scholarly journals Machine learning application to predict in-situ stresses from logging data

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Ahmed Farid Ibrahim ◽  
Ahmed Gowida ◽  
Abdulwahab Ali ◽  
Salaheldin Elkatatny
Keyword(s):  
Machine Learning ◽  
Horizontal Stress ◽  
Validation Dataset ◽  
Percentage Error ◽  
Well Log ◽  
Inference System ◽  
Log Data ◽  
In Situ Stresses ◽  
Overburden Stress

AbstractDetermination of in-situ stresses is essential for subsurface planning and modeling, such as horizontal well planning and hydraulic fracture design. In-situ stresses consist of overburden stress (σv), minimum (σh), and maximum (σH) horizontal stresses. The σh and σH are difficult to determine, whereas the overburden stress can be determined directly from the density logs. The σh and σH can be estimated either from borehole injection tests or theoretical finite elements methods. However, these methods are complex, expensive, or need unavailable tectonic stress data. This study aims to apply different machine learning (ML) techniques, specifically, random forest (RF), functional network (FN), and adaptive neuro-fuzzy inference system (ANFIS), to predict the σh and σH using well-log data. The logging data includes gamma-ray (GR) log, formation bulk density (RHOB) log, compressional (DTC), and shear (DTS) wave transit-time log. A dataset of 2307 points from two wells (Well-1 and Well-2) was used to build the different ML models. The Well-1 data was used in training and testing the models, and the Well-2 data was used to validate the developed models. The obtained results show the capability of the three ML models to predict accurately the σh and σH using the well-log data. Comparing the results of RF, ANFIS, and FN models for minimum horizontal stress prediction showed that ANFIS outperforms the other two models with a correlation coefficient (R) for the validation dataset of 0.96 compared to 0.91 and 0.88 for RF, and FN, respectively. The three models showed similar results for predicting maximum horizontal stress with R values higher than 0.98 and an average absolute percentage error (AAPE) less than 0.3%. a20 index for the actual versus the predicted data showed that the three ML techniques were able to predict the horizontal stresses with a deviation less than 20% from the actual data. For the validation dataset, the RF, ANFIS, and FN models were able to capture all changes in the σh and σH trends with depth and accurately predict the σh and σH values. The outcomes of this study confirm the robust capability of ML to predict σh and σH from readily available logging data with no need for additional costs or site investigation.

Generation of Synthetic Photoelectric Log using Machine Learning Approach

2021 ◽  
Author(s):  
Mohammad Rasheed Khan ◽  
Zeeshan Tariq ◽  
Mohamed Mahmoud
Keyword(s):  
Machine Learning ◽  
Mean Squared Error ◽  
Coefficient Of Determination ◽  
Percentage Error ◽  
Well Log ◽  
Log Data ◽  
Hydrocarbon Reservoir ◽  
Analysis Scheme ◽  
Data Points ◽  
Error Metric

Abstract Photoelectric factor (PEF) is one of functional parameters of a hydrocarbon reservoir that could provide invaluable data for reservoir characterization. Well logs are critical to formation evaluation processes; however, they are not always readily available due to unfeasible logging conditions. In addition, with call for efficiency in hydrocarbon E&P business, it has become imperative to optimize logging programs to acquire maximum data with minimal cost impact. As a result, the present study proposes an improved strategy for generating synthetic log by making a quantitative formulation between conventional well log data, rock mineralogical content and PEF. 230 data points were utilized to implement the machine learning (ML) methodology which is initiated by implementing a statistical analysis scheme. The input logs that are used for architecture establishment include the density and sonic logs. Moreover, rock mineralogical content (carbonate, quartz, clay) has been incorporated for model development which is strongly correlated to the PEF. At the next stage of this study, architecture of artificial neural networks (ANN) was developed and optimized to predict the PEF from conventional well log data. A sub-set of data points was used for ML model construction and another unseen set was employed to assess the model performance. Furthermore, a comprehensive error metrics analysis is used to evaluate performance of the proposed model. The synthetic PEF log generated using the developed ANN correlation is compared with the actual well log data available and demonstrate an average absolute percentage error less than 0.38. In addition, a comprehensive error metric analysis is presented which depicts coefficient of determination more than 0.99 and root mean squared error of only 0.003. The numerical analysis of the error metric point towards the robustness of the ANN model and capability to link mineralogical content with the PEF.

Predicting Formation Pore-Pressure from Well-Log Data with Hybrid Machine-Learning Optimization Algorithms

2021 ◽  
Author(s):  
Mohammad Farsi ◽  
Nima Mohamadian ◽  
Hamzeh Ghorbani ◽  
David A. Wood ◽  
Shadfar Davoodi ◽  
...  
Keyword(s):  
Machine Learning ◽  
Pore Pressure ◽  
Optimization Algorithms ◽  
Well Log ◽  
Log Data ◽  
Hybrid Machine ◽  
Formation Pore Pressure

A Novel Method and its Application to Define Maximum Horizontal Stress and Stress Path

2021 ◽  
Author(s):  
Jianguo Zhang ◽  
Karthik Mahadev ◽  
Stephen Edwards ◽  
Alan Rodgerson
Keyword(s):  
Fracture Initiation ◽  
Stress Path ◽  
Horizontal Stress ◽  
Geomechanical Model ◽  
Breakdown Pressure ◽  
In Situ Stresses ◽  
Fracture Closure ◽  
Minimum Horizontal Stress ◽  
Maximum Horizontal Stress

Abstract Maximum horizontal stress (SH) and stress path (change of SH and minimum horizontal stress with depletion) are the two most difficult parameters to define for an oilfield geomechanical model. Understanding these in-situ stresses is critical to the success of operations and development, especially when production is underway, and the reservoir depletion begins. This paper introduces a method to define them through the analysis of actual minifrac data. Field examples of applications on minifrac failure analysis and operational pressure prediction are also presented. It is commonly accepted that one of the best methods to determine the minimum horizontal stress (Sh) is the use of pressure fall-off analysis of a minifrac test. Unlike Sh, the magnitude of SH cannot be measured directly. Instead it is back calculated by using fracture initiation pressure (FIP) and Sh derived from minifrac data. After non-depleted Sh and SH are defined, their apparent Poisson's Ratios (APR) are calculated using the Eaton equation. These APRs define Sh and SH in virgin sand to encapsulate all other factors that influence in-situ stresses such as tectonic, thermal, osmotic and poro-elastic effects. These values can then be used to estimate stress path through interpretation of additional minifrac data derived from a depleted sand. A geomechanical model is developed based on APRs and stress paths to predict minifrac operation pressures. Three cases are included to show that the margin of error for FIP and fracture closure pressure (FCP) is less than 2%, fracture breakdown pressure (FBP) less than 4%. Two field cases in deep-water wells in the Gulf of Mexico show that the reduction of SH with depletion is lower than that for Sh.

scholarly journals In Situ Stress Prediction in Subsurface Rocks: An Overview and a New Method

Geofluids ◽  
2021 ◽  
Vol 2021 ◽  
pp. 1-11
Author(s):  
Yushuai Zhang ◽  
Shangxian Yin ◽  
Jincai Zhang
Keyword(s):  
Horizontal Stress ◽  
In Situ Stress ◽  
New Method ◽  
Alternative Methods ◽  
Tensile Fracture ◽  
In Situ Stresses ◽  
Stress Estimation ◽  
Minimum Horizontal Stress ◽  
Polygon Method

Methods for determining in situ stresses are reviewed, and a new approach is proposed for a better prediction of the in situ stresses. For theoretically calculating horizontal stresses, horizontal strains are needed; however, these strains are very difficult to be obtained. Alternative methods are presented in this paper to allow an easier way for determining horizontal stresses. The uniaxial strain method is oversimplified for the minimum horizontal stress determination; however, it is the lower bound minimum horizontal stress. Based on this concept, a modified stress polygon method is proposed to obtain the minimum and maximum horizontal stresses. This new stress polygon is easier to implement and is more accurate to determine in situ stresses by narrowing the area of the conventional stress polygon when drilling-induced tensile fracture and wellbore breakout data are available. Using the generalized Hooke’s law and coupling pore pressure and in situ stresses, a new method for estimating the maximum horizontal stress is proposed. Combined it to the stress polygon method, a reliable in situ stress estimation can be obtained. The field measurement method, such as minifrac test, is also analyzed in different stress regimes to determine horizontal stress magnitudes and calibrate the proposed theoretical method. The proposed workflow combined theoretical methods to field measurements provides an integrated approach for horizontal stress estimation.

scholarly journals Machine Learning: A Useful Tool in Geomechanical Studies, a Case Study from an Offshore Gas Field

Energies ◽  
2020 ◽  
Vol 13 (14) ◽  
pp. 3528 ◽  
Author(s):  
Seyedalireza Khatibi ◽  
Azadeh Aghajanpour
Keyword(s):  
Machine Learning ◽  
Shear Wave ◽  
Wave Velocity ◽  
Shear Wave Velocity ◽  
Linear Method ◽  
Earth Model ◽  
Learning Approaches ◽  
Gas Field ◽  
In Situ Stresses

For a safe drilling operation with the of minimum borehole instability challenges, building a mechanical earth model (MEM) has proven to be extremely valuable. However, the natural complexity of reservoirs along with the lack of reliable information leads to a poor prediction of geomechanical parameters. Shear wave velocity has many applications, such as in petrophysical and geophysical as well as geomechanical studies. However, occasionally, wells lack shear wave velocity (especially in old wells), and estimating this parameter using other well logs is the optimum solution. Generally, available empirical relationships are being used, while they can only describe similar formations and their validation needs calibration. In this study, machine learning approaches for shear sonic log prediction were used. The results were then compared with each other and the empirical Greenberg–Castagna method. Results showed that the artificial neural network has the highest accuracy of the predictions over the single and multiple linear regression models. This improvement is more highlighted in hydrocarbon-bearing intervals, which is considered as a limitation of the empirical or any linear method. In the next step, rock elastic properties and in-situ stresses were calculated. Afterwards, in-situ stresses were predicted and coupled with a failure criterion to yield safe mud weight windows for wells in the field. Predicted drilling events matched quite well with the observed drilling reports.

Valuation of hydraulic fracturing potentials of organic-rich shales from the Anambra basin using rock mechanical properties from wireline logs

2021 ◽  
Vol 19 (3) ◽  
pp. 45-44
Author(s):  
Homa Viola Akaha-Tse ◽  
Michael Oti ◽  
Selegha Abrakasa ◽  
Charles Ugwu Ugwueze
Keyword(s):  
Mechanical Properties ◽  
Hydraulic Fracturing ◽  
Horizontal Stress ◽  
Shale Formations ◽  
In Situ Stresses ◽  
Anambra Basin ◽  
Wireline Logs ◽  
Rock Mechanical Properties ◽  
Minimum Horizontal Stress

This study was carried out to determine the rock mechanical properties relevant for hydrocarbon exploration and production by hydraulic  fracturing of organic rich shale formations in Anambra basin. Shale samples and wireline logs were analysed to determine the petrophysical, elastic, strength and in-situ properties necessary for the design of a hydraulic fracturing programme for the exploitation of the shales. The results obtained indicated shale failure in shear and barreling under triaxial test conditions. The average effective porosity of 0.06 and permeability of the order of 10-1 to 101 millidarcies showed the imperative for induced fracturing to assure fluid flow. Average Young’s modulus and Poisson’s ratio of about 2.06 and 0.20 respectively imply that the rocks are favourable for the formation and propagation of fractures during hydraulic fracking. The minimum horizontal stress, which determines the direction of formation and growth of artificially induced hydraulic fractures varies from wellto-well, averaging between 6802.62 to 32790.58 psi. The order of variation of the in-situ stresses is maximum horizontal stress>vertical stress>minimum horizontal stress which implies a reverse fault fracture regime. The study predicts that the sweet spots for the exploration and development of the shale-gas are those sections of the shale formations that exhibit high Young’s modulus, low Poisson’s ratio, and high brittleness. The in-situ stresses required for artificially induced fractures which provide pore space for shale gas accumulation and expulsion are adequate. The shales possess suitable mechanical properties to fracture during hydraulic fracturing. Application of these results will enhance the potentials of the onshore Anambra basin as a reliable component in increasing Nigeria’s gas reserves, for the improvement of the nation’s economy and energy security. Key Words: Hydraulic Fracturing, Organic-rich Shales, Rock Mechanical Properties, Petrophysical Properties, Anambra Basin

Sign in / Sign up

Export Citation Format

Share Document