Carbon dioxide (CO2) is an acidic gas that causes significant production problems, thereby increasing the cost of operations and the cost of capital expenditures, especially for completion and surface equipment, and for flow lines. The presence of carbon dioxide may change the economic viability of an asset, even in low concentrations. Gas phase concentrations greater than 2 bar partial pressure are already considered highly corrosive, which may correspond to a liquid phase concentration as low as between 0.4 to 2.4 wt% for most oils. Corrosion alone costs the oil and gas industry USD 1.4 billion per year in 2013, with approximately 60% of failures directly related to carbon dioxide. Furthermore, lost production related to corrosion failures further costs the oil and gas industry tens of billions of dollars in lost revenue annually.
Multivariate optical computing (MOC) is an optical analysis technique that has been shown to match the sensitivity and accuracy of a Fourier transform infrared spectrometer using a partial least squares regression. MOC performs an analogue dot product regression in the optical domain, which uniquely suits MOC to enable high resolution mid-infrared spectroscopy at high temperature. This is particularly important for downhole carbon dioxide measurements. The near-infrared (1952 to 2080 nm) region is far less sensitive for carbon dioxide measurement than the mid-infrared region (MIR) (2686 to 2835 nm), with an intensity ratio of 43 between the bands of the two regions. Therefore, a new MOC sensor has been developed to access the MIR for highly accurate carbon dioxide measurements.
Although carbon dioxide-rich fluids may be sampled and shipped to a laboratory for analysis, choosing a location to sample can be problematic. With varying degrees of compositional grading in reservoirs, the carbon dioxide concentration will likely not be equivalent at all locations. Consequently, a short trial pumpout with extrapolation to reservoir concentration of carbon dioxide would be desirable to determine the relevant locations to sample. A new technique has been developed to accomplish this goal using the MOC sensor. In this study, a first observation of caustic drilling fluid filtrate suppressing carbon dioxide signature throughout the pumpout will also be shown. This suppression effect is reduced as the amount of filtrate present decreases during the pumpout cleaning period, but not at the same rate as the filtrate dilution effect. Therefore, a new technique has been developed to correct this effect in real time to deliver not only the instantaneous carbon dioxide concentration, but also the true reservoir concentration. Furthermore, the accuracy of the carbon dioxide measurements made in five wells, with nine formation tester pumpout stations, has been determined as +/-0.4 wt% over a concentration ranging from 1.5 to 23 wt%.
