Neural network seismic inversion based on lithofacies classification and seismic attributes to predict reservoir properties away from production wells

2021 ◽  
Author(s):  
Lilik Hardanto ◽  
Masako Robb
Keyword(s):  
Neural Network ◽  
Seismic Attributes ◽  
Seismic Inversion ◽  
Reservoir Properties ◽  
Production Wells ◽  
Lithofacies Classification

scholarly journals Petrophysical seismic inversion based on lithofacies classification to enhance reservoir properties estimation: a machine learning approach

2020 ◽  
Author(s):  
Amir Abbas Babasafari ◽  
Shiba Rezaei ◽  
Ahmed Mohamed Ahmed Salim ◽  
Sayed Hesammoddin Kazemeini ◽  
Deva Prasad Ghosh
Keyword(s):  
Acoustic Impedance ◽  
Reservoir Characterization ◽  
Seismic Inversion ◽  
Reservoir Properties ◽  
Well Location ◽  
The Neural Network ◽  
Machine Learning Approach ◽  
Lithofacies Classification ◽  
Elastic Inversion ◽  
Significant Match

Abstract For estimation of petrophysical properties in industry, we are looking for a methodology which results in more accurate outcome and also can be validated by means of some quality control steps. To achieve that, an application of petrophysical seismic inversion for reservoir properties estimation is proposed. The main objective of this approach is to reduce uncertainty in reservoir characterization by incorporating well log and seismic data in an optimal manner. We use nonlinear optimization algorithms in the inversion workflow to estimate reservoir properties away from the wells. The method is applied at well location by fitting nonlinear experimental relations on the petroelastic cross-plot, e.g., porosity versus acoustic impedance for each lithofacies class separately. Once a significant match between the measured and the predicted reservoir property is attained in the inversion workflow, the petrophysical seismic inversion based on lithofacies classification is applied to the inverted elastic property, i.e., acoustic impedance or Vp/Vs ratio derived from seismic elastic inversion to predict the reservoir properties between the wells. Comparison with the neural network method demonstrated this application of petrophysical seismic inversion to be competitive and reliable.

scholarly journals Porosity Prediction Using Neural Network Based on Seismic Inversion and Seismic Attributes

2019 ◽  
Vol 125 ◽  
pp. 15006
Author(s):  
Taufik Mawardi Sinaga ◽  
M. Syamsu Rosid ◽  
M. Wahdanadi Haidar
Keyword(s):  
Neural Network ◽  
Probabilistic Neural Network ◽  
Average Frequency ◽  
Seismic Attributes ◽  
Seismic Inversion ◽  
Carbonate Reservoir ◽  
Neural Network Method ◽  
Time Migration ◽  
The Neural Network ◽  
Network Method

It has done a study of porosity prediction by using neural network. The study uses 2D seismic data post-stack time migration (PSTM) and 2 well data at field “T”. The objective is determining distribution of porosity. Porosity in carbonate reservoir is actually heterogeneous, complex and random. To face the complexity the neural network method has been implemented. The neural network algorithm uses probabilistic neural network based on best seismic attributes. It has been selected by using multi-attribute method with has high correlation. The best attributes which have been selected are amplitude envelope, average frequency, amplitude weighted phase, integrated absolute amplitude, acoustic impedance, and dominant frequency. The attribute is used as input to probabilistic neural network method process. The result porosity prediction based on probabilistic neural network use non-linear equation obtained high correlation coefficient 0.86 and approach actual log. The result has a better correlation than using multi-attribute method with correlation 0.58. The value of distribution porosity is 0.05–0.3 and it indicates the heterogeneous porosity distribution generally from the bottom to up are decreasing value.

scholarly journals Porosity estimation in the Fort Worth Basin constrained by 3D seismic attributes integrated in a sequential Bayesian simulation framework

Geophysics ◽  
2017 ◽  
Vol 82 (4) ◽  
pp. M67-M80 ◽  
Author(s):  
Martin Blouin ◽  
Mickaele Le Ravalec ◽  
Erwan Gloaguen ◽  
Mathilde Adelinet
Keyword(s):  
Oil And Gas ◽  
Reservoir Characterization ◽  
Geophysical Methods ◽  
Seismic Attributes ◽  
Seismic Inversion ◽  
Fort Worth ◽  
3D Seismic ◽  
Reservoir Properties ◽  
Fort Worth Basin ◽  
Porosity Estimation

The accurate inference of reservoir properties such as porosity and permeability is crucial in reservoir characterization for oil and gas exploration and production as well as for other geologic applications. In most cases, direct measurements of those properties are done in wells that provide high vertical resolution but limited lateral coverage. To fill this gap, geophysical methods can often offer data with dense 3D coverage that can serve as proxy for the variable of interest. All the information available can then be integrated using multivariate geostatistical methods to provide stochastic or deterministic estimate of the reservoir properties. Our objective is to generate multiple scenarios of porosity at different scales, considering four formations of the Fort Worth Basin altogether and then restricting the process to the Marble Falls limestones. Under the hypothesis that a statistical relation between 3D seismic attributes and porosity can be inferred from well logs, a Bayesian sequential simulation (BSS) framework proved to be an efficient approach to infer reservoir porosity from an acoustic impedance cube. However, previous BBS approaches only took two variables upscaled at the resolution of the seismic data, which is not suitable for thin-bed reservoirs. We have developed three modified BSS algorithms that better adapt the BSS approach for unconventional reservoir petrophysical properties estimation from deterministic prestack seismic inversion. A methodology that includes a stochastic downscaling procedure is built and one that integrates two secondary downscaled constraints to the porosity estimation process. Results suggest that when working at resolution higher than surface seismic, it is better to execute the workflow for each geologic formation separately.

scholarly journals Reservoir properties estimation from 3D seismic data in the Alose field using artificial intelligence

2021 ◽  
Vol 11 (3) ◽  
pp. 1275-1287
Author(s):  
A. Ogbamikhumi ◽  
J. O. Ebeniro
Keyword(s):  
Neural Network ◽  
Correlation Coefficient ◽  
Feasibility Study ◽  
Cross Validation ◽  
Water Saturation ◽  
Seismic Inversion ◽  
Reservoir Properties ◽  
Before And After ◽  
Secondary Constraint ◽  
Static Modeling

AbstractIn an attempt to reduce the errors and uncertainties associated with predicting reservoir properties for static modeling, seismic inversion was integrated with artificial neural network for improved porosity and water saturation prediction in the undrilled prospective area of the study field, where hydrocarbon presence had been confirmed. Two supervised neural network techniques (MLFN and PNN) were adopted in the feasibility study performed to predict reservoir properties, using P-impedance volumes generated from model-based inversion process as the major secondary constraint parameter. Results of the feasibility study for predicted porosity with PNN gave a better result than MLFN, when correlated with well porosity, with a correlation coefficient of 0.96 and 0.69, respectively. Validation of the prediction revealed a cross-validation correlation of 0.88 and 0.26, respectively, for both techniques, when a random transfer function derived from a given well is applied on other well locations. Prediction of water saturation using PNN also gave a better result than MLFN with correlation coefficient of 0.97 and 0.57 and cross-validation correlation coefficient of 0.89 and 0.3, respectively. Hence, PNN technique was adopted to predict both reservoir properties in the field. The porosity and water saturation predicted from seismic in the prospective area were 24–30% and 20–30%, respectively. This indicates the presence of good quality hydrocarbon bearing sand within the prospective region of the studied reservoir. As such, the results from the integrated techniques can be relied upon to predict and populate static models with very good representative subsurface reservoir properties for reserves estimation before and after drilling wells in the prospective zone of reservoirs.

Comparison of linear regression and a probabilistic neural network to predict porosity from 3‐D seismic attributes in Lower Brushy Canyon channeled sandstones, southeast New Mexico

Geophysics ◽  
2001 ◽  
Vol 66 (5) ◽  
pp. 1349-1358 ◽  
Author(s):  
Daniel J. Leiphart ◽  
Bruce S. Hart
Keyword(s):  
Neural Network ◽  
New Mexico ◽  
Linear Regression ◽  
Regression Model ◽  
Seismic Data ◽  
Time Window ◽  
Probabilistic Neural Network ◽  
Seismic Attributes ◽  
Oil Field ◽  
Reservoir Properties

The Lower Brushy Canyon Formation of the Delaware Basin, New Mexico, consists of a series of overlying sand‐filled channels and associated fans separated by laterally extensive organic siltstone and carbonate interbeds. This laterally and vertically complex geology creates the need for precise interwell estimation of reservoir properties. In this paper we integrate wireline log and 3‐D seismic data to directly predict porosity in the area of an existing oil field in southeast New Mexico. The 3‐D seismic data were used to interpret the location of major stratigraphic markers between wells, and these seismic horizons were used to constrain a time window for a volume‐based attribute analysis. Stepwise regression and crossvalidation were used to combine seismic attributes to predict porosity in wells where the porosity was known from the well logs. The results of a linear regression porosity model showed good correlation [Formula: see text] between seven seismic attributes and the observed porosity logs at 11 wells in the study area, but the porosity volume created from the regression model did not display the known geologic features. A probabilistic neural network was then trained to look for a nonlinear relationship between the input data (the seven attributes) and the observed porosity at the 11 wells. The correlation was better [Formula: see text], but the biggest improvement over the linear regression model came in the more geologically realistic predicted porosity distribution.

Unsupervised machine learning using 3D seismic data applied to reservoir evaluation and rock type identification

2021 ◽  
pp. 1-69
Author(s):  
Marwa Hussein ◽  
Robert R. Stewart ◽  
Deborah Sacrey ◽  
Jonny Wu ◽  
Rajas Athale
Keyword(s):  
Machine Learning ◽  
Seismic Data ◽  
Rock Type ◽  
Seismic Attributes ◽  
Reservoir Quality ◽  
3D Seismic ◽  
Reservoir Properties ◽  
Unsupervised Machine Learning ◽  
Amplitude Data ◽  
3D Seismic Data

Net reservoir discrimination and rock type identification play vital roles in determining reservoir quality, distribution, and identification of stratigraphic baffles for optimizing drilling plans and economic petroleum recovery. Although it is challenging to discriminate small changes in reservoir properties or identify thin stratigraphic barriers below seismic resolution from conventional seismic amplitude data, we have found that seismic attributes aid in defining the reservoir architecture, properties, and stratigraphic baffles. However, analyzing numerous individual attributes is a time-consuming process and may have limitations for revealing small petrophysical changes within a reservoir. Using the Maui 3D seismic data acquired in offshore Taranaki Basin, New Zealand, we generate typical instantaneous and spectral decomposition seismic attributes that are sensitive to lithologic variations and changes in reservoir properties. Using the most common petrophysical and rock typing classification methods, the rock quality and heterogeneity of the C1 Sand reservoir are studied for four wells located within the 3D seismic volume. We find that integrating the geologic content of a combination of eight spectral instantaneous attribute volumes using an unsupervised machine-learning algorithm (self-organizing maps [SOMs]) results in a classification volume that can highlight reservoir distribution and identify stratigraphic baffles by correlating the SOM clusters with discrete net reservoir and flow-unit logs. We find that SOM classification of natural clusters of multiattribute samples in the attribute space is sensitive to subtle changes within the reservoir’s petrophysical properties. We find that SOM clusters appear to be more sensitive to porosity variations compared with lithologic changes within the reservoir. Thus, this method helps us to understand reservoir quality and heterogeneity in addition to illuminating thin reservoirs and stratigraphic baffles.

Prediction of Petrophysical Properties from Seismic Inversion and Neural Network: A case study

2021 ◽  
Author(s):  
Siddharth Garia ◽  
Arnab Kumar Pal ◽  
Karangat Ravi ◽  
Archana M Nair
Keyword(s):  
Neural Network ◽  
Bulk Density ◽  
Low Frequency ◽  
Seismic Inversion ◽  
Well Log ◽  
Petrophysical Properties ◽  
Seismic Section ◽  
Frequency Model ◽  
Seismic Reflection Data ◽  
Reflection Data

<p>Seismic inversion method is widely used to characterize reservoirs and detect zones of interest, i.e., hydrocarbon-bearing zone in the subsurface by transforming seismic reflection data into quantitative subsurface rock properties. The primary aim of seismic inversion is to transform the 3D seismic section/cube into an acoustic impedance (AI) cube. The integration of this elastic attribute, i.e., AI cube with well log data, can thereafter help to establish correlations between AI and different petrophysical properties. The seismic inversion algorithm interpolates and spatially populates data/parameters of wells to the entire seismic section/cube based on the well log information. The case study presented here uses machine learning-neural network based algorithm to extract the different petrophysical properties such as porosity and bulk density from the seismic data of the Upper Assam basin, India. We analyzed three different stratigraphic  units that are established to be producing zones in this basin.</p><p> AI model is generated from the seismic reflection data with the help of colored inversion operator. Subsequently, low-frequency model is generated from the impedance data extracted from the well log information. To compensate for the band limited nature of the seismic data, this low-frequency model is added to the existing acoustic model. Thereafter, a feed-forward neural network (NN) is trained with AI as input and porosity/bulk density as target, validated with NN generated porosity/bulk density with actual porosity/bulk density from well log data. The trained network is thus tested over the entire region of interest to populate these petrophysical properties.</p><p>Three seismic zones were identified from the seismic section ranging from 681 to 1333 ms, 1528 to 1575 ms and 1771 to 1814 ms. The range of AI, porosity and bulk density were observed to be 1738 to 6000 (g/cc) * (m/s), 26 to 38% and 1.95 to 2.46 g/cc respectively. Studies conducted by researchers in the same basin yielded porosity results in the range of 10-36%. The changes in acoustic impedance, porosity and bulk density may be attributed to the changes in lithology. NN method was prioritized over other traditional statistical methods due to its ability to model any arbitrary dependency (non-linear relationships between input and target values) and also overfitting can be avoided. Hence, the workflow presented here provides an estimation of reservoir properties and is considered useful in predicting petrophysical properties for reservoir characterization, thus helping to estimate reservoir productivity.</p>

Long-Term Forecasting and Optimization of Non-Stationary Well Operation Modes Through Neural Networks Simulation

2021 ◽  
Author(s):  
Roman Yurievich Ponomarev ◽  
Vladimir Evgenievich Vershinin
Keyword(s):  
Neural Network ◽  
Forecast Error ◽  
Optimal Solution ◽  
Hydrodynamic Modeling ◽  
Neural Network Modeling ◽  
Sharp Change ◽  
Technological Parameters ◽  
Production Wells ◽  
Long Term Forecasting

Abstract The article discusses the results of long-term forecasting of non-stationary technological modes of production wells using neural network modeling methods. The main difficulty in predicting unsteady modes is to reproduce the response of producing wells to a sharp change in the mode of one of the wells. Such jumps, as a rule, lead to a rapid increase in the forecast error. Training and forecasting of modes was carried out on the data of numerical hydrodynamic modeling. Two fields with significantly different properties, the number of wells and their modes of operation were selected as objects of modeling. Non-stationarity was set by changing the regime on one or several production wells at different points in time. The LSTM recurrent neural network carried out forecasting of production technological parameters. This made it possible to take into account the time-lagging influence of the wells on each other. It is shown that the LSTM neural network allows predicting unsteady technological modes of well operation with an accuracy of up to 5% for a period of 10 years. The solution of the problem of optimization of oil production is considered on the example of one of the models. It is shown that the optimal solution found by the neural network differs from the solution found by hydrodynamic modeling by 5%. At the same time, a significant gain in calculation time was achieved.

A Proxy Flow Modelling Workflow to Estimate Gridded Dynamic Properties and Well Production Rates by Deep Learning Algorithms

2021 ◽  
Author(s):  
Soumi Chaki ◽  
Yevgeniy Zagayevskiy ◽  
Terry Wong
Keyword(s):  
Neural Network ◽  
Deep Learning ◽  
Dynamic Properties ◽  
Reservoir Management ◽  
Flow Modeling ◽  
Static Properties ◽  
Production Rates ◽  
Reservoir Properties ◽  
Well Production ◽  
Reservoir Simulator

Abstract This paper proposes a deep learning-based framework for proxy flow modeling to predict gridded dynamic petroleum reservoir properties (like pressure and saturation) and production rates for wells in a single framework. It approximates the solution of a full physics-based numerical reservoir simulator, but runs much more rapidly, allowing users to generate results for a much wider range of scenarios in a given time than could be done with a full physics simulator. The proxy can be used for reservoir management tasks like history matching, uncertainty quantification, and field development optimization. A deep-learning based methodology for accurate proxy-flow modeling is presented which combines U-Net (a variant of convolutional neural network) to predict gridded dynamic properties and deep neural network (DNN) models to forecast well production rates. First, gridded dynamic properties, such as reservoir pressure and phase saturations, are predicted from static properties like reservoir rock porosity and absolute permeability using a U-Net. Then, the static properties and the dynamic properties predicted by the U-Net are input to a DNN to predict production rates at the well perforations. The inclusion of U-net predicted pressure and saturations improves the quality of the well rate predictions. The proposed methodology is presented with the synthetic Brugge reservoir discretized into grid blocks. The U-Net input consists of three properties: dynamic gridded reservoir properties (such as pressure or fluid saturation) at the current state, static gridded porosity, and static gridded permeability. The U-Net has only one output property, the target gridded property (such as pressure or saturation) at the next time step. Training and testing datasets are generated by running 13 full physics flow simulations and dividing them in a 12:1 ratio. Nine U-Net models are calibrated to predict pressures/saturations, one for each of the nine grid layers present in the Brugge model. These outputs are then concatenated to obtain the complete pressure/saturation model for all nine layers. The constructed U-Net models match the distributions of generated pressures/saturations of the numerical reservoir simulator with a correlation coefficient value of approximately 0.99 and above 95% accuracy. The DNN models approximate well production rates accurately from U-Net predicted pressures and saturations along with static properties like transmissibility and horizontal permeability. For each well and each well perforation, the production rate is predicted with the DNN model. The use of the constructed proxy flow model generates reservoir predictions within a few minutes compared to the hours or days typically taken by a full physics flow simulator. The direct connection that is established between the gridded static and dynamic properties of the reservoir and well production rates using U-Net and DNN models has not been presented previously. Using only a small number of runs for its training, the workflow matches the numerical reservoir simulator results with reduced computational effort. This helps reservoir engineers make informed decisions more quickly, resulting in more efficient reservoir management.

Sign in / Sign up

Export Citation Format

Share Document