Unreliable determination of in situ stress orientation by cross multipole array acoustic logging in fractured shale reservoirs

With the development of shale gas exploration in China, the use of conventional logging tools has been introduced, and cross multipole array acoustic logging tools have gradually been used to determine the stress orientation in shale. The direction of fast shear waves (FSWs) is generally parallel to the horizontal maximum principal compression stress (SHmax). However, the azimuth of FSWs is found to be parallel to the main strike (but not to the SHmax) direction of structural fractures in shale reservoirs. Outcrop and image logging data indicate that the natural fractures in this area strike NE‒SW. If the shear wave anisotropy is caused by only the stress around the borehole and the FSWs are known to be NE‒SW, SHmax should be parallel to NE‒SW; however, according to statistics of land movement in adjacent areas, anelastic strain recovery, earthquake focal mechanism, borehole breakouts, hydraulic fracturing data, deviated well data, and drilling‐induced fracture data in local regions, SHmax is oriented in the NW‒SE direction, and the directions of FSWs are generally parallel to the structural fracture direction. This contradiction indicates that the development of structural fractures may affect the orientation of FSWs. Therefore, it is not reliable to use XMAC (Cross‐Multipole Array Acoustilog) logging only to determine the direction of in situ stress in fractured shale reservoirs. In addition, the direction of the FSWs in the middle of thick mudstones is NW‒SE, which may represent accurate information about the in situ stress direction.


| INTRODUCTION
Rock complexity makes it impossible to accurately determine the value and direction of in situ stress; thus, we must address large variations and uncertainties in stress measurement results. Careful screening and diagnosis are needed in studies of the direction of in situ stress. [1][2][3][4][5][6] After nearly a century of development, a variety of comprehensive methods can be used to determine the stress state in reservoirs. [7][8][9][10] Stress evaluation has become an indispensable basic method for reservoir engineering [4][5][6] ; oil, gas, and coal exploration 8,[11][12][13][14][15] ; and oil and gas engineering design, drilling, and completion. [16][17][18][19] Stress evaluation also has important applications in the development of well pattern deployment, 4,6,20 prediction of reservoir fractures, [21][22][23][24] wellbore stability, 4,19,25 hydraulic fracturing, [26][27][28][29] and sand control in oil and water wells. 30, 31 Winterstein and Meadows 32 first determined subsurface stresses using shear wave polarization in the Lost Hills field. Sinha 33,34 and Norris 35 developed the theory of borehole acoustic elasticity and a method for inverting in situ stress using cross-dipole data in the case of the stress concentration of borehole stress. Tang and Patterson 36 further reviewed the research progress of cross-dipole acoustic logging measurements using shear wave anisotropy and provided an overview of crossdipole logging and data interpretation in three major areas: (1) fracture delineation; (2) hydraulic fracture evaluation; and (3) formation stress diagnosis and orientation determination. Then, some inversion methods used to obtain the stress direction and stress profile of a stratum under different conditions were proposed. 37 Due to the potential of cross-dipole logging in anisotropy and in situ stress detection, the current practice is to generally combine cross-dipole instruments with unipolar acoustic waves, which also contain the functions of conventional array acoustic logging. With the development and application of dipole array acoustic logging tools, three professional well-logging service companies have introduced their own cross-dipole acoustic logging tools, e.g., Schlumberger's dipole array acoustic logging tool (DSI), Baker Hughes' multipole array acoustic (MAC, XMAC) logging tool, and Halliburton's dipole logging tool (WaveSonic).
XMAC logging has been indicated to be a reliable method to determine the direction and magnitude of in situ stress. 36 The principle of using XMAC logs to determine the horizontal principal stress is shear wave splitting. 38,39 Shear waves mainly propagate in two directions; fast shear waves (FSWs) propagate along the fracture direction and spread faster, whereas slow shear waves (SSWs) propagate perpendicular to the fracture direction and spread slower. The existence of shear wave splitting in the local layer shows that shear wave anisotropy is related to changes in the fast and SSW velocities; thus, shear wave splitting can be expressed by shear wave anisotropy. 39 If shear wave anisotropy is caused only by stress, and the directions of FSWs and S Hmax are the same, the magnitude of stress can also be determined by the acoustic interval transit time.
This study provides an example that shows that XMAC logging cannot accurately measure the direction of in situ stress in fractured shale reservoirs. Based on XMAC logs, earthquake focal mechanisms, land movement in surrounding areas, borehole breakouts, hydraulic fracturing, deviated wells, and drilling-induced fracture data, discrepancies between the S Hmax values derived using multiple methods in the lower Cambrian Niutitang Formation of the Cen'gong block were described. According to analyses of land movement in adjacent regions, borehole breakouts, hydraulic fracturing, deviated wells, and drilling-induced fracture data in local regions, S Hmax is oriented NW-SE, and the directions of FSWs are generally parallel to the natural fracture direction. In this case, other data, such as the earthquake focal mechanism, land movement, borehole breakouts, hydraulic fracturing data, deviated well data, and DIF data, should be considered. In fractured reservoirs, anisotropic rocks can force the measured stress direction to turn to the direction of structural fractures or bedding. Therefore, XMAC logging should not be used alone to determine the stress direction. It is necessary to determine the direction of in situ stress in combination with other methods, including the analysis of drilling-induced fractures (DIFs), hydraulic fracturing, and elastic strain recovery. In the middle of thick mudstones, the direction of the structural fractures is not consistent with the direction of FSWs, and the direction of FSWs is consistent with the direction of S Hmax ; the FSWs in the middle of thick mudstones (>8.0 m) probably document accurate information about the present in situ stress direction.

| Location and structure
The Cen'gong block is in the northeastern part of Guizhou Province in southern China. This block is located in the trough-like fold belt in the southeastern part of the Upper Yangtze Plate ( Figure 1A). The Tuyun movement in the Late Ordovician brought the research area up to the surface and began its weathering and denudation. During the Yanshanian period, the Cen'gong block experienced strong tectonic movement, NE folding, large-scale fault development, and the formation of the region's current geological structure and landscape pattern. During the Himalayan period, the research area was subjected to nearly NE-oriented compressional stress, and its faults and folds were further complicated ( Figure 1B). The South China plate is mainly affected by the northwest thrusting of the Philippine plate, which is an extrusion-shear tectonic environment with rightward northwesterly faults and leftward northeasterly faults or leftward reverse faults.

| Stratigraphy
The thickness of the Niutitang shale is between 45 and 100 m ( Figure 2). Based on the analysis of 87 core plug samples, the source of rich organic matter in the Niutitang Formation is mainly algae and other aquatic organisms in a low saltwater environment, and the types of organic matter are mainly types I and II 1 , with good hydrocarbon generation capacity. The organic carbon content of the Niutitang Formation is generally greater than 1%, and the thickness of the organic shale layer (TOC > 2.0%) is greater than 45 m, which provides a good material basis for hydrocarbon generation. The TOC of the whole Niutitang Formation section is 4.15%, of which the average TOC value in the upper section is 1.33%, the average TOC value in the middle section is 4.80%, and the average TOC value in the lower section is 4.61%. Shales with high TOC contents are located in the middle and lower members of the Niutitang Formation. 15,40-42

| Natural fractures
In this paper, based on data from outcrop observations, cores from three wells, full-hole microresistivity scan imaging logs (FMI) of the TX-1 well, 40 thin sections, and field emission scanning electron microscopy of 39 samples, fractures exist in the Niutitang Formation. Most of the fractures in these cores are mainly (77.9%) either filled or half-filled with calcite ( Figure 3). Unsealed fractures (3.9%) mainly exhibit low dip angles. Unsealed secondary pores and fractures are well developed in microfractures ( Figure 3C); the pores in these fractures may be related to dissolution, and the secondary fractures may be related to late tectonic movement and multiple fracture openings. Unsealed secondary pores and fractures are suspected to be main factors controlling production. 40,43,44 The outcrop and FMI log data from the TX-1 well show that the strikes of fractures are NE-SW (40°-50°), with some striking N-S and NW-SE ( Figure 4A); the dominant dip angle of the fractures is vertical (or subvertical) ( Figure 3A,B), with some lowdip-angle tectonic fractures ( Figure 4B).

| METHODS
When XMAC logging is used to determine the direction and magnitude of stress, it is necessary to differentiate between shear wave anisotropy caused by stress and that caused by other reasons, such as fractured or high-angle formations. 36,45,46 Some researchers 37 have proposed theoretical methods (e.g., strong changes in the position of shear wave velocity or the intensity of shear wave anisotropy) to identify such differences, but these methods do not apply to the study area. To obtain more realistic and accurate compressional wave, shear wave, and Stoneley wave time differences, correlation analysis is used to calculate the point with the highest correlation coefficient on the entire wave train by selecting appropriate filtering parameters, window length, step size, and other parameters to determine the maximum possible time difference for each corresponding wave train. The advantages of this algorithm are as follows: first, it is not necessary to assume any physical model or limit the arrival order of various mode waves; second, it is not necessary to detect the first wave, thereby avoiding the phenomenon of cycle jumping; third, a high-quality time difference can be obtained in situations such as poor wellbore conditions, and a more realistic and reliable acoustic time difference can be obtained through the processing of the calculation center. Dipole acoustic logging mainly uses acoustic full wave train data to extract P-wave, S-wave, F I G U R E 2 Stratigraphic and lithological system in the study area (modified from Wang et al. 40 and Liu et al. 43 ). and Stoneley wave moveouts; it mainly uses dipole S-waves to extract S-wave moveouts and calculate anisotropy; and rock mechanics parameters are calculated according to the extracted P-wave moveout data in combination with density, porosity, shale contents and other data.
Each borehole breakout, to some degree, is a mechanical experiment. Hence, the long axes of borehole breakouts are generally oriented perpendicular to the S Hmax direction. 5,6,47 DIFs occur when the stress concentrated around the borehole exceeds the tensile strength of the wellbore wall. 48 Hence, DIFs develop parallel to the  S Hmax direction. 48 Hydraulic fracturing is a commercial procedure for oil and gas stimulation and reservoir stress estimation in the petroleum industry, which can be derived from microearthquake monitoring and impression orientations. 5 Hydraulic fracturing is often adopted for deep in situ stress measurements in tight reservoirs, shales, and coal seams. 5,6 The basic principle of the anelastic strain recovery (ASR) method for measuring stress is that the rock has a viscoelastic property, that is, elasticity and viscosity exist at the same time. 49,50 The experiment was completed in the ASR in situ stress measurement laboratory of the Chinese Academy of Geological Sciences. When the core is taken out of the corer, the rock begins to unload, elastic strain recovery occurs immediately, and then inelastic strain recovery occurs slowly. According to the inelastic strain recovery obtained in different directions after the core is removed, the stress of the core in the crustal stress field can be deduced. [50][51][52] The direction of the in situ stress field can also be determined from the direction of the ground displacement measured by GPS.

| RESULTS
Through on-site construction, array acoustic data were collected using the XMAC imaging logging tool SonicScanner ( Figure 5A,B). The horizontal principal stress directions derived from XMAC logs in the TX-1 well are presented in Figures 5C and 6. The results of shear wave F I G U R E 5 In situ stress in the TX-1 well. In cores from 1760 to 1816 m, natural fractures exist in shale, except in a short section A, which has a small number of fractures. The rock mechanical parameters were determined by the method of Liu et al. 43 See Figure 1 for well locations. AAM, azimuthal anisotropy map; AHMPS, azimuth of the maximum horizontal principal stress; FWV, fast wavetrace; GC, gas content; SWV, slow wavetrace; WDST, window start; WEND, window end. anisotropy show that those shale layers are strongly anisotropic, and the stress tensor in the TX-1 well is determined by using monopole full wave and shear wave spitting in the azimuthal anisotropy map. The statistical results of FSWs derived from XMAC logging show that S Hmax and S Hmin in the TX-1 well at depths ranging from 1760 to 1816 m are approximately 40 MPa (2.24 MPa/ 100 m) and 30 MPa (1.64 MPa/100 m), respectively; according to the principle of determining stress by XMAC logging, the directions of S Hmax and S Hmin are NE-SW (30°-40°) and NW-SE (120°-130°), respectively.
Based on XMAC logs, the direction of FSWs is NE-SW (30°-40°) (Figures 5C and 6). However, the local stress direction obtained from XMAC logs is inconsistent with those obtained from borehole breakouts, deviated wells, drilling-induced fractures, hydraulic fractures and GPS data, but it is consistent with the directions of natural fractures ( Table 1). The azimuth of S Hmax measured by the ASR method in the ST-1 well has good consistency, indicating that the dominant azimuth of the horizontal maximum principal stress is NW to WNW, and WNW is the main direction (Figure 7). Well diameter analysis was performed to determine the spatial pattern of the mean S Hmax orientation around the borehole; the results, which are based on six-arm caliper log data, indicate that the main borehole breakout orientation is NNE-SSW (10°-20°). Therefore, according to the principle of in situ stress determined by borehole breakouts, the orientation of S Hmax is WNW-ESE (100°-110°) ( Figure 8). We cataloged the deviated well information at depths ranging from 55 to 1885 m in the TX-1 well (Figure 9). The results based on drilling data indicate that the deviated well orientation and mean dip angle are NW-SE (120°-150°) and 0.75°, respectively. Therefore, according to the principle of determining in situ stress based on deviated well information, the S Hmax orientation is NW-SE (120°-150°).
The image logging results show that the DIFs are mainly dominated by high-angle fractures. After the statistical analysis of the DIF trends in the AY-1 well, we found that the strikes of DIFs were mainly WNW-ESE (100°-120°) and that their dip angles were mainly vertical fractures. Thus, the S Hmax orientation was Note: In the table, "Local" represents the in situ stress in a local area or in a particular well, and "Regional" represents the direction of stress in the area. Based on the detection of seismic waves, we determined the direction of the P axes from natural earthquakes, which are generally parallel to the S Hmax orientation. Based on 23 focal mechanisms of earthquakes in adjacent regions, the regional S Hmax is oriented NW-SE ( Figure 12). 54 Using this database and data from Wang et al. 55 and Niu 56 we collected current stress indicators. Determining the in situ stress based on this technology yielded results ( Figure 13) that were consistent with the direction of S Hmax determined by other methods (Figures 8-12). The data obtained from borehole breakouts, hydraulic fracturing data, deviated well data, and DIF data can be used to determine the direction of S Hmax . These results are consistent with the regional stress derived from the earthquake focal mechanism and land movement in adjacent areas. It is, therefore, unreliable to determine the direction of S Hmax by using only XMAC logging.

| DISCUSSION
XMAC logging is mainly used to measure the shear wave anisotropy of formations. In fractured formations caused by tectonic stress or other geological F I G U R E 7 Principal stress pattern obtained by the ASR method. See Figure 1 for the well location (modified from Chen 53 ). ASR, anelastic strain recovery.   factors, the shear wave velocity usually shows azimuthal anisotropy. The shear wave splitting phenomenon exists not only in fractured formations but also in nonfractured formations with unbalanced crustal stress, that is, in formations with strong anisotropy; when the dip angle of the local layer is large and the layer is weathered, it also exhibits strong anisotropy. In a homogeneous formation, the differences in horizontal principal stress are some of the most important factors causing shear wave anisotropy, which can be used to determine the orientations of S Hmax and S hmin . 36,45,46 Other nonstress factors in rocks, such as fractured formations or high-angle formations, also influence shear wave propagation. Therefore, in shale reservoirs with high-density fractures, it is not reliable to use XMAC logging to determine the direction of in situ stress. In this paper, we use the natural fracture orientation obtained by image logging and the P-and S-wave orientations obtained by XMAC logs to analyze the unreliability of reservoir inhomogeneity to determine the direction of S Hmax (Figure 14). In the middle of thick mudstones ( Figure 14A-C), the direction of natural fractures is not consistent with the direction of FSWs, and the direction of FSWs is consistent with the direction of S Hmax ; the underdevelopment of natural fractures is suspected to be the main factor affecting the change in the direction of FSWs. In those layers, shear wave anisotropy is mainly caused by in situ stress, so the FSWs in the middle of thick mudstones (>8.0 m) most likely document accurate information about the present in situ stress direction ( Figure 15). When the thickness of mudstone is less than 8 m ( Figure 14D,E) OR there is shale ( Figure 14F) AND limestone ( Figure 14G) in thick mudstone, it is unreliable to determine the direction of S Hmax using FSWs. In addition, in thick sandstones (6.36 m), shales (16.25 m), argillaceous siltstones (11.09 m) and limestones (14.31 m), the orientation of FSWs also cannot be used to determine the direction of S Hmax ( Figure 14H-K).
The main factors leading to shear wave anisotropy include the imbalance of horizontal crustal stress, open fractures, faults, high-angle stratigraphic bedding, and the influence of elliptical boreholes. The azimuth of the FSW corresponds to the direction of the maximum horizontal principal stress or the strike of the fault/fracture and the strike of stratigraphic F I G U R E 11 Induced fractures from the impresser orientation in the ST-1 well. See Figure 1 for well locations.
F I G U R E 12 Strikes of P axes from natural shallow earthquakes in adjacent areas (N = 23, data from Dong et al. 54 ). The P axes are oriented NW-SE and indicate that S Hmax is oriented NW-SE. bedding. Natural fractures may affect the accuracy of many measurement methods, such as hydraulic fracturing and stress release. As shown in Figure 16, α is the angle between the natural fracture orientation and the shear wave orientation. When the density of natural fractures is less than 1.5 fracs/m, the natural fractures have little effect on FSW propagation. When the density of natural fractures is greater than 1.5 fracs/m, with increasing fracture density, α gradually approaches 0; that is, the orientation of FSWs tends to be consistent with the direction of fractures, and the density of fractures has a significant impact on FSWs ( Figure 16A). When the dip angle of the natural fracture is less than 40°, the natural fracture has little effect on the propagation direction of the FSW. When the dip angle of the natural fracture is greater than 40°, as the fracture dip angle increases, α gradually approaches 0; that is, the orientation of the FSW tends to be consistent with the direction of the fracture, and the larger the dip angle of the fracture is, the more significant the impact of the FSW ( Figure 16B).
During logging, middle-and high-angle fractures can cause changes in the radial wave impedance of the formation, thus causing the amplitude of the shear wave energy to attenuate and allowing the degree of attenuation to increase as the degree of fracture development increases. 36,46 Because of the development of secondary pores and fractures in sealed fractures ( Figure 3C), when the effective stress increases, the contribution of fractures to the permeability is still great due to the filler supporting effect, and natural fractures still play a crucial role in reservoir seepage and propagation of FSWs. 57,58 The azimuth of shear waves is chiefly controlled by structural fractures and deviates from S Hmax . The azimuths of structural fractures yield results that correspond well to the direction of FSWs. Therefore, structural fractures striking NE-SW cause the azimuth of FSWs to deviate from the direction of S Hmax (NW-SE) to become NE-SW-striking.
Nie et al. 59 reported that the use of in situ stress directions determined from array acoustic logging is consistent with the direction of FSWs in a shale reservoir ( Figure 17A). This may be because shale exhibits a low Young's modulus, high Poisson's ratios, low brittleness, and little to no natural fracture development. In contrast, in a fractured reservoir, the results must be revised when XMAC logs are used to determine the direction of S Hmax ( Figure 17B). In these layers, we must search for thick mudstones, which probably document accurate information about the present in situ stress direction ( Figure 14A-C). The direction of FSWs is mainly affected by crustal stress and natural fractures, so we believe that this method is also applicable to other rocks with similar natural fracture development patterns. The orientation of FSWs is mainly affected by in situ stress and natural fractures, so this method is also applicable to other fractured reservoirs.
F I G U R E 15 Rose diagram of the FSW azimuths in the middle of thick mudstones in the TX-1 well (N = 470). The FSW azimuth is NW-SE (140°-150°) and indicates that S Hmax is NW-SE (140°-150°), which is approximately consistent with the direction of S Hmax derived from borehole breakouts, hydraulic fracturing data, deviated well data, and DIF data. FSW, fast shear wave.
F I G U R E 16 (A) Relationship between the linear fracture density and α; (B) relationship between the fracture dip angle and α. α is the angle between the direction of natural fractures and the direction of FSWs. FSW, fast shear wave.

| CONCLUSIONS
The FSWs of local XMAC logs are oriented NE-SW in the Niutitang Formation. The hydraulic fractures observed using impression orientations and DIFs observed using FMI log data are NW-SE. The S Hmax from local XMAC logs is inconsistent with those derived from earthquakes and the orientations of hydraulic fractures and DIFs. The XMAC log data no longer accurately indicate the local stress direction of S Hmax .
Structural fractures are widely developed in shale reservoirs and influence the direction of FSWs. When fractures are well developed in shale, structural fractures cause the azimuth of FSWs to deviate from the direction of S Hmax to become parallel to the fracture strike. Therefore, determining the direction of S Hmax by using only XMAC logging in fractured shale reservoirs is unreliable. In this case, other data, such as earthquake focal mechanism data, land movement data, borehole breakouts, hydraulic fracturing data, deviated well data, and DIF data, should be considered.
When the thickness of mudstone is less than 8 m or there is shale and limestone in thick mudstone, it is unreliable to determine the stress direction using FSWs. In addition, in thick sandstones (6.36 m), shales (16.25 m), argillaceous siltstones (11.09 m) and limestones (14.31 m), the orientation of FSWs also cannot be used to determine the stress direction. In the middle of thick mudstones, the directions of natural fractures are not consistent with the directions of FSWs, and the directions of FSWs are consistent with the direction of horizontal stress; the FSWs in the middle of thick mudstones (>8.0 m) most likely document accurate information about the present in situ stress direction.