Halliburton Energy Services, Inc.

Baker Hughes Corporation

Wei Shao

Songhua Chen

Sheng Fang

Daniel Lee Miller

Magdalena Sandor

Sonali Pattnaik

Mike Gillen

Ronnel Charles Balliet

Adly Helba

Yusheng Cheng

MingJen Wang

Mette Stengaard Munkholm

Huiwen Sheng

System and methods of petrophysical modeling are disclosed. Measurements of formation parameters are received from one or more measurement tools during a first stage of a downhole operation within a reservoir formation. A correlation between each of the formation parameters and a target parameter of the formation is determined based on the measurements. One or more formation parameters are selected as input parameters for a symbolic regression model, based on the correlation. A symbolic regression model is trained to generate candidate formation models representing the target parameter, based on the selected input parameters. One or more optimizations are applied to the candidate models to determine a target petrophysical model of the formation. Values of the target parameter are estimated for at least one formation layer, based on the target petrophysical model. A second stage of the downhole operation is performed within the formation layer(s) based on the estimated values.

Data driven approach to develop petrophysical interpretation models for complex reservoirs

System and methods for nuclear magnetic resonance (NMR) fluid substitution are provided. NMR logging measurements of a reservoir rock formation are acquired. Fluid zones within the reservoir rock formation are identified based on the acquired measurements. The fluid zones include water zones comprising water-saturated rock and at least one oil zone comprising rock saturated with multiphase fluids. Water zones having petrophysical characteristics matching those of the oil zone(s) within the formation are selected. NMR responses to multiphase fluids resulting from a displacement of water by hydrocarbon in the selected water zones are simulated. A synthetic dataset including NMR T2 distributions of multiphase fluids is generated based on the simulation. The synthetic dataset is used to train a machine learning (ML) model to substitute NMR T2 distributions of multiphase fluids with those of water. The trained ML model is applied to the NMR logging measurements acquired for the oil zone(s).

Nuclear magnetic resonance (NMR) fluid substitution using machine learning

A method includes generating a temperature-corrected nuclear magnetic resonance (NMR) measurement-derived value corresponding to a target temperature using a correlation model that is based on a difference between the target temperature and a sample temperature. The method also includes determining a formation property based on the temperature-corrected NMR measurement-derived value corresponding to the target temperature.

Temperature correction of NMR relaxation time distributions

A method for estimating a property of an earth formation is provided. At least one porosity dataset is received. The porosity dataset is analysed to determine pore-throat size distribution. A NMR measurements dataset is received from a NMR tool. The dataset is analyzed to determine a distribution of a first relaxation time versus a second relaxation time. Distribution of a ratio of the first relaxation time to the second relaxation time is determined by calculating a ratio between a geometric mean of the first relaxation and the second relaxation time. A threshold for the ratio of the first relaxation time to the second relaxation time is selected. The threshold indicates a value of the ratio corresponding to pore sizes not contributing to property of the formation. The property of the formation is determined using the determined threshold for the ratio of the first relaxation time to the second relaxation time.

Reservoir formation characterization from NMR T1/T2 ratio

Systems and methods of the present disclosure relate to processing nuclear magnetic resonance (NMR) data. An NMR downhole tool may include a housing, magnets disposed within the housing, an antenna extending along a circumference of the housing, and an information handling system configured to receive NMR signals via the antenna, wherein the NMR signals are based on an operating frequency, a static magnetic field B, and a radio frequency (RF) field B, which are defined by the antenna and the magnets. The information handling system may be further configured to project the NMR signals to a forward modeling space, and transmit vectors resulting from projecting the NMR signals to the forward modeling space.

Processing nuclear magnetic resonance signals in a downhole environment

An NMR logging system is disclosed which continues logging without interruption despite switching activation sets to adapt to changes in formation properties. Based on detection of an approaching or encountered geological boundary, an appropriate activation set is transmitted to the downhole NMR tool while the NMR tool continues logging. This system optimizes NMR data collection for each formation and associated formation fluid properties while reducing the need to stop the tool string movement while switching the activation set, and reduces incomplete collection of NMR and non-NMR logging tool data.

NMR data acquisition while switching NMR activation sets

Disclosed herein are methods and systems that determine a microfacies or a microfacies characteristic of a sample of a subterranean formation based on a segmented image of a petrographic image of the sample, wherein the segmented image is derived from the petrographic image using a machine-learning algorithm.

Automating microfacies analysis of petrographic images

An apparatus comprising a data acquisition tool including NMR sensors, a data acquisition processor communicatively coupled with the NMR sensors, and a first memory storing instructions that cause the data acquisition processor to perform operations comprising acquiring data of earth formation fluid, varying at least one of a magnetic field gradient and an inter-echo time, and acquiring additional data. The apparatus further comprises a data processing unit comprising a second memory storing instructions that cause the data processor to perform operations comprising receiving data acquired by the data acquisition tool, constructing a mathematical model of the data, conducting a first inversion of the mathematical model to obtain a first set of NMR responses, performing a forward model of the first set of NMR responses obtained from the first inversion, conducting a second inversion to obtain a second set of NMR responses, and determining earth formation fluid properties.

Apparatus and method for determining earth fluid formation

A method includes generating a temperature-corrected nuclear magnetic resonance (NMR) measurement-derived value corresponding to a target temperature using at least one of a dimension-reduction operation or a parameter-correlation operation based on a difference between the target temperature and a sample temperature. The method also includes determining a formation property based on the temperature-corrected NMR measurement-derived value corresponding to the target temperature.

A formation evaluation system reduces inversion matrixes used to determine formation properties, thereby increasing the memory management and processing efficiency of the evaluation system. NMR data is acquired from a wellbore and expressed mathematically by the system as a least squares solution to a linear system. The least squares solution is approximated using a numerical decomposition method and the evaluation system determines a formation property using the approximated least squares solution. Thereafter, a downhole operation may be planned, analyzed or conducted using the determined formation property.

Downhole formation evaluation using nuclear magnetic resonance inversion matrix reduction

Systems, methods, and software for predicting total organic carbon (TOC) values are described. A representative method includes obtaining nuclear magnetic resonance (NMR) data and training a radial basis function (RBF) model based on the NMR data and measured total organic carbon (TOC) values. The method also includes obtaining subsequent NMR data and employing the trained RBF model to predict TOC values based at least in part on the subsequent NMR data. The method also includes storing or displaying the predicted TOC values.

Predicting total organic carbon (TOC) using a radial basis function (RBF) model and nuclear magnetic resonance (NMR) data

A method for determining earth formation properties including a data acquisition tool and a data acquisition processor coupled with NMR sensors and a first memory; transmitting earth formation fluid data to a data processing unit comprising a processor and a second memory. Obtaining a fully polarized state echo train (E) and a partially polarized state echo train burst (E); inverting the Eto obtain an apparent transverse relaxation time (T) distribution; truncating the Tdistribution by discarding the partially polarized state echo train data; completing a forward model of the Eto obtain an additional echo train burst (E), performing a second inversion of the data set; and determining earth formation fluid properties.

Apparatus and method for obtaining T2 distribution

Systems, methods, and software for predicting a pore throat size distribution are described. A representative method includes obtaining a nuclear magnetic resonance (NMR) relaxation-time distribution data set. The method also includes training a radial basis function (RBF) model based on the NMR relaxation-time distribution data set and a measured pore throat size distribution data set. The method also includes obtaining a subsequent NMR relaxation-time distribution data set. The method also includes employing the trained RBF model to predict a pore throat size distribution data set based at least in part on the subsequent NMR relaxation-time distribution data set. The method also includes storing or displaying a predicted pore throat size distribution corresponding to the predicted pore throat size distribution data set. The predicted pore throat size distribution is associated with a rock sample or subsurface formation volume.

Methods and systems employing NMR-based prediction of pore throat size distributions

Systems, methods, and software for modeling subterranean formation permeability are described. In some aspects, a method of training a subterranean formation permeability model based on NMR data includes accessing relaxation-time distributions generated from NMR measurements associated with a subterranean region. Multiple sets of principal components are generated from the relaxation-time distributions. Each set of principal components represents a respective one of the relaxation-time distributions. Parameters for weighted radial basis functions are computed based on the sets of principal components. A subterranean formation permeability model that includes the weighted radial basis functions and the computed parameters is produced.

Modeling subterranean formation permeability

Systems, methods, and software for modeling subterranean formation viscosity are described. In some aspects, a method of training a subterranean fluid viscosity model based on NMR data includes accessing multiple relaxation-time distributions generated from NMR measurements of a fluid, normalizing each relaxation-time distribution to a common normalizing value, computing parameters for a plurality of weighted radial basis functions from the normalized relaxation-time distributions, and producing a subterranean fluid viscosity model that includes the weighted radial basis functions and the computed parameters.

Growing community of inventors

Conroe, TX, United States of America

Wei Shao