Shear wave prediction using committee fuzzy model constrained by lithofacies, Zagros basin, SW Iran

Highlights

• The geological constraint specially lithofacies information improves estimation shear wave velocity utilizing Fuzzy machine system.
• Predicting shear wave velocity from post-stack seismic attributes is an efficient and inexpensive strategy when real data of shear wave velocity are not accessible in some wells.
• Fuzzy inference systems provide a cost-effective and accurate way to estimate shear wave velocity from seismic attributes. Amongst fuzzy inference systems, OFIS has provided more accurate results.
• Combining the results of fuzzy inference systems in a power law structure of committee machine, improves the results of individual fuzzy models.
• The BA is a fast and exact method to optimize the OFIS and CFM methods. The grouping of data based on lithofacies significantly enhanced the prediction accuracy spacially with lower error in sand.

Abstract

The main purpose of this study is to introduce the geological controlling factors in improving an intelligence-based model to estimate shear wave velocity from seismic attributes. The proposed method includes three main steps in the framework of geological events in a complex sedimentary succession located in the Persian Gulf. First, the best attributes were selected from extracted seismic data. Second, these attributes were transformed into shear wave velocity using fuzzy inference systems (FIS) such as Sugeno’s fuzzy inference (SFIS), adaptive neuro-fuzzy inference (ANFIS) and optimized fuzzy inference (OFIS). Finally, a committee fuzzy machine (CFM) based on bat-inspired algorithm (BA) optimization was applied to combine previous predictions into an enhanced solution. In order to show the geological effect on improving the prediction, the main classes of predominate lithofacies in the reservoir of interest including shale, sand, and carbonate were selected and then the proposed algorithm was performed with and without lithofacies constraint. The results showed a good agreement between real and predicted shear wave velocity in the lithofacies-based model compared to the model without lithofacies especially in sand and carbonate.

Introduction

Shear wave velocity () plays a vital role in geomechanical studies and reservoir characterization in petroleum industry including petrophysical modeling, wellbore stability analysis, casing design, well planning, sand production and hydraulic fracturing. This parameter is obtained directly from core analysis in laboratory or downhole measurement such as dipole shear sonic imager (DSI) tools. Utilizing downhole measurement is an economical and general method in petroleum industries due to its advantages such as being nondestructive, continuous (in logging rock in reservoir condition), time-efficient and cost-effective compared to core analysis (Lacy, 1997). However, DSI is an expensive tool and is usually not recorded in many wells. Also, DSI information was not acquired in older wells due to lack of this technology. Therefore, finding out a quantitative formulation to estimate is a very important task. In this regard many researches have been focused on establishing a relation between and other rock properties obtained in laboratory or well logging (Castagna et al., 1985, Han, 1986, Anselmetti and Eberli, 1993, Eskandari et al., 2004, Brocher, 2005, Rezaee et al., 2006, Rajabi et al., 2009, Asoodeh and Bagheripour, 2012a, Asoodeh and Bagheripour, 2012b; Bagheripour et al., 2015). Bagheripour et al. (2015) show intelligence-based methods performed better than empirical correlation in the prediction of . Estimating from conventional well logs has a good correlation with the measured values. However, this strategy has the main limitation of using post-drilling data and cannot be used for drilling forecasts. For overcoming this limitation, estimation of from seismic data is a practical solution. Pre-stack inversion is used to convert seismic angle or offset data into shear impedance or velocity (Jin et al., 2000, Stewart et al., 2002, Hampson et al., 2005, Lu et al., 2015). Since pre-stack inversion methods are time-consuming and expensive and require specialist skills, using integrated post-stack data with an intelligence-based model could be an appropriate technique to estimate

.

The aim of this research is to introduce a novel, available, easy, and inexpensive method for estimating shear wave velocity from post-stack seismic data. In this research upper cretaceous geological complex formation was selected from an oil field located in the SW of Iran. For this purpose, a committee machine based on fuzzy inference system (FIS) was established for formulating post-stack seismic attributes to through a double-stage estimation. As can be seen in the flowchart shown in Fig. 1, initial estimations are achieved from elements of committee machine including Sugeno’s fuzzy inference system (SFIS), adaptive neuro-fuzzy inference system (ANFIS) and optimized fuzzy inference system (OFIS). The seismic attributes used as input data in this study were a set of physical attributes extracted from a post-stack seismic cube. Relative P-impedance was also considered as input since there was a pronounced correlation between shear and compressional impedance (Castagna et al., 1985). In this study, Relative P-impedance was obtained by a fast, easy and inexpensive inversion method called colored inversion (CI) based on Lancaster and Whitcombe (2000). The CI transforms the mean seismic spectrum into the mean impedance log spectrum by using an inversion operator. The CI’s results and other seismic attributes were extracted at the well locations. Among them, best attributes were determined by step-wise regression as input for

estimation using fuzzy inference systems. Finally, a committee fuzzy machine (CFM) was utilized to combine all the results of earlier steps into the final output. The optimization method used in CFM is a new and powerful nature-inspired algorithm named bat-inspired algorithm (BA).

The introduced strategy was applied in an Iranian oil field in a clastics and carbonate reservoir. The method was performed with and without data classification. The classification of data was done based on lithofacies of sand, shale, and carbonate sediments in Kazhdumi, Dariyan, and Gadvan formations, respectively.

Section snippets

Bat-inspired algorithm

The BA is a nature-inspired algorithm proposed by Yang (2010) which uses the echolocation behavior of bats. The bats typically determine the size and position of preys by emitting loud sound impulses forward to objects and hear back the response that comes from them. In this algorithm the probable solutions are the positions () over which the bats randomly fly with velocity (). Each bat releases sound and varies its loudness (A) and wavelength () to discover a prey. Different frequencies (

Geology setting

The oil and gas field of interest situated on the Hendijan-Nowrooz-Khafji fault, is an elongated symmetrical anticline with NNE-SSW axial trend, and is along with the nearby SW Iran oil and gas fields such as Khafji, Hout, Dorra-Arash, Hendijan and Bahregansar in Zagrous basin (Shiroodi et al., 2015). The study field is located in the northeastern margin of the Afro-Arabian plate in the Zagros basin. The classical stratigraphic chart presented by James and Wynd (1965) was used which is similar

Extraction of seismic attributes

Seismic attributes are mathematical transforms of seismic data which present enhanced information which might be invisible in the original seismic data (Chen and Sidney, 1997, Taner et al., 1994). These attributes have been divided into two classes including physical and geometrical attributes. Geometrical attributes are generally used in seismic stratigraphy and physical attributes are related to the physical properties of rock. In this study, 26 physical attributes (Table 1) which were mostly

Multi-attribute transform to shear wave velocity

All input/output data were normalized in the range of [-1 1] according to the following linear mapping function:where is the mapped values, is the original value of data, and

are maximum and minimum of original data, respectively. All input/output data were normalized in the range of [-1 1] and after modeling, the estimated shear wave values were back-transformed into the original range. The normalization task reduces the confusion of constructing the model (

Results and discussion

The graphical comparison between the measured and predicted shear wave velocity using the CFM model separately in each lithofacies for training and test data is shown in Fig. 13. According to this figure, the best correlation coefficient for testing the model was achieved in sand lithofacies. Fig. 14 shows the measured and predicted shear wave logs for the test data at the blind well location. The CFM model and its elements including SFIS, ANFIS and OFIS were evaluated in terms of statistical

Conclusion

Shear wave velocity is a critical parameter in geomechanical studies and reservoir characterization. Geological phenomena are always complex in many heterogeneous reservoirs. So that these events control physical and geometrical properties across the reservoir. In this study, a new intelligence-based method limited to geological lithofacies information has been proposed to predict shear wave velocity in pre-drilling phase from seismic data. The proposed method is a committee machine whose

Acknowledgments

Important supports by the Ferdowsi University of Mashhad Research Council are greatly acknowledged. The authors are especially indebted to the Iranian Offshore Oil Company (IOOC) for providing the dataset and interpretation facilities and permission to publish subsurface data used in this paper. We are grateful to Ali Faghih for his constructive suggestions and corporations. We also wish to thank Hossein Mohammadrezaei for his thoughtful suggestions.

References (50)

• H.R. AnsariUse seismic colored inversion and power law committee machine based on imperial competitive algorithm for improving porosity prediction in a heterogeneous reservoirJ. Appl. Geophys.(2014)
• H.R. Ansari et al.An improved support vector regression model for estimation of saturation pressure of crude oilsFluid Phase Equilibria(2015)
• M. Asoodeh et al.Estimation of bubble point pressure from PVT data using a power-law committee with intelligent systemsJ. Pet. Sci. Eng.(2012)
• M. Asoodeh et al.Fuzzy classifier based support vector regression framework for poisson ratio determinationJ. Appl. Geophys.(2013)
• M. Asoodeh et al.Asphaltene precipitation of titration data modeling through committee machine with stochastically optimized fuzzy logic and optimized neural networkFluid Phase Equilibria(2014)
• P. BagheripourCommittee neural network model for rock permeability predictionJ. Appl. Geophys.(2014)
• P. Bagheripour et al.Fuzzy ruling between core porosity and petrophysical logs, subtractive clustering vs. Genetic algorithm pattern searchJ. Appl. Geophys.(2013)
• A. Gholami et al.How committee machine with SVR and ACE estimates bubble point pressure of crudesFluid Phase Equilibria(2014)
• A. Kadkhodaie-Ilkhchi et al.Petrophysical data prediction from seismic attributes using committee fuzzy inference systemComput. Geosci.(2009)
• E.H. Mamdani et al.An experimental in linguistic synthesis with a fuzzy logic controlInt. J. Man. Mach. Stud.(1975)
• S. Mishra et al.A new meta-heuristic bat inspired classification approach for microarray dataProcedia Technol.(2012)
• L. ZadehFuzzy setsInf. Control(1965)
• M. Afshar et al.Genetic optimization of neural network and fuzzy logic for oil bubble point pressure modelingKorean J. Chem. Eng.(2014)
• H.R. Ansari et al.Drilling rate of penetration prediction through committee support vector regression based on imperialist competitive algorithmCarbonates Evaporites(2016)
• F.S. Anselmetti et al.Controls on sonic velocity in carbonatesPure Appl. Geophys.(1993)
• M. Asoodeh et al.Prediction of compressional, shear, and stoneley wave velocities from conventional well log data using a committee machine with intelligent systemsRock Mech. Rock Eng.(2012)
• P. Bagheripour et al.Support vector regression based determination of shear wave velocityJ. Pet. Sci. Eng.(2015)
• T.M. BrocherEmpirical relations between elastic wave speeds and density in the Earth’s crustBull. Seismol. Soc. Am.(2005)
• J.P. Castagna et al.Relationship between compressional and shear wave velocities in silicate rocksGeophysics(1985)
• Q. Chen et al.Seismic attribute technology for reservoir forecasting and monitoringLead. Edge(1997)
• R.B. Davies et al.Early to mid-cretaceous mixed carbonate-clastic shelfal systems; examples, issues and models from the Arabian plateGeoArabia(2002)
• N.R. Draper et al.Applied Regression Analysis(1966)
• D. Dubois et al.Fuzzy Information Engineering, a Guided Tour of Applications(1997)
• H. Eskandari et al.Application of multiple regression and artificial neural network techniques to predict shear wave velocity from wireline log data for a carbonate reservoir, south IranCSEG Rec.(2004)
• F. GhazbanPetroleum Geology of the Persian Gulf(2007)