Orbital conducted a H2S comparison testing between its VE Technology system and a conventional system, to demonstrate the benefits of VE Technology. Orbital tested the client’s chromatograph and the pipeline simulator.
The results were expanded/extrapolated so as to represent a test gas in excess of the entry limits. 3.3 ppm is the legal limit and if readings are found to be above this, then National Grid will shut the pipeline down immediately. The conventional probe never actually reaches the limit, therefore, this gas would be very dangerous and nobody would even know.
This enables a clear demonstration of how the sampling system can effect/mask the analysis results. When changes in concentration of contaminants occur around entry condition levels the results are never actually seen by the analyzer.
Like links in a chain, every component between the source stream and the analyzer is required to deliver the sample intact and in time, in order to achieve an accurate and representative measurement. Any one component can break this chain and distort readings if not appropriately designed or selected, and if multiple components contribute sampling error then there can be no link between readings and reality.
The level of error and uncertainty generated by the sample system alone can far outweigh the uncertainty of measurement of the analytical device. In this report, a ~12% error and significant time delay was registered between the real and recorded measurement of H2S. However, this may be much more depending on equipment and process/ambient conditions. Furthermore, this value will vary over time, forever casting doubt over the validity of readings obtained.
The ramifications of an uncertainty of this magnitude, as found in a typical sampling system, are profound. Some key specifications and legislation rely on measurements that would be affected by the phenomena described in this report. Such measurements include: hydrogen sulfide and total sulfur content, other impurity measurements such as mercury content and water content, hydrocarbon dewpoint and water dewpoint, calorific value, Wobbe number, and other such gas quality parameters. As well as the possible regulatory issues, there are some potentially very serious financial and safety implications for companies depending on these measurements.
Time alignment of data, as required for effective process control and for flow computers, also becomes a near impossibility due to the magnitude of delay introduced by the sampling system, and the unknown variance thereof in operation. There are many sources of delay (in responding to change) introduced by the sampling system components, with uncertain relative error contributions in any particular set of conditions (pressure, temperature, velocity, composition, ambient temperature, age of sampling equipment, and others). This has a huge impact on operational efficiency and ultimately revenue. In some cases, misaligned data can be more dangerous and costly than no data.
Effective sampling should be of interest and concern to users as sampling error may not been seen, may not be consistent and cannot be calculated in order to adjust analytical results.