We report here on a new membrane-based electrochemical sensor (MBES) that may provide an important utility in monitoring and characterizing internal corrosion of natural gas pipelines. Using this sensor, we have measured the corrosion rate of X65 steel exposed to H2S in humidified environments up to 60 °C. Consistent with our earlier CO2 study, the membrane’s conductivity did not change when exposed to H2S-contaning acidic gas. Introducing H2S consistently increased the measured corrosion rate between testing conditions, though corrosion rates were typically less than 2 μm y-1. At 30 °C, the corrosion rate doubled from 7.3 to 14 nm y-1 below a relative humidity of 30 %, and increased by an order of magnitude (0.19 μm y-1 to 1.9 μm y-1) at 55 % relative humidity, showing that the influence of H2S on corrosion increases dramatically with larger humidity. Trends with relative humidity match industry expectations: corrosion rate is low (<0.25 μm y-1) without the presence of a condensed aqueous phase, but increases as the water content of the system increases. The MBES was therefore able to captures relevant corrosion trends, even while the corrosion rates would not have presented a serious threat to any natural gas pipeline. As such, the MBES can be used to detect the onset of emerging corrosion threats before they occur. Field emission scanning electron microscopy and energy-dispersive X-ray spectroscopy confirmed that H2S reacted with the metal covered by the membrane phase, showing evidence of sulfur-rich sites on the X65 surface. In addition, finite element analysis confirmed that electrochemical measurements and data analysis techniques could be successfully used for this membrane-based sensor, despite its unconventional cell geometry.
Improvement of the machine learning-based corrosion rate prediction model through the optimization of input features
