Trace moisture analyzer instrument, gas sampling and analyzing system, and method of detecting trace moisture levels in a gas

Abstract

An analyzer instrument, gas sampling and analyzing system, and method are disclosed for detecting water vapor in gas using tunable diode laser absorption spectroscopy at a wavelength of 1871 nm plus or minus 2 nm.

Claims

1. An analyzer instrument for detecting water vapor within a natural gas and/or methane gas stream comprising: a light source positioned to emit light and pass said light through the gas stream, and a detector is positioned to detect the intensity of light passing through the gas stream, said light source configured to emit light consisting of a frequency corresponding to an absorption line of water at a wavelength of 1871 nanometers plus or minus 2 nanometers, and an electronic circuit coupled to said detector for determining the level of water vapor in the gas based on the detected intensity.

2. An analyzer instrument for detecting water vapor within a natural gas and/or methane gas stream comprising: a cell member having a laser diode for holding the gas stream, said laser diode having a temperature controller that maintains the laser diode at a predetermined set-point temperature, a mirror system configured to reflect the light internally within the cell member so the light makes multiple passes prior to exiting the cell member, a photo-detector positioned at an end of the cell member to detect the intensity of light passing through the gas stream and exiting the cell member, and a control circuit coupled to said detector for determining the level of water vapor in the gas stream based on the detected intensity including a microprocessor configured to set the parameters of the laser diode and the operation thereof so the laser diode emits light consisting of a wavelength of 1871 nanometers plus or minus 2 nanometers, and a display device for showing the amount of water vapor detected in the gas.

3. In an absorption spectroscopy method for detecting water vapor within a natural gas and/or methane gas stream comprising: passing through the gas stream light consisting of a frequency corresponding to an absorption line of water at a wavelength of 1871 nanometers plus or minus 2 nanometers, and detecting the intensity of light passing through the gas sample and determining the level of water vapor in the gas stream based on the detected intensity.

Description

DESCRIPTION OF THE DRAWING

(1) Some embodiments of my instrument and method are discussed in detail in connection with the accompanying drawing, which is for illustrative purposes only. This drawing includes the following figures (Figs.), with like numerals and letters indicating like parts:

(2) FIG. 1 is a gas (methane and higher molecular weight hydrocarbons and water) enlarged absorption spectra in the 1870-1880 micron region with methane absorption lines shown in dotted lines and water vapor absorption lines shown in solid lines, contrasting the wavelengths of 1871 microns verses 1877 microns.

(3) FIG. 2 is the 2f spectrum at the 1871 nm region used in my instrument and method for different concentrations of mixtures of methane and water vapor.

(4) FIG. 3 is a schematic diagram of one embodiment of my analyzer instrument.

(5) FIG. 3A is a schematic diagram of an alternate embodiment of my analyzer instrument.

(6) FIG. 4 is a front view of a system employing my analyzer instrument contained in one housing and in a separate housing a gas sampling device.

(7) FIG. 5 is a side view of the system shown in FIG. 4 taken along line 5-5 of FIG. 4.

(8) FIG. 6 is a side view of the gas-sampling device used with my system shown in FIGS. 4 and 5.

(9) FIG. 7 is a schematic diagram of the fiber optical connection between a laser within a laser housing and the explosion resistant gas-sampling device in the other housing.

GENERAL

(10) Natural Gas:

(11) Main line natural gas consists primarily of methane (CH.sub.4, 85-98%), with typically 0-4% carbon dioxide (CO.sub.2), 0-6% ethane (C.sub.2H.sub.6), and smaller amounts of higher hydrocarbons such as propane, butane, etc. In terms of spectral interference when using second harmonic (2f) detection, the most troublesome compounds are the lower molecular weight molecules because these gases have sharper absorption features that are more separated in wavelength than the heavier gases (higher molecular weight) the more densely packed the absorption lines are, and at typical measurement pressures near 1-2 bar the absorption spectrum consists of broad humps rather than sharp and narrow absorption features for which 2f detection is optimized. So for butane (C.sub.4), pentane (C.sub.5) and higher hydrocarbons spectral interference issues are usually not a problem as long as the wavelength chosen for measurement of H.sub.2O is well away from a strong band of these gasses. CO2 has no absorption bands near 1871 nm so there is zero spectral interference under any circumstances. The 1871 nm band is located within a gap in the spectrum where only a very weak band is present.

(12) Wavelength Considerations:

(13) FIG. 1 shows examples of Fourier transform absorption spectra for CH.sub.4 and H.sub.2O in the region most suitable for measuring H.sub.2O in natural gas. Interference by CH.sub.4 is a major problem throughout the infrared and near infrared wavelength region, and it is the key consideration when choosing an H.sub.2O feature to measure. The region short of 1830 nm is not suitable for measuring H.sub.2O because the CH.sub.4 absorption is so strong that nearly 100% of the light is absorbed for even modest absorption path lengths. Beyond about 1940 nm the H.sub.2O band starts to tail off and the absorption lines get very weak. The sweet spot for H.sub.2O measurement is therefore between about 1830-1940 nm.

(14) FIG. 1 is a blowup of the 1850-1900 nm region with several target H.sub.2O lines indicated and the companies using them based on published information. The results of my improved instrument (AMI Model 4010BR FIGS. 4 and 5) are designated as AMI. None of these target H.sub.2O lines are completely clear of CH.sub.4, and in principle any of them could be used to measure H.sub.2O (there is no interfering CO2 absorption in this region to worry about). No matter which H.sub.2O line is used, the spectrum processing algorithms have to manage the interfering absorption by CH.sub.4, and this can be done in similar ways for any of these target H.sub.2O lines. Ametek has chosen 1854 nm (presumably) because it is the strongest H.sub.2O line in the region, but it is also the most interfered with by CH.sub.4. SpectraSensors is using 1877 nm, which is a slightly weaker line but less interfered with by CH.sub.4. The 1871 nm line is of similar strength to 1877 nm. Moreover, the 1871 nm line consists of two adjacent H.sub.2O lines that overlap. This is an advantage for area under the curve processing and produces a greater net signal to work with than simply using the amplitudes of the signals. Nevertheless, using the amplitudes of the signals is also a viable approach (and both can be done in software simultaneously).

(15) 2f Spectroscopy:

(16) A common technique used for laser-based gas sensing is called second harmonic detection, or 2f detection. This technique provides increased detection sensitivity at the expense of a more complicated control electronics subsystem, and a somewhat more difficult spectrum analysis process. It has been widely used since the 1970's with tunable diode lasers, and is the approach taken here.

(17) FIG. 2 shows examples of 2f spectra at the 1871 nm region in a small multi-pass absorption cell built for AMI. The horizontal axis is proportional to wavelength, which changes smoothly as the laser current is swept. The vertical axis is the 2f signal amplitude, and this spectrum contains absorption lines of both H.sub.2O and CH.sub.4. This is the raw data that is processed for the H.sub.2O concentration. On the left shows a region where there is no H.sub.2O absorption. This region can be used to determine the CH.sub.4 concentration, as well as to provide a reference signal to compare against the region to the right that contains the target H.sub.2O lines (as well as underlying CH.sub.4 lines). The different shades of gray depict different levels of H.sub.2O, and it is the difference in the amplitudes and/or areas of these features that are used to quantify the amount of H.sub.2O in the gas sample.

(18) Signal processing of these types of 2f spectra can proceed via two basic methods. One is to use the difference in amplitudes of the various bumps across the spectrum and to relate those to the H.sub.2O concentration. The lower detection limit is determined by the smallest signal differences that can be reliably measured. The second is to integrate the separate oval regions shown in FIG. 1 to get the area under the curve of each section, then to ratio these to derive the H.sub.2O concentration. Since both of these signal or spectrum processing methods are implemented purely in the software program 12a (FIG. 3), both can be utilized and compared to determine which is most appropriate for my instrument. Both pressure (in particular) and temperature (to a lesser extent) impact the exact shape of the 2f spectral features. Since neither of these parameters pressure (P) and temperature (T) are held constant in the AMI instrument, it is necessary to build into the software program 12a calibration polynomials that quantify the changes in the 2f spectrum with pressure and temperature so that the correct H.sub.2O value can be extracted.

(19) From measurements to date, the lower detection limit for H.sub.2O is in the range 0.25 pounds of H.sub.2O per million standard cubic feet of CH.sub.4 (lb./mmscf). This standard unit used in the natural gas industry equals 20.5 parts per million by volume (ppmv). This is without any active control of the pressure (P) and temperature (T). The spectrum signal-to-noise ratio (SNR) produces a lower detection limit of approximately 0.25 lb. at a fixed P and T. Measurements still need to be made over ranges in P and T before this can be confirmed over the full expected operational range, but if the base spectrum SNR is sufficient for a lower detection limit of 0.25 lb. or better, then it should be possible via appropriate lab calibrations to easily maintain this level in the field. The typical specification for a base instrument such as this is a 0-20 lb. range with a lower detection limit of 0.5 lb. My analyzer instrument cuts the minimum detection limit in half

(20) Methane Measurement:

(21) As shown in FIG. 1, there is a peak at the left side of the 2f spectrum that is caused by the absorption of laser light from methane (CH.sub.4) gas in the sample cell. Because this isolated CH.sub.4 peak is within the same laser scan as the desired water absorption peak at 1871 nm it is possible to accurately measure the CH.sub.4 concentration in addition to the water concentration.

(22) Peak Subtraction:

(23) As FIG. 1 depicts, as the water concentration is increased the H.sub.2O peak height increases. The H.sub.2O peak height is directly proportional to the water concentration. Note, however, that even with dry CH.sub.4 there is still some absorption at the location of the H.sub.2O peak due to the underlying CH.sub.4 absorption. Typically, a constant value is subtracted from the H.sub.2O peak height to account for the underlying CH.sub.4. However, if the amount of CH.sub.4 changes that constant peak height is no longer the correct value to subtract. Variation in the underlying CH.sub.4 peak height adds an uncertainty to the calculation of the water concentration and reduces the accuracy of current instruments. Since it is desirable to have the highest possible accuracy, the uncertainty in the underlying CH.sub.4 peak height should be reduced. The spectrum includes a peak at the left side that does not change with water concentration. The height of this peak as well as the height of the water peak is measured by a microcontroller or microprocessor 12 that has the program 12a that processes a signal from light detector 7 as shown in FIG. 3. It is known that the various CH.sub.4 peak heights are always in a constant ratio to each other. Therefore, by measuring the left hand CH.sub.4 peak one can calculate the exact height of the CH.sub.4 peak that lies at the water peak location. Subtracting out this exact peak height instead of a constant value reduces uncertainty to the calculation of the water concentration, and therefore, the desired increase in accuracy. This calculation is automatically conducted by the microprocessor 12 according to its program 12a.

(24) Integrated Laser Package:

(25) Typical TDLAS analyzers use a laser that is packaged alone in a TO type can or package P (FIG. 3). The temperature of the laser is controlled by controlling the temperature of a heat sink to which the laser package is attached. In practice, there is always some temperature between the heat sink and the actual laser chip. Since the laser wavelength is strongly dependent on the laser chip temperature there will be some variation in the temperature control of the laser chip. The instrument I has a laser chip that is mounted directly on top of a thermoelectric cooler 8 inside of the laser package. This tight coupling between the thermoelectric cooler 8 and the laser chip provides a more stable laser temperature (and therefore instrument stability) over a wide range of ambient operating temperatures.

DETAILED DESCRIPTION OF SOME ILLUSTRATIVE EMBODIMENTS

FIG. 3

(26) As depicted in FIG. 3, one embodiment of my instrument is designated by the letter I. In my instrument I, near-infrared light from a laser diode 1 is collimated by an aspheric lens 2 and passes through a Herriott cell 3. The light makes multiple passes through the Herriott cell 3 because of reflections between concave mirrors 4 and 5. The light exiting the Herriott cell 3 is focused by another aspheric lens 6 onto an extended InGaAs photo-detector 7. When the wavelength of light from the laser corresponds to an absorption line of the gas in the Herriott cell 3, the intensity of light reaching the InGaAs detector 7 will be reduced.

(27) The wavelength of the laser is determined by its temperature and drive current. The temperature of the laser is determined by a thermoelectric cooler 8 that is in contact with the laser substrate and mounted internally to a laser package P. The thermoelectric cooler 8 is controlled by the output of the PID temperature controller circuit 9. Feedback from the thermistor temperature sensor 10, also in thermal contact with the laser's substrate, maintains the laser diode 1 to better than 0.1 C. of the set-point temperature. The thermoelectric cooler 8 will also heat when the controller circuit's 9 output voltage is reversed. By using an H-bridge circuit for the output of the temperature controller 9 the laser diode 1 can be maintained at a precise temperature, ideally around 30 C., in ambient temperatures between 10 C. and 50 C. The exact temperature to be maintained depends on the specific laser characteristics.

(28) The laser current is provided by an emitter follower output of the laser drive circuit 11. This current is the sum of a current ramp 13 that scans the laser output wavelength over a narrow range, typically less than 1 nm, in a few tenths of a second and a small sine-wave modulation at a high frequency on the order of 10 KHz. The sine wave modulation is produced by dividing down a square-wave signal at twice the frequency. All of the laser current parameters are set by the microprocessor controller 12. Those parameters include the starting and ending currents of the current ramp, the speed of the current ramp, and the amplitude and frequency of the square-wave. Since the sine-wave is derived from the square-wave that too is controlled by the microprocessor controller 12.

(29) When the laser temperature and modulated current ramp are properly adjusted, the laser output wavelength will periodically sweep across the desired absorption line of the gas in the Herriott cell 3. The resulting signal from the detector is amplified by a trans-conductance pre-amplifier 16. The signal from the pre-amplifier 16 looks similar to the laser current drive signal, except that it is reduced in strength when the laser wavelength equals the absorption line wavelength. Because the modulation frequency is much higher than the ramp rate the two signal components can be separated. This is done by putting the pre-amp output to a low pass filter 17 and a band pass filter 19. The output of the band pass filter is digitized by the A/D converter 18, usually at 256, 512, or 1024 points across the scan. The signal from the low pass filter 17, usually called the DC signal, looks like a ramp with a dip in the middle where the absorption from the gas in the Herriott cell 3 is greatest. The area between the dip and straight line along the top of the ramp is proportional to the natural log of the density of the absorbing gas. Although this signal can be used to calculate the density of the absorbing gas it is noisy and has a high lower limit of detection.

(30) The band pass filter 19 is tuned to two times the modulation frequency. The amplitude modulation of this second harmonic signal is proportional to the second derivative of the absorption strength. Further noise reduction is accomplished by using a synchronous detector 20 to filter out any components of the second harmonic signal that are within the pass band but not in phase with the laser modulation. This signal is usually called the 2f signal. A second A/D converter 21 converts the signal into data points that can be analyzed by the microprocessor controller 12. Instead of a Lorentzian shape like the DC signal has, the second derivative is a peak with a dip on each side of it. Because the dips are usually asymmetric the average of the two dips is subtracted from the peak. The density of the absorbing gas in the Herriott cell 3 is proportional to the dip-to-peak height of the signal. The dip-to-peak value is normalized by the value of the DC signal to account for gain factors in the band pass filter and the synchronous detector.

(31) Since the density of the absorbing gas in the cell 3 is only the partial pressure in the cell, it is necessary to divide the partial pressure by the total pressure measured by a pressure sensor 22 to get the concentration of the absorbing gas. The pressure sensor 22 needs to measure the absolute pressure in the Herriott cell 3. The final concentration value can be calculated by the microprocessor controller 12, converted to an analog output signal by a D/A converter 23, and shown as a percentage reading on a liquid crystal display device LCD.

(32) Details of the spectrum analysis performed by the microprocessor controller 12, include, for example, the pressure in the cell 3 causes collisions between the absorbing gas molecules. The collisional energy can add or subtract to the absorption energy and cause the absorption peak to broaden and shorten. The area under the DC peak stays constant, but the 2f signal is reduced by the broadening. The 2f peak height is also affected by the gas temperature in the cell 3. The measured dip-to-peak height needs to be adjusted for cell pressure and cell temperature measured by a gas temperature sensor 24. The pressure broadening is dependent on the relative sine-wave modulation amplitude and is calibrated along with temperature dependence for every analyzer during final test. At high concentrations there can be non-linearity in the response that must also be corrected for at final test with another second or third order polynomial function.

FIGS. 4 Through 7

(33) As illustrated in FIG. 7, a commercial embodiment of my gas sampling and analyzing system is designated by the alphanumerical symbol 12. In my system 12, two separate housings are employed: housing H1 for the laser package P and housing H2 for a gas sampling device GSD. The gas-sampling device GSD is the subject of my Utility Patent application Ser. No. 15/782,697, entitled Gas Sampling Device and Method, filed Oct. 12, 2017, copy attached in Appendices A1 and A2. Because of the voltage connections to the electrical components of the laser package P, there is a risk of sparking that could ignite any natural gas being sampled and tested. By using a separate a gas sampling device GSD that does not employ any electrical components, the likelihood of the natural gas flowing into the gas sampling device GSD is essentially eliminated.

(34) As depicted in FIG. 7, the laser light from an output 1a of the laser diode 1 is directed to the Herriott gas cell 3 by means of a fiber optic coupler or cable FOC. The fiber optic cable FOC has a first segment A feeding light from the output 1a to an explosion proof fiber optic cable FOC. The second segment B of the fiber optic cable FOC feeds the laser light into a lens L, which is mounted in the gas sampling device GSD. As depicted in FIG. 6 sample gas enters the gas sampling device GSD through a port P1, flows along a flow path FP provided by an internal passageway shown by arrows a, and exits an exhaust port P2. A bypass port P3 may also be employed. The body of the gas sampling device GSD may be formed by a plurality of metal blocks fastened together with conventional fasteners.

FIG. 3A-Alternate Embodiment

(35) In an alternate embodiment of my analyzer illustrated in FIG. 3A, only one lens, a gradient index (GRIN) lens, may be used. In this embodiment, the lens 2 is replaced by the GRIN lens (11) and the lens 4 is eliminated.

(36) As depicted in FIG. 3A an alternate embodiment of my instrument is designated by the letter 12. In my instrument 12, near-infrared light from a laser diode in the Laser Mount with TEC cooler (9) is collimated by a Grins Lens (11) and passes through a Herriott cell (1). The light makes multiple passes through the Herriott cell (1) because of reflections between concave mirrors in the Cell Cap Near and Far Included with the Herriot Cell (1). The light is then detected by an extended InGaAs photo-detector (12). When the wavelength of light from the laser corresponds to an absorption line of the gas in the Herriott cell (1), the intensity of light reaching the InGaAs detector (12) will be reduced.

(37) The wavelength of the laser is determined by its temperature and drive current. The temperature of the laser is determined by a thermoelectric cooler that is in contact with the laser substrate and mounted internally to a laser package (9). The laser package contains both the laser and the thermoelectric cooler. The thermoelectric cooler is controlled by the output of the PID temperature controller circuit (6). Feedback from the thermistor temperature sensor in the laser package (9) that is in thermal contact with the laser's substrate, maintains the laser diode in the Laser Package (9) to better than 0.1 C. of the set-point temperature. The thermoelectric cooler in the laser package (9) will also heat when the controller circuit's (6) output voltage is reversed. By using an H-bridge circuit for the output of the temperature controller 6 the laser diode in the laser package (9) can be maintained at a precise temperature, ideally around 30 C., in ambient temperatures between 10 C. and 50 C. The exact temperature to be maintained depends on the specific laser characteristics.

(38) The laser current is provided by an emitter follower output of the laser drive circuit (8). This current is the sum of a current ramp created by the laser engine (5) that scans the laser output wavelength over a narrow range, typically less than 1 nm, in a few tenths of a second and a small sine-wave modulation at a high frequency on the order of 10 KHz. The sine wave modulation is produced by dividing down a square-wave signal at twice the frequency. All of the laser current parameters are set by the microprocessor controller (3). Those parameters include the starting and ending currents of the current ramp, the speed of the current ramp, and the amplitude and frequency of the square-wave. Since the sine-wave is derived from the square-wave that too is controlled by the microprocessor controller (3).

(39) When the laser temperature and modulated current ramp are properly adjusted, the laser output wavelength will periodically sweep across the desired absorption line of the gas in the Herriott cell (1). The resulting signal from the detector is amplified by a trans-conductance pre-amplifier (4). The signal from the pre-amplifier (4) looks similar to the laser current drive signal, except that it is reduced in strength when the laser wavelength equals the absorption line wavelength. Because the modulation frequency is much higher than the ramp rate the two signal components can be separated. This is done by putting the pre-amp output to a low pass filter (17) and a band pass filter in the laser engine (5). The output of the band pass filter is digitized by the A/D converter in the microprocessor system (3), usually at 256, 512, 1024, 2048, or 5000 points across the scan. The signal from the low pass filter from laser engine (5), usually called the DC signal, looks like a ramp with a dip in the middle where the absorption from the gas in the Herriott cell (1) is greatest. The area between the dip and straight line along the top of the ramp is proportional to the natural log of the density of the absorbing gas. Although this signal can be used to calculate the density of the absorbing gas it is noisy and has a high lower limit of detection.

(40) The band pass filter in the Laser Engine is tuned to two times the modulation frequency. The amplitude modulation of this second harmonic signal is proportional to the second derivative of the absorption strength. Further noise reduction is accomplished by using a synchronous detector in the Laser Engine to filter out any components of the second harmonic signal that are within the pass band but not in phase with the laser modulation. This signal is usually called the 2f signal. A second A/D converter in the microprocessor controller (3) converts the signal into data points that can be analyzed by the microprocessor controller (3). Instead of a Lorentzian shape like the DC signal has, the second derivative is a peak with a dip on each side of it. Because the dips are usually asymmetric the average of the two dips is subtracted from the peak. The density of the absorbing gas in the Herriott cell (1) is proportional to the dip-to-peak height of the signal. The dip-to-peak value is normalized by the value of the DC signal to account for gain factors in the band pass filter and the synchronous detector.

(41) Since the density of the absorbing gas in the Herriot cell (1) is only the partial pressure in the cell, it is necessary to divide the partial pressure by the total pressure measured by a pressure sensor in the Herriot Cell (1) to get the concentration of the absorbing gas. The pressure sensor in the Herriot Cell (1) needs to measure the absolute pressure in the Herriott cell (1). The final concentration value can be calculated by the microprocessor controller (3), converted to an analog output signal by a pressure temperature conditioning (7) and the microprocessor D/A converter (3), and shown as a percentage reading on a display device (10).

(42) Details of the spectrum analysis performed by the microprocessor controller (3), include, for example, the pressure in the Herriot cell (1) causes collisions between the absorbing gas molecules. The collisional energy can add or subtract to the absorption energy and cause the absorption peak to broaden and shorten. The area under the DC peak stays constant, but the 2f signal is reduced by the broadening. The 2f peak height is also affected by the gas temperature in the Herriot cell (1). The measured dip-to-peak height needs to be adjusted for cell pressure and cell temperature measured by a gas temperature sensor in the Herriot Cell (1). The pressure broadening is dependent on the relative sine-wave modulation amplitude and is calibrated along with temperature dependence for every analyzer during final test. At high concentrations there can be non-linearity in the response that must also be corrected for at final test with another second or third order polynomial function.

(43) The personal computer (2) communicates with the microprocessor controller (3) to verify correct operation of the system, download information for analysis, upload firmware to the Microprocessor Controller, and change internal variables in the microprocessor controller (3) that relate to the operation of the system.

SCOPE OF THE INVENTION

(44) The above presents a description of the best mode I contemplate of carrying out my instrument, system, and method for detecting water vapor in gas, and of the manner and process of making and using them, in such full, clear, concise, and exact terms as to enable a person skilled in the art to make and use. My instrument, system, and method are, however, susceptible to modifications and alternate constructions from the illustrative embodiments discussed above which are fully equivalent. Consequently, it is not the intention to limit my instrument, system, and method to the particular embodiments disclosed. On the contrary, my intention is to cover all modifications and alternate constructions coming within the spirit and scope of my instrument, system, and method as generally expressed by the following claims, which particularly point out and distinctly claim the subject matter of my invention: