It’s a gas

If you’re using natural gas to cook your dinner, heat the water in your boiler or even power your car, then you want that natural gas to contain a high concentration of methane, usually above 85% mol/mol. This is because methane is the main provider of energy in natural gas, which can also contain higher alkanes such as butane and propane, carbon dioxide and nitrogen.

Up to now, gas chromatography (GC) has been the standard method for determining methane levels in natural gas, but it’s time consuming, expensive and requires bulky equipment, preventing it from being deployed out in the field. Now, however, Brazilian chemists have shown that NIR spectroscopy can make a perfectly acceptable alternative, able to distinguish natural gas containing methane above 85% from natural gas containing methane below 85%. Even better, it can still do this when the methane concentration is only just below 85%.

Led by Mário Araújo at the Federal University of Paraíba, the chemists took 55 samples of natural gas from vehicle fuelling stations in north-eastern Brazil. In addition, using a specially-developed gas mixing system, they mixed methane with different concentrations of nitrogen, compressed air, ethane and lighter fluid, which is a mixture of propane and butane. As reported in the Microchemical Journal, this allowed them to produce 23 analogues of natural gas in which the concentration of methane varied between 30% and 85%.

Using GC, they determined the methane concentrations of the 55 natural gas samples from vehicle fuelling stations, finding that 45 had methane concentrations above 85% while 10 had methane concentrations slightly below 85%. Using NIR spectroscopy, they then analysed 35 of the natural gas samples with methane concentrations above 85% and used the multivariate statistical technique known as soft independent modelling of class analogies to produce a predictive model from the spectral data.

When they analysed the other natural gas samples with NIR spectroscopy, they found this model could correctly distinguish the 10 samples with methane concentrations above 85% from the 33 samples with methane concentrations below 85% based on the spectral data. Some of these samples, especially those from the vehicle fuelling station, had methane concentrations of between 75% and 85%, showing that this technique is sensitive enough to distinguish samples with concentrations just below 85%.