Compressed Air Flow Meters and Dew Point Sensors an Instrumentation

The U.S. DOE Compressed Air Challenge program has assessed thousands of industrial facilities since 1998. Fewer than half had permanent flow metering. Dew point monitoring was rarer.

Thermal Mass Flow Meters

Thermal mass flow meters are calibrated against dry air. The CS Instruments VA 520 manual puts humidity correction factors in an appendix that most buyers never open. When the gas passing the sensor carries more moisture than the calibration reference assumed, thermal conductivity shifts and the meter overreads. At around +10°C pressure dew point on a 7 bar system the overreading is around 4%, which on a 2,000 Nm³/h header is 80 Nm³/h that doesn't exist, sitting in the historian, getting trended, getting reported. A dew point sensor in the same pipe section catches this. Without one, every number downstream of that meter inherits the bias and nobody can tell.

Refrigerated dryers don't alarm when they lose capacity. The +3°C PDP nameplate rating assumes rated inlet conditions. Push the compressor room past 38°C and the evaporator can't hold temperature. This happens every summer in facilities with un-air-conditioned compressor rooms. The flow meter keeps logging.

Now the part that most compressed air articles cover in a sentence or two and shouldn't.

POE Contamination of Capacitive Polymer Dew Point Sensors

POE contamination of capacitive polymer dew point sensors has cost more money in misdiagnosed dryer maintenance than any other single instrumentation failure mode in compressed air systems. That claim comes from accumulated field service data at Vaisala and CS Instruments, not from a published study, because nobody has funded a published study on it. The mechanism is specific enough to trace step by step.

POE is hygroscopic. This is well documented in ASHRAE literature and refrigeration engineering textbooks, where the concern is acid formation inside HFC refrigerant circuits from moisture absorbed by the lubricant. The compressed air industry encounters the same chemistry from a different angle. POE is the factory lubricant fill in Atlas Copco SF scroll compressors, Hitachi SRL scroll compressors, and various other machines. The lubricant enters the compressed air as a fine aerosol. Coalescing filters rated at 0.01 micron capture most of it. Not all. Even below 0.01 mg/m³ of residual carryover, over weeks and months, a film accumulates on surfaces downstream. One of those surfaces is the polymer sensing chip of whatever dew point sensor is installed in the line.

The Vaisala DMT152 is probably the most widely installed capacitive polymer dew point sensor in compressed air service worldwide. It is also the sensor on which POE contamination has been most thoroughly documented by Vaisala's own field service organization. When POE residue sits on the DMT152's HUMICAP polymer element, the oil film attracts water molecules from the air independently of the polymer underneath. The sensor's electronics cannot distinguish moisture held in the oil from moisture in the air. The reading shifts wet. Vaisala's field service records include cases where the bias exceeded 7°C, visible under 10x magnification as a faint yellowish discoloration on the sensor chip. The discoloration is subtle enough that someone unfamiliar with what a clean HUMICAP chip looks like would not notice it.

The bias is stable. It doesn't drift, doesn't fluctuate with load changes, doesn't respond to dryer maintenance. It looks like a dryer that's performing below spec. Maintenance teams, operating from the reasonable assumption that the sensor is reporting the air condition, go after the dryer. On a typical 1,000 cfm heatless desiccant unit, desiccant replacement runs $8,000 to $15,000 depending on adsorbent type and tower size. Switching valve overhaul adds labor. Compressed air service contractors in the UK have estimated total unnecessary maintenance spend triggered by this specific misdiagnosis at £20,000 to £30,000 per incident before someone inspects the sensor.

Cleaning is isopropanol and a lint-free wipe, 30 minutes dry-down before reinstallation. If the contamination has been in place longer than about six months, the POE may have caused partial swelling of the polymer matrix. At that point the sensor goes back to Vaisala in Helsinki or CS Instruments in Tannheim for factory recalibration. Lead time 2 to 4 weeks.

Mineral oil from legacy rotary screw compressors isn't hygroscopic and doesn't cause this problem.

7 °C

Max observed POE bias

£20–30k

Unnecessary spend per incident

2–4 wks

Factory recalibration lead time

Sampling Tubing

This is the other area where compressed air dew point measurement goes wrong far more often than the sensor itself, and it gets almost no coverage in application guides because it falls in the gap between the sensor manufacturer's scope (the sensor) and the compressor manufacturer's scope (the compressor and dryer). Nobody owns the two meters of tubing between the process pipe and the sensor.

Nylon tubing is permeable to water vapor. This isn't a compressed air observation. It's a material property documented in polymer engineering references and in permeation data published by Swagelok and Parker for their tubing product lines.

Over two meters of 6 mm OD nylon at a sample flow below 0.5 L/min, ambient moisture permeating inward through the tubing wall biases the measured dew point by more than the accuracy specification of any capacitive sensor on the market. In a 35°C compressor room at 60% relative humidity, the bias can reach 10°C. In an air-conditioned instrument room the bias is smaller. The Vaisala DM70 portable dew point meter ships with a direct insertion probe as standard for field verification work. Vaisala's application engineers moved to direct insertion specifically because sampling-system artifacts were corrupting their own field verification measurements often enough to be a recurring problem. That tells you what you need to know about whether nylon sampling tubing is acceptable for permanent monitoring.

Stainless steel tubing eliminates permeation. Direct sensor insertion through a ball valve retraction fitting eliminates the sampling path. Both Vaisala and CS Instruments sell the ball valve assemblies. For permanent installations, direct insertion.

Condensation inside sampling tubing during a wet transient is a separate failure from permeation. Liquid water coats the inner wall and re-evaporates into subsequent dry samples for hours.

Sensor Fouling and Filter Schedules

Sensor fouling from compressor lubricant on thermal flow meter elements causes the reading to drift low over time as the oil film adds thermal resistance. How fast varies enormously between installations. A clean system with fresh Parker Domnick Hunter OIL-X elements might drift less than 1% per year. A system where filter changes got deferred drifts faster. How much faster depends on compressor discharge temperature, lubricant type, gas velocity across the sensor, and whether the replacement filter elements are OEM or pattern parts with different fiber packing. There is no useful general number.

The instrumentation verification schedule and the filter maintenance schedule are the same problem managed by different people in different software systems. When the filter element differential pressure climbs, the instruments downstream are already being exposed to elevated contamination. Connecting these two schedules is an organizational problem more than a technical one.

Sensor Response Time and Desiccant Dryer Cycles

The Vaisala DMT152 specifies T63 response at about 10 seconds at +20°C dew point. At -40°C PDP, T63 stretches to minutes and final settling takes 20 to 30 minutes. Water enters a dry polymer readily. Getting it out at low concentrations is slow because the vapor pressure gradient driving diffusion is shallow.

This matters on heatless desiccant dryers with short cycle times. A Parker PNEUDRI on a 5-minute half-cycle switches towers every 5 minutes. Each switch produces a brief moisture spike. The DMT152 starts responding, hasn't settled, and the next switch hits. The displayed value never represents the steady-state dew point between switches. It sits above it. On PNEUDRI units with 5-minute half-cycles the offset compared to chilled mirror readings at the same measurement point has been observed at 3°C to 5°C. Heated desiccant dryers with 4-hour cycles don't produce this effect because 30 minutes of settling is a small fraction of the cycle.

Operators who see transient dew point excursions self-correct after a few minutes develop the habit of waiting them out instead of responding immediately. When a sustained dryer failure produces an elevated reading that won't self-correct, the first 10 minutes look identical to every transient they've been ignoring. The fix is procedural: any alarm not clearing within a defined time window triggers investigation regardless of past false alarm history.

Galvanized Piping as a Moisture Source

Galvanized carbon steel piping corrodes internally into porous zinc carbonate and zinc hydroxide deposits when moisture contacts the zinc coating. These deposits absorb moisture during wet transients and release it back into the air stream for hours after the dryer has recovered. The CAGI Compressed Air and Gas Handbook, 7th edition, covers this in the distribution systems chapter. Sensor at the dryer outlet reads -40°C. Sensor 200 meters downstream in galvanized piping reads -25°C. The dryer is working. The pipe is a secondary moisture source. Without measurement at both points, the downstream reading gets attributed to the dryer. Replacement with aluminum piping (Transair, AIRnet) eliminates the moisture memory.

Desiccant dust: sub-micron alumina fines from adsorption dryers, hygroscopic, pass through after-filters, embed in polymer sensor surfaces. Twenty to thirty pipe diameters of spacing between the after-filter outlet and the sensor reduces exposure. This comes from field practice. No manufacturer publishes it.

Temperature stratification in DN200 headers running through unheated spaces creates a gradient of several degrees between pipe top and bottom during winter. Sensors end up at 12 o'clock on the pipe when nobody specifies the clock position in the installation drawing, because 12 o'clock is where the ladder reaches. They should go at 3 or 9 o'clock.

Twenty to thirty pipe diameters of spacing between the after-filter outlet and the sensor reduces exposure. This comes from field practice. No manufacturer publishes it.

Flow Measurement, Leakage, and Calibration

Vortex meters can't quantify overnight leakage because they lose signal at low flow. DP devices waste energy. Clamp-on ultrasonics from Flexim or Katronic suit temporary surveys.

Flow during zero-production periods minus dryer purge equals system leakage. Purge fraction per model size is in Atlas Copco CD series and Parker PNEUDRI documentation, 15 to 20% of rated throughput on heatless units. Trend the result monthly. ISO 11011:2013 requires both flow and quality measurements in supply-side assessment.

Calibration verification intervals set by observed drift at each installation, not by calendar. Dew point verification below -20°C PDP requires chilled mirror reference (MBW 473, Michell S8000) or dew point generator (Michell ADG400, Thunder Scientific 3900). Spare sensor rotation avoids gaps.

Measurement station: flow, dew point, line pressure, temperature at the same location. Pressure for PDP-to-ADP conversion, temperature for normalizing flow to Nm³.