A5000790855- Reservoir Engineering and Simulation Methods
- Hydraulic Fracturing and Reservoir Analysis
- Drilling and Well Engineering
- Model Reduction and Neural Networks
- Enhanced Oil Recovery Techniques
- Rough Sets and Fuzzy Logic
- Language and cultural evolution
- Hydrocarbon exploration and reservoir analysis
- Geomagnetism and Paleomagnetism Studies
- Seismic Imaging and Inversion Techniques
- Natural Language Processing Techniques
- Geological Studies and Exploration
- Authorship Attribution and Profiling
- Earthquake Detection and Analysis
- Sports Science and Education
- Physical Education and Training Studies
- Oil and Gas Production Techniques
- Ionosphere and magnetosphere dynamics
Skolkovo Institute of Science and Technology
2019-2023
Kazan Federal University
2014-2017
Abstract This paper considers the development of a computationally fast model for simulation multiphase flow in porous media heterogeneous reservoir with unlimited number wells characterized by different type completion. solution has been obtained means replacing differential equation governing approximate equations which are parametrized convolutional neural networks. The matching dynamic properties original and reduced models is ensured conservation spatial invariance property equations....
Summary The increased interest for geological and reservoir simulation model construction the old fields raises issue of reinterpretation well logging data stock, taking into account concepts geology, specified during period field production. This work shows results mathematical approach development automatic interpretation stock. is aimed to solve problem fast a standard set using machine learning algorithms. solutions obtained determination stratigraphic boundaries with use logistic...
In this paper the method for modelling of word usage frequency time series is proposed. An artificial feedforward neural network was used to predict frequencies. The trained using maximum likelihood criterion. Google Books Ngram corpus analysis. This database provides a large amount data on specific forms 7 languages. Statistical allows finding optimal fitting and filtering algorithm subsequent lexicographic analysis verification trend models.
In this paper we describe results of the principal components analysis dynamics Total Electronic Content (TEC) data with use global maps presented by Jet Propulsion Laboratory (NASA, USA) for period 2007-2011. We show that result decomposition in essentially depends on method used preprocessing data, their representation (the coordinate system), and centering technique (e.g., daily seasonal extracting). The momentarily co-moving frame reference other special techniques provide opportunity...
Abstract In adaptation of reservoir models a direct gradient backpropagation through the forward model is often intractable or requires enormous computational costs. Thus one have to construct separate that simulate them implicitly, e.g. via stochastic sampling solving adjoint systems. We demonstrate if neural network, becomes naturally involved both in training and adaptation. our research we compare 3 strategies: variation variables, network latent space discuss what extent they preserve...
Let us consider some set of points on the Cartesian plane. Each point is a part one few curves describing dependency between abscissas and ordinates. In our case these are dependencies rock occurrence depth oil saturation described by Skelt-Harrison equation. this work problem distributing into clusters corresponding to different being investigated. Our original method based presenting data as elements partial ordered sets with coordinate order proposed. Thus solve clustering needs find all...
This paper introduces a method for constructing adaptive reduced-order reservoir simulation models based on the POD-DEIM approach field development optimization and assisted history matching problems. The is adapting orthogonal decompositions bases to varying model configuration. utilizes information contained in of original supplements them with new components instead from scratch. Adapting significantly reduces computational costs building allows application such tasks requiring multiple...
Reservoir simulation and adaptation (also known as history matching) are typically considered separate problems. While a set of models aimed at the solution forward problem assuming all initial geological parameters known, other adjust under fixed model to fit production data. This results in many difficulties for both reservoir engineers developers new efficient computation schemes. We present unified approach A single neural network allows pass from 3D through dynamic state variables...
Summary In this work, the adaptive approach for reservoir simulation, based on POD-Galerkin ROM, is discussed. The proposed technique idea of utilizing information contained in POD basis constructed specific model configuration to build a new setup. adaptation performed with use small set snapshots from updated configuration. required number significantly smaller than one constructing scratch. allows us reduce computational resources needed offline stage thus enabling ROM variety production...