Gas detection device and method

Abstract

A method of identifying the presence of a first gas, such as methane, within a sample, the first gas comprising, for example, natural gas. A detector is provided having a sensor responsive to a first wavelength, a sensor responsive to a second wavelength, and a sensor for collecting reference readings. The gas sample is analyzed to obtain a first absorption reading corresponding to the first wavelength, a second absorption reading corresponding to the second wavelength, and a reference reading. A first absorption number is calculated using the first absorption reading and the reference reading, and a second absorption number is calculated using the second absorption reading and the reference reading. A linearizer function is applied to each of the first and second absorption numbers to calculate first and second concentration numbers. The sensor for each wavelength is calibrated for detecting the first gas so that the data collected at each wavelength gives the same reading when only the first gas is present in the sample. A ratio of the first concentration number to the second concentration number is then calculated and used to identify whether only the first gas is present in the sample.

Description

[0001] This application is a divisional application. The name of the invention of the parent application is “Gas Detection Device and Method”. The application date is July 27, 2016, and the application number is 2016800440687.

[0002] The present invention relates to gas detection devices and, more particularly, to devices for distinguishing between different types of gases.

[0003] Devices and methods for detecting the presence of gas within an environment and also for discerning the composition of a gas mixture are known in the art. Many of these processes and equipment are laboratory-based, but in many applications, the delay associated with collecting gas samples and sending them to a laboratory for analysis is undesirable or unacceptable. For example, a gas utility involved in the distribution of natural gas must act quickly on any report of a gas leak from a gas supply line. Most gas distribution pipeline work is underground, and therefore, there are significant costs associated with digging pipeline work, which may be under roads, etc. However, there may be situations where the reported gas is not in fact leaking from a pipeline but from some other source. Natural gas is a mixture of hydrocarbons such as methane, ethane, propane, butane, etc. However, basic detectors will also respond to other sources of gas such as landfill gas, which is primarily methane. Then in order to avoid unnecessary digging, there is a need for a system for distinguishing between different types of gas before digging.

[0004] As already mentioned above, laboratory based equipment is known in the art which can analyse gas samples and differentiate between different gas types / identify components. However, collecting samples and sending them for laboratory analysis will cause significant delays and, depending on the severity of the leak, may not be an option.

[0005] One known solution is to provide a micro gas chromatograph in a portable gas detection instrument. The instrument collects a sample of the gas and provides an indication of the composition of the gas over a period of several minutes.

[0006] Another solution known in the art has been to use laser based detectors. Although these work very effectively, they are prohibitively expensive.

[0007] US-A-4958076 discloses a selective gas detection device that can be used to determine the concentration and type of hydrocarbon gas in a gas sample. It does this by using at least two infrared radiation absorption channels, each of which responds to different wavelengths (particularly, 3.2 microns and 3.4 microns). The infrared absorption channels work by measuring the amount of infrared radiation absorbed by the gas sample at a specific wavelength. The data from each channel will independently give an indication of the concentration of the specific hydrocarbon to which it is tuned, but it is also sensitive to other hydrocarbons. In US'076, the absorption data from each channel is used to calculate an absorbance number for each channel, and then the ratio of the absorbance numbers for the two channels is used to infer the type of hydrocarbon present. In order to minimize the effect of variance in the output of different sensors tuned to the same wavelength, US'076 further proposes taking the logarithm of the ratio of the outputs of paired channels, because this will reduce the effect of any such variance. This is an acceptable solution when using an analog system of the type taught in US'076, but it is less acceptable with respect to the more accurate digital systems that are now more commonly utilized.

[0008] Accordingly, there is a need for an improved gas detection system that can provide both more accurate information about the identification of the gas present in a sample and information about the concentration of the gas. It is a further object of the present invention to provide such a system that reduces the effect of compositional changes on the measurement readings. A further need is that such a system should be implementable in a portable device to enable rapid analysis to be performed at the site of a suspected gas leak.

[0009] According to a first aspect of the present invention, there is provided a method for identifying the presence of a first gas in a sample, comprising the steps of: providing a detector having a sensor responsive to a first wavelength, a sensor responsive to a second wavelength, and a sensor for collecting reference readings; analyzing the gas sample to obtain a first absorption reading corresponding to the first wavelength, a second absorption reading corresponding to the second wavelength, and a reference reading; calculating a first absorption number using the first absorption reading and the reference reading, and calculating a second absorption number using the second absorption reading and the reference reading; applying a linearizer function to each of the first and second absorption numbers to calculate first and second concentration numbers; calibrating the sensor for each wavelength for detecting the first gas so that when only the first gas is present in the sample, the data collected at each wavelength gives the same reading; calculating a ratio of the first concentration number to the second concentration number, and using the ratio to identify whether only the first gas is present in the sample.

[0010] The method of the first aspect of the invention differs from the prior art in that the sensors for each wavelength are calibrated for detecting the first gas so that if only the first gas is present in the sample, the sensors give the same reading. Furthermore, in the method of the invention, the readings for each wavelength are first processed using a linearizer function to calculate a concentration reading, and then the two concentration readings are used to calculate a ratio. This is in contrast to the prior art, in which the absorbance numbers for each channel are processed using a logarithmic function to produce absorbance numbers, which are then divided by each other to produce a number that is the ratio of the two absorbance numbers.

[0011] The method according to the first aspect of the invention thus has the advantage that by applying a lineariser function to the readings and then calculating the ratio of the concentrations, the ratio figure obtained is less susceptible to sensor variations such as variance in the filter in the sensor due to manufacturing tolerances. Thus, a more reliable identification can be made.

[0012] The present invention is particularly useful in distinguishing between methane and higher hydrocarbons such as propane and butane, and is particularly useful for distinguishing methane from natural gas. Preferably, then, the first wavelength is 3.3 microns plus or minus 0.1 microns, and the second wavelength is 3.4 microns plus or minus 0.1 microns. Calibration is then performed for detecting methane so that the same reading is obtained at each wavelength when a sample containing methane but not other hydrocarbons is analyzed. In this case, the ratio figure calculated according to the present invention will be 1, so that a calculation of a ratio of 1 indicates that methane is in the sample.

[0013] Natural gas includes a range of hydrocarbons that includes methane as well as many other hydrocarbons. Due to their different operating wavelengths, when a sample containing natural gas is analyzed by the sensors, the readings from the two sensors will be different, and the difference between the two readings will increase with increasing concentration of natural gas. Accordingly, calculation of a ratio greater than 1 indicates that a gas other than methane is present in the sample. The way the sensor is calibrated with methane means that it will indicate that the gas is not methane if the sampled gas is, for example, natural gas or any higher hydrocarbon.

[0014] Preferably, compensation is performed separately on the readings at each wavelength in order to eliminate errors due to changes in environmental parameters. In particular, at least one of temperature compensation and pressure compensation may be performed, wherein preferably temperature compensation is performed in two stages (zero drift correction and span drift correction).

[0015] When utilizing sensors of the type used in the present invention that are all calibrated for detecting the same first gas (preferably methane), when the sample being measured is actually a second gas (preferably natural gas) or any other hydrocarbon, the increased absorption relative to the first gas means that the reading given will be higher than the actual concentration. In the case where volumetric levels of natural gas are present, the concentration indicated by the linearizer can be hundreds of percent higher than the actual concentration.

[0016] Accordingly, a need exists to compensate for the increased absorption so as to enable a more accurate value of the actual concentration of the gas in the sample to be obtained from the readings taken from the sensor.

[0017] Preferably, two sensors are used, one for each wavelength. However, it is also possible within the scope of the invention to utilize a single sensor that can be operated in different modes (such as with a variable or changeable filter) to allow it to take readings at different wavelengths simultaneously or at different times, the important feature being that data can be collected from the same sample corresponding to two different wavelengths.

[0018] The or each sensor may be an infrared radiation absorption sensor, in particular a broadband IR source (such as, for example, a flickering incandescent light bulb) together with a suitable spectral filter(s) and detector applicable to the required wavelength. However, the sensor may alternatively incorporate an emitter which emits light at the required wavelength, thereby avoiding the need to use a source in conjunction with a filter. In particular, an LED emitting light at a first and a second wavelength may be utilized, or possibly an LED operable in two modes corresponding to the first and second wavelengths.

[0019] According to another aspect of the present invention, there is provided a method of measuring the concentration of a first gas in a sample, comprising the steps of: providing a detector having a sensor responsive to a first wavelength, a sensor responsive to a second wavelength, and a sensor for taking a reference reading; applying a linearizer function to each of first and second absorption figures to calculate first and second concentration figures; calibrating the sensor for each wavelength for detecting the first gas so that the data at each wavelength gives the same reading when only the first gas is present in the sample; analyzing the gas sample to obtain a first absorption reading at the first wavelength, a second absorption reading at the second wavelength, and a reference reading; calculating a first concentration figure (c1) and a second concentration figure (c2) using the readings at the first wavelength, the readings at the second wavelength, and the reference reading; and using equation c cor = c1 – X·(c2 – c1) to calculate a corrected concentration number (c2) indicative of the actual concentration of one of the first gas and the second gas in the sample.cor ), where X = A·(c1 / c2) B , where A and B are constants for a particular pair of filter wavelengths chosen depending on the gas being sought in the sample.

[0020] The method according to the second aspect of the invention has the advantage of providing a correction for errors in the concentration reading arising from the differential readings at two wavelengths due to the different gases being sought in the sample.

[0021] Preferably, the first gas is methane and its concentration is measured in a sample containing natural gas or another higher hydrocarbon, in which case A = 2.5577 and B = 1.869. However, for different values ​​of the first and second wavelengths, the values ​​of A and B will be different (and these may be different if, for example, the method is used in association with other gases). In this case, the values ​​of A and B may be calculated empirically as set out below.

[0022] Although this correction has been found to produce significantly more accurate values ​​than prior art solutions that do not utilize this correction method, it has been found that the correction algorithm does introduce noise into the measurement results when only methane is present. Preferably, therefore, if the gas present is methane, the algorithm is not used and instead, only the output of the first sensor is used to calculate the concentration reading.

[0023] In a particularly preferred embodiment, the equation is used only if the ratio of the readings of the first sensor to the second sensor (c1 / c2) is above a first threshold, only the output of the first sensor is used if (c1 / c2) is below a second threshold, and a transition algorithm is used if (c1 / c2) is between the first and second thresholds.

[0024] In particular, the first threshold is 1 and the second threshold is 0.9. In this case, the correction equation used is: cor = M2·(c1 – X·(c2 – c1)) + M1·c1, where M1 = (z – 0.9) / (1 – 0.9) and M2 = (1– z) / (1 – 0.9), where z = 0.9 when (c1 / c2) < 0.9, z = 1 when (c1 / c2) > 1, and z = (c1 / c2) when 0.9≤(c1 / c2)≤1.

[0025] This has the advantage of removing the effect of the noise introduced by the correction algorithm by not using that noise when the underlying readings from the sensor indicate a concentration within the noisy range (ie by not using the algorithm when only methane is present).

[0026] In a further development of the method of the second aspect of the invention, it has been found that the accuracy of the corrected concentration value can be improved by introducing a correction factor into the equation dependent on the measured value of the concentration taken from the first sensor D1 .

[0027] Accordingly, the present invention further provides a method of measuring the concentration of a first gas in a sample, comprising the steps of: providing a detector having a sensor responsive to a first wavelength, a sensor responsive to a second wavelength, and a sensor for taking a reference reading; applying a linearizer function to each of first and second absorption figures to calculate first and second concentration figures; calibrating the sensor for each wavelength for detecting the first gas so that the data at each wavelength gives the same reading when only the first gas is present in the sample; analyzing the gas sample to obtain a first absorption reading at the first wavelength, a second absorption reading at the second wavelength, and a reference reading; using the readings at the first wavelength, the readings at the second wavelength, and the reference reading to calculate a first concentration figure (c1) and a second concentration figure (c2); and using equation c cor = c1 – X·(c2 – c1) to calculate a corrected concentration number (c cor ), where X = D·A·(c1 / c2) B , where A and B are constants for a particular pair of filter wavelengths selected depending on the gas being sought in the sample, and D is a correction factor that depends on the concentration of the first gas in the sample.

[0028] Preferably, the value of D is extrapolated from calibration readings taken using the detector to measure the concentration of a first gas mixture of a known component at at least two known concentrations and a second gas mixture of a known component different from the composition of the first gas mixture at at least two known concentrations, the calibration readings preferably being taken using the same concentration value for each of the first and second gas mixtures, and in particular, the readings being taken at 100% of each gas mixture and at 50% of each gas mixture in the case of nitrogen.

[0029] Preferably, the use of the equation The calibration value of D is calculated from the readings taken from a sample of known composition, where C1 and C2 are the sensor readings taken for the sample, and C cor is the actual concentration of the gas in the known sample. Preferably, the value of D for the sample of unknown composition is then extrapolated from the calibration value of D using linear extrapolation between the calibration values ​​to extrapolate to the value of D for the measured value of C1 for the sample.

[0030] In a particularly preferred embodiment, D is calculated for a particular sample of an unknown component using the following equation:

[0031]

[0032] in

[0033]

[0034] as well as

[0035]

[0036] in ; R @100%NG1 = a calibrated determined ratio measured with at least one known concentration of a first gas mixture NG1 of known composition, R @100%NG2 = a calibrated determined ratio measured with at least one known concentration of a second gas mixture NG2 of known composition, and D NG1 is a linear equation extrapolated between the calibrated values ​​of D for the first gas mixture, and D NG2 is a linear equation extrapolated between the calibrated values ​​of D for the second gas mixture.

[0037] It is also possible to utilize the method according to this third aspect of the invention, wherein preferred options are explained above and below in conjunction with the second aspect of the invention.

[0038] In a particularly preferred embodiment of the invention, the correction algorithm of the second or third aspect of the invention is utilized in combination with the identification method according to the first aspect of the invention.

[0039] Again, while a separate sensor is used for each wavelength in the preferred embodiment, a single sensor configured to collect data at three different wavelengths (including a reference reading) may also be used.

[0040] The present invention further provides a selective gas detection device, comprising: a sensor responsive to a first wavelength, a sensor responsive to a second wavelength, and a sensor for collecting reference readings; and a processing component, the processing component being programmed to: calculate a first absorption number using the reading for the first wavelength and the reference reading and calculate a second absorption number using the reading for the second wavelength and the reference reading; calculate first and second concentration numbers by applying a linearizer function to each of the first and second absorption readings; the sensor for each wavelength is calibrated for detecting the first gas so that the readings for each wavelength are the same when only the first gas is present in the sample; calculate a ratio of the first concentration number to the second concentration number; and identify the gas present in the sample based on the calculated ratio.

[0041] The present invention further provides a selective gas detection device, comprising: a sensor responsive to a first wavelength, a sensor responsive to a second wavelength, and a sensor for collecting reference readings; the sensor for each wavelength is calibrated for detecting a first gas so that when only the first gas is present in the sample, the readings for each wavelength are the same; and a processing unit, the processing unit being programmed to apply a method according to the second aspect of the present invention.

[0042] Preferably, a separate sensor is used for each of the first wavelength, the second wavelength and the reference reading.

[0043] For each of the above aspects of the invention, the or each sensor may be an infrared radiation absorbing, in particular, a broadband IR source (such as, for example, a flickering incandescent light bulb) used with a suitable spectral filter(s) and detector applicable to the required wavelength. However, instead, the sensor may incorporate an emitter that emits light at the required wavelength, thereby avoiding the need to use a source in conjunction with a filter. In particular, an LED emitting light at the first, second and reference wavelengths may be utilized, or possibly a single LED operable in three modes corresponding to the first, second and reference wavelengths.

[0044] In order that the invention may be better understood, some embodiments of the invention given as examples will now be described, with reference to the accompanying drawings, in which:

[0045] Figure 1 is a graph showing an example of an indication algorithm of a sensor according to a preferred embodiment of the present invention with respect to methane and natural gas;

[0046] Figure 2 is a graph showing the distribution of linearizer errors for a sample batch of methane and hydrocarbon sensors;

[0047] Figure 3 is a graph showing the variation of gas determination ratios for different sensor batches;

[0048] Figure 4 is a graph showing the effect of correction according to the present invention on both methane and hydrocarbon sensor data;

[0049] Figure 5 is a graph showing the effect of applying the linearizer correction algorithm of the present invention on the gas determination ratio;

[0050] Figure 6 is a graph showing methane and hydrocarbon sensor linearizer outputs incorporating the method of the present invention for different gases;

[0051] Figure 7 is a graph showing the linearizer response to natural gas using the correction algorithm of the present invention;

[0052] Figure 8 is a graph showing the variation of the algorithm correction factor for different gases;

[0053] Fig. 9 is a graph showing the variation of the algorithm correction factor with respect to the discrimination ratio;

[0054] Fig.10 is a graph showing measurement errors in a methane sensor compared to errors caused by a hydrocarbon correction algorithm for methane according to the present invention;

[0055] Fig.11 is a graph showing a change in the modified correction factor with respect to the discrimination ratio;

[0056] Fig.12 is a graph showing the variation of the determination ratio with respect to the concentration of different gases for a detector using the method of the present invention;

[0057] Fig.13 is a graph showing calibration readings of a methane sensor plotted against a ratio of the output of the methane sensor to the output of another hydrocarbon sensor taken over a plurality of samples of known but different compositions and concentrations; and

[0058] Fig.14 is shown relative to the Fig.13 A graph of the output of a methane sensor plotted against the values ​​of the correction factors calculated from the calibration data shown in FIG.

[0059] NDIR spectroscopy detection of methane is a widely used technique with sensors available from a variety of suppliers. However, there are no sensors of this type that have been successfully used to identify natural gas. Such sensors can be specified to measure different parts of the IR spectrum in order to produce a sensor where the target gas is methane or a sensor where the target gas is a higher hydrocarbon (propane, butane, etc.). Regardless of the target gas, however, they will still respond to all hydrocarbons to varying degrees.

[0060] The preferred embodiment of the invention now described is based on the provision of a detector having a methane sensor and a hydrocarbon sensor. Although the detailed description of the invention relates to methane and hydrocarbons, it will be appreciated that this is for illustrative purposes only and the invention is generally applicable to gases and is not necessarily limited to the detection of methane.

[0061] Accordingly, the detector utilized as part of the present invention effectively has two hydrocarbon sensors. They both respond to hydrocarbons, but to different degrees. If both are calibrated for methane and methane is the only gas present, then both will produce the same result. However, if methane is present as well as another hydrocarbon (e.g., ethane, as would be the case for natural gas), then the two sensors will produce different results. It is this imbalance of the signal in the presence of natural gas that enables the discrimination between methane and natural gas in the present invention.

[0062] The sensor utilized in the present invention is a non-dispersive IR sensor, which uses the basic principles of infrared spectroscopy to measure the concentration of a gas. In short, a gas absorbs light at a specific wavelength (spectral fingerprint). The amount of light absorbed will depend on the concentration of the target gas, the higher the concentration, the greater the absorption.

[0063] This process can be modeled by the Beer-Lambert equation,

[0064]

[0065] where I0 = light intensity in air; α = absorption coefficient; c = gas concentration; and l = optical path length.

[0066] Because a non-dispersive system looks at many numbers of absorption lines instead of just a single line of the gas (as in laser spectroscopy), the product "αl" is not constant but depends on c.

[0067] The sensor used in the preferred embodiment of the present invention consists of three pyroelectric detectors and an infrared source. Each of the three detectors has a spectral filter for detecting a different part of the mid-IR spectrum.

[0068] In the present invention, the specifications of these three filters are as follows:

[0069] Filter 1 (Methane): 3.33μm + / - 0.1μm

[0070] Filter 2 (higher hydrocarbons): 3.40μm + / - 0.1μm

[0071] Filter 3 (reference): 3.95μm + / - 0.1μm

[0072] The bandwidth of each filter is in the order of nanometers or hundreds of nanometers, such as, for example: 160 nm for filter 1, 120 nm for filter 2, and 90 nm for filter 3. However, these bandwidth values ​​are not exclusive and will vary depending on the specifications of the sensor used.

[0073] The infrared source is flashed at 4 Hz. This generates a 4 Hz sinusoidal waveform at the output of the detector circuit, the amplitude of which is measured to provide an indication of the amount of light reaching the detector ("l" in Equation 1).

[0074] The method to identify natural gas is as follows:

[0075] First, the absorbance in each channel is calculated (using the reference detector as a measure of the signal strength in air).

[0076]

[0077] in,

[0078] I m = normalized methane channel intensity;

[0079] I hc = normalized hydrocarbon channel intensity;

[0080] I r = Normalized reference channel intensity.

[0081] The linearizer equation is then used to calculate the concentration measured in each channel.

[0082] c = c(abs) (4)

[0083] where c(abs) = linearizer function.

[0084] There is a separate linearizer for each channel.

[0085] The ratio of the two calculated concentrations is then calculated,

[0086]

[0087] If R = 1, the gas is methane.

[0088] Due to the spectral filter positions and relative absorption intensities for methane and other hydrocarbons, the hydrocarbon channel will see greater absorption than the methane channel in the presence of natural gas (or any hydrocarbon other than methane). For natural gas, this means C2>C1. Therefore, if R>1, it can be inferred that the gas is not only methane.

[0089] The gas determination ratio (Equation 5) is fairly straight forward in theory. However, because the ratio of two numbers is taken, if these numbers are small, as in the case for low gas concentrations, it has been found that noise can play a significant role in the calculation. It has also been found that it is desirable to account for errors in linearity. Therefore, in a preferred embodiment, limits are set to reduce the likelihood of incorrect determinations.

[0090] The preferred limits are:

[0091] If C1 < Y, the gas is attenuated; (the signal is too low to make a reliable decision)

[0092] If C1 > Y and R < 1.025, the gas is methane;

[0093] If C1 > Y and 1.025 < R < 1.05, the gas is indeterminate; (this is a buffer to account for noise in linearity and sensor-to-sensor variations)

[0094] If C1 > Y and R > 1.05, the gas is natural gas;

[0095] where Y is constant and is chosen to provide acceptable results depending on specific calibration values, etc. The lower the value of Y, the greater the sensitivity of the system but the more susceptible it is to noise. Preferably, Y is less than or equal to 1.5, and more particularly equal to 1.0, especially for calibration at 1% concentration of methane. An example indicating the operation of the algorithm is shown in Figure 1 in.

[0096] It is well known to apply environmental compensation to compensate for changes in the sensor output due to environmental conditions such as temperature and pressure.

[0097] For the preferred embodiment of the present invention, temperature compensation of the sensor must be done separately for each absorption channel. It also needs to be done in two stages.

[0098] 1. Zero drift correction;

[0099] 2. Span drift correction.

[0100] Data should be compared for a large number of sensors (about 100 sensors) in order to examine how similar / different the sensors are to each other after the linearizer is applied.

[0101] Although Figure 2 the small errors in linearity shown in are not errors involved in the calculation of the gas concentration per se, they can introduce significant errors when calculating the gas determination ratio as shown in Figure 3 in.

[0102] Figure 3 It is shown that: for a batch of 100 examples of the preferred embodiment, many of the measured data points lie in the indeterminate region of 1.025 < R < 1.05. When the gas is actually methane, there are also a few errors where R > 1.05, which would give an indication of natural gas.

[0103] To improve Figure 3In the scenario in Figure 1, a method has been created to calibrate the sensor with 1% and 5% methane as well as 100% methane.

[0104] The corrected equation for the linearized data is:

[0105] 0-5% Methane:

[0106] 5-100% Methane:

[0107] in

[0108]

[0109]

[0110] c 1% and c 5% is the linearizer output for 1% and 5% methane before correction. The result of this correction algorithm is Figure 4 and Figure 5 Shown in.

[0111] For the particular sensor of the example, a predominantly "undecided" decision was produced prior to calibration. After calibration, a significant improvement in the calculation of the gas discrimination ratio was achieved, and calibration can be performed across the full range. This three point calibration need only be a one-time event at the initial calibration, and serves as a way to characterize each individual sensor. Subsequent calibrations should only need to be completed with 100% methane.

[0112] Because the linearizer used in the instrument according to the method of the preferred embodiment of the present invention is calibrated for methane, if the instrument measures natural gas or any other hydrocarbon, the increased absorption relative to methane means that it will read as high. In the case of volume levels of natural gas, the concentration indicated by the linearizer can be hundreds of percent higher than the actual concentration present. An example of this is shown in Figure 6 Shown in.

[0113] To correct for this, a preferred embodiment of the method of the present invention uses Equation 19 (see also Figure 7 ) to combine the two channels in order to estimate the actual concentration of the gas being measured.

[0114]

[0115] from Figure 6 The data shown in are used to calculate the correction factor "X" for each gas, such as Figure 8 as shown in .

[0116]

[0117] Figure 8 Each gas type is shown to have its own profile for the correction factor "X". When "X" is plotted against the discrimination ratio, there appears to be a trend related to these two values, such as Fig. 9 As shown in the figure.

[0118] Therefore, the correction factor "X" can be calculated using the following equation:

[0119] .

[0120] While this correction factor works reasonably well for natural gas, it does not work as well for propane and butane. However, if the instrument is intended for propane or butane measurement, it will be calibrated as such. This approach ensures that the instrument will give realistic readings regardless of the gas being detected. The natural gas discrimination sensor is only intended to give a "yes / no" decision on whether the gas is methane. It is not intended to be a completely accurate measurement tool for all gases at all times.

[0121] With a hydrocarbon correction algorithm in place, it is possible to introduce noise into the measurements when recording methane. Fig.10 It is shown that despite the presence of very little noise on the single methane channel, the hydrocarbon correction algorithm appears to amplify any measurement errors.

[0122] To counteract this effect, the correction algorithm is applied only if natural gas is detected. If the gas is methane, only the methane channel output should be used as the displayed measurement. To achieve a smooth transition between using a single channel and switching to a combination of all two measurement channels, a set of switching limits is used.

[0123] To achieve this, Equation 19 is modified to:

[0124] .

[0125] The limits and conditions for the calculation of M1 and M2 are as follows:

[0126] (Microsoft Excel (RTM) Syntax for writing an "if" statement)

[0127] .

[0128] Equation 22 effectively means that if the gas being measured is methane, then:

[0129] c = c1.

[0130] If the gas being measured is natural gas, then:

[0131] c = c1 – X·(c2 – c1).

[0132] Therefore, the issue of noise being added to the methane measurements disappears.

[0133] To simplify the software required to implement the hydrocarbon correction algorithm other than calculating "X", a lookup table can be used to replicate Equation 21. This can be taken even further and a lookup table can be created for Equation 22 based on the following equation:

[0134] .

[0135] This produced Fig.11 The data shown in .

[0136] Therefore, Equation 22 can be replaced by:

[0137] .

[0138] The method and associated detector of the present invention enable the measurement of the concentration of a detectable gas (i.e., methane) and the discrimination of methane from natural gas (and other hydrocarbons). The detector houses at least three individual pyroelectric detectors, each with a different spectral filter attached. The first detector is used to detect primarily methane, the second detector is used to detect higher hydrocarbons, and the third detector is used as a reference.

[0139] The preferred filter specification is:

[0140] Filter 1 (Methane): 3.33μm + / - 0.1μm

[0141] Filter 2 (higher hydrocarbons): 3.40μm + / - 0.1μm

[0142] Filter 3 (reference): 3.95μm + / - 0.1μm

[0143] The bandwidth of each filter is in the order of nanometers or hundreds of nanometers, such as, for example: 160 nm for filter 1, 120 nm for filter 2, and 90 nm for filter 3. However, these bandwidth values ​​are not exclusive and will vary depending on the specifications of the sensor used.

[0144] The combination of a methane detector and a reference detector and the combination of a hydrocarbon detector and a reference detector serve to create a methane channel and a hydrocarbon channel, respectively. By creating a methane linearizer for each channel and monitoring the ratio of the concentrations measured by the two channels, we can set limits to enable identification of the type of gas being monitored.

[0145]

[0146] The preferred limits are:

[0147] If C1 < Y, the gas is attenuated; (too low a level to make a reliable decision)

[0148] If C1 > Y and R < 1.025, the gas is methane;

[0149] If C1 > Y and 1.025 < R < 1.05, the gas is indeterminate; (this is a buffer to account for noise in linearity and sensor-to-sensor variation)

[0150] If C1 > Y and R > 1.05, the gas is natural gas;

[0151] Where Y is constant and is chosen to give acceptable results depending on specific calibration values etc. Preferably, Y is less than or equal to 1.5, and more particularly equal to 1.0, especially for calibration at 1% concentration of methane.

[0152] Now referring to Fig.12 , which shows how the gas determination ratio (R) varies for different gas mixtures for a detector incorporating the sensor arrangement of the present invention. This implies that if R is greater than or equal to 2, it is likely to be propane / butane (the two are indistinguishable), also known as the presence of LPG. However, based on this information alone, it does not follow that if R is less than 2 the gas is natural gas, because for high concentrations of gas, the data shows that the ratio for LPG drops below 2. Accordingly, the present invention can be used to detect the presence of LPG by applying different algorithms, according to which if R ≥ 2, the gas is LPG, and if R < 2 and the concentration is determined to be less than a defined figure (according to Fig.12 , less than 20% and more particularly less than 10% to account for variation in sensor readings between detectors), the gas is not LPG.

[0153] Problems regarding linearity due to sensor-to-sensor variation are minimized by implementing a three-point calibration. Preferably, the sensor is calibrated with 100%, 5% and 1% methane. Calibration at 1% and 5% methane only needs to be done once for each sensor.

[0154] If the gas being detected is natural gas, the method of the present invention also provides an estimated indication of the gas concentration. If natural gas is detected, the linearizer for methane will give a high reading. The method ensures that the instrument does not give unrealistic readings in the presence of natural gas or higher hydrocarbons.

[0155] Although it has been found that regarding the above (where A and B are constants for a particular pair of filter wavelengths selected depending on the gases being sought in the sample) The use of equation 19 produces satisfactory results in many applications, but the accuracy of the corrected concentration derived using this equation has been found to be reduced with some sensors. According to a further development of the invention, then, it has been found that the accuracy of the concentration readings can be further improved by modifying the value of X using a variable multiplier D, the value of which depends on the composition of the gas being measured.

[0156] In particular, it has been found that the value of the constant A calculated as set out above is necessarily constant for a given sensor used with a given gas, but may vary depending on the specific composition of the gas being sampled. In a further embodiment of the invention, therefore, depending on the actual concentration readings taken from the two sensors, if the constant A is modified by a new variable D, then an improved value may be obtained so that X in equation 19 becomes:

[0157] .

[0158] In this alternative embodiment to improve accuracy over a range of natural gas compositions and concentrations, the detector is first calibrated by taking concentration readings of two natural gas mixtures, NG1 and NG2. These can be any two mixtures, but to get better accuracy across the range, it is preferred that they represent the limits of the compositions that might in practice be analysed by the detector. To this end, it has been found that it works well if the compositions are as follows:

[0159] NG1 is a mixture consisting mainly of methane, with minimal additions of higher hydrocarbons expected to be found in use. In this example, a mixture of 98% methane and 2% ethane is used, but other compositions are possible as long as the actual composition is known.

[0160] • NG2 is similar but has a maximum additional higher hydrocarbons expected to be found in use. In this example a mixture of 90% methane and 10% ethane is used, but again other compositions are possible as long as the actual composition is known.

[0161] It will also be appreciated that more than two known components may be used if desired, for example if greater accuracy is required.

[0162] Measurements are taken on the detector for each gas mixture, preferably at least two measurements - one with a 50% mixture of the gas with another gas (e.g. nitrogen) and one with a 100% natural gas mixture (NG1 or NG2), each giving a reading for each sensor (c1 and c2).

[0163] Nitrogen is preferred as the further gas as it reduces the risk of explosion, but other gases such as air may also be used.

[0164] Furthermore, without departing from the scope of the invention, more than two measurements may be taken for each gas mixture, each measurement being taken with a different mixture composition with another gas. Two measurements are preferred as this has been found to give a good level of accuracy without making the calibration process too difficult and time consuming. However, if higher accuracy is a requirement, more than two sets of measurements may be taken.

[0165] Fig.13 A graphical plot of example data generated by the example is shown, where the methane sensor output reading (c1) is plotted against the ratio of the two sensor output readings (c2 / c1). The x-axis is % methane equivalent, so when tested at 50% and 100% methane, the ratio is 1. When tested at either of the two concentrations of NG1, the ratio is higher, and when tested at NG2, the ratio is even higher; as is the % methane equivalent reading.

[0166] This analysis gives four data sets, two for NG1 (one for 50% concentration and one for 100% concentration) and two for NG2 (at 50% and 100% concentration). Each data set includes the value of C1, the value of C2 and the value of the actual concentration Ccor, which is known because the composition of NG1 and NG2 is known.

[0167] Combining equations 19 and 28 and rearranging them for D gives D as:

[0168] .

[0169] The values ​​of A and B are known from the first embodiment described above, and the values ​​of C1, C2 and Ccor obtained from these four data sets can be used to generate four values ​​of D, two for NG1 and two for NG2. These can be plotted on a graph as Fig.14 , where the coefficient D is on the y-axis and the methane sensor output value (C1) is on the x-axis.

[0170] The straight line approximation of the relationship between C1 and D2 for a particular composition has been found to be a good approximation, and therefore, as Fig.14 The data shown in is used to generate D for each of NG1 and NG2 (D NG1 and D NG2 ) and C1.

[0171] The data are then further extrapolated between the two linear approximations for concentrations that are neither NG1 nor NG2. If the measured ratio is lower than the ratio for NG1, the values ​​are interpolated between the line for NG1 and the line for methane (correction factor = 1). If the measured ratio is between NG1 and NG2, the values ​​are interpolated between the line for NG1 and the line for NG2. If the ratio is still higher, the line for NG2 is used.

[0172] Use the following equation in the instrument to calculate the actual correction factor, D, that should be applied for the specific component being analyzed:

[0173] .

[0174] Equation 30 relates the measured value from the unknown gas mixture to Fig.13 and 14 The correction factor NG1 is related to the known value shown in cor and NG2 cor The position of the measured determined ratio relative to the known determined ratio values ​​at 100% NG1 and NG2 is determined and calculated using the following formula:

[0175]

[0176] The ratio is determined , and R @100%NG1 = 100% Determined ratio at NG1, R @100%NG2 = 100% Determination ratio at NG2.

[0177] D NG1 and D NG2 yes Fig.14 The equation shown in .

[0178] Therefore, each detector is calibrated using this process and, using R @100%NG1 , R @100%NG2 and the characteristic equation D derived for this detector during the calibration process NG1 and D NG2 The processor is programmed with the value of C1 and C2 obtained for the sample being analyzed. The values ​​of C1 and C2 obtained for the sample being analyzed are then used in Equation 30 to generate the value of D, which is then used in Equations 28 and 19 to generate a correction value for the actual concentration of the sample.

[0179] Example 1

[0180] As an example, using the graph above, if the methane reading is 100 and the ratio is 1.1 (curve Figure 1 ), the ratio for NG1 is approximately 1.2, so the methane line (y1) and the NG1 line (DNG1 = -0.0003729x +1.1014806). This gives a correction factor of approximately 1.025. Fig.14 This calculation is shown above as a dashed line.

[0181] Example 2

[0182] If the methane sensor gives a reading of 80 and the ratio is 1.3 (i.e. the correction factor is between NG1 and NG2), then Fig.13 , the ratio for 100% NG1 is 1.2 and for 100% NG2 is 1.45. Therefore, the correction factors for NG1 are (1.45-1.3) / (1.45-1.2) and for NG2 are (1.3-1.2) / (1.45-1.2). This gives a figure of 0.6× correction factor for NG1 and 0.4× correction factor for NG2.

[0183] The reading from the methane sensor is then used in the equation for NG1 and NG2, namely:

[0184] D NG1 : -0.0003729×80+1.1014806= 1.072;

[0185] D NG2 :-0.0003773×80+1.2146882= 1.185.

[0186] The final correction factor (D) in this example is then given by 0.6 × 1.072 + 0.4 × 1.185 = 1.1172. This number can be found from the curve Figure 2 confirm.

[0187] Of course, it will be understood that although various parts and features of the method have been described in combination, the advantages provided by different parts of the method can be achieved at least partially in isolation from each other, and the description of the entire method is not intended in any way to limit the teachings of the present application to the use of different aspects in combination with each other. Equally, although the method has been described primarily in conjunction with the detection of methane and natural gas, other gas combinations are possible.

Claims

1. A method of identifying the presence of a first gas in a sample, comprising the steps of: providing a detector having a sensor responsive to a first wavelength, a sensor responsive to a second wavelength, and a sensor for collecting reference readings; analyzing a gas sample to obtain a first absorbance reading corresponding to the first wavelength, a second absorbance reading corresponding to the second wavelength, and a reference reading; calculating a first absorption number using the first absorption reading and the reference reading, and calculating a second absorption number using the second absorption reading and the reference reading; applying a linearizer function to each of the first absorption number and the second absorption number to calculate first and second concentration numbers; calibrating the sensor for each wavelength for detecting the first gas so that data collected at each wavelength gives the same reading of the first and second concentration numbers when only the first gas is present in the sample; calculating a ratio of the first concentration number to the second concentration number, and using the ratio to identify whether only the first gas is present in the sample.

2. The method of claim 1, wherein the first gas is methane and the sample comprises higher hydrocarbons.

3. The process according to claim 2, wherein the higher hydrocarbons are propane and butane mixed with methane.

4. The method of any one of claims 1 to 3, wherein the first wavelength is 3.3 microns plus or minus 0.1 microns, and the second wavelength is 3.4 microns plus or minus 0.1 microns.

5. The method of claim 4, wherein the step of calibrating the sensor comprises: The sensor is calibrated for detecting methane so that identical readings of the first and second concentration figures are obtained at each wavelength when analyzing a sample containing methane but no other hydrocarbons.

6. A method according to any one of claims 1 to 3 or 5, wherein compensation is performed on the readings of the first and second concentration figures at each wavelength respectively, in order to eliminate errors due to changes in environmental parameters. The method of claim 6 , wherein at least one of temperature compensation and pressure compensation is performed.

8. The method of claim 7, wherein temperature compensation is performed in two stages, the two stages being zero drift correction and span drift correction.

9. The method of any one of claims 1 to 3, 5, 7 or 8, wherein the sensor responsive to the first wavelength is a first sensor and the sensor responsive to the second wavelength is a second sensor different from the first sensor.

10. A method according to any one of claims 1 to 3, 5, 7 or 8, wherein a single sensor operable in different modes is used to analyse the gas sample to obtain absorption readings corresponding to both the first and second wavelengths.

11. A method according to any one of claims 1 to 3, 5, 7 or 8, wherein each sensor is an infrared radiation absorption sensor.

12. A selective gas detection device comprising: a sensor responsive to a first wavelength, a sensor responsive to a second wavelength, and a sensor for collecting reference readings; and a processing component programmed to: calculate a first absorption number using the reading for the first wavelength and the reference reading and to calculate a second absorption number using the reading for the second wavelength and the reference reading; calculate first and second concentration numbers by applying a linearizer function to each of the first and second absorption readings; the sensor for each wavelength is calibrated for detecting a first gas so that the first and second concentration numbers read the same for each wavelength when only the first gas is present in the sample; calculate a ratio of the first concentration number to the second concentration number; and identify the gas present in the sample based on the calculated ratio.

13. The selective gas detection apparatus of claim 12, wherein a separate sensor is used for each of the first wavelength, the second wavelength, and the reference reading.

Images