Julia C.Shapiro, Ph.D.
Laboratory Manager Gerald R.Burkett, N.B., B.S.
Plant Manager
William A.Crowley, M.B., B.S.
Operations Manager Scott Specialty Gases, Inc.
Troy, Michigan 48083
ABSTRACT
The United States Natural Gas Industry requirements for high accuracy analysis of natural gas pointed out a need for interlaboratory comparison of compositional analysis. Scott Specialty Gases, in response to this
need, has developed a Cross Reference Service to correlate analytical results between laboratories. The statistically summarized results of the interlaboratory correlations have identified strong and weak points of
existing analytical techniques. The accuracy of analysis of specific components varies from a relative standard deviation of 2% to as high as 20%. Although the Natural Gas Industry would like to analyze energy to ±1 BTU, the results of the Cross Reference Service indicate an accuracy of ±2.5 BTU at a 95%
confidence level. The problems causing inter!aboratory disagreement are discussed with particular emphasis on differences in analytical techniques, instrumentation, and absence of uniform standards throughout the
Industry.
The United States Natural Gas Industry’s goal for precision in measurement of BTU is less than one BTU.
The need for such extreme accuracy equal to ±0.1% can be understood from the magnitude of United States natural gas consumption (Table 1).
TABLE 1
UNITED STATES 1984 NATURAL GAS MARKET
Energy, BTU, Quads (Billions) Dollar Estimate
Residential 7.4 43.73
Commercial 7.0 27.51
Industrial 3.2 11.62
Miscellaneous .5 0
Energy, BTU, Quads (Billions) Dollar Estimate
18.1 82.86
1 Quad—1015 BTU
Estimates of United States gas production in 1984 are approximately 18 quadrillion BTU or 83 billion dollars. In this case an error of one BTU has a potential of gains and losses to transmission and pipeline companies and public utilities of 83 million dollars a year.
The extreme accuracy requirements of measuring BTU encouraged development of compositional analysis of natural gas and calculation of the theoretical BTU value which is now the standard practice. The majority of the Natural Gas Industry uses gas chromatography as a fast, efficient and reliable analytical technique.
There are several analytical procedures published; these include American Society for Testing and Materials, Gas Processors Association, as well as those developed by various G.C. manufacturers and users.
Many G.C. manufacturers have developed instruments specially designed for natural gas analysis.
As it is known, one of the most essential factors in G.C. analysis is the existence of a reliable standard. Many specialty gas companies, including Scott, provide gravimetric compositional standards. The accuracy of standards available varies. A primary standard reference from the National Bureau of Standards has not existed until recently.
Thus, all three factors mentioned above, various analytical techniques, different instrumentation and absence of a primary standard reference material can cause a significant difference in analytical results between laboratories.
The objective of implementation by Scott of the Natural Gas Cross Reference Service was to improve interlaboratory correlation by allowing laboratories to compare themselves to others in the Industry.
The Natural Gas Cross Reference Service is performed on a quarterly basis. The process consists of:
1. Manufacture of the Natural Gas mixtures.
2. Transfer of the mixture to multiple, small cylinders.
3. Analysis of samples to insure their homogenity or uniformity.
4. Supply of samples to subscribers.
5. Compilation of the reported data.
6. Preparation of a statistical report of results submitted.
7. Submission of the report to participants.
The BTU Cross Reference Service cylinders are manufactured from a master cylinder containing at least 8–
9 natural gas components in a Methane balance gas. This mixture is then transferred to 25 or more smaller cylinders.
In the process of manufacturing, the major problems appear to be the same as has been met in the field:
transferring of natural gas mixtures without condensation, separation of heavy hydrocarbons, and pressure and temperature changes which occur inside of the sample system.
Each sample cylinder was analyzed, in order to assure their homogenity, using a Varian Vista 6000 Gas Chromatograph. When developing this technique we encountered the common G.C. problems: separation of Ethane from Methane in the presence of 90% Methane, separation of Isobutane from Butane, Isopentane from Pentane and condensation of higher hydrocarbons (Hexane, Heptane) on intermediate lines before injection.
All hydrocarbons were analyzed using an 80/100 mesh n-octane on porasil C stainless steel column (6′×1/
8″) and a FID detector with temperature programming. The temperature is programmed from 30°C (2 minutes) to 110° at 10°C/min.
Nitrogen and Carbon Dioxide are analyzed on a Carbosieve S column (10′×1/8″) at 30°C isothermally using a TCD detector. Both analyses used Helium carrier gas at 30 ml/min., a 0.25 cc loop and a heated valco valve. An external gravimetric standard was used for calibration throughout the process of analysis.
A sample of the chromatogram is shown in Figure 1.
This analytical procedure, along with proper manufacturing technique, gave us a guarantee of homogenity of the gas with spread less than 0.5% relative standard deviation of each component from cylinder to cylinder. This deviation includes uncertainties in transferring mixtures into cylinders and uncertainties in the analysis (Table 2).
Each participant returns his analytical results to Scott for compilation. We normalize all of these results to 100%. All BTU values are recalculated using the GPA procedure. This process of compiling results is necessary to make them comparable.
The data is then statistically summarized and the resulting report issued to subscribers. The report consists of the following: All participants data normalized to 100% (Table 3), the statistical analysis of reported results (Table 4). The statistical summary includes the mean, median, estimate of standard deviation, estimate of standard error, number of observations and number of outliers.
The balance of the report includes histograms of the participants data. Figures 2–3 show sample distributions for some components.
Besides the regular Cross Reference Service provided by Scott on a quarterly basis, we arranged in 1985 a Round Robin testing program among nine laboratories in the Natural Gas Industry. In this program a cylinder with synthetic natural gas mixture was manufactured and analyzed by Scott, then sent to participating laboratories for analysis and returned back to Scott for final analysis. The results were published. Table 5 shows the precision of analysis by Scott and by the nine laboratories.
The results of all Cross Reference services (Table 6&7) which has been provided during the last two years can be summarized in the following conclusions:
1. There is a gradual improvement in the analysis of most components in the last series of BTU Cross Reference Services compared to the first. Especially significant are changes for Ethane, Propane, Carbon Dioxide and Nitrogen.
2. The error in component analysis increases with carbon content of the molecule from 0.18% Relative Standard Deviation for Methane to 52% for Hexane.
Table 2
Relative Standard Deviation from Cylinder to Cylinder for Each Component in One Series of BTU Service (Analyzed by Scott).
Methane 0.38%
Ethane 0.31%
Propane 0.37%
Butanes 0.52%
Pentanes 0.46%
Hexane 0.53%
Nitrogen 0.24%
Carbon Dioxide 0.23%
FIGURE 1
Sample of Chromatograms
TABLE 4
Statistical Summary of Reported Analytical Results Methane
(Vol %)
Ethane (Vol
%)
Propane (Vol %)
Butane (Vol
%)
BTU (Vol
%) No. of
Observ.
14 15 15 15 14
Min.
Observed Value
88.5243 4.8824 1.0362 .1620 1024.5400
Max.
Observed Value
89.0780 5.1699 1.1406 .2601 1030.4699
Mean Value 88.8917 4.9838 1.0725 .1931 1027.9697
Median Value
88.9406 4.9670 1.0670 .1840 1028.3449
Scott Value 88.7845 5.0884 1.0691 .1768 1029.0400
Est. Std.
Dev.
.16105 .07412 .02668 .02619 1.64478
Est. Std.
Error
.04304 .01914 .00689 .00676 .43959
Coeffic. of Variance (%)
.18118 1.48730 2.48729 13.56396 7.65446
No. of Outlying Values
1 0 0 0 2
TABLE 3
Tabulation of Reported Analytical Results
Precision of Analysis by Scott and Industry (Round Robin)
Components Analysis by Scott % Diff. Between First &
Final
Relative Standard Deviation of Industry % (n=9)
First Last
Methane 95.5280 95.5220 0.006 0.22
Ethane 1.6140 1.6140 0 3.08
Propane 0.4973 0.5019 0.925 1.17
FIGURE 2
Frequency Distribution of Reported Analyses—Hexane (Vol %)
FIGURE 3
Frequency Distribution of Reported Analyses—N2 (Vol %) TABLE 5
Components Analysis by Scott % Diff. Between First &
Final
Relative Standard Deviation of Industry % (n=9)
First Last
i-Butane 0.1676 0.1676 0 1.72
n-Butane 0.2156 0.2149 0.326 1.75
i-Pentane 0.0547 0.0551 0.730 18.09
n-Pentane 0.0571 0.0572 0 69.52
Hexane 0.0240 0.0245 2.100
Heptane 0.0179 0.0180 0.560 32.74 (C6+)
Octane 0.0045 0.0040 11.100
N2 1.0500 1.0500 0 4.39
CO2 0.7400 0.7426 0.350 16.97
He 0.0212 0.0225 6.100 14.14
BTU 1029.64 1029.69 0 0.06
3. The largest errors were observed for components requiring accurate G.C. separation technique, as isobutane and n-butane, iso-pentane and n-pentane. The combined error for the sum of Iso-butane and n-butane and sum of iso-Pentane and n-Pentane is significantly less than for the individual components.
The high relative standard deviation for Hexane is explained by the fact that some customers analyze heavy hydrocarbons as Hexane+ mixtures, some analyze them separately.
4. Despite the relatively high error in minor components approaching 20–30% relative standard deviation, the process of calculating BTU value from compositional analysis results in a relative standard deviation of 0.16%.
5. The analysis of BTU value show a significant improvement in relative standard deviation from 0.47%
two years ago to 0.16% most recently.
Table 6
Relative Standard Deviation % for All Components in Seven Cross Reference Services
BTU BTU BTU BTU BTU BTU BTU
Component #1 #2 #3 #4 #5 #6 #7
Methane 0.21 0.25 0.36 0.17 0.32 0.16 0.18
Ethane 8.25 3.68 2.45 1.65 2.26 2.22 1.49
Propane 6.41 9.11 2.68 2.19 6.40 1.84 2.49
Iso-butane 5.22 7.70 30.91 3.11 10.86 13.56
n-butane 20.86 8.57 20.23 23.56 7.65
Butanes 7.8 4.8 7.6 4.60
Iso-Pentane 18.09
n-Pentane 28.20 9.14 26.35 69.52
Pentanes 14.50
Hexane 19.03 10.87 50.29 20.37 52.10
BTU BTU BTU BTU BTU BTU BTU
Component #1 #2 #3 #4 #5 #6 #7
Heptane 25.20 123.3 24.64
Hydrogen 75.88 9.13
Carbon Dioxide
7.45 8.87 1.92 3.26 1.99 3.33 4.91
Nitrogen 2.45 2.32 1.13 2.35 3.01 3.19 1.95
BTU 0.47 0.51 0.34 0.24 0.19 0.34 0.16
6. All observed Industry’s errors were much higher than Scott’s estimated accuracy. In an attempt to rationalize the high errors of compositional analysis with the relatively low errors in BTU analysis the following model of contributing BTU errors from component analysis was developed (Table 7). The Table shows that at 95% Confidence level (assumed two standard deviations), Scott’s precision for mixtures of 1028 BTU is estimated at 0.40 BTU compared to the Industry’s 2.5 BTU.
7. Although Nitrogen has a relative standard deviation of 2% which is significantly less than many of the heavy hydrocarbons, it was the largest contributor in BTU error because it has no heating value.
Table 7
Comparison of Standard Deviation of BTU for Scott’s and Industry Analysis of Natural Gas
Industry Precision Scott Precision
Mean
%
Std.
Dev.
Contr ib. in BTU
Contri b. in BTU of Std.
Dev.
Mean
%
Std.
Dev.
Contr ib. in BTU
Contri b. in BTU of Std.
Dev.
C1 88.
93
1009 .7
88.
77±
1009 .7
C2 4.98
±
0.
074
37.
80 ±
0.56 5.09
±
0.
0193
38.
64 ±
0.
146
C3 1.07
±
0.
027
16.
18 ±
0.40 1.07
±
0.
0033
16.
13 ±
0.
050
C4 0.40
±
0.
021
8.99
±
0.47 0.39
±
0.
0020
8.77
±
0.
045 C6+ 0.03
±
0.
012
1.34
±
0.49 0.06
±
0.
0004
2.43
±
0.
015
N2 4.03
±
0.
079
−40.
65 ±
0.79 4.05
±
0.
0122
−40.
89 ±
0.
123
CO2 0.56
±
0.
027
−5.
61 ±
0.27 0.57
±
0.
0015
−5.
75 ±
0.
015
BTU 1027
.81 ± 1.
28*
1028 .33 ±
0.
20*
Estimated Accuracy of BTU Analysis (95%
Confidence at ±2% Standard Deviation)
2.5 BTU
0.4 BTU
Industry Precision Scott Precision Mean
%
Std.
Dev.
Contr ib. in BTU
Contri b. in BTU of Std.
Dev.
Mean
%
Std.
Dev.
Contr ib. in BTU
Contri b. in BTU of Std.
Dev.
* These numbers are calculated:
8. The fact the Industry precision is much lower than Scott’s can be explained by three factors contributed to it: various analytical instrumentation, different analytical techniques and absence of a “Primary Standard”. Table 8 shows the technique and instrumentation employed by the participants. Some participants used regular G.C.’s, some dedicated natural gas chromatographs. There were analysis done at isothermal, as well as T-programming conditions. Most participants analyzed the mixtures on two columns, but some used as many as four columns.
Table 8
Analytical Technique Used by Participants
Number of Participants Instruments Amount of Columns Detectors Conditions
3 G.C. 2 TCD Isotherm
8 G.C. 2 FID & TCD T—Progr.
1 G.C. 4 FID & TCD T—Progr.
3 G.C.Carle In Series
1 Gas Monitor Dedicated BTU Analyzer
CONCLUSIONS
Two years of providing the Cross Reference Service has proved the fol lowing:
A) Clear evidence of improving correlation between laboratories.
B) Industry precision is approximately ±2.5 BTU.
C) Industry should be able to achieve ±1 BTU or better accuracy.
D) The Cross Reference Services is justified as a method of obtaining improved interlaboratory correlation.
Edward J.Dahn Director—Sales/Marketing
Flow Technology, Inc.
Phoenix, AZ 85040
ABSTRACT
The paper describes a new and unique design of a primary Gas Flow Calibrator. It is an automatically controlled precision volumetric flowmeter calibrating device specifically designed for use with a variety of
gases over a wide range of pressures. The Aerotrak is a closed loop positive displacement system able to minimize both the quantity of gas used and the energy required to create the desired flow rate. The entire operation of the Aerotrak is under the automatic control of a uniquely programmed Personal Computer (PC). When initiated by the operator, the pre-selected program ensures the calibration data is taken after the
pressure and temperature surge created by the initial piston motion has completely decayed and the flow rate is constant. This is achieved by the computer comparing successive calibration points and accepting
data only when it has repeated within specified limits. As the actual data is being generated, it is continuously displayed for the test operator and stored in memory. Upon completion of the test, all data can
be viewed on the computer screen by the operator and immediately evaluated. The completed data is then stored on a diskette which can be used to generate hard copy printouts and the plots of curves.
DESCRIPTION
The Aerotrak Gas Calibrator design, not a modification of some existing calibration technique, is a new and truly unique approach to Gas Calibration. It is a closed loop positive displacement calibrator, combined with this new design concept is a powerful marriage with a Personal Computer and menu driven software program. The program automatically controls the total operation of the calibrator. In addition to rapid data reduction the PC Screen displays Final Test Report and Curves. This new Closed loop principle affords the user an opportunity to calibrate both gas and liquid with the same basic system. However, this paper will address only the Gas Application.
OPERATION
The system is charged with the selected gas ranging in pressure from 5 psia through 1440 psig @ 100 degrees F. The gas flow is created by a solid sealed piston which is moved through the length of a bored and honed flow tube with 12 RMS finish. The preselected velocity of the piston is established by a servo controlled motor. A rotary encoder mounted on the motor shaft provides the feedback necessary for constant rate control of the advancing piston. (Figure 1) The flow transducer under test is installed in an external flow path with the upstream and downstream legs geometrically symmetrical. As the piston is advanced it creates the gas flow through the flow transducer and then to the upstream side of the piston in the flow tube. At the time of initial motion of the piston a pressure/temperature surge is experienced by the flow transducer. As the pressure/temperature surge decays a steady state flow condition exists at the flowmeter. The Computer samples the variables required to make the flow rate calibrations such as transducer frequency, encoder frequency, pressure, temperature, delta pressure, and delta temperature.
Successive calibration points are calculated by the PC and when repeatability at the desired flowrate is achieved in a selected number of points the data is stored and the velocity of the piston is automatically changed to the next desired calibration point. This pattern is repeated until the required number of preselected consecutive points are gathered and are repeatable within the desired limit.
When the piston reaches the extreme downstream portion of the flow tube, a motor driven ball valve in the return line is opened and the gas is returned to the upstream side of the sealed piston while the piston is being moved to its original starting position. It is then ready for another calibration run.
SYSTEM CHARACTERISTICS The advantages of the closed loop gas calibration are as follows :
A. Minimizes the amount of gas required for test. The system is charged to the required pressure and remains at that pressure level throughout the test.
B. Minimizes the amount of energy required to create desired flow rate at working pressure.
C. Capable of calibrating flowmeters from partial vacuum to 1440 psig.
D. Capable of utilizing a variety of calibration gases.
E. Capable of calibrating at required operating densities with a selected gas.
F. Easily checked for seal leaks.
G. Time required for calibration is reduced to a few minutes versus hours for the present conventional methods.
H. Rapid change from one density to another.
I. Does not require correction for changes in atmospheric pressure.
J. Provides a high degree of consistency in Test Program through standardization of technique.
K. Improved ease of operation by less skilled people.
L. The symmetrical design of the flow paths allows flowmeters to be placed in a position which minimizes the amount of pressure/temperature correction required.
M. The displacement is determined by conventional water draw technique using test measures traceable to National Bureau of Standards.
SOFTWARE PACKAGE
Upon completion of the automated calibration runs the raw and calculated data is displayed on the Computer Screen. The operator is afforded the opportunity to immediately view and approve the test results. The completed calibration data is stored on a diskette and then used to generate hard copy print-outs and plots of curves. (Figure 2)
The Aerotrak Gas Calibrator is capable of determining the changes in flowmeter calibration characteristics, where the calibration charge pressure is changed from one level to another. The test data is stored, resulting from a test at one pressure level. The calibrator is pressurized at another level and an identical test is conducted. The PC has the capability of plotting both curves on the PC Screen, or generating a hard copy via a plotter. (Figures 3, 4 and 5) The operator has the option to view a variety of plots on the FIGURE 1
PC Screen. The variables includes frequency, K-factor, flow rate, density, linearity limits, and time.
Additionally the operator may select his units of volume and time.
TYPES OF TRANSDUCER
The Aerotrak has an inherent capability to calibrate a variety of flow transducers, such as turbine, variable area, nozzles, vortex shedders, orifice plates, laminar, thermal mass flow controllers.
The type of inputs include both frequency and non-pulsing analog inputs such as 0 to 5 VDC, 4 to 20 MA or 10 to 50 MA. Manual inputs are required for the variable area flowmeters.
FUTURE DEVELOPMENTS
Development work is being carried on to drastically increase the maximum flow rate capability. This increase in flow rate is required to support the needs generated for the Natural Gas Industry.
FIGURE 2
FIGURE 3
FIGURE 4
FIGURE 5
FIGURE 6