2.1 Pressure, Level, and Temperature Sensors
Key Takeaways
- Use absolute pressure whenever the equation is referenced to vacuum (VLE, gas-law mass, compressor maps); gauge pressure is not interchangeable by adding 14.7 psi.
- Closed-tank DP level on condensable vapor needs a wet leg, condensate pot, or remote seals; a dry leg that fills with condensate reads high.
- A 4-wire Class A Pt100 is the 20–200 °C custody-like choice; Type K or N belongs on oxidizing furnace exhaust near 850–1100 °C, not on the custody stream.
- Non-contact radar and ultrasonic level fail first on foam, low dielectric, and vapor-space speed-of-sound errors; GWR and nuclear are the usual backups, not a second hydrostatic tap.
- Grounded thermocouples are faster but inject sheath noise; specify ungrounded junctions beside VFDs and electrically noisy furnaces.
Why this is a PE specification problem
NCEES PE Control Systems (April 2027 specifications) puts Measurement at 16–24 questions. Domain 1.A is not intro physics. The exam asks you to specify a transmitter that survives the fluid, meets the accuracy and rangeability the loop actually needs, and does not hide a hydrostatic, density, or electrical error. If you cannot name the reference (atmosphere versus vacuum), the wet-leg live zero, or why a Type K couple is the wrong 150 °C custody sensor, you will miss items that look like 'sensor trivia' and are really installation and uncertainty problems.
Gauge, absolute, and differential pressure
Gauge pressure is referenced to local atmosphere. Most process 'pressure' transmitters are gauge capsules unless the datasheet says absolute. Use gauge for vessel maximum allowable working pressure (MAWP) in psig, pump discharge, utility headers, and hydrostatic level in open tanks.
Absolute pressure is referenced to a full vacuum. Use it whenever the engineering equation is written in absolute units: vapor–liquid equilibrium, vacuum-column overheads, condenser hot-well pressure, compressor suction for surge maps, and gas-law mass from P·V = m·R·T/M. Adding 14.7 psi to a gauge reading is an exam trap at altitude, in a weather-changing barometer, and in any vacuum service. Sealed-gauge capsules (internal vacuum reference, output still labeled in gauge units) are not a substitute for a true absolute measurement in VLE work.
Differential pressure (DP) is the difference between two process connections. Filter ΔP, orifice ΔP, and closed-tank level are differential measurements. Do not specify a single gauge transmitter and 'subtract atmosphere in software' when the two legs see different elevations, condensable vapor, or unequal fill fluid.
A smart DP transmitter is specified by calibrated span, upper range limit (URL), turndown, static-pressure rating, and accuracy as percent of span (or of URL—read which one the vendor quotes). A 150 inH2O URL capsule used on a 20 inH2O span is 7.5:1 turndown; percent-of-span error grows as you suppress the range. Overrange protection, a three-valve or five-valve manifold, and impulse-line slope are part of the specification, not accessories.
DP cells, diaphragm seals, capillaries, bellows, and Bourdon tubes
DP cells sense diaphragm or silicon-chip deflection between high (H) and low (L) ports and typically output 4–20 mA with Highway Addressable Remote Transducer (HART) digital superposition. Diaphragm seals (remote seals) isolate the sensor from corrosive, plugging, sanitary, or hot fluid. Capillary length, inner diameter, and fill fluid (silicone or a high-temperature fill) set ambient-temperature zero shift and response time: long capillaries add lag. Vacuum can vaporize the fill; high temperature can cook it. Two seals on a DP level or DP flow application must be matched—unequal capillary volumes look like a level change when the sun hits one side of the pipe rack.
Bellows and Bourdon tubes remain on local gauges and some mechanical recorders. They are not the default for a 4–20 mA control loop. Specify them for local indication, not for cascade temperature or compressor antisurge.
Hydrostatic and displacer level
Open-tank hydrostatic level is h = P / (ρ g). Density is the specification: a 10% density swing is a 10% inferred-level error if you freeze ρ. Closed tanks require DP. A dry leg ties the low side to the vapor space and works only if that vapor does not condense in the impulse line. Wet legs on condensable vapors (steam drums, hydrocarbon fractionators) keep the low-side leg full of a known-density liquid; the live zero is ρ_leg · g · H_leg and must be calibrated out. A dry leg that slowly fills with condensate reads high, and the operator stops trusting the glass.
Displacer transmitters (Archimedes) infer level or interface from buoyant force. They are strong on clean interfaces with a known density pair and weak on fouling, density-varying, or two-phase services. Range is limited by displacer length; do not stretch a 14-inch displacer across a 10-foot boot.
Radar, ultrasonic, capacitance, and nuclear level
Guided-wave radar (GWR) is time-domain reflectometry along a probe. It is largely density-independent and often sees through light foam better than non-contact radar, but coating, bridging between probe and stilling well, and very low dielectric constants (εr ≈ 1.4–2 for light hydrocarbons) can swallow the echo unless you use a coaxial probe. Non-contact radar (frequency-modulated continuous-wave) looks down from a nozzle. Heavy foam, agitation, and low εr are the classic misses. Ultrasonic level uses speed of sound in the vapor space; temperature, vapor composition, dust, and foam all bias the path—do not aim it into a steaming condensate tank without path compensation and a clear vapor. Capacitance level needs a stable dielectric and hates conductive coating. Nuclear (gamma) level or density is a last-resort through-wall measurement: high installed cost, source licensing, and leak-test programs, justified on molten, coking, or extreme-pressure vessels where no wetted sensor can touch the process.
Temperature: RTDs, thermocouples, and the rest
A resistance temperature detector (RTD), almost always Pt100 (100 Ω at 0 °C, International Electrotechnical Commission (IEC) 60751 α = 0.00385 /°C), is the default from about −200 °C to 600 °C when you need stability. Class A tolerance is ±(0.15 + 0.002|t|) °C; at 200 °C that is ±0.55 °C. Two-wire connections add lead resistance (on the order of 0.4 Ω per 100 ft of 18 AWG copper, about 1 °C). Three-wire cancels equal lead resistances and is the plant standard. Four-wire Kelvin sensing cancels unequal leads and is what you specify for custody-like, laboratory, or compressor-efficiency tests.
Thermocouples generate a millivolt from the Seebeck effect. Specify type by atmosphere and temperature, not habit:
| Type | Why it wins | Where it loses |
|---|---|---|
| J (iron–constantan) | Cheap, reducing atmospheres, healthy millivolts | Iron rusts in oxidizing/humid service; practical use often to ~750 °C |
| K (chromel–alumel) | Workhorse to ~1100–1200 °C in oxidizing gas | Drift and 'green rot' in reducing/sulfur service |
| T (copper–constantan) | Tight limits, cryogenic to ~350 °C | Upper range |
| E (chromel–constantan) | Highest common base-metal sensitivity | Spare-parts rarity versus K |
| N (nicrosil–nisil) | Better high-temperature stability than K | Cost versus K |
| S / R (Pt–Rh / Pt) | Accuracy and oxidizing service to ~1600 °C | Cost, slow, reducing atmospheres attack platinum |
| B (Pt30Rh–Pt6Rh) | Very high T; weak room-temperature emf so cold junction is less critical | Poor below ~600 °C |
Thermistors give a large resistance change over a narrow band (bearings, battery rooms)—not furnace control. Filled systems (liquid- or vapor-filled capillaries) add lag and leak risk; mercury-filled systems are largely retired on environmental grounds. Infrared (IR) pyrometers need known emissivity and a clear line of sight; steam, dust, and scale fool them. Bimetal thermometers are local indication only. Grounded thermocouple junctions are faster and couple sheath noise into the input card; use ungrounded junctions on variable-frequency-drive (VFD) skids and electrically noisy furnaces, accepting a slower time constant. Always check thermowell wake frequency (American Society of Mechanical Engineers (ASME) PTC 19.3 TW) on high-velocity gas or steam; a snapped well is a process-safety event, not a measurement footnote.
Sensor comparison (selection, not catalog shopping)
| Duty | First-choice sensor | Cost class (instrument only, typical U.S. 2026) | Exam trap |
|---|---|---|---|
| Open-tank level, constant density | Hydrostatic gauge / DP | $800–2,500 transmitter | Ignoring density swing |
| Steam-drum / condensable vapor level | DP with wet leg, pots, or remote seals | Transmitter plus $2,000–6,000 seals | Dry leg that fills |
| Changing density, light foam | GWR | $3,000–8,000 | Coaxial probe omitted on low εr |
| Heavy foam, no nozzles | Nuclear gamma | Tens of thousands plus licensing | Using non-contact radar 'because it is cheaper' |
| 20–200 °C accuracy | 4-wire Class A Pt100 | $80–400 plus transmitter | Type K 'because we stock it' |
| 850–1100 °C oxidizing flue | Type K or N in metal/ceramic sheath | Couple cheaper than RTD; well and transmitter dominate | Pt100 out of range; Type B at 400 °C |
| Local utility pressure | Bourdon gauge | $50–200 | Using it as the cascade PV |
Worked example: wet-leg span, then RTD versus Type K
Steam-drum DP. The high-side tap is at the lower (water) nozzle. The low-side tap is in the vapor space. A wet leg stands 50 in of condensate. The useful water-level span is 40 in. At 0% level, ΔP = 0 − 50 = −50 inH2O. At 100% level, ΔP = 40 − 50 = −10 inH2O. Calibrate the transmitter −50 to −10 inH2O (elevated/reverse zero), not 0–40 inH2O as if it were a dry-leg open tank. If the wet leg partially empties, the live zero moves and the indicated level is wrong even though the capsule is healthy.
Custody-like 20–200 °C versus furnace exhaust. A finished hydrocarbon stream, 20–200 °C, target ±0.5 °C, 3-year drift under 0.2 °C, 4-wire, thermowell: specify a Class A Pt100. Type K special limits are the greater of ±1.1 °C or ±0.4%; at 200 °C even special-limits K is about ±1.1 °C before cold-junction and extension-wire error. The RTD meets ±0.5 °C; the couple does not. The same plant's furnace exhaust at 850–1050 °C in oxidizing flue gas is the opposite: a Pt100 is out of range (thin-film often ≤600 °C; industrial wire-wound typically ≤850 °C and not for flue). Specify Type K in a ceramic or Inconel sheath, or Type N if the heater cycles and K has already drifted. Reducing or sulfurous atmospheres are not 'just use K'—move to N or a noble-metal couple with the correct protection tube.
Exam traps
Using gauge pressure in PV = nRT; dry legs on steam; radar into heavy foam; grounded TCs next to VFDs; hydrostatic level without a density spec; 2-wire RTDs on a 200 ft run; Type B below 600 °C; unmatched remote-seal capillaries on opposite sides of a pipe rack.
A vacuum distillation overhead pressure is used in a vapor–liquid equilibrium calculation and in a gas-law mass balance. Which pressure type must the transmitter report?
A steam-drum level transmitter uses a DP cell with the low-side tap in the vapor space. What installation keeps the measurement valid on condensable steam?
A finished-product line must hold ±0.5 °C from 20–200 °C. A second point is oxidizing furnace exhaust at 850–1050 °C. Which pairing is the PE-correct specification?