Reliable assessment of petrophysical, compositional, and mechanical properties is critical and yet challenging for formation characterization of organic-rich mudrocks. The formation evaluation results are the key to build an accurate geological model and are the starting point for reservoir characterization. This paper aims to (a) develop a method to integrate an iterative formation evaluation workflow with geological modeling and (b) improve the reliability of field-scale geological modeling by reducing the average relative error and uncertainty in well-log-based estimates of petrophysical and geochemical properties using the integrated workflow.
The first step includes joint inversion of well logs to obtain depth-by-depth estimates of formation properties such as porosity, fluid saturations, total organic content (TOC), and volumetric concentrations of minerals, which are inputs to the well-log-based rock classification algorithm. Then the model parameters are updated in each rock-class and the multi-mineral analysis results are cross validated with core measurements in wells where core measurements are available. The iterative procedure is repeated until agreement with core data is achieved. We use the geostatistical analysis to extend the workflow obtained in the cored wells to the neighboring non-cored wells where the developed model in each rock class is reliable. Finally, we use the well-log-based estimates of the petrophysical and geochemical properties as an input to the geological model to estimate reservoir properties such as original-hydrocarbon-in-place.
We successfully applied the method to more than 300 wells in the Midland Basin. Results showed that the average relative error in well-log-based estimates of porosity and water saturations, improved by 22% and 35%, respectively, compared to a conventional non-iterative, non-integrated method which results in 76% improvement in the calculated hydrocarbon-in-place. A sensitivity analysis was performed to evaluate the impact of optimizing and calibrating the model parameters throughout the basin to reduce average relative error and uncertainty in well-log-based estimates of porosity and water saturation, as key petrophysical inputs to the geological model. Results indicated that applying a South Midland Basin TOC model to a North Midland Basin well, causes 56% and 16% increase in average relative error in estimates of water saturation and porosity, respectively, which results in 47% increase in average relative error of original-oil-in-place calculations.
Coupling the integrated formation evaluation workflow, geostatistical analysis, and geological modeling is a novel approach that not only incorporates formation heterogeneity and spatial variations of reservoir properties, but also yields dependable reservoir characterization by quantifying the uncertainty associated with hydrocarbon-in-place estimation. This method enables a reliable field-scale formation characterization in the Midland Basin which is critical for field development planning in organic-rich mudrocks.