Laboratory analysis of formation-tester samples provides critical information for exploration and production activities. The physical properties, chemical properties, and composition of samples are used to confirm and resolve reservoir architecture questions, including compartmentalization and compositional grading, which is then used to design completion and production strategies. The laboratory data are used to calibrate equation-of-state models employed in reservoir simulations to project the lifetime recovery of an asset under different scenarios. This yields both an optimal production strategy and reduced capital expenditure costs for a project. Furthermore, the data are used to identify flow assurance problems and to estimate the operational costs of production. Lastly, the data provide crude value, as required, to book reserves. However, these benefits are predicated on the requirement that samples be representative of the actual formation fluid. In this aspect, contamination of samples with near-wellbore drilling fluid filtrate remains the most common reason that samples are not fit-for-purpose. Therefore, contamination estimation must be improved.
The ubiquitous real-time downhole contamination estimation procedure for filtrate in petroleum uses pumpout trend fitting. The trend fitting attempts to regress a model to a change in the formation tester sensor data during a pumpout in which fluid grades exist from filtrate to formation fluid. Sensor responses of the pure filtrate or formation fluid endmembers are estimated and the contamination calculated. When contamination is deemed sufficiently low, the sample is captured and shipped to a surface laboratory. Because sampling is often the last activity before cementing a section of well, there is not a second sampling opportunity if the laboratory determines the sample is not fit-for-purpose. Too often, the contamination estimate does not match the laboratory due to the breakdown of three key trend-fitting assumptions:
that the model is sufficient to describe reservoir complexity;
that the model can be extrapolated to determine endmember responses; and
that the asymptote of the pumpout is not falsely representing steady-state contamination.
Improvement in current methods of contamination estimation are nearly all driven by improving trend fitting to existing sensor data as opposed to the development of new sensors. This work, however, describes the development of a new direct contamination sensor, designed to detect synthetic drilling fluid using optical measurements. Nearly all synthetic drilling fluids contain additives such as olefins, esters, ketones, alcohols, and amines which are not naturally present in geologic formations. These compounds look like other petroleum hydrocarbons in the conventional visible and near-infrared optical ranges of existing wireline tools. However, in the midinfrared optical range, the signature of olefins and other synthetic filtrate additives is distinct, which has allowed for the construction of a synthetic drilling fluid (SDF) filtrate-specific detector that can be used to directly determine the contamination level of drilling fluid filtrate without the limitations of trend-fitting assumptions. A comprehensive study of eight pumpout stations from five wells has validated the performance to greater than +/− 2.5 wt%, consistently delivering results superior to those obtained via conventional trend-fitting methods.