(1. China Petroleum Exploration and Development Research Institute, Beijing100083;

2. China Youshi University Postdoctoral Mobile Station (Beijing), Beijing 102249)

The Lower Jurassic in western Hubei and eastern Chongqing is a typical continental shale gas reservoir in Sichuan Basin. The mineral composition and organic carbon content of the reservoir are important indicators to determine the difficulty and effectiveness of this kind of gas reservoir engineering development. In order to break through the multi-solution problem of conventional reservoir logging evaluation methods in complex mineral reservoir evaluation, this study fully excavates the geological information contained in conventional logging data, and comprehensively evaluates the composition and content of shale including organic carbon with nonlinear inversion and optimization algorithm as the core idea, and obtains good logging evaluation results. The research results improve the logging evaluation method of shale gas reservoir and play a positive role in promoting the development of shale gas exploration and development related technologies.

Conventional logging response; Nonlinear inversion optimization method; Logging evaluation of organic carbon content in mineral components

Logging evaluation of mineral composition

Total organic matter content of oil shale

Lu Jing 1, 2, Li Jun 1

(1. China Petrochemical Exploration and Development Research Institute, Beijing100083;

2. Postdoctoral Research Center of China Shiyou University, Beijing 102249, China)

The Lower Jurassic strata in western Hubei and eastern Chongqing are typical continental shale gas reservoirs in Sichuan Province. Mineral composition and total organic matter content are important indicators to reflect the engineering difficulty and effectiveness of this kind of gas reservoir. In order to break through the multi-solution problem that often occurs when conventional reservoir logging evaluation methods are used to evaluate oil shale reservoirs, this study fully excavates the hidden geological information in conventional logging response, and evaluates the mineral composition content and TOC of oil shale with the core idea of nonlinear joint inversion and optimization, and achieved good results. This study supplements the gas shale logging evaluation method and plays a positive role in the development of gas shale exploration and development related technologies.

Keywords conventional logging; Mineral composition; TOC nonlinear joint inversion; Optimization; Logging evaluation

Western Hubei and eastern Chongqing are one of the favorable targets of shale gas reservoirs around Sichuan Basin. Jiannan structure is located in the north-central part of Shizhu syncline in the eastern Sichuan fold belt of Sichuan Basin. The deep-semi deep lake shale developed in the Lower Jurassic Ziliujing Formation is a typical continental shale gas reservoir. This set of shale has stable regional distribution, large thickness and shallow burial, but it has more frequent phase transformation characteristics than marine shale, and the mineral composition of the reservoir is complex and changeable. Accurately mastering the mineral composition and organic carbon content of shale gas reservoir is an important basis for subsequent evaluation of key reservoir parameters-brittleness and gas bearing, and it is also an important and difficult problem to be solved urgently in shale gas logging evaluation. Depending on a fixed interpretation model and using a few logging curves to determine the mineral content of the reservoir, the evaluation effect is good in conventional reservoirs with single lithology and minerals, but it can not properly solve the problem of multi-solution of complex mineral composition and content of shale gas reservoirs. The author deeply digs the rich geological information contained in various conventional logging data, analyzes and establishes the reservoir mineral composition model, and evaluates the complex rock composition content including organic carbon through nonlinear inversion and optimization algorithm, which breaks through the conventional logging reservoir evaluation idea and expands the logging evaluation method of unconventional shale gas reservoir mineral composition and organic carbon content. Through the verification of the whole rock composition data of the core in the laboratory, this method has achieved good evaluation effect.

Lithology and rock mineralogy characteristics of gas-bearing shale in 1 Dongyuemiao section

The mud shale in Dongyuemiao section of the Lower Jurassic Ziliujing Formation in the target formation has a stable lateral distribution and a large thickness, and the dark mud shale is about 60 ~ 100 m thick. The lithology of the reservoir is mainly mud shale, with common gray silty shale, middle clastic shale and middle clastic limestone interlayer (Figure 1). The reservoir minerals are mainly clay minerals, quartz and calcite (average contents are 22.49%, 55.95% and 65,438+07.5% respectively), with a small amount of feldspar and pyrite. The existence of authigenic minerals indicates that the sedimentary environment of Dongyuemiao section is a reducing environment conducive to the enrichment and preservation of organic matter. The results of laboratory analysis show that kerogen ⅱ is the main organic carbon in the reservoir, with an average content of 2% ~ 3%. The pore structure of the reservoir is dominated by mineral intergranular pores, while a small number of intragranular pores and dissolution fractures are developed, and a large number of nano-pores are generated by pyrolysis of organic matter, which makes the reservoir have good natural gas adsorption and storage performance.

Figure 1 Typical lithology of shale in Dongyuemiao section of the study area

2 conventional logging response evaluation of reservoir rock composition

Logging response is the macroscopic expression of the physical properties of the measured formation [1]. Under the condition of excluding the influence of borehole and mud invasion, the logging response is essentially a comprehensive performance of the physical characteristics of all rock micro-components within the detection range of logging instruments, so all kinds of logging responses actually cover the petrophysical information of all components of the measured formation. Using the reservoir information contained in the conventional logging response to fully excavate and evaluate shale components provides a reservoir evaluation idea other than laboratory analysis and ECS logging, and at the same time makes up for the problems that the core laboratory analysis cannot be continuous in the whole interval and the ECS logging data collection, interpretation and evaluation cost is high [2, 3].

2. 1 conventional curve nonlinear joint optimization inversion algorithm

2. 1. 1 objective function

Different from the method of establishing the functional relationship between a single or a few logging curves and a certain mineral content in the reservoir to evaluate its content, the method and steps of evaluating the mineral composition of the reservoir by nonlinear joint optimization inversion using conventional logging information can be briefly summarized as follows: firstly, the measured response needs to be preprocessed to obtain a corrected logging response close to the real physical properties of the original reservoir; Secondly, according to the preliminary understanding obtained from core observation and conventional evaluation results, the types of rock components existing in this interval are delineated, explained and evaluated, and their initial contents are determined, forming a complete physical volume model of reservoir rock based on the original hypothesis; Thirdly, according to the regional experience or theoretical parameters, reasonably select the logging response skeleton values of each component, forward the conventional logging responses with the nonlinear logging response equation, and calculate the objective function T (XJ) about the calibration curve and the forward simulation curve, as shown in equation (1); Finally, the objective function T(X) is minimized by repeatedly adjusting the content of each mineral component, and the rock composition and content model at this time is taken as the final result of inversion, that is, the problem of rock composition and content in complex mineral reservoirs is solved by solving the optimization problem shown in Figure 2 [4].

Oil and gas accumulation theory and exploration and development technology (5)

Fig. 2 Schematic diagram of nonlinear joint optimization inversion algorithm

Where: loggings is the positive curve group generated after the j-th iteration; Loggingc is a calibration curve group generated by calibrating the measured curve; Xj is the content of each rock component determined in the j-th iteration; W is the weight of each logging curve in the objective function; α is an iterative stability control parameter; T(Xj) is an objective function reflecting the similarity between the forward curve and the correction curve. When the function reaches the minimum value, it means that the forward curve is close to the correction curve. At this time, it can be considered that the rock composition and content obtained by the model are closest to the real situation of the stratum. It should be noted that the multi-solution of inversion algorithm can be reduced to a greater extent by using more abundant logging response information and the preliminary understanding of strata obtained from core analysis and conventional reservoir evaluation.

2. 1.2 *** yoke gradient optimization algorithm

From the above analysis, it can be seen that solving the logging evaluation problem of shale complex rock components has been transformed into solving the optimization problem of the minimum value of the objective function T(Xj). In this study, the objective function belongs to multivariate function, and the nonlinear relationship of logging response determines the nonlinear characteristics of the objective function, so the yoke gradient method is used to solve the optimization problem of the objective function [5].

For the objective function T(Xj), Taylor expansion is carried out at the extreme point X*. When the effective term is ignored, there are

Oil and gas accumulation theory and exploration and development technology (5)

Where: h = ▽ 2T(X *) is the second-order partial derivative matrix of T(X) at X *. Because X* is the extreme point, so ▽ t (x *) = 0, so

Oil and gas accumulation theory and exploration and development technology (5)

It can be seen that the function T(X) of any degree has the characteristics of quadratic function near its extreme point. Let T(X) be expressed as a quadratic function as shown below.

Oil and gas accumulation theory and exploration and development technology (5)

It can be proved that an n-dimensional quadratic function with an n-order positive definite matrix A can find at most n * * * yoke directions (vectors) about A in the n-dimensional space. Starting from any initial point, one-dimensional search along these n * * * yoke directions for no more than n times can find the minimal point of the objective function T(X) in the n-dimensional space. By using the above-mentioned * * * yoke gradient algorithm, the huge amount of calculation caused by Newton's method and its improved algorithm which need to calculate the inverse matrix of the second-order partial derivative matrix is avoided, and the shortcoming that the steepest descent method converges slowly when approaching the minimum point is overcome, and the problem of solving the nonlinear inversion algorithm established by the research is properly solved.

2.2 Inversion of Shale Composition in Dongyuemiao Profile

2.2. 1 initial model assumptions

Figure 3 shows the conventional logging response of shale in Dongyuemiao section of a well in the study area. The purpose of rock composition evaluation of gas-bearing shale in this well is to clarify the specific content of important rock components including organic carbon. The establishment of initial model hypothesis needs to determine the reservoir rock composition to be solved and its initial content, as well as the logging curve involved in rock composition evaluation.

According to the laboratory whole rock analysis results described in the previous section (Figure 3), the initial model assumes that there is no other solid organic carbon except kerogen in shale; Brittle minerals include quartz, calcite and feldspar, and plastic minerals are clay. In addition, related research shows that spontaneous pyrite often crystallizes between layered interfaces of reservoirs during shale diagenesis, which is beneficial to hydraulic fracturing to form network fractures to some extent. Pyrite has excellent conductivity, extremely high photoelectric capture cross-section index and high density, even if the content is small, it has obvious influence on logging response such as resistivity, photoelectric cross-section index and volume density. Therefore, pyrite, as an important mineral affecting the mechanical properties and petrophysical properties of shale, can not be ignored in the rock composition model. Finally, due to the high clay mineral content, low effective porosity and low salinity of formation water, free water has little effect on logging response, so the model only considers the existence of clay bound water and assumes that all effective pores of shale reservoir are occupied by free gas. Based on the above considerations, it is finally determined that the shale in Dongyuemiao section of this well needs inversion calculation, as shown in Figure 4, which includes clay (including clay bound water), calcite, feldspar, pyrite, pore (free gas) and organic carbon (kerogen) in turn.

Fig. 3 conventional logging response characteristics and core analysis results of shale in dongyuemiao section of a well in Jiannan area

Fig. 4 shale volume model

By comprehensively investigating the number of logging curves that can be referenced by this well and the types of components that need to be covered by the above shale volume model, it is determined to use photoelectric cross section index (PEF), natural gamma (GR), neutron porosity (NPHI), bulk density (DEN), acoustic time difference (DT), shallow lateral resistivity (LLS), deep lateral resistivity (LLD), uranium (URAN) and thorium. It can be seen that if the underdetermined solution is not considered, the number of logging curves participating in nonlinear inversion can theoretically solve the problem of 10 rock component content at most, which is greater than the number of rock components solved by the model, so the model solution belongs to nonlinear overdetermined solution, which can effectively reduce the multiplicity of evaluation results and ensure that the evaluation results are closer to the real situation of shale gas reservoirs.

By means of conventional reservoir evaluation methods, such as natural gamma argillaceous content Vsb evaluation method [6], density neutron porosity Phi evaluation method [7] and Pessay organic carbon TOC content evaluation method [8], the preliminary evaluation results of clay, porosity and kerogen content can be obtained, which can be attributed to the initial content of rock components in this well. The initial content of the remaining components-timely content is distributed in proportion according to the average content (timely 55.9%, calcite 17.5%, feldspar 7.2% and pyrite 4.5%) determined by core analysis in the laboratory. The results are shown in solid lines in lanes 2-8 of fig. 5. It can be noted that the initial content of each rock component (solid brown line) is different from the core analysis result (black dot). Among them, the content deviation of calcite and feldspar is the most obvious; The porosity of neutron-density porosity evaluation is also obviously higher. In addition, the clay mineral content calculated by natural gamma argillaceous content evaluation method and the organic carbon content calculated by resistivity-acoustic wave overlapping Parsi method still have some errors in local depth. In this study, these errors will be gradually reduced through the subsequent inversion calculation, so as to obtain the evaluation result closest to the real formation rock composition.

Fig. 5 Initial nonlinear inversion model of shale in Dongyuemiao section of a well in Jiannan area.

2.2.2 Model inversion results

After nonlinear inversion calculation, the rock composition content of this well is finally determined, as shown in Figure 6. The second to eighth lanes in the figure are the evaluation results of clay minerals (including clay bound water), time, calcite, feldspar, pyrite, porosity and organic carbon content (solid lines) and the laboratory analysis results of corresponding components (black dots). Track 9 and 10 in the figure are the nonlinear inversion results of shale composition and the laboratory analysis results of cores, respectively.

Fig. 6 Nonlinear inversion results of shale rock composition in Dongyuemiao section of a well in Jiannan area.

Through the comparative analysis of the initial evaluation results (black squares) of each component in Figure 7 and the nonlinear inversion calculation results (triangles), it can be found that the nonlinear inversion results have better linear correlation with the laboratory analysis results, and are more concentrated near the 45 diagonal than the initial evaluation results. As can be seen from Figures 6 and 7, the nonlinear inversion algorithm significantly improves the evaluation accuracy of time and calcite content; The porosity evaluation results are closer to the laboratory analysis results; In addition, the local errors of clay minerals and organic carbon content have also been well corrected; In the initial model, the rough estimation of feldspar and pyrite content based on the average content has also been further refined here, and the evaluation results are more consistent with the laboratory analysis results in the overall trend. So far, this study has completed the logging evaluation of complex mineral composition and organic carbon content of shale gas reservoir in Dongyuemiao section of the study area by using the established nonlinear inversion method, and achieved high evaluation accuracy. This study will further quantitatively analyze the logging evaluation results to verify the reliability and effectiveness of this method.

2.2.3 Analysis of nonlinear inversion results

Considering that the essence of each measured logging response is the macroscopic physical characteristics of the measured reservoir rock composition in various physical fields, in order to verify the reliability and effectiveness of the nonlinear inversion algorithm and its inversion results, this study also analyzes the nonlinear inversion results and simulates the logging response error under the inversion results.

Fig. 8 shows the simulated logging response (dotted line) under the nonlinear inversion result and the logging response after environmental correction (solid black line). The logging items involved are natural gamma GR, uranium uran, thorium th, neutron porosity Nphi, bulk DENsity den, macro profile index u, acoustic time difference DT, flushed zone conductivity CXO and undisturbed formation conductivity CT. From the comparison of two sets of logging responses, the simulated logging response under nonlinear inversion results is in good agreement with the measured logging response. In table 1, the correlation coefficient between two sets of logging responses is quantitatively evaluated and analyzed. The correlation coefficient of each logging response is between 0.867 and 0.996, and the average correlation coefficient reaches 0.92 1, which fully reflects the similarity between the macroscopic physical properties of rock components and the real reservoir physical properties, that is to say, the rock components and their contents obtained by nonlinear inversion are very close to the actual situation of shale gas reservoirs. In addition, based on the laboratory analysis results, Table 2 statistically analyzes the correlation coefficient between the initial evaluation results and the nonlinear inversion results in Figure 8. The comparison of two groups of correlation coefficients shows that the nonlinear inversion algorithm established in this study obviously improves the accuracy of shale composition evaluation. Therefore, the above two aspects fully prove the reliability and effectiveness of the nonlinear inversion algorithm established in this study in solving the problem of complex rock composition and content evaluation of shale reservoirs.

This method can solve two major problems in the evaluation of shale gas reservoir rock mineral composition and organic carbon content at the same time. The successful solution of these two problems provides scientific basis and technical guarantee for the subsequent evaluation of important reservoir parameters such as reservoir brittleness and adsorbed gas content.

Fig. 7 Comparison between initial evaluation results of shale components and nonlinear inversion calculation results

Table 1 Correlation coefficient between simulated logging response and measured response

Fig. 8 inversion quality control of shale complex rock components in dongyuemiao section of a well in Jiannan area

Table 2 Correlation comparison between initial evaluation and nonlinear inversion evaluation and core analysis results

3 Conclusion

In this study, the logging evaluation method of shale gas reservoir rock composition based on nonlinear inversion optimization algorithm has achieved good evaluation results in Dongyuemiao section of Jiannan structure in western Hubei and Yudong area. This method fully excavates the rich geological information contained in conventional logging data, solves the two major problems of shale reservoir evaluation with important minerals and organic carbon content, makes up for the shortcomings of discontinuous core analysis depth and high ECS logging cost, greatly improves the accuracy of logging evaluation results, and provides scientific basis and important technical support for subsequent comprehensive evaluation of reservoir brittleness and gas-bearing property.

refer to

Hong [1]. Logging principle and comprehensive interpretation. Dongying: Petroleum University Press, 1993: 1 ~ 365.

[2]Pemper R R, Sommer A, Guo P, et al. A new pulsed neutron detector for deducing formation lithology and mineralogy. SPE annual technical conference and exhibition in San Antonio, Texas, USA, 2006: 1~ 13.

[3]Sondergeld C H, Newsham K E, Comisky J T, et al., Petrophysical Factors of Evaluating and Exploiting Shale Gas Resources. SPE Unconventional Natural Gas Conference, Pittsburgh, Pennsylvania, USA, 20 10: 1~34.

[4]Heidari Z, Torres-Verd C, Preeg W.E. Quantitative method for estimating total organic carbon and porosity and diagnosing mineral composition from shale gas formation logging data.2011:1~15.

Forever and ever. Optimize logging interpretation. Dongying: Petroleum University Press,1995: 87 ~119.

Pan, Wu Yuan,. Preliminary study on geochemical indexes of shale gas exploration and logging analysis methods. Petroleum Exploration in China. 2009, 3: 6 ~ 28.

, river, edge, etc. Evaluation method system of shale gas resources and its application--taking west sichuan depression as an example. Natural gas industry. 29, 2009 (12):130 ~134.

[8] Percy ·Q·R, Criner S, Kula ·J·B, etc. Practical model of organic abundance in porosity and resistivity logging. AAPG Bulletin.1990,74:1777 ~1794.