Reconciling MICP Curves into 3D Static Models to Understand Sources of Uncertainty in Water Saturation Modelling
Static models used in the oil industry is the process of generating three-dimensional representation of reservoir characteristics for a better understanding of the available observations. Well log data is usually the most abundant source of data in terms of reservoir properties while core data and rock analysis are usually reduce to a minimum quantity. The process of modeling is accomplish with the constant will to make consistent representation of a mix of data observations involving different disciplines, scales and nature of observations. Thus, the three-dimensional environment and more specifically the static grid is the common platform to integrate geological understanding with geophysical and petro-physical and fluid-dynamic interpretations. While most of the log-derived reservoir properties such as lithology, porosity and permeability are easy to propagate in 3D space using geo-statistical methods, a very different situation occurs when estimating water saturation (SW) using static models. At the log scale, water saturation calculation starts by deciphering the water saturation fraction as a function of the following parameters: porosity, connate water resistivity and the resistivity measurements form wireline data. Calibration in this process is highly recommended, as core oil saturation (if available) has to be in ranges with those calculated form the electric logs. At this point is worth to mention that water saturation is a variable that requires many different inputs and/or assumptions predict actual saturation ranges in a given reservoir. Therefore, all the combined uncertainties per individual parameter are lumped into a combined uncertainty envelope at the log scale. When dealing with water saturation estimation in static grids, simple propagation of log values cannot be applied as fluid accommodation into the porous-media do not follow the same logic as reservoir properties (e.g. porosity, permeability, etc.). Consequently, integrated interpretation of: pressure gradient within the reservoir, fluid density and height about Free-Water-Level (FWL) becomes part of the equation and a more complex procedure has to be established. Given the fact that carbonate reservoirs are susceptible of strong alteration because of diagenesis, resulting in property heterogeneities of a high degree. If this is the case, discretization of the whole dataset results mandatory to define trends by independent rock classes or rock-types, the process of classifying rocks according to their similarities in reservoir properties is well known in reservoir characterization as rock-typing. Bearing in mind that water saturation modeling in heterogeneous carbonate reservoirs is a process that incorporate all the multiple topics described above, coupled with all the intrinsic uncertainties associated with data sampling, log measurement and tree-dimensional property distribution in static models. Then, it is highly expected that the output from the SW modeling process in static models is rather uncertain. More over the common practice in terms of quality check of the predicted model output is to compare it with a water saturation profile calculated from a resistivity log that is also carrying uncertainty as any other indirect measurement taken from the subsurface. Regardless all of these challenges, a reservoir characterization process should look for a realistic and practical solution in terms of quantification of water saturation. Business decisions of investments are commonly related to STOIIP ranges, so does the dynamic simulation models used to predict ultimate recovery factors and field development plans in carbonate reservoirs.
AAPG Datapages/Search and Discovery Article #90370 ©2020 AAPG Middle East Region Geoscience Technology Workshop, 3rd Edition Carbonate Reservoirs of the Middle East, Abu Dhabi, UAE, January 28-29, 2020