Estimation of porosity from seismic attributes using a committee model with bat-inspired optimization algorithm

Gholami, Amin; Ansari, Hamid Reza

doi:10.1016/j.petrol.2017.03.013

Cited by 49 publications

(19 citation statements)

References 44 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One of the most important tools for reservoir evaluation and description of reservoir parameters is well log data . Information such as porosity, p-wave velocity, shale volume, water saturation, permeability, lithology, and production zones can be obtained from the processing and interpretation of well logs (Gholami and Ansari 2017). Although this type of data has higher resolution than seismic data, it relates to a small part of the reservoir or the well environment, and considering the complexities of the geology, errors will occur in generalizing the data to the whole reservoir (Somasundaram et al 2017).…”

Section: Introductionmentioning

confidence: 99%

Estimation of reservoir porosity using analysis of seismic attributes in an Iranian oil field

Soleimani

Hosseini

Hajivand

2020

J Petrol Explor Prod Technol

View full text Add to dashboard Cite

The most commonly used data for reservoir description are well and seismic data. Well data such as logs typically provide sufficient vertical resolution but leave a large space between the wells. Three-dimensional seismic data, on the other hand, can provide more detailed reservoir characterization between wells. However, the vertical resolution of seismic data is poor compared with that of well data. Conventionally, seismic data have been used to delineate reservoir structure; however, seismic data can be used for reservoir characterization such as porosity. Therefore, we can combine these two types of data to obtain reservoir parameters such as porosity and saturation. It is available the desired parameter (such as porosity) of the number of wells in the reservoir and seismic cube. And we are looking for the parameter estimation in the whole reservoir. To do this, there are several methods including multiple linear regression, neural networks, and geostatistical methods. Therefore, by determining the reservoir properties and correctly estimating these parameters, optimization can be performed with fewer wells, and the costs of exploration and production are reduced. In this paper, we apply these methods on the available data for an oil field in southwest Iran to obtain the porosity in a total reservoir cube, and these methods are then compared with one another. The results clearly show the superiority of neural networks compared with the other methods in estimating the reservoir parameter. The results also show that although estimation accuracy is increased significantly with the use of the geostatistical approach, this method requires that a sufficient number of well logs, representing all the fields under investigation, be provided in order to improve the geological model obtained by the multi-attribute and neural network methods.

show abstract

Section: Introductionmentioning

confidence: 99%

Estimation of reservoir porosity using analysis of seismic attributes in an Iranian oil field

Soleimani

Hosseini

Hajivand

2020

J Petrol Explor Prod Technol

View full text Add to dashboard Cite

show abstract

“…Moreover, machine learning and deep learning methods have been utilized in the seismic domain for different tasks such as inversion and interpretation (AlRegib et al, 2018). For example, seismic inversion has been attempted using supervised-learning algorithms such as support vector regression (SVR) (Al-Anazi and Gates, 2012), artificial neural networks (Röth and Tarantola, 1994;Araya-Polo et al, 2018), committee models (Gholami and Ansari, 2017), convolutional neural networks (CNNs) (Das et al, 2018), and many other methods (Chaki et al, 2015;Yuan and Wang, 2013;Gholami and Ansari, 2017;Chaki et al, 2017;Mosser et al, 2018;Chaki et al, 2018). More recently, a sequencemodeling-based machine learning workflow was used to estimate petrophysical properties from seismic data (Alfarraj and AlRegib, 2018), which showed that recurrent neural networks are superior to feed-forward neural networks in capturing the temporal dynamics of seismic traces.…”

Section: Introductionmentioning

confidence: 99%

Semi-supervised Sequence Modeling for Elastic Impedance Inversion

Alfarraj,

AlRegib

2019

Preprint

View full text Add to dashboard Cite

Recent applications of machine learning algorithms in the seismic domain have shown great potential in different areas such as seismic inversion and interpretation. However, such algorithms rarely enforce geophysical constraints; the lack of which might lead to undesirable results. To overcome this issue, we propose a semi-supervised sequence modeling framework based on recurrent neural networks for elastic inversion from multi-angle seismic data. Specifically, seismic traces and elastic impedance traces are modeled as time series. Then, a neural-network-based inversion model comprising convolutional and recurrent neural layers is used to invert seismic data for elastic impedance. The proposed workflow uses well-log data to guide the inversion. In addition, it utilizes seismic forward modeling to regularize the training, and to serve as a geophysical constraint for the inversion. The proposed workflow achieves an average correlation of 98% between the estimated and target elastic impedance using 10 well-logs for training on a synthetic dataset.

show abstract

“…Several attempts have been made using machine learning and statistical learning tools such as artificial neural networks, and support vector regression to solve the RC problem [2][3][4][5]. The literature shows great promise for machine learning algorithms for property estimation.…”

Section: Introductionmentioning

confidence: 99%

Petrophysical-property estimation from seismic data using recurrent neural networks

Alfarraj

AlRegib

2018

SEG Technical Program Expanded Abstracts 2018

View full text Add to dashboard Cite

Reservoir characterization involves the estimation petrophysical properties from well-log data and seismic data. Estimating such properties is a challenging task due to the non-linearity and heterogeneity of the subsurface. Various attempts have been made to estimate petrophysical properties using machine learning techniques such as feed-forward neural networks and support vector regression (SVR). Recent advances in machine learning have shown promising results for recurrent neural networks (RNN) in modeling complex sequential data such as videos and speech signals. In this work, we propose an algorithm for property estimation from seismic data using recurrent neural networks. An applications of the proposed workflow to estimate density and p-wave impedance using seismic data shows promising results compared to feed-forward neural networks.

show abstract

Estimation of porosity from seismic attributes using a committee model with bat-inspired optimization algorithm

Cited by 49 publications

References 44 publications

Estimation of reservoir porosity using analysis of seismic attributes in an Iranian oil field

Estimation of reservoir porosity using analysis of seismic attributes in an Iranian oil field

Semi-supervised Sequence Modeling for Elastic Impedance Inversion

Petrophysical-property estimation from seismic data using recurrent neural networks

Contact Info

Product

Resources

About