Test Method: 1998-06-18 Automated Compressed Gas Certifications - Technical Paper
This page last reviewed April 29, 2010
Automated Method for Establishing National Institute of Standards and Technology Traceable Certifications of Secondary and Tertiary Compressed Gas Standards.
Gerhard H. Achtelik, Kitty A. Howard, Elbert V. Lawrence III
California Air Resources Board
1927 13th Street
Sacramento CA. 95812-7107
The mission of the Air Resources Board (ARB) Standards Laboratory is to provide ozone, compressed gas, and flow transfer standards traceable to the highest attainable primary standards. In early 1980, staff of the ARB was faced with the challenge of providing calibration and precision compressed gases for an increasing number of field air monitoring stations located over a large geographic area. The demand for traceable gases which could be used to calibrate field instruments included requests from most of California's local air pollution control agencies. A fully equipped station needs cylinders of carbon monoxide (CO), methane (CH4), nitric oxide (NO), hydrogen sulfide (H2S), and sulfur dioxide (SO2). Federal regulations require that these gases be either Standard Reference Materials (SRM) or traceable to a SRM. SRM's are available from the National Institute of Standards and Technology (NIST). Since the cost of a single SRM is over $500, it is not practical to provide one for every air monitoring station. The cost problem was addressed by purchasing primary compressed gas standards that allowed the ARB to establish SRM traceability for the field gases. As a consequence, an increasing amount of staff time was spent establishing SRM traceable gases. In 1980, typically one person could assay eight single gas cylinders or up to five multiple gas cylinders in a ten hour work day. 1 Today, the automated gas analyses system can assay up to 23 gas cylinders, in typically 14 hours.
The ARB's automated gas assay system was developed to provide compressed gas secondary and tertiary standards traceable to primary standard SRMs in an expedient fashion at reasonable cost. Primary standards are standards that have the highest possible accuracy obtainable. SRMs sold by NIST are generally recognized as the highest level of accuracy. Secondary standards have been calibrated or assayed against primary standards. Finally, tertiary standards have been calibrated or assayed against secondary standards. 2
The term assay is used to describe "the measurement of a physical property in a material" 3. For example the determination of carbon monoxide concentration in a compressed gas cylinder is an assay. The term calibration is used to describe the "establishment of a relationship between a standard(s) and the response of a measurement system". 4 For example, challenging a CO analyzer with a SRM CO cylinder is a calibration. The term certification is used to describe the process of repeated assays to verify that a standard is reproducible and/or stable. 5
Some basic concepts and processes are common to both the manual and automated system. Standard air monitoring analyzers are used to assay the gases in both systems. In both systems the gas analyzers are calibrated, gases are assayed in lowest to highest concentration order, and a post assay calibration check is performed. Both the manual and automated system use mass flow controllers (mfc) to allow a wider range of gases to be assayed.
The ARB instrument calibration procedure consists of seven points: a zero or clean air point, four upscale concentration points, a post zero point, and post upscale point. A linear regression (best fit slope) is calculated using the seven calibration points. The slope of the response is required to be linear with a correlation coefficient of 0.9999 or greater. The zero or clean air gas can be supplied from either a clean air generator or a clean air compressed gas cylinder. The clean air generator is used when gases are diluted, while the clean air compressed gas cylinder is used when gas samples are assayed without dilution. The two types of zero gases are used to eliminate any bias one type of clean air might have on the analyzer. The five upscale points fall at approximately 90%, 65%, 40%, and 15% of full scale. (The 90% point is repeated.) The upscale points can be obtained by diluting a gas cylinder to the four levels or by using four gas cylinders fed to the analyzer without dilution. The choice to dilute is dependent on the availability of SRMs and the analyzer operating range.
The manual system, as it existed in 1980, required the operator to be present at all times. Each cylinder would have to be connected and disconnected individually. All mfc settings would have to be manually calculated and set. A typical day would start for the operator with calibrating the required analyzer(s). The various analyzers were connected to strip chart recorders. The operator was required to read the strip chart and calculate a linear regression for each analyzer. The guest cylinder concentration was derived by reading the analyzer response from the chart recorder and applying the regression equation. Gases of higher concentration than the analyzer range could be assayed by diluting the gas. A dilution factor would be calculated and applied to the analyzer response.
The current automated gas assay system has minimized the amount of direct operator involvement. A 386/20 personal computer (Everex Systems, Inc., Fremont, CA) has been programmed to control, monitor, and record the analyzer responses, route the gases, set the various required flows, and perform all required calculations. The operator collects the cylinders to be assayed and connects the cylinder to the automated system with a regulator and quick disconnect fittings. The cylinders are isolated from the analyzers with electric solenoids. The operator follows program prompts and enters into the computer the location of the gas cylinder, type of gas, concentration of gas, the cylinder number, cylinder pressure, and owner. This information will be used to dilute the gas if needed, route the gas to the proper analyzer, and keep a record of assays for the cylinder. After the operator has entered the required gas cylinder information, a program start time and the number of program runs for this batch of cylinders is selected. For example, the program can be set to start at 4:00 PM, three days in a row, which allows for unattended weekend assay runs.
The automated system consists of two separate subsystems--an ambient level and a source level system. The ambient system is designed to meet the needs of the ambient air monitoring programs. Analyzer ranges are fixed and blended gas cylinders are purchased with combinations that allow various gases to be analyzed at the same time. At the ARB, a typical blended gas cylinder consists of 5500 ppm CO, 2500 ppm CH4, 110 ppm NO, and 55 ppm SO2. This allows a field use for analyzers set at 50 ppm CO, 20 ppm CH4, 1.0 ppm NO, 0.5 ppm SO2. A blended gas cylinder is used in the field for nightly calibration checks and bi-weekly precision checks.
Twelve ambient level cylinders can be assayed during one run of the automated system. Five analyzers are used and can be calibrated at the same time using three different gas paths. CO and CH4 gas are blended into one cylinder allowing the supply of an upscale point to both the CO and CH4 analyzer. CO and CH4 are calibrated using four cylinders for the four upscale points. NO is calibrated using a single gas cylinder that is diluted to four concentration levels. The dilution is performed using a 200 cubic centimeter (cc) mfc for the gas flow and a 30 liter per minute (lpm) mfc for the dilution flow. H2S and SO2 are calibrated using two SO2 permeation tubes. The flow through the permeation tubes is fixed with a pressure regulator and a flow orifice. The calibration points are obtained with a 5 lpm mfc. H2S gas is assayed with a SO2 analyzer and a thermal oxidizer. The thermal oxidizer converts H2S to SO2.
Unlike the ambient system, the source system allows only one type of gas to be assayed per batch of cylinders. Eleven source level cylinders can be assayed during one run of the automated system. In general, the gas concentrations are higher, and conditioning or contamination among different types of gases becomes a larger concern. In addition to CO, CH4, NO, and SO2, carbon dioxide (CO2), oxygen (O2), and propane (C3H8) can be assayed by the source system.
The automated system performs error checks and quality assurance functions that previously had to be completed by the operator. For example, the slope of an analyzer calibration is required to meet a correlation coefficient of 0.9999 or better. Any calibration data that do not meet this criterion are automatically flagged and require additional review. In addition, the analyzer responses collected for a calibration point or a cylinder assay are checked for stability. The controller scans the analyzers and mfc outputs 2000 times at one minute intervals, determines the mean of the scan values and stores the data. The mean and standard deviation of the most recent nine stored values and the preceding nine values are calculated. Each nine-minute mean must have a standard deviation of less than 0.03 and the absolute difference between the two nine-minute means must be less than 0.015. These two criteria must be met before the point is accepted as a valid assay value and the controller assays the next gas. Assay responses that do not stabilize are flagged for additional review. For the ambient level system, the gas sample must run a minimum of 25 minutes before the most recent two nine-minute segments are compared. The maximum allowable stabilizing time is 75 minutes before the data is flagged for additional review and the controller assays the next gas. The source level gases, except NO and SO2, run a minimum of 21 minutes before the most recent two nine-minute segments are compared. The NO and SO2 samples run a minimum of 35 minutes before the two most recent nine-minute segments are compared. A maximum assay run time for NO and SO2 is 90 minutes and 75 minutes for the other gases. In general, the minimum assay time was determined empirically to be the minimum length of time required to ensure a reliable analyzer response. The maximum assay time limits were set to prevent an excessive use of gas.
The automated system will also generate certification values for the assayed gases. The certification requirements are that the standard deviation of the most recent three valid assay values have a relative standard deviation of less than 1.5 percent. The operator can review the gas assay history of a particular cylinder, mark the valid assay results, and request a certification report to be printed on a local printer.
The ranges of the 200 cc mfc and 30,000 cc mfc were selected to give a wide dilution range and avoid excessive use of gas. Using the two mfcs the automated system can dilute by a factor ranging from 20 to 1,200. The flows are selected to favor a minimum amount of residence time of the gas while en route to the dilution point. The 5 lpm mfc is used for gases not requiring dilution. To minimize contamination problems, the gas lines are purged with nitrogen after each cylinder assay is completed. The controller interfaces with the instruments through Digital(D)-to-Analog(A), A-to-D converter boards, and optically isolated relays (MetraByte Corp. Taunton, MA). Each analyzer is set up with a zero-to-ten volt output range. The converter boards have 4096 bits for the 10 volt scale giving a resolution of 2.5 mV. The D-to-A boards are used to set the gas and dilution flow levels of the mfcs. The A-to-D boards are used to scan the analyzers and mfcs. Optically isolated relays are used to switch solenoids that control the gas flow. The current automatic system program is written using a combination of Quick BASIC (Microsoft Corporation, Redmond, WA) and dBASE IV (Ashton-Tate Corporation, Torrance, CA). Quick BASIC is used to communicate with the interface boards and to control the progression of the assay run. The dBASE IV software is used to enter initial data, manage the final data and generate the assay calibration and cylinder analyses reports.
The automated system came on line in January 1983. The initial planning started in August 1980, with full-fledged development beginning April 1982. The first program was written in FORTRAN IV language using an Eclipse 120 computer (Data General Corporation, Westboro, MA). The cost of the first automated system exceeded $115,000 (1981 dollars). The following is a general cost break down of the major components: the computer was $43,000, a line printer was $10,500, D/A and A/D boards were $9,500, thirty-two solenoids were $1220, three mass flow controllers were $4,400, a pure air supply generator was $7,800, and five analyzers were $38,500. In 1987 a major renovation and upgrade of the system was started. The source level system was added which included six analyzers at $50,700, three mass flow controllers at $4,500, and 24 solenoids at $1150. The computer currently in operation, an Everex 386/20, was purchased in 1988 for $5,400. At the same time in 1988 the D/A, A/D, and solenoid control boards were upgraded for $4,400, a letter quality dot matrix printer was added for $600, and all mass flow controllers were replaced in 1989 for $8,000.
The automated system has allowed the operator to spend less time actually running cylinder assays. As a consequence, quality control procedures have been improved ensuring the integrity of the assay values. The number of program runs performed to achieve the three required valid assays for a certification has declined. The precision of the data has improved. By converting from strip chart recorder values to computer scanned values a tenfold gain in resolution was realized. The strip chart recorder, at best, provided a resolution of 25 mV. The automated system has a resolution of 2.5 mV. During 1992, 271 source level tertiary gases were certified, requiring 1028 assays. In addition, 221 ambient level tertiary gases were certified, requiring 1180 assays, and 48 secondary calibration standards were established.
The automated gas assay system was first developed at the California Air Resources Board by the late Kevin L. Kalthoff and Rudy Abangan with support from the section manager Bill Oslund and branch manager Don Crowe. The first software program was written by Mr. Kalthoff. With the guidance of Mr. Abangan a number of electrical engineering students aided in the building and design of the electrical infrastructure of the system.
1. R. Abangan, California Air Resources Board, Sacramento, personal communication, 1993.
2,3,4,5. K. L. Kalthoff, B. Reismann, Technical Report on Criteria Pollutant Standards Type, Hierarchy, Definitions, ADD-87-010, California Air Resources Board, Sacramento, April 1987, pp 2-4.
Return to the Standards Lab Page