Elsevier

Fuel

Volume 266, 15 April 2020, 116934
Fuel

Full Length Article
Hydrocarbon saturation in shale oil reservoirs by inversion of dielectric dispersion logs

https://doi.org/10.1016/j.fuel.2019.116934Get rights and content

Abstract

Hydrocarbon saturation is an important petrophysical parameter in reservoir evaluation and reserve calculation. Dielectric dispersion logs have been widely used to estimate the hydrocarbon saturation, especially in high salinity shale formations. However, there is still room to improve the accuracy of interpretation of dielectric dispersion logs. In this paper, we used two interpretation models combined four dielectric permittivity and four conductivity logs measured by Array Dielectric Tool (ADT) at four frequencies to predict the hydrocarbon saturation in shale oil reservoirs, Triassic Yanchang Formation, Ordos Basin, China. The logs measured at the highest frequency are interpreted by the Complex Refractive Index (CRIM) model. And the other three frequencies measured logs are processed by the shaly sandstone (SHSD) model. Based on the two models, simulated annealing algorithm is selected to calculate water saturation and rock textual parameter, the dispersive phase fraction, and water salinity. Besides, accurate porosity, dielectric permittivity of rock matrix, and temperature are input information to the proposed method. The proposed method is applied to shale oil reservoirs of Triassic Yanchang Formation in two wells. The effectiveness and reliability of proposed model is verified by synthetic log responses. The mean relative error between synthetic and field logs is small. The predicted hydrocarbon saturation is consistent with core data, and its variation is consistent with characteristics of NMR T2 distributions, verifying the accuracy and reliability of proposed method and inversion results. Our estimated hydrocarbon saturation of the two wells is about 30%, showing a good oil generation potential of the studied shale oil reservoirs.

Introduction

Shale oil has gained a significant interest in exploration and development of unconventional resources in the past ten years [1], [2], [3], [4]. In addition to North America, a large number of shale reservoirs are found in the basins in China, including Ordos, Bohai Bay, Junggar, Songliao Basin, etc. that are a huge source of oil production [5], [6]. Although unconventional shale oil plays have been studied for many years, some of the characterization methods especially those focused on fluid properties still need further investigation. Hydrocarbon saturation is a vital petrophysical parameter in play evaluation and reserve estimation. However, it is difficult to calculate hydrocarbon saturation in unconventional shale plays due to variabilities in lithology, high clay content, presence of organic matter and pyrite, low porosity and complex wettability [7], [8]. Dielectric logging is a method that distinguishes reservoir fluid properties through the difference between relative dielectric permittivity of hydrocarbon/ rock minerals and water [9], [10]. The relative dielectric permittivity of hydrocarbon and most rock-forming minerals ranges from 2 to 10, while this value for water varies from 50 to 80 [7], [8]. Therefore, the dielectric permittivity of any reservoir depends largely on the water content of the rock. The Array Dielectric Tool (ADT) developed by Schlumberger provides permittivity and conductivity at four different megahertz range frequencies [11]. Considering the application of ADT logging in shale oil plays, since these reservoirs are extremely tight and mud intrusion in open wells is almost negligible, the saturation in the detection range of dielectric logging would be close to the saturation of an uninvaded formation. This makes dielectric logging suitable for reservoir evaluation and fluid saturation estimation in unconventional reservoirs, in particular.

In recent years, dielectric dispersion logs have been widely used in the evaluation of unconventional reservoir, such as shale oil. Seleznev et al. [12] measured the dielectric permittivity of shale rocks of Green River Formation, further inverted the relative dielectric permittivity of different minerals by (Complex Refractive Index) CRIM method, and applied them to process real logging data. Combining elemental capture spectroscopy (ECS), nuclear magnetic resonance (NMR) with dielectric log data, Musharfi et al. [13] comprehensively evaluated the petrophysical properties of shale gas reservoirs by CRIM methods. Almarzooq et al. [14] measured the dielectric permittivity of shale samples from a certain area located in Saudi Arabia and found that pyrite has a great impact on the dielectric responses of shale. Wang and Popitt [15] believed that at higher frequencies, there is still an evident interfacial polarization in shale reservoirs with high clay content, thus CRIM model may not be applied in such cases. Chen and Heidari [16] studied the pore-scale simulation of dielectric constant for estimating hydrocarbon. In their study, the effects of both organic matter and pyrite content on the relative dielectric constant of the formation as a whole were studied. They found that total organic carbon (TOC) content would affect the dielectric constant, while for pyrite both distribution and content would show a significant impact on the dielectric constant. As a result, they proposed an improved CRIM model by incorporating water distribution in tortuous pathways in the formation. Based on the similarity in dielectric properties between organic matter and hydrocarbon, Zhao et al. [17] improved the pseudo-Archie model and applied it to the saturation calculation of Lucaogou shale oil reservoirs in Jimusar Sag, Junggar Basin.

Pratiksha et al. [18], [19] utilized dielectric dispersion logs measured at four different megahertz range frequencies to evaluate the Wolfcamp and Bakken shale reservoirs, respectively. Based on the Waxman-Smith (WS) equation, CRIM and Stroud-Milton-De (SMD) model [20], Pratiksha [18], [19] obtained water saturation, formation water conductivity, Archie parameters m and n by joint inversion of resistivity and dielectric dispersion logs. They pointed out that the relative error is high in the intervals with higher pyrite content. However, SMD model is more suitable for clean sandstone reservoirs, and the dielectric dispersion is enhanced in pyrite and clay-rich shale formations. In addition, the Levenberg-Marquardt algorithm they used was sensitive for initial values for the unknown parameters. Han et al. [7] calculated the hydrocarbon saturation in a Lower-Paleozoic organic-rich shale gas formation based on the Markov-chain Monte Carlo (MCMC) stochastic inversion of broadband electromagnetic logs with the interfacial polarization IP model [21], [22]. It is noteworthy that defining equations of the IP model would vary based on different grain shapes, however, the real clay or pyrite shapes cannot be characterized by simple shapes. Besides, the MCMC inversion requires prior information about the reservoir properties.

Simulated annealing (SA) algorithm is a general probabilistic algorithm, which is used to find the global optimal solution of a proposition in a large search space [23]. It has been widely used in geosciences and engineering field [24], [25], [26], [27]. Compared with Markov-chain Monte Carlo (MCMC) stochastic inversion, it is not sensitive to the initial guess and does not require any prior information. In addition, inspired by the SMD model, Han et al. [28] proposed a shaly sandstone (SHSD) model that considers the effect of shale on dielectric dispersion. This model provides more accurate hydrocarbon saturation in shaly sandstone reservoirs.

Based on what was above, in this paper, a method that combines SHSD and CRIM models to calculate the hydrocarbon saturation of shale oil reservoirs by the simulated annealing algorithm of dielectric dispersion logs is proposed. The method is applied to the Triassic Yanchang Formation, Ordos Basin, China, which is a prominent shale oil reservoir. The effectiveness and reliability of the proposed model is verified by synthetic log responses and core data from the lab.

Section snippets

Data

For this study, the conventional, NMR and dielectric dispersion logs along with data from core plus retrieved from two wells were collected and analyzed. Conventional and NMR logs are used to calculate accurate porosity and mineral volume contents. The dielectric dispersion logs provide four permittivity and conductivity, which were measured at 22 MHz, 100 MHz, 350 MHz and 960 MHz. The core data includes porosity, and hydrocarbon and water saturations. The core and logging data were taken from

Interpretation models for dielectric dispersion logs

CRIM is a commonly used model for interpreting the complex dielectric constant of dielectric dispersion logs at around 1 GHz. It does not include the variation of texture because the texture affects the interfacial polarization but the interfacial polarization around 1 GHz is negligible. The formulation of the CRIM for a three-phase model is [11]:ε=(1-ϕ)εm+ϕ(1-Sw)εh+ϕSwεwwhere ε is the complex relative permittivity of the rocks; ϕ is the total porosity of the rock; Sw is water saturation; εm

Proposed inversion method to synthetic logs with assumed parameters

The proposed inversion method is first applied to the synthetic logs to test its robustness. We synthesize four permittivity and four conductivity logs at 22 MHz, 100 MHz, 350 MHz, 960 MHz by assuming formation properties close to the Chang 7 member summarized in Table 1. The porosity is set as 10%, wet-clay dispersive phase parameters pd, ε εs are set as 0.05, 5, and 650, respectively, while the salinity is assumed 20ppk. These values are utilized to generate the permittivity and conductivity

Conclusions

In this paper, we proposed a method utilizing two interpretation models combined with simulated annealing algorithm to process four dielectric permittivity and four conductivity logs measured at four frequencies in shale oil reservoirs. The output parameters are the hydrocarbon saturation, rock textual parameter, volume fraction of clay-induced dispersive phase, and water salinity. The proposed method is applied to shale oil reservoirs of Triassic Yanchang Formation, Ordos Basin, China. The

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

This paper is supported by the National Science and Technology Major Project of China (2016ZX05050-008) and the National Natural Science Foundation of China (41574121, 51874262).

References (39)

  • L. Ingber

    Very fast simulated re-annealing

    Math Comput Modeling

    (1989)
  • R.J. Hill et al.

    Oil and gas geochemistry and petroleum systems of the Fort Worth Basin

    AAPG Bull

    (2007)
  • Jarvie DM. Shale resource systems for oil and gas: Part 2-Shale-oil resource systems....
  • M.A. Kirschbaum et al.

    Controls on the deposition and preservation of the cretaceous mowry shale and frontier formation and equivalents, rocky mountain region, Colorado, Utah, and Wyoming Controls on the mowry shale and frontier formation, rocky mountain region

    AAPG Bull

    (2013)
  • K. Liu et al.

    The impact of pore size distribution data presentation format on pore structure interpretation of shales

    Adv Geo-Energ Res

    (2019)
  • L. Zhu et al.

    An improved theoretical nonelectric water saturation method for organic shale reservoirs

    IEEE Access

    (2019)
  • Y.M. Hong
  • T. Han et al.

    Dielectric properties of fractured carbonate rocks from finite-difference modeling Dielectrics of fractured carbonate rocks

    Geophysics

    (2019)
  • J.M. Donadille et al.
  • Cited by (0)

    View full text