1. Introduction
Lacustrine carbonate reservoirs are extensive in hydrocarbon-bearing basins in China. Such reservoirs have tremendous hydrocarbon resource potential, which has been proven by numerous recent oil–gas discoveries [
1,
2,
3,
4,
5,
6,
7,
8,
9,
10,
11,
12,
13,
14]. In the Mahu area and the Jimusar area of the Junggar Basin, several oilfields were discovered in the Permian strata, with more than 230 million tons oil of proven reserves. In the Central Sichuan region of the Sichuan Basin, 86 million tons oil of proven reserves were discovered in the Jurassic strata. Additionally, in the Kuqa Depression of the Tarim Basin, more than 24 billion cubic meters of proven gas reserves were discovered in the Paleogene strata. However, exploration practice demonstrates that these carbonate reservoirs are mostly characterized by complex fabrics, high heterogeneity, and high terrigenous detrital content, which leads to high difficulty in lithofacies identification and constraints on the exploration up-scale and efficient development. The reservoir formation mechanism and distribution law need to be further investigated.
At present, the methodology of identifying lithology or lithofacies of carbonate rocks using well logs mainly includes the qualitative analysis of curve characteristics, crossplot method, apparent rock fabric number (ARFN) method, and relevant mathematical analysis methods [
15,
16,
17,
18,
19,
20,
21,
22,
23,
24]. Qualitative analyses identify lithofacies according to well log response features, curve shape, values of well logs, and imaging characteristics of different rocks [
15,
16]. This method can only describe lithology in a rather coarse way. The crossplot method is used to qualitatively distinguish lithofacies by crossplotting the pairs of natural gamma ray, interval transit time, density, and neutron and photoelectric absorption cross-section index [
17]. This method features simple operations but cannot accurately identify lithofacies with complex mineral composition and diverse rock fabrics. The ARFN method, based on both cores and well logs, develops different formation models and calculation formulas, according to well logging characteristics [
18]. This method delivers satisfactory application performance in marine carbonate rocks with uncomplicated mineral composition and high porosity (more than 15%), yet inferior performance in lacustrine carbonate rocks with high terrigenous detrital content.
Furthermore, there are also some lithofacies or lithology identification methods based on machine-learning techniques and mathematical algorithms, such as neural network [
19], cluster analysis [
20], principal component analysis [
21], grey correlation [
22], support vector machine [
23], and Naive Bayesian method [
24]. The key to these methods is to build a well-trained and suitable model which can be applied to well log measurements from wells with poor core data to predict lithofacies.
This research, based on data such as cores, thin sections, and well logs, systematically investigates the lithofacies and well log response characteristics of carbonate reservoirs in the Yingxi area. Moreover, a well logging lithofacies identification method based on the rock fabric factor (RFF) is developed. The findings of this research are of great significance for the favorable reservoir distribution prediction of well deployment and reservoir evaluation.
2. Geological Background
The Yingxi area is located at the western margin of the Yingxiongling structural belt within the Qaidam Basin (
Figure 1a). The area contains complete Cenozoic sedimentary sequences, which are, from bottom to top, the Paleogene Lulehe Formation (E
1+2) and Lower Ganchaigou Formation (E
3), Neogene Upper Ganchaigou Formation (N
1), Lower Youshashan Formation (N
21), Upper Youshashan Formation (N
22) and Shizigou Formation (N
23), and the Quaternary Qigequan Formation (Q
1+2) (
Figure 1b). In the Upper Member of the Lower Ganchaigou Formation (E
32), a set of lacustrine carbonate rocks has developed, including five lithofacies, which are pack-wackestone, mudstone, laminated carbonate, muddy gypsum, and limy claystone. During the period of decreasing lake level, the suspended components of terrigenous mud and silty sand are high, and they are rapidly mixed with calcite precipitated by evaporation and concentration to form limy claystone. Later, under the quiet water environment of decreasing terrigenous debris, seasonal laminae of calcite and clay are formed, which form laminated carbonate after compaction. Then, with the enhancement in evaporation and concentration, the lake level decreases and salinity rises continuously. In this period, carbonate and sulfate successively enter the oversaturation state and precipitate, and then pack-wackestone, mudstone, and muddy gypsum are deposited upon laminated carbonate. Due to the frequent changes in sediment supply and water desalination, the complete sedimentary process mentioned above is often interrupted and enters a new cycle.
During the deposition of the Upper Member of the Lower Ganchaigou Formation (E
32), the area, on an overall basis, successively experiences three complete sedimentary evolution stages of the saline lacustrine basin, namely, semi-salinization, salinization, and salt lake, under the arid paleo-climate [
25,
26,
27,
28]. During the semi-salinization stage, the sediment supply is sufficient, the salinity is low, the lake level is relatively high, and the deposited sediments are mainly limy claystone, based on well core observations. This stage is the development stage of the main source rock in the area. During the salinization stage, the salinity is generally high, due to the intermittent sediment supply and intensive evaporation; the lake level is relatively low and frequently rises and falls. This stage is associated with the main development stage of mixed carbonate reservoirs. The last stage is the salt lake stage, which features the lowest sediment supply and intensive evaporation. Multiple sedimentary cycles are developed, in which the lake evolves into a dry salt lake via evaporation. The deposition is predominated by evaporite deposits and some carbonate reservoirs. This stage is the development stage of the regional high-quality cap rock (
Figure 1c).
3. Materials and Methods
This research focused on the Upper Member of Lower Ganchaigou Formation (E32) and performed a comprehensive analysis of well logs, cores, thin sections, and nuclear magnetic resonance (NMR) data. Log data from 55 wells and thin sections from 78 samples were analyzed. Cores were collected and observed from 14 wells, with a cumulative length of about 300 m. The porosity data from 96 samples, X-ray diffraction (XRD) data from 96 samples, and NMR data from 9 samples were studied. The thin section pictures were taken from 83 samples under conditions of both plane-polarized light and cross-polarized light. All data were offered by the PetroChina Qinghai Oilfield Company.
Geophysical well logs are measurements of rock physical properties at given depths in well points. For each well, four parameters measured by well logging are incorporated:
Gamma ray (GR)—a measure of the total natural radioactivity emitted by strata;
Bulk density (ρb)—the bulk density of rocks measured from the electron density in strata;
Matrix density (ρma)—calculated from rock mineral composition measured by high-definition spectroscopy log (Litho Scanner);
Mineral content—the content of several major minerals in strata, calculated from Litho Scanner logs, including siliceous minerals (quartz, feldspar, mica), calcite, dolomite, and anhydrite. Litho Scanner can provide a detailed description of complex reservoirs by measuring both the inelastic and capture spectra, which can be converted to elemental weight fractions via a modified geochemical oxides closure model. Then, lithologic fractions are generated by an iterative inversion technique, such as the Quanti.Elan multicomponent inversion solver in the Techlog 2018 software.
Geophysical well logs represent single-point measurements of rock physical properties with depth recorded in a well.
The analysis of well log response characteristics suggested that natural gamma ray, matrix density, and bulk density were sensitive to lithofacies. Then, for the first time, the rock fabric factor (RFF) method was proposed, and the lithofacies identification plot was based on the calculated RFF and high-definition spectroscopy log. The mathematical expression of the rock fabric factor (RFF) was developed using the natural gamma ray value, bulk density, and matrix density to identify lithofacies (Equation (1)):
where
RFF is the rock fabric factor;
GRmax is the maximum value of natural gamma ray of the target interval (API);
GR is the value of the natural gamma ray (API);
ρma is the matrix density (g/cm
3); and
ρb is the value of the bulk density well log (g/cm
3).
In this study, all of the log data were measured by a MAXIS 500 system, and Geolog, Techlog software was used to process log data and calculate RFF. Thin section pictures were captured by a Leica DM6 M under both cross-polarized light and plane-polarized light. The porosity data were obtained from a UPore 300 Instrument, and the core sample size was 1 inch in diameter and 3 inches in length. The NMR data were obtained from a RECCORE-3010; the measurement conditions were echo spacing of 0.1 millisecond, waiting time of 2 s, and scanning time of 64 times, and the sample size was 1 inch in diameter and 2 inches in length. The XRD data were obtained from an Empyrean X-ray Diffractometer and the sample specification was 10 g of powder.
4. Results
4.1. Lithofacies
The lacustrine carbonate rocks in the Yingxi area contained terrigenous debris, carbonate minerals, and evaporate minerals, indicating mixed deposition. The observation of thin sections from 78 samples showed that the lacustrine carbonate rocks had different rock structure characteristics, mineral composition, and formation environments. Given this, the petrographic division of the carbonate reservoirs of the Yingxi area was performed, and five lithofacies were identified, which were pack-wackestone, mudstone, laminated carbonate, muddy gypsum, and limy claystone. The pack-wackestone and mudstone comprised the main reservoirs, and the laminated carbonate was the high-quality source rock. Characteristics of all lithofacies are presented below:
(a) Pack-wackestone. This lithofacies develops at the top of the underwater uplift and the upper part of the slope, which is of the shallow-water and high-energy beach deposition, and cross-bedding is observed on the cores. Under the microscope, it presents a sparry granular texture; the grains are mainly fine-silty carbonate debris; in the absence of anhydrite, it develops low-angle cross-bedding, which is destructed by the presence of anhydrite and changed into a porphyritic texture. The debris sometimes forms superficial ooids (
Figure 2a,b);
(b) Mudstone. This lithofacies occurs in the middle of the slope and is of the shallow-water, low-energy, and stable deposition. The cores show a relatively homogeneous massive texture. The composition is dominated by micritic carbonates, followed by clays. In the absence of anhydrite, horizontal bedding is observed, which is destructed by the presence of anhydrite and changed into a porphyritic texture. The thin section microscopy shows a micritic texture formed by a mixture of carbonates, terrigenous debris, and clays (
Figure 2c);
(c) Laminated carbonate. Its sedimentary environment is relatively deep and calm water with rhythmic supply. The cores have shale-like texture characteristics, and typical laminated features are found under the microscope. This lithofacies is mainly composed of muddy-silty terrigenous debris lamination and pure carbonate lamination, which stack over each other. The pure carbonate lamination is composed of anhedral calcite grains, with grain sizes generally of 0.03–0.06 mm (
Figure 2d);
(d) Muddy gypsum. This lithofacies is of the shallow-water evaporate deposition and mainly consists of anhydrite, followed by clays and carbonate minerals. Anhydrite is found to be snowflake or layered at the core scale, and snowflake or porphyritic under the microscope (
Figure 2e);
(e) Limy claystone. It is developed in the deep sub-sag and is of the deep-water deposition with rapid terrigenous supply. Its composition is mainly clays, followed by micritic carbonates. With no notable horizontal bedding and some porphyritic anhydrite, this lithofacies has a blocky texture (
Figure 2f).
4.2. Well Log Response Characteristics
Due to different mineral composition and rock fabrics, different lithofacies have varied well log responses. The natural gamma ray can reflect the change of clay mineral content. Bulk density and matrix density can represent characteristics of the pore development and rock matrix, respectively. The well log response characteristics shown in
Figure 3 demonstrate that the logs of natural gamma ray, bulk density, and matrix density are sensitive to the variation in lithofacies and, thus, can be used to identify lithofacies. The reason for this phenomenon is that the sedimentary environment of the five lithofacies was different, resulting in differences in their mineral composition and porosity (
Table 1). Compared to mudstone, the sedimentary environment of pack-wackestone exhibits stronger hydrodynamic forces, allowing for more thorough elutriation of terrigenous debris and resulting in higher carbonate content. The pores of the reservoir in the study area were mainly from the intercrystalline pores of carbonate minerals, so the porosity of pack-wackestone was higher than that of mudstone. The laminated carbonate was affected by the seasonal and frequent changes within the sedimentary environment, resulting in the content of carbonate minerals being basically equal to that of terrigenous debris, and the porosity was also smaller than that of pack-wackestone and mudstone. Muddy gypsum had a high content of anhydrite due to its high salinity during sedimentation. These features can be reflected in natural gamma ray, bulk density, and matrix density curves. Natural gamma ray, bulk density, and matrix density can, respectively, reflect changes in terrigenous debris content, porosity, and mineral composition. Therefore, the above three logging curves can be used to identify lithofacies.
The results calculated via Litho Scanner logs (
Figure 4a,b) and XRD measurements (
Table 1) can adequately capture the differences in mineral composition between different lithofacies. The carbonate mineral content of pack-wackestone was the highest. The mudstone had the second highest carbonate mineral content, and, in some cases, contained a small amount of anhydrite. The contents of terrigenous debris and carbonates in laminated carbonate were comparable with each other. The limy claystone presented the highest clay mineral content. The muddy gypsum contained a large amount of anhydrite.
Based on sensitive parameters such as natural gamma ray, bulk density, and matrix density, the typical characteristic chart of well log responses was developed for the five lithofacies in the Yingxi area (
Figure 5). Moreover, the refined lithofacies identification was performed, based on these sensitive parameters.
4.3. Well Logging Lithofacies Identification Method Based on Rock Fabrics
In view of the mixed deposition characteristics and complex well log responses of lacustrine carbonate rocks in the Yingxi area, the focus of lithofacies evaluation was shifted from the calculation of mineral composition and content to the identification of rock fabrics, and the rock fabric factor (RFF) method was developed for well logging lithofacies identification.
Analysis of core data and well logs revealed the following three characteristics of lacustrine carbonate rocks in the study area:
(1) Higher radioactivity is associated with finer grains and higher clay mineral content.
Among the five lithofacies developed in the study area, the clay mineral content of the limy claystone was the highest, associated with the finest grains and the highest natural gamma ray value. The clay mineral contents of the laminated carbonate, mudstone, and pack-wackestone were lower than that of the limy claystone. Moreover, the clay mineral content of laminated carbonate, mudstone, and pack-wackestone declined successively, while their grain sizes grew and natural gamma ray values reduced. In the presence of anhydrite, the clay mineral content and natural gamma ray value of muddy gypsum were low, just slightly higher than those of pack-wackestone. In view of lithofacies variation, natural gamma ray values and clay mineral content are positively correlated with each other (
Figure 6).
(2) Higher clay mineral content represents smaller porosity.
Clay minerals usually fill pores among grains in the form of interstitial fillings, which reduces the porosity of rocks. Therefore, the higher the content of clay minerals, the smaller the porosity of rocks. The measurements of XRD and porosity showed that the clay mineral content and porosity of lacustrine carbonates in the study area were negatively correlated with each other, as lithofacies changed (
Figure 7).
(3) Larger grains stand for better pore structure.
The grains gradually became finer, from pack-wackestone to mudstone and limy claystone, while the total porosity gradually decreased. The NMR data of cores revealed that, with coarser grains, the T2 spectrum energy grows, and the spectrum peak gradually shifts rightward, which indicates that the pore radius gradually becomes larger, the pore structure is improved, and the porosity rises (
Figure 8).
Based on the above three characteristics, it is considered that the clay mineral content, porosity, and grain size can represent the change in lithofacies in the study area. These three parameters can be obtained from well logs—natural gamma ray reflects the change in clay mineral content and grain size, and the bulk density and matrix density capture the change in porosity. Therefore, the mathematical expression of the rock fabric factor (RFF) was developed by using the natural gamma ray value, bulk density, and matrix density to identify lithofacies (Equation (1)). Strata consist of the rock matrix and pore fluids. The average density of the rock matrix is referred to as the matrix density. A high-definition spectroscopy logging tool can measure the content of main minerals in strata. Because each mineral corresponds to a constant density, the matrix density of strata can be calculated according to the type and content of minerals in the strata (Equation (2)):
where
ρma is the matrix density, g/cm
3;
m is the mineral component in strata;
Mi is the mass fraction of the
i-th mineral component in strata; and
ρi is the density corresponding to the
i-th mineral component.
By combining the high-definition spectroscopy logs, the crossplots of RFF versus natural gamma ray and carbonate minerals versus terrigenous debris were prepared (
Figure 9). The lithofacies boundaries in the two charts were determined by the distribution of each lithofacies data points, but data points of different lithofacies may overlap in the charts. For example, in the RFF–natural gamma ray crossplot, some overlap was found between the laminated carbonate and muddy gypsum; in the carbonate minerals–terrigenous debris crossplot, the limy claystone and laminated carbonate also had some overlap. The reason for this overlap is the similarity in log response characteristics and mineral components of some lithofacies. A portion of muddy gypsum with high clay content has similar natural gamma ray values to laminated carbonate, resulting in the overlap of these two lithofacies in the RFF–natural gamma ray crossplot. Furthermore, a portion of limy claystone with high carbonate content has similar mineral composition to laminated carbonate, resulting in the overlap of these two lithofacies in the carbonate minerals–terrigenous debris crossplot. Nonetheless, the five lithofacies can be distinguished from each other using the two crossplots simultaneously. For example, in the RFF–natural gamma ray crossplot, the laminated carbonate and muddy gypsum are seen with considerable overlap, but they are well separated from each other in the carbonate minerals–terrigenous debris crossplot; therefore, the laminated carbonate and muddy gypsum can be distinguished from each other.
5. Discussion
The lithofacies identification results of Well S43 are shown in
Figure 10. The first track is the natural gamma ray log; the second, the well depth; the third, the matrix density and bulk density logs; the fourth, the RFF; the fifth, the core profile; and the sixth, the lithofacies profile identified using well logs. The cored section of Well S43 was 19.8 m long and included the pack-wackestone, mudstone, muddy gypsum rock, and limy claystone. The lithofacies identification method based on the RFF delivered satisfactory identification performance, except for thin layers (less than 0.2 m thick), due to the limited vertical resolutions of well logs. The interpretation accuracy of Well S43 was 87%. The methodology was applied to 55 wells in the study area, and the average accuracy of lithofacies interpretation in 14 cored wells reached 82.4% (
Figure 11), which demonstrates good application performance.
In this study, the RFF, which integrates various well logs, reveals the relationship between different lithofacies and well logs, and helps to accurately identify lithofacies of mixed lacustrine carbonate rocks, thus providing suggestions for exploration plans. Moreover, well logs are common data in any oilfield, and such a complex lithofacies identification method based on well logs may show greater potential, and also show its importance for carbon capture and storage projects in the future.
Although the method proposed in this paper has improved the accuracy of lithofacies identification, there are still some shortcomings in this method, such as the overlap problem and lithofacies boundaries influenced by human factors in identification charts. It can be noted that the overlapping areas in the two charts are related to laminated carbonate, which indicates that it is difficult to accurately identify laminated carbonate. The full bore formation microimager (FMI) log may be a solution to this problem. With 80% borehole coverage in 8 in boreholes and 0.2 in image resolution in the vertical and azimuthal directions, imaging with the FMI microimager is an approach for determining lithofacies in laminated sediments and mixed depositional environments. We will also suggest to operators that FMI data may be helpful to identify lithofacies, and, with the acquisition of FMI, they can determine whether FMI can improve the accuracy of lithofacies identification, especially for laminated carbonate.
The method can also be applied to identify lithofacies of other complex lacustrine carbonate rocks, such as the pre-salt lacustrine carbonate in the Santos Basin, Brazil. The tectonic evolution of the Santos Basin is related to the disintegration of the Gondwana continent and the expansion of the South Atlantic since the Mesozoic, with three stages of rifting, transition, and drift. Three ultra-thick sedimentary sequences were developed correspondingly, i.e., the rifting mega-sequence, transitional mega-sequence, and drift mega-sequence [
29]. The rifting mega-sequence includes the Picarras Formation, Itapema Formation, and Barra Velha Formation. The Picarras Formation is mainly composed of deep lacustrine mudstone and shale, and the lacustrine shale is one of the main hydrocarbon source rocks in the Santos Basin. The Itapema Formation is dominated by lacustrine shale, marl, and coquina. The coquina stratum is one of the main pre-salt reservoirs. The Barra Velha Formation is mainly composed of stromatolites and spherulites, which are the main pre-salt reservoirs. Differently from the lacustrine carbonate rocks in the Yingxi area, the pre-salt lacustrine carbonates in the Santos Basin have undergone complex superimposed modifications of volcanism, diagenesis, and tectonism, etc., resulting in strong reservoir heterogeneity. In addition, the accurate prediction of lithofacies is very challenging due to the high content of supercritical CO
2 [
30]. The idea of identifying complex lacustrine carbonate lithofacies proposed in this paper may be helpful to solve these problems of pre-salt lacustrine carbonate in the Santos Basin, and this work can also be applied to carbon capture and storage projects.
6. Conclusions
The E32 lacustrine carbonate reservoir in the Yingxi area is a key block for exploration and development. The following conclusions are drawn from this research:
(1) Due to the sedimentary environment, the lithofacies in the study area can be divided into five types, according to data from cores and thin sections. The resultant classification shows that the rock fabric and mineral composition are key factors affecting well log responses in the target reservoir;
(2) Based on the analysis of well log response characteristics, the rock fabric factor (RFF) is defined, based on natural gamma ray, bulk density, and matrix density. It is combined with high-definition spectroscopy log to effectively identify the five lithofacies in the study area;
(3) The lithofacies classification scheme and identification method presented in this research have explicit geological implications. They effectively bridge the geological characteristics and well log parameters, and improve the currently used correlation between well log parameters and conventional lithofacies classification. They can be applied to identify lithofacies in other complex lacustrine carbonate rocks such as pre-salt lacustrine carbonate in the Santos Basin.