic 'density' inferred from two pressure taps unless you actually have height and a valid two-tap calibration.

Moisture and dew point in gases use chilled-mirror (reference-grade), metal-oxide or polymer impedance, and optical absorption methods. Specify dew point versus ppmv, wet versus dry basis, and whether the sample must stay above dew point in the line. A gas chromatograph (GC) is only as good as the sample: representative phase, no lost heavies, known lag, and a fast loop so the analyzer sees a flowing stream, not a dead volume.

Sampling systems: lag, drop-out, adsorption, phase change

Transport lag for plug flow is line volume divided by sample flow, t = V/Q, plus filter bowls, knockout pots, and the analyzer cell. Mixed volumes add first-order lags. A fast loop runs a high flow past a short, small-bore takeoff so the analyzer lag is the takeoff, not the 200-foot header. Drop-out is condensable liquid falling out in a cold spot—your GC then analyzes a stripped vapor. Adsorption of water, H2S, or polar organics on tubing walls (especially long stainless runs) makes the analyzer slow and biased; treated tubing or PTFE-lined sample lines are a materials spec. Phase change (a sample that was liquid in the process flashing in the regulator, or a gas dropping below dew point) is a composition error, not an analyzer drift. Filtration protects the cell and can strip aerosols you meant to measure. Heated lines keep the sample above dew point with margin; cooling a hot wet sample 'to protect the analyzer' without a proper condenser/knockout is how you measure the wrong stream.

Worked example: pH failure modes and sample lag versus in-situ

A neutralization reactor, 4,000 gal, acid/base reagents, process time constant about 1–2 minutes. Two bids: (1) retractable in-situ pH in a recirculation spool; (2) extractive sample, 80 ft of 3/8-inch tubing (about 0.30 in ID), filter bowl 0.15 gal, analyzer flow 0.05 gpm.

Tubing volume: ID 0.305 in → area ≈ 0.073 in² ≈ 5.1×10⁻⁴ ft². Eighty feet holds about 0.041 ft³ ≈ 0.30 gal. Plug-flow lag in the line alone is 0.30 / 0.05 = 6 minutes. Add the filter bowl (another 3 minutes if mixed as 0.15 gal at 0.05 gpm) and the cell. The extractive loop is several process time constants late. You cannot tune a 2-minute plant with a 10-minute PV. Fix: a fast loop at 1–2 gpm in short large-bore return, then a 5–10 ft small-bore takeoff, or put the electrode in situ with a retractable housing for cleaning.

Failure-mode walk-down when in-situ still reads wrong. If the process contains H2S or cyanide, suspect reference poisoning (junction clogged or silver converted)—slope may look odd and the reading is sluggish or stuck. If the probe sat in air over a shutdown or was stored dry, suspect glass dehydration—buffers will not span. If the reading is slow only after a latex or oily campaign, suspect coating. If the loop has no RTD compensation, a 10 °C swing is about 2 mV/pH-unit slope error, which is several tenths of a pH at mid-scale. None of these is 'buy a new analyzer model'; they are electrode and sample-system specifications.

GC analog. A 150-foot unheated sample line to a shelter GC on a condensing hydrocarbon vapor will drop out heavies, adsorb water, and report a light-biased composition with minutes of lag. Specify tap location, phase (vaporize at the tap if the process is two-phase), fast loop, filtration that does not strip the aerosol you care about, and heated lines above dew point. Then size the GC cycle time so it is still fast relative to the unit disturbance.