09 April 2001

Repeatability of online natural gas analysis

by Teresa Lechner-Fish, Yang Xu

New instruments save money with improved design.

A new gas chromatograph design can return on the order of $14,000 savings per year by combining the valve and detector systems. Further, the design improves measurement repeatability by nearly an order of magnitude, boosts the durability of the valves to as much as 38 years, and significantly reduces maintenance requirements.

Engineers must design online gas chromatographs (GCs) for continuous operation in remote locations, with negligible maintenance. Most commercial GCs for natural gas analysis are advertised as repeating ±0.5-1 British thermal units (Btus) per 1,000 readings (±0.05% to 0.1%), which is usually verified by tests of a single unit over a single midrange (i.e., room) temperature; the specification typically improves to ±0.25 Btu per 1,000 readings if the GC is housed in a shelter and calibrated weekly.

We believed we could obtain better performance from a GC and initiated a multiyear development project with the following objectives:

Improve repeatability to ±0.1 Btu per 1,000 over the full temperature range
Increase the GC valve and solenoids durability to 10 million cycles
Reduce carrier consumption by at least 40% without requiring air actuation
Enable parallel chromatography for expanded applications
Establish application modules for field upgrades
Simplify valve maintenance

We have designed a GC system that meets or exceeds these requirements. To validate our improvements to the design, we evaluated the system performance using a prototype Danalyzer Pro GC with a C6+ (single detector) application installed. During a 24-hour period, the system repeated within ±0.02 Btu per 1,000 at room temperature. The testing simulated rapidly changing weather conditions by dynamically evaluating the GC over an ambient temperature range of 0° to 130°F (-18° to +55°C). In a 24-hour period, the instrument repeated within ±0.12 Btu per 1,000. The system repeated within ±0.03 Btu per 1,000 when subjected to the maximum temperature for an extended period and within ±0.09 Btu per 1,000 when subjected to the minimum temperature for an extended period. An order of magnitude improvement in repeatability is especially valuable for custody transfer applications, where a measurement error as small as 0.05% (0.5 Btu/standard cubic feet, or scf) results in more than $14,000 per year of lost and unaccounted energy (20 million scf per day, 1,030 Btu/scf, $4 per million Btu).

Design considerations

To obtain valid measurements, the online GC must be compatible with the physical state of the sample. In particular, the sample must be transported as a single phase from the sample point to the sample loop. Accidental conversion from one phase to another can seriously alter the sample composition and the resulting analytical data. The dew point, the temperature at which a vapor mixture begins to condense at a given pressure, goes down with decreases in pressure. A vapor sample must be at least 10°C (50°F) above its dew point to prevent partial condensation.

A stream-switching system selects a single stream from multiple streamlines and introduces it into the GC system. Typically, multiple individual valves assume a double block-and-bleed structure. Because the configurations are usually quite large, they can be difficult to heat uniformly, which can result in partial condensation. In addition, they introduce significant dead volume, which can result in cross contamination. For reliable analyses, it's essential that the current stream doesn't carry over into the next stream.

To minimize these problems, the design (patents pending) has the heating flexibility to prevent sample condensation and greatly reduce system dead volume. The heated zone is composed of a 20-watt cartridge heater, located in an aluminum block mounted on the valve system, and an insulation cover. The insulation cover is made of Rogers Poron BF-1,000 silicon material covered with a steel sheet.

The stream-switching system prevents cross contamination with a double-block and double-bleed configuration. Each stream has two flow blocks and two bleed paths for system purging. The stream-switching valves and the sample shutoff valves occupy a single-block structure, which reduces the system dead volume. The stream-switching system is approximately 7 inches long, 1 inch wide, and 2 inches high.

Low-wattage electrical solenoids drive the stream-switching valves. In the event of a power failure, the stream-switching system and sample shutoff valves will remain closed to provide a fail-safe system. Because the ratio of actuating pressure to the sample gas pressure is better than 1:3, the system will withstand up to 240 pounds per square inch (psi) sample pressure when 80 psi actuating gas is used. The standard system has four streams, including calibration and two sample shutoff valves for two-channel analysis. The system can handle up to five streams, including calibration, if the double-block configuration isn't required.

Multivalve system

The valve system must be easy to maintain because upset conditions can contaminate an online GC system. With a traditional GC system, an hour or more may be required to replace the diaphragms of individual valves. To minimize this problem, the new design integrates five valves into a single block structure. Replacing the diaphragms of the five valves in the multivalve system as a single structure takes less than 30 minutes. The screw patterns of the multivalve system allow removal of the upper sample/carrier flow portion while the lower, actuating portion remains intact.

Another problem with traditional GC systems is the large system dead volume, due to the interconnecting tubing between components (i.e., valves and detectors). Two thermal conductivity detectors in the block structure minimize the problem. The multivalve system is approximately 5 inches in diameter and 3 inches in height. This compact design reduces the system dead volume by approximately 60%. To allow "drop-in and tighten" installation, the tubing and columns are preshaped and assembled into kits; the tubing kit can be replaced in 15 minutes, and the column kit can be replaced in less than an hour.

Many valve designs suffer from a relatively high-pressure drop across the valve. In the multivalve system, the internal valve structure (i.e., flow path, pistons) is designed to maximize mechanical advantage and minimize the pressure drop (i.e., flow restriction). As a result, less than 5 psi, gauge, pressure drives flow through the valve. To improve the internal sealing of the valve, the actuating side piston area is much larger than the sample side. Because the ratio of actuating pressure to sample gas pressure is more than 1:3, the system will withstand up to 240 psi sample pressure when 80 psi actuating gas is used.

Microheat sink oven

Integration of the valve and the detector systems improves the temperature stability of the system, due to the thermal mass of the valve system. In addition, the valve system provides a very stable environment for the column module. To allow flexibility, the valve system heats independent of the column module. The first heated zone is composed of a 31-watt band heater around the valve system, insulation cover, and plastic manifold. The insulation cover is made of Thermal Ceramics' Btu block material covered with steel sheet. The Btu material minimizes heat leakage out of the system.

The second heated zone, the column module, is located in the center of the multivalve system block. To ensure uniform heating, the columns wind around an aluminum spool with a 15-watt cartridge heater in the center. A nylon-GP insulation cup surrounding the column spool is thermally isolated from the valve and detector block with an air gap. With this arrangement, the temperature differential between the two zones can range up to 15°C. To increase the accuracy of the temperature control, the 100-ohm platinum remote terminal units are located near the control points. Using proportional, integral, derivative temperature controllers, the valve system, including the detectors and the column module, are controlled within ±0.05°C at room temperature and ±0.2°C over an ambient temperature range of 0° to 130°F (-18 to +55°C).

Typically, a solenoid located outside the temperature-controlled oven, due to the temperature limitation of the solenoids, drives a valve. When used as actuation gas, the large dead volume of tubing used to connect the valve to the solenoid increases the carrier consumption. To minimize this problem, a plastic manifold interfaces the electrical solenoids to the actuation portion of the valve. The manifold is made of Commercial Plastics & Supply Ultem 1000 polyetherimide. It functions as a thermal insulation layer as well as the gas connection path in the system. The low dead volume has reduced carrier consumption nearly 40%.

The multivalve system can easily be adapted for a wide variety of chromatography applications. For very simple applications, the multivalve system uses three valves with a single detector; for more complex applications, it uses five valves and two detectors. A technician can replace the application module in less than an hour. In addition, the GC system can be field upgraded with a new chromatography application by replacing the application module and downloading a new software application.

Natural gas applications

Industry analyzes natural gas to determine physical properties, to identify the source, and to measure gas quality. Analyzers typically measure a composite of the heavier hydrocarbons for calculation of calorific value, relative density, and compressibility. The heavier components should be further characterized, however, to calculate phase properties. Small variations in the composition, especially in the heavier hydrocarbons, can result in large variations of hydrocarbon dew point.

The designers tested the GC system for natural gas (C6+ application) and rich gas (C9+ application) analyses, heating the valve/detector to 80°C and the stream-switching system to 85°C. Heating the sample cylinder and associated transport tubing to 50°C prevented condensation. For the C6+ application, a single column train was installed in the system and used to analyze nitrogen through C6 and heavier. For the C9+ application, two column trains were installed. The first detector analyzed nitrogen through pentane, and the second was used for hexane through C9 and heavier.

The natural gas composition used for the C6+ application was a vapor calibration blend of nitrogen, carbon dioxide, methane, ethane, propane, isobutane, butane, neopentane (2,2-dimethyl propane), isopentane, pentane, isohexane (2,2-dimethyl butane), and hexane. The sample size (250 microliters) minimized methane tailing from column overload while retaining the required precision. A 14-inch porous polymer column separated C6s for back flush to the detector, which measured the isohexane and heavier components as a C6+ composite.

The system transferred the lighter components from the back-flush column to a 4-foot boiling point column coupled to a 3-foot porous polymer column for further separation. The critical separation for the coupled column was the ethane/propane separation for the series bypass valve switch. The thermal conductivity detector then quantified the propane, isobutane, butane, neopentane, isopentane, and pentane. The third column, a 7-foot porous polymer column, separated nitrogen, methane, carbon dioxide, and ethane. Finally, 50 centimeters of 0.01-inch ID capillary tubing dampened pressure pulses at the thermal conductivity detector from valve switches.

The lighter components, nitrogen through pentane, were measured as described above. The second detector was used to further characterize the heavier components (C6+). The natural gas composition used for the C9+ application was a vapor calibration blend of nitrogen, carbon dioxide, methane, ethane, propane, isobutane, butane, isopentane, pentane, hexane, heptane, octane, and nonane. The sample size (745 microliters) optimized detection of octane.

A 12-inch boiling point column separated C9s for back flush to the detector, which measured the C9 and heavier components as a C9+ composite. The system transferred the lighter components from the back-flush column to a 12-foot boiling point column to separate the C6, C7, and C8 components. As before, 50 centimeters of 0.01-inch ID capillary tubing dampened pulses at the detector.

Long-term performance

In a related experiment, we evaluated the long-term performance of the system (valves, solenoids, and columns) using an earlier mechanical prototype, and it repeated within 0.13 Btu per 1,000 before long-term testing was completed. After 15 million cycles (equivalent to 38 years for this application), the system still repeated within ±0.13 Btu per 1,000 without any deterioration in performance.

Researchers evaluated the stream-switching system by alternating between a stream of C6+ natural gas and a stream of helium. Using a sample purge time of 120 seconds, the system repeated within ±0.05 Btu per 1,000 for the natural gas stream and measured 0.0 Btu for the helium stream throughout a 24-hour period. While switching between a stream of C9+ rich gas and a stream of helium, the system repeated within ±0.15 Btu per 1,000 for the natural gas stream and measured 0 Btu for the helium stream throughout a 24-hour period. IT