No abstract

No abstract

Formation water saturation is a critical target property for any comprehensive well log analysis program. Most techniques for computing saturation depend heavily on an analyst’s ability to accurately model resistivity measurements for the effects of formation water resistivity and rock texture. However, the pre-requisite knowledge of formation water properties, particularly salinity, is often either unknown, varying with depth or lateral extent, or is difficult to derive from traditional methods. A high degree of variability may be present due to fluid migration from production, water injection, or various geological mechanisms. In unconventional reservoirs, the complexity of the rocks and pore structure further complicates traditional interpretation of the available well logs. These factors introduce significant uncertainties in the computed fluid saturations and therefore can substantially affect final reserves estimates. A novel technique in geochemical spectroscopy has recently been introduced to distinguish the chlorine signals of the formation and borehole. The new, quantitative measurement of formation chlorine enables a direct calculation of bulk water volume for a given formation water salinity. When integrated into a multi-physics log analysis workflow, the chlorine-derived water volume can provide critical information on fluid saturations, hydrocarbon-in-place, and producibility indicators. This additional information is especially useful for characterizing challenging and complex unconventional reservoirs. We present the new technique through several full petrophysical evaluation case studies in organic shale formations across the U.S., including the Midland, Delaware, Marcellus, and DJ basins. We solve for formation-specific water salinity and bulk water volume through an optimization that combines chlorine concentration with resistivity and dielectric measurements. These outputs are integrated into comprehensive petrophysical evaluations, leveraging a suite of advanced well log measurements to compute final fluid and rock properties and volumetrics. The evaluations include geochemical mineralogy logs, 2D NMR analyses, dielectric dispersion analyses, basic log measurements, and multi-mineral models. The results underscore the utility of the new spectroscopy chlorine log to reduce petrophysical model uncertainties in an integrated workflow. While this workflow has been demonstrated here in several U.S. organic shale case studies, the fundamental challenges it addresses will make it a valuable solution for a range of unconventional reservoirs globally.

This paper describes an innovative machine learning application, based on variational autoencoder frameworks, to quantify the concentrations and associated uncertainties of common minerals in sedimentary formations using the measurement of atomic element concentrations from geochemical spectroscopy logs as inputs. The algorithm comprises an input(s), encoder, decoder, output(s), and a novel cost function to optimize the model coefficients during training. The input to the algorithm is a set of dry-weight concentrations of atomic elements with their associated uncertainty. The first output is a set of dry-weight fractions of fourteen minerals, and the second output is a set of reconstructed dry-weight concentrations of the original elements. Both sets of outputs include estimates of uncertainty on their predictions. The encoder and decoder are multilayer feed-forward artificial neural networks (ANN), with their coefficients (weights) optimized during calibration (training). The cost function simultaneously minimizes error (the accuracy metric) and variance (the precision or robustness metric) on the mineral and reconstructed elemental outputs. Training of the weights is done using a set of several-thousand core samples with independent, high-fidelity elemental and mineral (quartz, potassium-feldspar, plagioclase-feldspar, illite, smectite, kaolinite, chlorite, mica, calcite, dolomite, ankerite, siderite, pyrite, and anhydrite) data. The algorithm provides notable advantages over existing methods to estimate formation lithology or mineralogy relying on simple linear, empirical, or nearest-neighbor functions. The ANN numerically capture the multi-dimensional and nonlinear geochemical relationship (mapping) between elements and minerals that is insufficiently described by prior methods. Training is iterative via backpropagation and samples from Gaussian distributions on each of the elemental inputs, rather than single values, for every sample at each iteration (epoch). These Gaussian distributions are chosen to specifically represent the unique statistical uncertainty of the dry-weight elements in the logging measurements. Sampling from Gaussian distributions during training reduces the potential for overfitting, provides robustness for log interpretations, and further enables a calibrated estimate of uncertainty on the mineral and reconstructed elemental outputs, all of which are lacking in prior methods. The framework of the algorithm is purposefully generalizable that it can be adapted across geochemical spectroscopy tools. The algorithm reasonably approximates a ‘global-average’ model that requires neither different calibrations nor expert parameterization or intervention for interpreting common oilfield sedimentary formations, although the framework is again purposefully generalizable so it can be optimized for local environments where desirable. The paper showcases field application of the method for estimating mineral type and abundance in oilfield formations from wellbore logging measurements.

No abstract