Rock core-based pre-stress evaluation experimental validation : A case study on Yutengping Sandstone as CO 2 storage reservoir rock

Yutengping Sandstone in Tieh-chan-shan, Taiwan is a potential reservoir for geological CO2 storage. Cyclic loadings were applied to rock samples taken from an outcrop to create artificial pre-stress. The pre-stress evaluation accuracies using two core-based techniques, acoustic emission (AE) and deformation rate analysis (DRA), were investigated under different pre-stresses, delay times and curing temperatures. The experimental results validate the pre-stress evaluations using AE and DRA. The delay time and curing temperature were shown to have minor impacts on the measurement accuracy. However, although both axial strain and lateral strain can be used in DRA, the stress memory fades as the delay time increases. Therefore, delay time, which represents the time from the borehole drilling to the DRA test, must be carefully considered when applying these techniques to evaluate the in situ stress of Yutengping sandstone. Article history: Received 2 February 2012 Revised 10 June 2012 Accepted 21 September 2015


IntROduCtIOn
The injection pressure and rate at a CO 2 injection site must be sufficiently high to accommodate the injection of a desired yearly mass of CO 2 .The ability of the reservoir rock to contain the injected CO 2 is determined by measuring artificial pressure exceeding the initial reservoir pressure.During CO 2 injection the increasing reservoir fluid pressure induces mechanical deformations and stresses in and around the injection formation.If the reservoir pressure becomes too high the induced stresses may create new cracks or reactivate existing faults (Streit and Hillis 2004;Cailly et al. 2005;Rutqvist et al. 2007).
Figure 1 shows the relationship between the failure criteria of intact/jointed rock and the Mohr circles of the in situ stresses.In the analytical shear-slip analysis when the fluid CO 2 is injected into the rock mass the increasing pore pressure moves the effective-stress Mohr-circle with an identical diameter to the left, approaching the failure criteria (Fig. 1a).However, if thermal-or poro-elastic stressing is assumed during CO 2 injection, the diameter of the effective-stress Mohr-circle increases while the effective stress decreases (Fig. 1b).Therefore, the in situ stress of the reservoir rock is an essential parameter for evaluating the injection pressure during CO 2 injection.
Although tectonic stresses in Taiwan, which can be regarded as far field stresses, have been investigated extensively using GPS observations and tectonic data (Chang et al. 2003;Kaus et al. 2009;Hsu et al. 2010), local geological structures at a potential CO 2 injection site govern the local stresses in the rock mass to a considerable degree (Goodman 1989;Wu et al. 2004;Do et al. 2017).Hydraulic fracturing (HF) and overcoring (OC) are conventional techniques used in Taiwan to measure in situ stresses (Fig. 2).However, Villaescusa et al. (2002) noted that HF and OC techniques are time-consuming, very expensive and require specialized personnel.Additionally, OC cannot be used where no underground access is available and HF generates fractures in Terr. Atmos. Ocean. Sci., Vol. 28, No. 3, 193-207, June 2017 the rock mass that may damage the sealing geologic structure of the CO 2 storage site.
Acoustic emission (AE) and deformation rate analysis (DRA) are new methods to measure in situ stresses using borehole drilling cores.When the compressive stress applied to the rock exceeds its historical maximum, the stress memory can be measured as AE (Goodman 1963) and the inelastic strain behaviour as DRA (Yamamoto et al. 1990).These techniques also carry the advantages of being inexpensive and simple.However, AE and DRA are not widely practiced because of the poor reliability of the measured in situ stress (Ljunggren et al. 2003).
To increase the reliability of these methods the stresses were validated as being close to those measured by HF and OC (Villaescusa et al. 2002(Villaescusa et al. , 2003;;Tuncay and Ulusay 2008).However, to date, a number of aspects remain problematic, such as (1) the applicability of AE and DRA to different rock types; (2) influence of delay time on the prestress evaluation accuracy in a specimen, with delay time representing the elapsed time from borehole drilling to the pre-stress tests in the laboratory; and (3) the impact of temperature change due to geothermic gradient (Goodman 1963;Li and Nordlund 1993;Seto et al. 2001;Kramadibrata et al. 2011).When the rock core is taken from a great depth, the temperature change affects the characteristics of the cracks in the rock.Therefore, the impact of the curing temperature on the pre-stress evaluation technique accuracy must also be investigated.
The Pliocene Yutengping Sandstone in the Kueichulin Formation is a potential CO 2 storage reservoir rock at Tieh-chan-shan (Lu et al. 2008).The AE and DRA techniques were applied to samples from the Yutengping Sandstone for the first time.Recent borehole drilling cores are unavailable.Therefore, instead of measuring the in situ stress of Yutengping Sandstone at Tieh-chan-shan, we investigate the stress memory phenomenon in rock samples from outcrops by conducting thorough laboratory tests with the following objectives: (1) to identify applicability of AE and DRA on Yutengping Sandstone and (2) to quantify the influences of the pre-stress level, the delay time and the curing temperature on the pre-stress evaluation accuracy.The in situ stress memorization process can be considered as creep because the rock has been buried underground subjected to a specified stress for millions of years.Therefore, in the laboratory tests, the pre-stress can be memorized into a rock using static and dynamic approaches as creep and cyclic loads, respectively (Haimson 1978).The static stress memorization applies a specified stress applied continuously to the rock sample (Li and Nordlund 1993) but pre-stress memorization takes a long time to complete for each rock sample during the tests.Alternatively, the dynamic stress memorization applies cyclic loads to the rock sample and is an alternative approach to remember the artificial pre-stress within a short experimental time.Therefore, artificial prestresses are applied to each sample in this study using uniaxial cyclic loads (Seto et al. 2002;Wu and Jan 2010).

GeOlOGIC bACkGROund
Tieh-chan-shan is a major geologic structure and a potential geologic CO 2 storage site (  1. Major geological structure of potential reservoir rocks on Taiwan (Lu et al. 2008).
Futoukeng Fault to the north, Tunglo Syncline and Sanyi Fault to the east, Changhua Fault to the south and the Taiwan Strait to the west (Fig. 3).The area is 13 km long and 4 km wide.The local rock mass was generated during the Miocene, Pliocene, Pleistocene and Holocene epochs.The rock formation strike is NE -SW.The Tapingting Fault is a tear fault striking at a high angle in the fold axis dividing Tieh-chan-shan into two segmented anticlines, the Tunghsiao Anticline in the north and Tieh-chan-shan Anticline in the south.The balanced cross sections (Yang et al. 2007) show that the anticlines at Tieh-chan-shan are asymmetrical.In the north section the west wing is steep, but the east wing is gentle.In the middle section the anticline is roughly symmetrical.In the south section the west wing is gentle, but the east wing is steep.Figure 4 shows the seismic interpretation cross sections of A -A' (Fig. 4a), B -B' (Fig. 4b), and C -C' (Fig. 4c) shown in Fig. 3.In Fig. 4, the solid lines are the rock formation boundaries and the dashed lines are the thrusts.The depth of each rock formation is verified by drilling cores taken from the gas wells T-38 in Fig. 4a and T-2, T-20, and T-37 in Fig. 4b.Chuang (2011) noted that based on a seismic survey, Tieh-chan-shan is composed of two folds slipping along the thrusts at opposite dip directions.These two folds are softly linked in the transfer zone.Tieh-chan-shan is currently the biggest gas field and underground gas storage reservoir in Taiwan.The Miocene Talu Formation, at a depth of 2750 m below sea level (Wang et al. 2011), is the main gas production sandstone.Ts'eng et al. (2003) noted that faults cut through the gas field and generate complex geological structures.The interference test data show that the thrust that parallels the anticline axis of the Tieh-chan-shan structure is a sealing fault but that other faults are fluid connected.
In addition to the geological structure, in situ stress is a key issue that has been investigated at Tieh-chan-shan to guarantee the safety of the underground gas storage reservoir.Wu et al. (2008b, c) highlighted the impact of in situ stresses on the stability of Tieh-chan-shan to securely store natural gas and CO 2 .Stresses change significantly at the interfaces of different rock formations based on the value and direction of the maximum horizontal stress, evaluated using borehole breakouts (Chen and Ding 2005).Based on the in situ stress investigations using the petroleum exploration and exploitation database, fluid injection into the Talu Formation has been found not to reactivate the faults at Tiehchan-shan (Wang et al. 2011).Both Chen and Ding (2005) and Wang et al. (2011) assumed that the vertical stress is the principle stress.However, by comparing the computational results using FLAC 3D to the borehole deformation measured using callipers at a depth of 2700 -2750 m, Tsai et al. (2006) indicated that the complex geological structure at Tieh-chan-shan significantly disturbs the local stress.Vertical stress may therefore not be the principle stress.Further studies are required to increase the reliability of the three-dimensional in situ stress data.

PhYSICAl And meChAnICAl PROPeRtIeS Of the YutenGPInG SAndStOne
The Yutengping Sandstone, a rock member of the Kueichulin Formation, consists of lithic graywacke, subgraywacke and propoquartzite with intercalcated gray shale.Ripple marks, coal fragments, cross-stratifications, plant fossils and sand pipes are frequently observed in this member (Chou 1980).The Pliocene Yutengping Sandstone, at a depth of 1495 -1697 m below the ground surface, is a potential reservoir rock for future CO 2 storage, as it may offer advantages over the well-investigated Talu Formation.
At the time of this writing recent borehole drilling cores at Tieh-chan-shan were unavailable.Some samples were taken from a rock outcropping in the eastern Tieh-chanshan geologic structure for the AE and DRA investigations.Rock cylinders were drilled from rock blocks (Fig. 5a) with diameters of 50 mm and height-to-diameter ratios of 2-to-2.5 (as recommended by ASTM4543-04) (Fig. 5b).After sample preparation each sample was air-dried for at least 7 days.Table 2 lists the uniaxial compressive strength, water content and Young's modulus of the air-dried samples.The average uniaxial compressive strength was q u = 8.32 MPa.The average unit weight and void ratio of the sandstone are 25.48 kN m -3 and 0.027, respectively.The dynamic mechanical properties of the air-dried sandstones are shown in Table 3.

Acoustic emission
AE detects the pre-stress of a material by measuring acoustic emissions as the material is loaded and, when it is reloaded (Fig. 6a), measuring the degree to which the new stress exceeds the previously applied stress, at which point the acoustic emission increases significantly (Fig. 6b).This phenomenon is called the Kaiser Effect.In ASTM E 610-77 the Kaiser Effect is defined as the absence of detectable acoustic emission until previously applied stress levels are exceeded.Goodman (1963) first validated that the Kaiser Effect also exists in rocks.
The mechanism and theory of the Kaiser Effect are not well known.Rock type, delay time and the level of stress govern the fracture propagations in the rock and control the accuracy of pre-stress evaluations using AE (Goodman 1963;Scholz 1968 a, b, c).Table 4 shows that some rocks even show a poor Kaiser Effect (Li and Nordlund 1993).Although the in situ stresses evaluated using AE are validated close to those measured using the conventional HF and OC techniques (Villaescusa et al. 2002(Villaescusa et al. , 2003;;Tuncay and Ulusay 2008), the reliability of using AE in in situ stress evaluations remains low (Ljunggren et al. 2003).Therefore,   quantifying the accuracy of pre-stress AE evaluations is a key issue, especially when AE is applied to rock for the first time.

deformation Rate Analysis
The deformation rate analysis (DRA) method is based on the deformation memory effect.This method is applies more than twice the cyclic uniaxial compressible load to pre-stressed cores (Fig. 7a) and measures the corresponding strain differences during loads (Fig. 7b).Yamamoto et al. (1990) proposed the following equation [Eq.( 1)] to calculate the difference in axial strain ( ij f D ) in conventional DRA: where, ( ) i f v is the axial strain at the σ stress of the ith load.The variable ( ) j f v is the axial strain at the v stress of the jth load.Yamamoto et al. (1990) noted that the strain difference function in Fig. 7b can be approximated using a straight line with a positive gradient at stresses less than the pre-stress subjected to the rock sample and it sharply bends down near the pre-stress of two successive loads.The positive gradient at the applied stresses less than the pre-stress indicates that the rock sample is more compliant under the second load than in the first load because the increase in crack density at high stresses from the first load.On the contrary, the negative gradient in the strain difference function (Fig. 7b) means that the rock is easier to deform in the first load than in second load because the rock is ready to enlarge pre-existing cracks or to create new cracks when it first experi-ences high applied stresses.
Uniaxial DRA successfully evaluated the in situ stress (Yamamoto and Yabe 2001) and correlated well with other in situ stress evaluation techniques (Seto et al. 2001;Shimada et al. 2001;Stacey and Wesseloo 2002).However, the DRA application to practical in situ stress evaluations is still problematic for the following reasons: (1) low reliability of pre-stress estimation, (2) unclear stress memory mechanism and (3) impacts of critical environmental factors and (4) effects of delay time on the pre-stress evaluation accuracy (Seto et al. 2001).
The pre-stress of a rock sample can be evaluated using AE and DRA simultaneously.Furthermore, the test results comparison conducted sing AE and DRA increases the evaluated pre-stress confidence.However, using both AE and DRA to detect the pre-stress of a rock is only in its preliminary stages in Taiwan.Previous work includes that of Lin et al. (2007), who investigated the AE and DRA characteristics of Mushan Sandstone; Wu et al. (2008a), who assessed the accuracy of AE in evaluating the pre-stress applied to anisotropic block schist; and Wu and Jan (2010) and Wu and Pan (2013), who evaluated the pre-stresses of Changchikeng sandstone, the main rock in the Tseng-wen Reservoir trans basin water tunnel.
Using AE and DRA for CO 2 reservoir rock storage in situ stress determination is very attractive for the following reasons (modified from Stacey and Wesseloo 2002): (1) AE and DRA use original cores obtained for other purposes, such as exploration, which makes the methods cost effective.(2) Cores obtained remotely can be used; therefore, the methods do not interrupt the borehole drilling procedure.(3) Small cores are used for the tests.Therefore, many tests can be carried out using limited original borehole core lengths, which again makes the method cost effective.Additionally, the more cores tested the greater the confidence in the results obtained.(4) AE and DRA do not damage the rock mass; preserving the geologic structure integrity needed to seal the CO 2 underground.

unIAxIAl CYClIC lOAd teStInG PROCeduRe fOR Ae And dRA teStS
Samples from Yutengping Sandstone outcrops were investigated to determine the reliability of assessing the rock pre-stresses using AE and DRA. Figure 8 shows the uniaxial loading-unloading time history of our pre-stress program.The time history consists of three parts: (1) first stage of cyclic loading, (2) delay time, and (3) second stage of cyclic loading.The first stage involves 200 uniaxial 0.2 Hz load cycles to memorize pre-stress in the rock samples.Four different delay times (0, 4, 14, and 28 days) were then applied to simulate the impact of transportation time from the  (Li and Nordlund 1993).
in situ site to the laboratory and sample preparation because the elapsed time from the in situ borehole drilling to the in situ stress measurements is about 14 days based on personal discussions with senior engineers from CPC, Taiwan.The Yutengping Sandstone depth at Tieh-chan-shan is between 1495 and 1697 m (Lu et al. 2008), and the sandstone unit weight is 25.48 kN m -3 .Therefore, the Yutenping Sandstone overburden is in the 38 -43 MPa range, which exceeds the uniaxial compressive strength of the rock, q u = 8.32 MPa.The maximum pre-stress for the uniaxial compressive loads is set to 5.82 MPa (0.7q u ).The 5.82 MPa prestress is certainly much less than the in situ overburden of 38 -43 MPa.However, the uniaxial compressive loading is a simple approach to evaluate the possibility of conducting AE and DRA for Yutengping sandstone pre-stress assessment.
The maximum geothermal gradient at Tieh-chan-shan is 4.17°C 100 m -1 (Huang 1990) and the average ground temperature at Taichung, which is the nearest Central Weather Bureau temperature station is 23.3°C from 1981 to 2010 (http://www.cwb.gov.tw/V7/climate/monthlyMean/Taiwan_ tx.htm).Therefore, the temperature of the Yutenping Sandstone at the depth of 1495 and 1697 m underground at Tieh-chan-shan is between 85.6 -94.0°C.Suppose the rock cores are drilled at great depth and will be immediately subjected to unloading and a decrease in temperature once extracted.The temperature change impact on the Yutengping sandstone cores must therefore also be considered.Three curing temperatures (5, 30, and 105°C) were applied in this study to the pre-stressed samples during the delay time to investigate the impact of temperature change.
In the second cyclic load stage three uniaxial cyclic loads were applied to the samples, with AE and core deformation measured simultaneously.The P L and P U in the first stage (Fig. 8) are the lower and upper bounds of the cyclic load stresses, respectively.The P U is also the pre-stress of the sample in this study.Each sample must exhibit permanent deformation after cyclic load in the first stage to guarantee successful stress ''memorization''.The maximum load (P m in Fig. 8) in the second stage must exceed P U in the first stage, but must not induce uniaxial compressive failure.The average uniaxial compressive strength of the sandstone was q u = 8.32 MPa.Table 5 lists
samples used in this study.

PRe-StReSS evAluAtIOn uSInG ACOuStIC emISSIOn
The pre-stress applied to core samples in the first stage (Fig. 8) was assessed by measuring the AE signals from the first loading in the second stage cyclic load.When the load exceeded the previous maximum stress parallel to the load direction, significant AE signals occurred, which is known as the Kaiser Effect.The rock core axis does not necessarily coincide with the direction of any principal stress axis.Villaescusa et al. (2002) and Wu and Pan (2013) proposed algorithms to generated three-dimensional stresses using the pre-stress evaluations from oriented cores.The testing procedure (Figs. 6 and 9) is discussed below.
(1) Core drilling was conducted in Yutengping Sandstone blocks to obtain core samples (Fig. 9a).( 2) We assumed that the in situ stresses acting on the rock samples, which were taken from an outcrop rock block, were much lower than the specified pre-stresses in Table 5. Hence 200 uniaxial cyclic loads were applied to the cored samples to simulate the imposed pre-stress (Fig. 9b).
(3) A pre-stressed sample was installed on a loading system with two AE sensors that monitored the AE signals during the load testing (Fig. 6a).A miniature PICO sensor was selected because of its size and appropriate operation frequency, which is in the 200 -750 kHz range (http:// www.pacndt.com/index.aspx?go=products&focus=/ sensors/miniature.htm).The frequency range of the geomaterials in these laboratory tests is in the 100 kHz -1 MHz range (Scholz 1968a, b, c) and 100 Hz -50 kHz range (Hardy 1972).The PICO sensor has a diameter of 5 mm and a height of 4 mm, and can be easily attached to the cylindrical surface of the sample.The pencil lead break technique (ASTM E976-10) was used to check for proper equipment operation.(4) AE cumulative count versus loading stress was plotted (Fig. 6b).The pre-stress value, σ pre , was obtained by projecting the tangent, Lines A and B, at the turning point suggested by Lavrov (2003).(5) The pre-stress evaluated using the AE signals was compared with the memorized pre-stress, P U , in Table 5.

PRe-StReSS evAluAtIOn bY defORmAtIOn RAte AnAlYSIS
Wu and Jan (2010) noted that both axial strain and lateral strain can be used in DRA.Therefore, three load cycles were imposed in the second cyclic loading stage to assess the pre-stress, P U (see Fig. 8), with an axial strain gauge and a lateral strain gauge attached on opposite sides of the sample after the pre-stress memorization shown in Fig. 7a.
The basic DRA concepts are explained as follows (Seto et al. 2001): (1) Cracks in the rock sample, which are caused by prestress, deform steadily under low stresses.
(2) The rock deformation rate changes when the stress applied to the rock sample exceeds the pre-stress to generate new cracks.
Figure 7b shows that the difference between strains in the cyclic loads is a curve.Therefore, in a method similar to that proposed by Lavrov (2003) for AE, tangents to the maximum curvature are used to determine the pre-stress value, σ pre .

exPeRImentAl ReSultS Of the PRe-StReSS evAluAtIOn
Four delay times (0, 4, 14, and 28 days) and three curing temperatures (5, 30, and 105°C) were applied to the samples during pre-stress evaluations.When the delay time = 0 days, the cyclic loads in the second stage (Fig. 8) were applied to the pre-stress samples subsequent to the cyclic loads in the first stage.When the delay time > 0 days, the pre-stress samples were placed in a chamber with a fixed curing temperature for the specified delay time minus two days.The samples were then stored at room temperature for two days before applying the cyclic loads in the second stage.
The pre-stress evaluation is successful in most Yutengping Sandstone samples using AE (Table 6).However, the AE signals in Fig. 10 bend slightly downward and are different from the conventional upward curve shown in Fig. 6b.We therefore judged that Fig. 10 fails to evaluate the pre-stress.The error in Table 6 is calculated using the following Eq.( 2): The error is positive when the estimated stress overestimates the pre-stress and negative when the estimated stress underestimates the pre-stress.The estimated stresses errors using AE are in the range from -36.04 -15.14% (Table 6).
In conventional DRA pre-stress was estimated using the strain difference between the 2 nd load and the 1 st load in the second stage.However, the concave curve of the strain difference in the pre-stress sandstones (Fig. 11), which is caused by zero-stress memory (Yamamoto et al. 1990), is not the typical convex curve (Fig. 7b).Zero-stress memory can be caused due to the cracks produced during/ the cyclic loads in the first stage and destroyed by the first load in the second stage (Fig. 8) (Yamamoto et al. 1990).The typical convex DRA curve in this study was obtained using the strain difference between the 3 rd load and the 2 nd load in the second stage based on the suggestions by Yamamoto et al. (1990).Additionally, DRA applies uniaxial strain   6. Pre-stress evaluations using AE.
Note: --: fail to evaluate the pre-stress using AE data.(conventional DRA) and lateral strain to evaluate the rock pre-stress (Table 7).The estimated pre-stress errors range from -19.93 -17.42%.The data in Tables 6 and 7 show that many pre-stress evaluations under-estimate the pre-stresses applied to the sandstone samples by AE and overestimate the pre-stresses with DRA.The pre-stress underestimation mechanisms by AE and over-estimation by DRA are explained as follows: (1) AE is very sensitive to the crack propagation in the rock sample.The crack distribution and characteristics in rocks is complex with tiny crack propagations that occur before the pre-stress contributing to the underestimation.(2) In DRA the pre-stress is detected by the significant change in deformation in the rock sample.The strain difference in this study was calculated between the 3 rd load and the 2 nd load in the second stage.The first load in the second stage destroys the zero-stress memory (Yamamoto et al. 1990) and causes new cracks resulting in the pre-stress overestimation because the 1 st load in the second stage exceeds the maximum stress of the loads in the first stage (Lin et al. 2006).

Impact of different delay times and Pre-Stresses
Pre-stress evaluated using AE with a curing temperature of 30°C shows insignificant trends under different delay times and pre-stresses (Fig. 12a).The turning point in the AE curve is indistinct.Developing new methodologies to increase the pre-stress estimation accuracy with AE is essential for future studies.
In DRA the pre-stress evaluations with a curing temperature of 30°C shows that the stress memory fades as the delay time increases in both the axial strain (Fig. 12b) and the lateral strain (Fig. 12c).

Impact of different Curing temperature
Curing temperature does not significantly affect the prestress evaluations using AE (Fig. 13).Again, to increase the pre-stress evaluation accuracy further studies are required to discuss the techniques that determine the sample pre-stress and the crack distributions in the pre-stressed rocks.
The sandstone pre-stress evaluation using DRA is governed by delay time (Figs.14 and 15).That is, in DRA studies, stress memory fades as the delay time increases.Although the temperature change may generate new cracks in the rock sample, curing temperatures from 5 -105°C have a very minor impact on the pre-stress evaluation accuracy.
Based on Amadei and Stephansson (1997), 10 -20% scatter in magnitude is expected in a series of pre-stress measurements in typical rock conditions.Although the maximum error in AE is -36.04% (Table 6), in this study, most data are less than 20%.In addition, the magnitude error in DRA pre-stress evaluations scatters is less than 20%.Therefore, this study validates that both AE and DRA are useful techniques to evaluate the in situ stress of Yutengping Sandstone with acceptable accuracy.However, although Chen (2000) noted that the high pre-stress in a rock can also be investigated under high triaxial stresses, further studies must be conducted to investigate the impact of confining pressure on pre-stress evaluation accuracy using AE and DRA since the assumed pre-stress in this study is far below the in situ overburden.

COnCluSIOnS
Uniaxial cyclic load tests validate that AE and DRA are useful tools to evaluate pre-stresses in Yutengping Sandstone.Pre-stresses are underestimated in many AE processed samples because AE is very sensitive to crack propagation in the rock sample.Delay time and curing temperature have insignificant impacts on the pre-stress estimations.Remarkable crack propagation causes changes in rock deformation resulting in DRA overestimating the pre-stress.Both axial strain and lateral strain can be used in DRA.The stress memory of Yutengping Sandstone fades as the delay time increases.Curing temperatures between 5 -105°C have a very minor impact on the pre-stress evaluation accuracy.Special attention must be paid to the sample preparation, experimental procedure to increase the experimental results reliability when we practically apply AE and DRA to measure three-dimensional in-situ stresses.Samples must be drilled at different directions because AE and DRA evaluate the rock sample pre-stress parallel to the load direction.The rock sample fails before the pre-stress level when the in-situ stresses exceed the rock's uniaxial compressive strength.The triaxial compressive test increases the rock's compressive strength and may be an alternative procedure for in-situ stress measurements using AE and DRA.However, the unclear impact of confining pressure on the pre-stress evaluation accuracy and the requirement for waterproof techniques under high confining pressure to the strain gauge and AE sensors complicate the pre-stress evaluation procedure.The in-situ stresses evaluated using AE and DRA must compare to the experimental results obtained by conventional techniques to increase the data reliability.

structure Reservoir rock depth at the top of the rock (m) depth at the bottom of the rock (m)
Table 1) located in northwest Taiwan.The Tieh-chan-shan boundaries are the

Table 2 .
Physical and mechanical properties of the air-dried Yutengping Sandstone.

Table 3 .
Dynamic mechanical properties of the Yutengping Sandstone.

Table 4 .
The Kaiser Effect of different rocks