Comment

01 June 2003

Avoid Process Poison

By Mark McErlean

Natural gas challenges the dimensions of contemporary oxide water detection.

Measuring water vapor in gases is difficult even in pure gases—the best of circumstances.

Technologies such as automatic or manually cooled (chilled) mirrors, Karl Fisher titration, oscillating crystals, infrared absorption, metal oxide or polymer capacitive film, electrolytic, and others are used to perform water-vapor content measurement.

The prices range from many tens of thousands of dollars to under a thousand dollars. The feasibility of each technology for an application depends on many factors: measuring range, response time, accuracy, long term stability, temperature coefficient, susceptibility to contaminants, approvals for use in hazardous areas, minimum detection limits, and cost of ownership.

Various manufacturers have addressed the above-mentioned factors by methods best suited for their specific technologies. However, susceptibility to contaminants, including interference from the gas under measurement, still presents a significant problem for these instruments.

Major water interference

Here is an approach, borrowed from other measurement technologies, to accurately and reliably measure the water-vapor content in a potentially contaminated gas using a hyper-thin film (HTF) aluminum-oxide sensor.

It is desirable for the measurement to have National Institute of Standards and Technology (NIST) traceability. Here, we'll look at specific examples for measuring moisture in natural gas after it passes through a glycol drying station.

We use natural gas as the example because it presents challenges to all of the technologies mentioned above.

The hygroscopic nature (readily taking up and retaining moisture) of glycol provides an effective method for drying natural gas, but it also provides a significant interference with the water-vapor equilibrium when utilizing the capacitive and electrolytic technologies.

Other contaminants in natural gas can also interfere or, in the case of corrosives, even damage most of the sensing technologies.

The infrared absorption of water vapor occurs at many wavelengths, however methane's absorption is always nearby and thus causes interference.

The presence of hydrocarbons, which may coalesce at higher temperatures than water, interferes with an automatic chilled mirror's ability to perform accurate measurements, as do particulate contaminants.

Oscillating crystals are also very sensitive to contamination, and are best suited for in-lab use.

Karl Fisher titration will perform accurate measurements, but is best suited for laboratory or single sample use. It is not practical as an online continuous-duty measurement.

Accepted alarming point

The strategy used to provide a cost effective and traceable quality water-vapor measurement is borrowed from other process instrumentation.

A valve appears in the sample system, such that under the instrument's control, the sensor can switch from the process gas to a NIST-traceable nitrogen/water blend bottle.

The known water content of the bottle enters into the instrument's memory, together with a time schedule for verification and recalibration. The unattended instrument follows this schedule and performs the programmed task of recalibrating.

Thus the measurement near the water content of the bottle essentially has NIST-traceable accuracy even if the contaminants in the process gas have caused the sensing element to drift.

As the measurements get further and further from the calibration point, the accuracy diminishes slightly, but by using appropriate sensor technology, these variations are manageable and the instrument remains within the specified tolerance.

This method is particularly attractive for uses where there is an accepted alarm point, which for natural gas, for example, is seven pounds of water per million standard cubic feet.

When an alarm generates, one can have confidence in the accuracy of the measurement and quickly resolve suspicions by verification and recalibration against the traceable bottled standard.

Eliminate the ignition source

There were five challenges in constructing this system:

Challenge one: Provide a consistent nitrogen/water blend throughout the pressure life of the bottled standard. The higher the water content, the more difficult it becomes to release a constant concentration from the standard.

Solution: A manufacturer who produces specially lined high-pressure bottles that will release a constant nitrogen/water concentration came on board. Two bottles were tested with a NIST-traceable chilled mirror and proved to provide a consistent blend throughout the specified bottle pressure, for ambient temperatures as low as 60°F. For colder ambient temperatures a bottle warmer was added, and the bottle continued to release a constant concentration.

Challenge two: Find and use an extremely low-leakage electrically actuated three-way valve, which is approved for use in Class 1 Division 2 areas. Solenoid valves are not sufficiently tight and can corrupt the measurement or calibration. Air actuated valves may not be practical for areas that do not have instrument air. Many natural gas measurements take place in hazardous classified areas, while commercially available motorized three-way valves do not have approval for use in these areas.

Solution: Engineers developed a motorized actuator to operate a Swagelok SS-43XS4 three-way ball valve. This valve is essentially leak-free for these measurements. This actuator utilizes a 24-volts-direct-current stepper motor, thus eliminating the ignition source of motor commutator brushes and facilitating hazardous area use approval.

Challenge three: Assure that the contaminants do not damage the sensor or degrade it such that even the recalibration is not sufficient to provide an accurate measurement. Conventional aluminum-oxide capacitive sensors are highly affected by glycol's hygroscopic properties. Contaminants easily damage them, and they are prone to drift due to metal migration.

Solution: Utilizing an HTF aluminum-oxide sensor afforded several orders of magnitude greater sensitivity compared to conventional aluminum-oxide sensors. The greater sensitivity prolongs the sensor life by reducing the effects of the glycol contamination. Testing showed that several years of high performance usage is normal for these sensors in typical natural gas measuring environments.

Challenge four: Assure that instrument drift is minimal between recalibration. Conventional aluminum-oxide sensors can have significant drift with temperature changes. Periodic recalibration will not assure that the instrument is consistently reporting the correct measurement over daytime to nighttime temperature variation.

Solution: Testing showed that HTF sensors have negligible temperature drift and thus are ideally suited for these kinds of harsh environment applications.

Challenge five: Assure a low-maintenance requirement and ease of use. Measurements in natural gas typically require significant maintenance at frequent intervals. For example, conventional aluminum-oxide sensors may need monthly replacement. Other technologies have inherent drifts that also require monthly service and recalibration by well-trained personnel.

Solution: Testing showed that recalibrating once every two weeks is sufficient to maintain the specified accuracy. The calibration gas bottle can provide in excess of 50 calibrations worth of gas. The manufacturer guarantees the bottle for only one year, and thus the bottle can serve for a year at a once-per-week calibration rate.

Oxide layer style advances

The success of this approach to moisture measurement centers on the high-sensitivity HTF sensor.

This self-calibrating natural gas sample system takes advantage of the uniquely large and quasi-linear response to moisture of Xentaur HTF sensors. The large response to moisture of HTF sensors is an absolute requirement for adjusted moisture computations based on a single point calibration.

HTF and all other aluminum-oxide sensors share the same operating principle. The capacitance measured between the sensor's aluminum core and a gold film deposited on top of the oxide layer varies with the amount of water vapor in the pores of the oxide layer.

Three fundamental structural improvements in the oxide layer give HTF sensors greatly increased sensitivity and stability: a much thinner oxide layer, a better defined barrier layer between the aluminum and the aluminum oxide, and a unique pore geometry enhancing the entrapment of water molecules.

Hyper-thin layer: With HTF technology, making sensors with hyper-thin oxide layers—without compromising insulation strength—is possible. The thinner oxide layer of HTF sensors results in more pronounced capacitance changes, because capacitance is inversely proportional to the distance of the capacitor's plates from each other.

The thinner layer also means that water molecules will travel faster in and out of the pores. HTF aluminum-oxide sensors therefore respond several times faster than conventional sensors.

Barrier layer: In HTF sensors, the transition between the aluminum oxide and the aluminum is sharp and clearly defined. This thinner barrier layer produces a capacitor with its electrodes very close together, which in turn causes the sensor's wet-to-dry capacitance ratio to be high.

The benefit of a high wet-to-dry capacitance ratio is that drift in capacitance due to undesirable factors is less significant. This is important in combating temperature sensitivity problems and aging drift.

The sharp transition from aluminum to aluminum oxide also reduces metal migration, one of the major causes of aging drift in conventional sensors.

Pore geometry: The most significant difference between HTF sensors and conventional sensors is, however, their pore geometry. Although conventional sensors rely on hygroscopic aluminum-oxide structures to attract water, HTF sensors rely on a pore geometry that slows the Brownian motion of the water molecules when entering the pores.

Brownian motion is the random movement of microscopic particles suspended in liquids or gases that results from the impact of molecules from the fluid surrounding the particles.

The freed energy becomes part of the mass of the sensor, and the decreased entropy of the water molecules equalizes as to increase their total number. This results in more dielectric in the pores and consequently a higher capacitance.

The HTF pore geometry does not significantly change over time, while conventional hygroscopic aluminum-oxide structures are not stable and collapse slowly into nonhygroscopic structures.

Thus, while conventional sensors are subject to drift and need to recalibrate frequently, HTF sensors need no recalibration when used in clean, noncorrosive gasses.

No technology solves all problems or is perfect for all applications. However, for moisture content measurements in harsh environments that require a high degree of accuracy near a single point, with good accuracy over a wide range, this automatic system is an excellent solution.

Recalibration alone cannot compensate for all sensor weaknesses, but it does perform with distinction within acceptable parameters using HTF aluminum-oxide sensors. ST

Behind the byline

Mark McErlean is a manager at Cosa Instrument Corporation (www.cosaic.com).