Method of analysing gas chromatography data

Abstract

A method of analysing gas chromatography data is described. During the method, a first response factor data set acquired from a gas chromatograph (GC) apparatus during a procedure on a calibration or reference gas sample at a first time is received. One or more additional response factor data sets acquired from the gas chromatograph apparatus during a procedure on a calibration or reference gas sample from one or more later times are received. The method comprises calculating a measure of uncertainty for at least one compound of the reference gas sample from the first and additional response factor data sets. The one or more later times are during an operational period of the gas chromatograph apparatus. The measure of uncertainty may be used to, for example, identify the necessity to perform a maintenance action in the GC or to assess whether the GC is in a healthy or unhealthy condition.

Claims

1. A method of determining an operating condition of a gas chromatograph comprising: receiving a first response factor data set acquired from a gas chromatograph (GC) apparatus during a procedure on a calibration or reference gas sample at a first time; receiving one or more additional response factor data sets acquired from the gas chromatograph apparatus during a procedure on a calibration or reference gas sample from one or more later times, wherein the one or more later times are during an operational period of the gas chromatograph apparatus; calculating a measure of uncertainty for at least one compound of the reference gas sample from the first and additional response factor data sets; wherein calculating a measure of uncertainty comprises calculating a relative sensitivity of a desired gas property to a change in concentration of at least one compound; and comparing the calculated measure of uncertainty to a predetermined threshold value to determine an operating condition of the GC apparatus.

2. The method according to claim 1, wherein the calibration or reference gas sample is a working reference mixture.

3. The method according to claim 1, wherein the operational period is a period in which the gas chromatograph apparatus is in normal use.

4. The method according to claim 1, wherein the one or more additional response factor data sets comprises a plurality of data sets acquired over the operational period.

5. The method according to claim 4, wherein the one or more additional response factor data sets comprises data sets acquired at regular or irregular intervals interspersed between performing tests on unknown gas samples, and wherein the method comprises calculating an updated measure of uncertainty data at regular or irregular intervals during the period of operation.

6. The method according to claim 1, comprising calculating the measure of uncertainty data after every calibration of the GC.

7. The method according to claim 1, comprising calculating the measure of uncertainty data using historical reference data generated when the GC is known or assumed to be functioning correctly.

8. The method according to claim 1, wherein calculating a measure of uncertainty comprises calculating normalised compositions derived from calculated peak areas and/or historical response factors.

9. The method according to claim 1, wherein calculating a measure of uncertainty comprises calculating a standard relative uncertainty.

10. The method according to claim 1, wherein calculating a measure of uncertainty comprises calculating a combined relative uncertainty value.

11. The method according to claim 1, comprising calculating the combined relative uncertainty value by combining two or more of: a relative uncertainty of the calibration gas, an uncertainty of at least one compound derived from the GC reproducibility and/or a repeatability relative uncertainty.

12. The method according to claim 1, comprising calculating a combined relative uncertainty value for multiple compounds.

13. The method according to claim 1, wherein calculating a measure of uncertainty comprises calculating a combined standard uncertainty of a gas property by combining a calculated relative sensitivity of a desired gas property with a standard relative uncertainty.

14. The method according to claim 1 wherein calculating a measure of uncertainty comprises calculating a combined relative uncertainty value, and wherein calculating a combined standard uncertainty of a gas property comprises combining a calculated relative sensitivity of the desired gas property with a combined relative uncertainty.

15. The method according to claim 1, wherein calculating a measure of uncertainty comprises calculating a combined expanded uncertainty of a gas property.

16. The method according to claim 1, comprising identifying or scheduling a requirement for one or more maintenance operations on the GC apparatus.

17. The method according to claim 16, wherein the one or more maintenance operations is selected form the group consisting of: a general troubleshooting to identify components of the GC which are not in a correct functional state; a general maintenance service in which at least one common maintenance action is performed; changing a GC valve; replacing the calibration gas; or changing one or more columns.

18. A method of determining an operating condition of a gas chromatography comprising: receiving a first response factor data set acquired from a gas chromatograph apparatus during a procedure on a calibration or reference gas sample at a first time; receiving one or more additional response factor data sets acquired from the gas chromatograph apparatus during a procedure on a calibration or reference gas sample from one or more later times; wherein the one or more later times are when the gas chromatograph apparatus is in situ or on site in a gas sample analysis facility; calculating a measure of uncertainty for at least one compound of the reference gas sample from the first and additional response factor data sets; and comparing the calculated measure of uncertainty to a predetermined threshold value to determine an operating condition of the GC apparatus.

19. A computerised method of analysing gas chromatography data comprising performing the method of claim 1 in a computer system.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) There will now be described, by way of example only, various embodiments of the invention with reference to the drawings and examples, of which:

(2) FIG. 1 shows schematically a typical three-column gas chromatograph;

(3) FIG. 2 is a block diagram of the method of GC Condition Based Monitoring using historical calibration data for updated combined uncertainty calculation according to a first embodiment of the invention;

(4) FIG. 3 is a block diagram of the method of GC Condition Based Monitoring using historical calibration data for updated combined uncertainty calculation according to a second embodiment of the invention; and

(5) FIG. 4 is a schematic representation of a computer performing the method of GC Condition Based Monitoring using a combined uncertainty calculation.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS

(6) By way of example only, embodiments of the invention are described in applications of gas chromatography to the analysis of hydrocarbon-containing gas samples, for example natural gas samples. Applications to C.sub.6+GC systems are described, but the invention is not so limited and its principles may be applied to other GC systems. The embodiments are generally described as being implemented in a computer system and it will be appreciated that the invention may be implemented in software, hardware, firmware, or a combination thereof.

(7) Referring firstly to FIG. 2, there is shown a block diagram representing the steps of a method 200 in accordance with a first embodiment of the invention. This is a method 200 of GC Condition Based Monitoring for calculating updated or live uncertainty using historical calibration data. This method can also be described as an expert system for identifying maintenance requirements.

(8) In this example, the GC has been calibrated in the factory using a conventional multilevel calibration technique, and a repeatability relative uncertainty value for the apparatus is calculated using standard methodology. This repeatability relative uncertainty value may then be used in the assessment of uncertainty in calorific value calculations on data acquired from tests of gas samples, as is known in the art.

(9) With the GC apparatus in situ, the first step 201 of the method is to obtain the composition (Comp), which in the industry standard is referred to as (x.sub.i, wrm) of a calibration gas which is to be used for the periodical calibrations. This is obtained when the GC apparatus is assumed to be in a good condition. The response factors from this initial calibration gas measurement (RF.sub.i,FP) are determined and are used as reference or footprint data (step 202). The footprint data is data acquired from the gas chromatography apparatus in a known healthy condition such as after a multilevel calibration.

(10) The third step 203 is to obtain the response factors (RF.sub.i,cal) of one or more calibration reports, acquired throughout the operational period of the GC apparatus and after the initial calibration gas measurement. Historical response factors (RF.sub.i,cal) obtained from regular calibrations of the GC may be used, and can be obtained from a GC maintenance software installed in a computer connected to the GC which automatically or manually receives periodical calibration data. Alternatively these data can also be obtained from a written register of periodical calibration data.

(11) The previous data is received in a computer system, which is preferably a personal computer programmed to execute the steps of the method. The computer system can be interfaced with the GC. Alternatively the computer system can be connected via the internet to the GC operational system. The method can also be performed using data collected and stored in a GC at an earlier time

(12) The next step 204 is to calculate a peak area (PA), which in the industry standard is referred to as (R.sub.i,cal), for every component of the calibration gas from the last calibration data using the formula:
PA=RF.sub.calCalibration gas composition (% mol)(Eq.3)
which in the industry standard terminology is given by:
R.sub.i,cal=RF.sub.i,calx.sub.i,wrm(Eq.4)
where PA or R.sub.i,cal is peak area and RF.sub.i,cal is response factor from a periodical calibration. In the next step 205, the unnormalised composition (Un-Comp, which in the industry standard is referred to as (x.sub.i*)) of the calibration gas is calculated using the calibration peak areas obtained in the previous step and the reference (or footprint) response factors, according to the formula:
Unnormalised composition=PA/RF.sub.foot(Eq.5)
which in the industry standard terminology is given by;

(13) $\begin{array}{cc}{x}_i^{*}=\frac{R_{i,\mathrm{cal}}}{{\mathrm{RF}}_{i,\mathrm{FP}}}& \left(\mathrm{Eq}.6\right)\end{array}$

(14) The present method uses normalised compositions for calculating the uncertainty of the GC measurements (or derived gas properties) from the repeatability and from the reproducibility data obtained during an operational period of the GC. It has been found that this is more effective than using unnormalised compositions for calculating the uncertainty of the GC measurements (or derived gas properties) from the repeatability data, as the present method of calculating the uncertainty of the GC measurements does not reflect an overestimation of uncertainty due to changes in the composition of the calibration gas. The use of unnormalised compositions in calculating the uncertainty could be more sensitive to changes in the calibration gas composition.

(15) The next step 206 is to normalise the compositions obtained in the previous step. In this context, normalise means to express in % mol the composition of a gas containing the quantities of gas expressed by the unnormalised compositions figures. The normalised composition (N-Comp, which in industry standard is referred to as (x.sub.i)) of a component can be expressed as:
N-Comp (% mol)=(Un-Comp/sum of all unnormalised compositions of the components in the calibration gas)100(Eq.7)
which in industry standard terminology is given by:

(16) $\begin{array}{cc}{x}_i=\frac{x_i^{*}}{\underset{i=1}{\overset{q}{.Math.}}{x}_i^{*}}100& \left(\mathrm{Eq}.8\right)\end{array}$

(17) The previous steps are preferably done on a regular basis, for example most preferably on a daily basis. These method steps may be preferably carried out after every calibration, and therefore may use all available calibration data. However, it is not necessary to use data from each calibration, and the measure of uncertainty data may be updated at regular or irregular intervals during a period of operation. For example, the steps may be carried out after a selected number of calibrations or after a selected operational period.

(18) In order to calculate a live or updated uncertainty in the Calorific Value (CV) of the GC reproducibility measurements in the actual GC state, the following steps are followed.

(19) Firstly, it is necessary to calculate for each component of the calibration gas the standard deviation of the composition of the calibration gas using the historical data from regular calibrations. The standard deviation is a well known statistical parameter, which may be calculated according to the following formula:

(20) $\begin{array}{cc}s\left(x_i\right)=\sqrt{\frac{\underset{n=1}{\overset{N}{.Math.}}\left(x_{i,n}-{\stackrel{\_}{x}}_i\right)}{N-1}}& \left(\mathrm{Eq}.9\right)\end{array}$

(21) Here, N represents the number of historical calibrations since the last footprint data was taken; x.sub.i represents the i-th measured normalised mol fraction of a given component using the i-th measured response/peak area and the response factor from the footprint data; and x.sub.i represents the average measured normalised mol fraction of each component when measured using the response factor from the footprint data.

(22) In the next step 207 the reproducibility relative uncertainty U.sub.rpd(x.sub.i) for a component is calculated by dividing the standard deviation of the normalised historical calibration data by the concentration of that component in the working reference mixture according to:

(23) $\begin{array}{cc}{U}_{\mathrm{rpd}}\left(x_i\right)=\frac{s\left(x_i\right)}{x_{i,\mathrm{wrm}}}& \left(\mathrm{Eq}.10\right)\end{array}$

(24) Calculation of the calorific value (CV) of a gas of known composition is performed by a conventional method. The next step 208 is to calculate the relative sensitivity of the calorific value of natural gas of standard composition for every component of the calibration gas. This is done by dividing the relative change in calorific value produced by a change in a certain component's concentration value by the relative change in the certain component's concentration.

(25) The third step 209 is to calculate the Combined Standard Uncertainty of the Calorific Value of the natural gas by adding the squares of the products of each component's CV sensitivity by each component's relative uncertainty, and square-rooting the result. The Combined Standard Uncertainty is the sum of the previous calculations for all the components of the gas.

(26) The next step 210 is to calculate the Combined Expanded uncertainty of the Calorific Value by multiplying the Combined Standard Uncertainty of the Calorific Value by a numerical coverage factor, which may be for example a factor of 2.

(27) In the final step 211 the Combined Expanded Uncertainty value obtained is compared to a threshold value. If the calculated combined expanded uncertainty value is above of the threshold value, then a signal output in the form of, for example a sound, light, alarm, colour change is triggered indicating a maintenance action requirement 212.

(28) According to this embodiment, the gas chromatography data is analysed and transformed to produce an output signal indicative of a healthy or unhealthy condition of the GC. By analysing this information, it can be determined when to intervene and perform maintenance on the GC system before it enters an unhealthy state.

(29) The method described herein uses historical reference or footprint data generated when the GC is known or assumed to be functioning correctly. Data such as oven temperature, carrier gas pressure, carrier gas flow rate, response factor etc. are recorded. These footprint values can be used as a tool to analyze historical calibration results. This is in contrast to the prior art techniques, which take in account uncertainty only from a simplified repeatability analysis. The present method considers uncertainty derived also from reproducibility data obtained from periodical calibrations with the footprint data as reference.

(30) The previously described example is an embodiment of the invention in which the uncertainty calculation is associated to the calculation of the Calorific Value of the gas, but this must not be taken as a limitation of the principles of the invention. It will be appreciated that other gas properties can be calculated. Other desired gas properties that could be calculated include (without limitation) density, thermal conductivity, compressibility, and molecular weight.

(31) FIG. 3 depicts a method, generally referred to at 300, according to an alternative embodiment of the invention in which the uncertainty calculation is not associated with the calculation of any specific gas property. The method 300 is similar to the method 200 and will be understood from FIG. 2 and the accompanying description. In the method of FIG. 3 the steps refer to any desired output of a gas property that can be calculated and is dependent on the gas composition.

(32) The initial steps 301 to 306 of the method of FIG. 3 are equivalent to the initial steps 201 to 206 of the method of FIG. 2 and will not be described again for reasons of brevity.

(33) In the method 300, the step of calculating a reproducibility relative uncertainty is shown as two 307a and 307b (as opposed to the single step 207 in FIG. 2). In sub-step 307a, the standard deviation of each component is calculated. In step 307b, the standard relative reproducibility uncertainty is calculated.

(34) The method 300 of FIG. 3 comprises the additional step 310 of calculating a combined relative uncertainty of each component gas from the standard relative reproducibility uncertainty (from 307a, a calibration gas or working reference mixture relative uncertainty 308; and a repeatability relative uncertainty 309.

(35) To obtaining the calibration gas relative uncertainty U.sub.wrm(x.sub.i) (step 308) it is normally sufficient to check the United Kingdom Accreditation Service (UKAS) certification accompanying the gas cylinder which provides the working reference mixture relative uncertainty.

(36) To calculating the repeatability relative uncertainty in step 309 the procedure described in ISO 10723 can be followed. The standard deviation of the response for each component is then expressed as:
s.sub.i=a+bx.sub.i*+cx.sub.i*.sup.2+dx.sub.i*.sup.3(Eq.11)

(37) s.sub.i is standard deviation

(38) a, b, c, d are the coefficients of linear regression of s.sub.i on x.sub.i*

(39) x.sub.i* is un-normalised concentration of component i

(40) The repeatability of each measured component is a function of the repeatability of the response on the working reference mixture (calibration gas) and the repeatability of the response at the concentration being measured. The calculation of the standard deviation of the sample mixture is described in both ISO 6974-2 and ISO 10723. The following equation is defined in ISO 10723:

(41) $\begin{array}{cc}{\left[\frac{s\left(x_i^{*}\right)}{x_i^{*}}\right]}^2={\left[\frac{s\left(y_{\mathrm{is}}\right)}{y_{\mathrm{is}}}\right]}^2+{\left[\frac{s\left(y_{\mathrm{istd}}\right)}{y_{\mathrm{istd}}}\right]}^2& \left(\mathrm{Eq}.12\right)\end{array}$

(42) Where:

(43) y.sub.is and y.sub.istd are the instrument responses to component i in the sample and standard s(y.sub.is) and s(y.sub.istd) are the respective standard deviations;

(44) x.sub.i* is the un-normalised concentration of component i; and

(45) s(x.sub.i*) is the standard deviation of the un-normalised components x.sub.i*

(46) Having calculated the standard deviation of each un-normalised component, the standard deviation of the normalised mole fractions as defined by ISO 6974-2 is given:

(47) $\begin{array}{cc}s\left(x_i\right)=x_i\sqrt{\frac{1-2x_i^{*}}{x_i^{*2}}{s\left(x_i^{*}\right)}^2+\underset{w=1}{\overset{q}{.Math.}}{s\left(x_w^{*}\right)}^2}& \left(\mathrm{Eq}.13\right)\end{array}$

(48) Where:

(49) x.sub.i is the normalised mole fraction of component i; and

(50) x.sub.i* is the un-normalised mole fraction of component i

(51) The repeatability associated relative uncertainty U.sub.rpt(x.sub.i) is then calculated according to the following expression:

(52) $\begin{array}{cc}{U}_{\mathrm{rpt}}\left(x_i\right)=\frac{s\left(x_i\right)}{x_{i,\mathrm{wrm}}}& \left(\mathrm{Eq}.14\right)\end{array}$

(53) The combined relative uncertainty U.sub.com(x.sub.i) is then calculated by combining the repeatability relative uncertainty U.sub.rpt(x.sub.i); from the calibration gas relative uncertainty U.sub.wrm(x.sub.i) and the previously calculated reproducibility relative uncertainty U, using the following expression:
U.sub.com(x.sub.i)={square root over ((U.sub.wrm(x.sub.i)).sup.2+(U.sub.rpt(x.sub.i))+(U.sub.rpd(x.sub.i)).sup.2)}(Eq.15)

(54) The output of step 310 in FIG. 3 is equivalent to the output of step 207 of FIG. 2, and from this point until the last step, both methods are again similar and have equivalent method steps which will not be repeated for reasons of brevity (i.e. steps 311 to 315 of FIG. 3 are equivalent to steps 208 to 2012 of the method of FIG. 2).

(55) The method 300 may provide a more balanced value of the live or updated uncertainty compared to the method 200, because it takes into account different sources of uncertainty to provide a combined uncertainty measure. Therefore the identification of a maintenance action requirement may be more consistent with the actual state of the GC.

(56) The methods of the present invention are preferably implemented in software and executed in a computer system. FIG. 4 shows schematically a computer 400 performing the method 300 of GC Condition Based Monitoring (shown in FIG. 3) using a combined uncertainty calculation.

(57) The computer 400 is configured to execute a software program to perform the calculations and comparisons of the method 300. According to the method, the software generates output data displays and generates a signal to the computer to display an alert message that reads maintenance action required. This informs the operator (not shown) that it is necessary to and schedule an appropriate maintenance action. There may be an audible signal 402 for making the alert message more prominent or noticeable.

(58) This embodiment of the invention is described by way of example only and it will be understood that other variations can be implemented without departing from the principles of invention. For example the alert message could be an SMS message sent to a distant mobile phone or an e-mail sent to an e-mail account. Other types of alert messages could be used.

(59) The invention provides a method of analysing gas chromatography data. The method uses historical calibration data collected during an operational period of the GC or when the GC is on-site to calculate an uncertainty value representative of the current condition of the GC, i.e. a live or updated uncertainty value. The live uncertainty value is linked to a calculated gas property, for example its calorific value, density, or compressibility. and may be compared with a threshold value in order to, for example, identify the necessity to perform a maintenance action in the GC or to assess whether the GC is in a healthy or unhealthy condition.

(60) The present invention provides improved methods of analysing gas chromatography data, and in particular, improved methods of uncertainty monitoring for gas chromatography apparatus. The invention provides methods of calibrating, monitoring and/or maintaining gas chromatography equipment which permit monitoring a GC whilst also providing an on-line estimate of the overall uncertainty in the natural gas composition measurements. The methods are improved with respect to the prior art by providing an accurate and up to date uncertainty calculation which facilitates monitoring and maintenance scheduling methods for on-site gas chromatographs.

(61) Various modifications may be made within the scope of the invention as herein intended, and embodiments of the invention may include combinations of features other than those expressly claimed. Although embodiments of the invention are described with reference to three-column gas chromatographs, the principles of the invention can be applied to other types of gas chromatography system.

REFERENCES

(62) [1] S. Fosse, R. Sakariassen, and F. Teknikk, Online gas chromatography: A technical and historical overviewdesign and maintenance advices to achieve an accurate end result, North Sea Flow Measurement Workshop, 2010. [2] J. Witte, Further developments in the design & implementation of an advanced online condition based monitoring system & a dirty meter prediction model for custody transfer ultrasonic gas flow meters, North Sea Flow Measurement Workshop, 2011. [3] M. Rensing and T. J. Cunningham, Coriolis flowmeter verification via embedded modal analysis, IMAC xxvIII, 2010. [4] M. Skelton, S. Barrons, J. Ayre, and R. Steven, Developments in the self-diagnostic capabilities of orifice plate meters, North Sea Flow Measurement Workshop, 2010. [5] A. Sutan, C. Johnson, and J. Laidlaw, Three columns gas chromatograph analysis using correlation between component's molecular weight and its response factor, North Sea Flow Measurement Workshop, 2009. [6] D. Pettigrw, Advances in on-line gas chromatography in the natural gas industry, North Sea Flow Measurement Workshop, 2000. [7] C. Cowper and R. Mounce, Evaluation of on-line chromatograph performance, North Sea Flow Measurement Workshop, 1997. [8] M. Crane, R. Beaty, and W. Lake, Calibration of gas chromatograph for improved performance, North Sea Flow Measurement Workshop, 1995. [9] ISO 10723 Natural gas: performance evaluation for on-line analytical systems. International Organization for Standardization [10] ASTM D1945 Standard test method for analysis of natural gas by gas chromatography. American Society for testing and Materials. [11] ISO 6974 Determination of composition with defined uncertainty by gas chromatography. International Organization for Standardization.