A multifunctional rock testing system for rock failure analysis under different stress states:Development and application

Shui Li,Jie Hu,Florin Amnn,Liping Li,Hongling Liu,Shoshui Shi,Pooy Hmdi

a School of Qilu Transportation,Shandong University,Jinan,250061,China

b School of Mechanical Engineering,Nanjing University of Science and Technology,Nanjing,210094,China

c Department of Engineering Geology and Hydrogeology,RWTH Aachen University,Aachen,52064,Germany

Keywords:Rock testing system Compressive-shear test Tensile test Tensile-shear test Failure behavior Multiple variable evolutions

A B S T R A C T The stress state in a rock mass is complex.Stress redistribution around underground excavation may lead to various failure modes,including compressive-shear,tensile-shear,and tensile failures.The ability to perform laboratory tests with these complex stress states is significant for establishing new strength criteria.The present paper introduces a new rock testing system with“tensile-compressive-shear”loading functions.The device includes bi-directional and double-range hydraulic cylinders,auxiliary loading equipment,and roller rows that can perform direct compressive-shear tests,direct tensile tests,and direct tensile-shear tests.The testing system provides maximum vertical and lateral loading forces of 2000 kN and allows testing cubical rock specimens with dimensions of 0.5 m×0.5 m×0.5 m.The performance of the testing machine was evaluated by testing a rock-like material based on cement mortar under compressive-shear,tensile,and tensile-shear stress states.The failure process and deformation characteristics were monitored during loading using acoustic emission(AE)transient recorder,piezoelectric AE sensors,a high-speed camera,and a thermal infrared camera.The failure mechanism was investigated by analyzing AE counts,AE amplitude,strain,and temperature changes on the rock specimen surface.The test results confirmed that the testing system could successfully simulate the abovementioned stress path.The AE counts and amplitude responses were influenced by different failure modes.The temperature response during the compressive-shear test indicated the development of a high-temperature band on the rock specimen surface.In contrast,a negligible temperature change was observed during the tensile and tensile-shear tests.The newly developed multifunctional rock testing system allows laboratory tests under various failure modes.The monitoring results of multiple variables during rock failure tests provide valuable information on failure characteristics.

1.Introduction

As a result of continuous investments in the construction of transportation infrastructure and energy development,underground engineering applications are gradually shifting from shallower to more considerable depths,wherein the mechanical characteristics of rock mass become more challenging(Zhang et al.,2018a,b).In underground engineering,it is generally believed that discontinuities mainly control the failure of the surrounding rock mass under low-stress conditions(Han et al.,2013;Huang et al.,2016;Liu et al.,2016).At more considerable depths,however,rock mass failure may also involve failure of intact rock associated with compressive-shear,tensile,and tensile-shear conditions(Etheridge,1983;Lin et al.,2015a,b;Zhou et al.,2018).Rock mass failure criteria and the evolution of multiple variables,such as stress,strain,temperature,acoustic emission(AE)signal,electromagnetic(EM)radiation during the rock mass failure have been the research focus in rock mechanics and disaster control(Ghazvinian et al.,2012;Meng et al.,2016;Zhang et al.,2016;Mehranpour et al.,2018).In rock mechanics,rock mass failure criteria are typically used to evaluate the stability.In disaster control,the focus is put on the identification of quantitative indicators to detect rock mass failure processes for effective monitoring and early warning system(Lin et al.,2015a,b;He et al.,2017;Li et al.,2018;Zhao et al.,2018).

At present,the commonly used failure criteria in rock mechanics are the Mohr-Coulomb shear failure and the empirical Hoek-Brown failure criterion(Hoek and Brown,1980,2019;Zheng,2003;Eberhardt,2012;Labuz and Zang,2012;Peng et al.,2014).Both criteria are proposed by finding the best fit to a series of laboratory testing data,including tension and uniaxial and triaxial compression tests.Under these testing conditions,the specimens typically exhibit a compressive-shear failure mode(i.e.at least at higher confining pressures).Some scholars have considered tensile and tensile-shear deformation mechanisms to modify the Mohr-Coulomb criteria(Engelder,1999;Wang et al.,2012;Wang and Shen,2017).However,the reliability of such modified approaches remains uncertain due to the difficulty in carrying out direct tests.Similarly,AE radiations,EM radiation,and thermal effects associated with intact rock failure are mainly obtained from compressive and compressive-shear failure modes(Liu et al.,2018;Song et al.,2018;Wang et al.,2018).The evolution of these variables during tensile and tensile-shear failure tests has not yet attracted widespread attention.However,as engineering construction continues to develop into deeper underground spaces,it is necessary to understand the mechanical behavior and failure characteristics of intact rock under tensile and tensile-shear stress states(Zhou et al.,2018).

The tensile strength of rocks is estimated using either indirect(Brazilian)or direct tensile tests.The Brazilian test is an indirect test method,and the corresponding stress state may be vastly different from that in a direct tensile(uniaxial)test.The direct tensile test method is mainly divided into two types:clamping type and bonding type.The former often generates stress concentrations at the clamping position and causes partial rupture(Erarslan and Williams,2012).Regarding the bonding type,the loading plate is bonded to the rock specimen with high-strength glue,and then the rock specimen exhibits tensile failure under the action of a tensile load.This test configuration should try to prevent eccentric tension of the rock specimen.At present,research on direct tensile test devices for rock specimens has mainly focused on eccentricity prevention(Zhang et al.,2014).

The rock specimens used for direct tensile tests are mainly cylindrical(ISRM,1978;ASTM D5607-08,2008;Perras and Diederichs,2014).Chinese specifications for rock tests do not specify specimen dimensions for laboratory direct tensile tests on cylindrical specimens(DL/T 5368-2007,2007).However,for cubical rock specimens used in compressive-shear tests,the side length should not be less than 150 mm.Therefore,various specimen geometries and testing set-ups are required for different compression and tension laboratory tests.The direct tensile-shear test is mainly replaced by direct tensile test performed on rock specimens containing prefabricated inclined joints due to the lack of direct test equipment(Huang et al.,2011).There is a gap between the indirect test method and the direct tensile-shear test because the indirect test cannot acquire the stress-strain or stress-time curves.To overcome these shortcomings,Zhou et al.(2018)developed a multifunctional testing system that can be used for direct tensile tests,tensile-shear tests,and compressive-shear tests of cubic rock specimens.However,in their system,the rock specimen dimensions are limited to 50 mm×50 mm×50 mm,and the load capacity does not allow tests at high pressures typically expected in deep underground excavations.In addition,the loading box in their system is nearly wholly capsuled,which makes it extremely difficult to continuously monitor the rock surface(i.e.using a thermal infrared camera,a high-speed camera,and other optical equipment).To further study the failure mechanism and the associated evolution of AE,EM,and temperature under tensile,tensile-shear,and compressive-shear stress states,it is necessary to develop a new multifunctional testing system that can be combined with multiple monitoring equipment.

Aiming at overcoming some of the shortcomings mentioned above,a new rock testing system with“tensile-compressive-shear”loading functions was developed.The testing system adopts a new type of open loading box,which leaves the unstressed surface of the rock specimen exposed,enabling the rock specimen failure process to be captured in real time(i.e.using a thermal infrared camera and a high-speed camera).The newly developed testing system effectively integrates various test functions,saving testing resources and costs.

2.Description of the testing system

2.1.Main innovations

The testing system can meet the requirement of multi-type tests of rock specimens of different sizes.It allows for(i)uniaxial compressive test,(ii)direct tensile test,(iii)direct compressiveshear test,and(iv)direct tensile-shear test.Besides,specially designed open loading boxes make it possible to utilize optical equipment to monitor the progressive failure of rock specimens during testing.

2.2.Testing system and characteristics

The testing system(Fig.1)comprises a test bench,a loading system,a hydraulic system,a servo control system,a data measurement and acquisition system,and analysis software.The testing system can be utilized with different auxiliary loading boxes to perform tests under different stress states.

2.2.1.Test bench and loading system

The test bench(Fig.1)comprises two columns and upper and lower beams mounted on the two columns.A high-strength steel plate is used for the overall structure of the test bench,and the rigidity of the plate reaches 10 GN/m;hence,there is negligible bench deformation during testing.The overall dimension of the test bench is 2.5 m×1.2 m×3 m(length×width×height).The maximum allowable dimension of cubic rock specimen is 0.5 m×0.5 m×0.5 m.Two vertical and lateral hydraulic cylinders and two corresponding reaction force devices are installed inside the bench.An adjustable platform was designed in the lower part of the bench for rock specimen installation and position adjustment.The tensile and compressive forces can be applied through the newly designed roller rows.Each part of the test bench is shown in Fig.1.

The testing system has two vertical and lateral hydraulic cylinders with an independent oil supply.The vertical and lateral loading devices are designed in series,including a large-tonnage hydraulic cylinder with a 2000 kN maximum force and a smalltonnage hydraulic cylinder with a 500 kN maximum force.Aiming at the high loading requirement for compressive and compressive-shear tests of large-scale rock specimens,largetonnage hydraulic cylinders with a 2000 kN maximum force are utilized.Small-scale rock tests and tensile strength tests do not require very large forces(i.e.approximately＜20 kN).In this case,a large-tonnage hydraulic cylinder has an insufficient resolution and loading ramp control.Therefore,a highly accurate hydraulic cylinder with a maximum force of 500 kN is connected in series at the end of the large-tonnage hydraulic cylinder,and independent stress transducers are installed.The measurement ranges of the independent transducers are 0-2000 kN and 0-500 kN,respectively.The connection between the components for the vertical loading device is shown in Fig.2a.The design principle of the lateral loading device is the same as the vertical loading device.A small-tonnage hydraulic cylinder is used for the small-scale rock failure test.For large-scale rock specimens,the small-tonnage hydraulic cylinder can be substituted by a large-tonnage hydraulic cylinder.The vertical loading device adopts a two-way,double-rod cylinder for tensile and compressive forces.

To meet the installation and positioning requirements of largescale rock specimens,an adjustable platform is installed on the base of the test bench,as shown in Fig.2b.There is a slide track on the left and right sides of the platform,respectively.The two adjustable plates can move back and forth through the slide tracks.The lifting plate is connected with the adjustable plates using four lifting bolts,and the height of the lifting plate can be adjusted by rotating four bolts.A multifunctional block is installed above the lifting plate,mainly used to connect with different auxiliary moulds used for the different rock failure tests.A set of roller rows under the lifting plate is used to reduce the friction resistance in the shear direction.Before testing,the following steps are required to install the rock specimen:

(1)The lifting bolts are rotated clockwise to separate the lifting plate from the roller rows.

(2)The lifting plate is pulled out of the bench.

(3)The rock specimen is placed on the multifunctional block.

(4)The lifting plate is pushed into the designated position on the bench.

(5)The four lifting bolts are rotated anticlockwise to make the lifting plate fall on the upper surface of the roller row and make close contact.

For shear tests,the rock specimen is sheared by the lateral loading device.Conventional compressive-shear testing systems commonly utilize a roller row under the shear box to reduce the frictional forces between the shear box and the bottom plate during shearing.However,for a tensile-shear test,a single roller cannot effectively reduce lateral frictional forces due to tensile stresses.Therefore,this testing system utilizes independent double rollers suitable for compressive-shear and tensile-shear tests.The double roller row consists of an upper and a lower roller row and a connecting block.The bottom plate of the connecting block is located between the upper and lower roller rows,as shown in Fig.2c.When carrying out a compressive-shear test,the lifting plate is adjusted to fall on the upper surface of the connecting block of the roller row.Under the action of vertical compressive stress,the lifting plate and the connecting block are in tight contact and move together.Moreover,the lower bottom plate of the connecting block transmits the vertical compressive stress to the lower roller row and moves laterally along the lower roller row,as shown in Fig.2c.When a tensile-shear test is performed,the lifting plate is still adjusted to fall on the upper surface of the connecting block of the roller row,and the lifting plate is connected with the bottom plate of the connecting block using bolts to form a solid structure.For vertical tensile stresses,the bottom plate of the connecting block transfers the vertical tensile stress to the upper roller row and moves laterally along the upper roller row,as shown in Fig.2c.

2.2.2.Auxiliary loading equipment

The testing system is mainly used to perform failure tests of cubic rock specimens with dimensions of 150 mm×150 mm×150 mm.For this purpose,the auxiliary loading equipment described in the following was designed.However,the auxiliary loading equipment can be adjusted to large-scale rock specimens.

Fig.2.Design of hydraulic cylinders and auxiliary equipment:(a)Schematic diagram of the series structure of a vertical 2000 kN hydraulic cylinder and a vertical 500 kN hydraulic cylinder;(b)Diagram of the adjustable platform of the testing system;and(c)Schematic diagram of the roller rows and function principle.

The auxiliary loading equipment for the direct compressiveshear test is shown in Fig.3a.The design principle enabled the specimen surface to be visible during compressive-shear test.This allows a thermal infrared camera,a high-speed camera,and other optical measurement instruments to continuously capture thermal changes and crack formation and propagation during shearing.The auxiliary loading equipment includes an upper and a lower shear box.During the test,the position of the rock specimen is adjusted so that the central axis of the rock specimen coincides with that of the vertical hydraulic cylinder.The position of the lead screw is adjusted so that it is in contact with the left side of the upper shear box.When the lateral hydraulic cylinder applies the lateral shear load,the lead screw provides an equal reaction force until the rock specimen is completely sheared.

The auxiliary loading equipment for the direct tensile-shear test needs to maintain a stable vertical tension and a shear load application.The application of the lateral shear load is similar to that in the compressive-shear test.For the application of vertical tension,it is important to provide a solid connection of the vertical system components.To achieve this connection,a high-strength glue(i.e.up to 15 MPa tensile strength)and multiple bolts are utilized.The components of the auxiliary loading equipment for the direct tensile-shear test are shown in Fig.3b.In contrast to the compressive-shear test,the rock specimens installed between the upper and lower shear boxes need to be preinstalled with tension components.First,four side backing plates,each with a thickness of 5 mm,are bonded onto the front and back sides of the rock specimen with high-strength glue.After the glue is cured,the upper and lower sides of the rock specimen are evenly coated with high-strength adhesive,and then two C-shaped connecting plates are glued to the specimen.After the glue is cured,the bolts of the C-shaped plate on the front and back sides of the rock specimen are adjusted,and the inner side backing plate is slightly tightened to ensure that the rock specimen and the tension component are tightly connected(Fig.3b).After the rock specimen preparation is completed,the upper and lower shear boxes are bolted to the upper and lower Cshaped connecting plates.Then,the rock specimen can be rotated into the multifunctional block through the anti-tension bolt under the connecting plate.The upper shear box of the rock specimen is bolted to the upper connecting plate and subsequently bolted to the plate with a universal spherical hinge.After this step,the preparation and installation for the tensile-shear test are completed.

In the tensile-shear test,a low constant tensile load is applied vertically.However,the required tensile forces to crack the rock specimen are significantly higher for the direct tensile test.Therefore,the capacity of the tension component of the auxiliary loading equipment needs to be increased.In addition,the auxiliary loading equipment for direct tensile tests adopts a combination of bonding and clamping to minimize common problems with eccentricity(Fig.3c).Firstly,four side backing plates with a thickness of 5 mm are bonded to the rock specimen near the upper and lower surfaces with high-strength glue.The upper and lower backing plates are also bonded to the upper and lower surfaces of the rock specimen.When the adhesive is completely cured,the connection boxes are installed on the upper and lower sides of the rock specimen,which are bolted to the upper and lower backing plates.Then,the side bolts on the connection boxes are slightly tightened until contacting the side backing plate,and a pre-tightening force is applied to achieve a clamping effect.Afterward,the lower connecting box and the lower connecting plate are bolted together,and the rock specimen can be rotated into the multifunctional block through the anti-tension bolt under the connecting plate.Finally,the upper connecting plate,the upper connecting box,and the plate with a universal spherical hinge are also bolted together.After this step,the preparation and installation of the rock specimen for the direct tensile test are completed.

Fig.3.Schematic diagram of the auxiliary loading equipment:(a)For the direct compressive-shear test;(b)For the direct tensile-shear test;and(c)For the direct tensile test.

The shear gap zone between upper and lower specimen holders was set as 10 mm width in the direct shear tests,which follows the ISRM recommendation(Ulusay,2014).The loading designs can confirm the resultant shear force acting on specimens passes through the centroid of the shear plane surface to minimize adverse bending moment effect,which is also suggested by the ISRM(Ulusay,2014)and ASTM(ASTM D2936-08,2008).The highstrength glue has the largest tensile strength of 15 MPa,and it is conservatively suggested that the lateral clamping force need not be applied if the specimen’s tensile strength is lower than 10 MPa.In contrast,the lateral bolts should be tightened appropriately to provide a specific clamping force to the specimen ends.The lateral force is applied on the specimen side surface through the backing plate in our test system,which could make the load more uniform.Referring to the direct tensile strength of typical rock provided by Zhang et al.(2018a,b),the glue bonding’s tensile force can meet the direct tensile test requirements for the most common rock.Aiming at the lower bound rock strength that can be tested on our apparatus,it is mainly decided by the loading stabilization accuracy of the test apparatus.In the direct tensile tests,the hydraulic cylinder with a maximum force of 500 kN was used to apply normal tensile stress.Its loading stabilization accuracy is 0.5% full scale,i.e.the minimum normal tensile force for high-accuracy loading is 2.5 kN.Hence,the lower bound of the tensile strength that can be tested on our apparatus is 0.1 MPa for the cubic specimen with a size of 150 mm×150 mm×150 mm,while the upper bound could be more than 15 MPa when the bonding and clamping forces were both applied.

2.2.3.Servo control loading and data measurement systems

The servo control system uses two multi-controllers to control two valves and two sets of servo motors.This allows controlling all vertical and lateral hydraulic cylinders.The photoelectric encoder is used to control displacement and the hydraulic transducers are used to control the load.There are two loading methods,including displacement-and stress-controlled types.Four linear variable differential transformers(LVDTs)can measure deformations in the vertical(2 LVDTs)and lateral(2 LVDTs)directions.

2.2.4.Main technical characteristics of the testing system

The main technical parameters of the testing system are summarized as follows:

(1)The maximum vertical tensile and compressive loads of the large-tonnage hydraulic cylinder are 2000 kN,whereas the maximum vertical tensile and compressive loads of the small-tonnage hydraulic cylinder are 500 kN.

(2)The piston stroke of the hydraulic cylinder is 0-150 mm,and the deformation measurement accuracy is 0.5% full scale.

(3)The loading stabilization accuracy of the hydraulic cylinder is 0.5% full scale.

(4)The load-controlled loading rate of the hydraulic cylinder is 0.01-20 kN/s.

(5)The displacement-controlled loading rate is between 0.01 mm/min and 10 mm/min.

(6)The allowable range of cubic specimen dimensions is between 100 mm×100 mm×100 mm and 500 mm×500 mm×500 mm.

(7)The measurement range of the displacement transducer is 0-12 mm.

(8)The frequency of the controller can reach 1000 Hz,and the sampling rate is 50 Hz.

3.Test cases

To verify the performance of the testing system,direct compressive-shear,direct tensile-shear,and direct tensile tests were first performed on cubic specimens made of cement mortar with dimensions of 150 mm×150 mm×150 mm.Regarding the tests on the large-scale specimens,they will be conducted to investigate the scale effect in future research.During each test,AE sensors,a high-speed camera(for digital image correlation(DIC)),and a thermal infrared camera were used to compare the different failure processes.

3.1.Specimen preparation

The rock-like specimens were selected to compare the failure behavior and monitoring data evolution of the nearly identical specimens under different stress states in this study.All specimens were fabricated by the same rock-like material with a cement:sand:water mass ratio of 2:3:1 via mold casting,hence they have similar homogeneity and anisotropy characteristics.After curing,the specimens were stored in a curing room under constant temperature(i.e.20°C)and humidity(i.e.95%)for 28 d.The mechanical parameters of the specimen are shown in Table 1.The strain evolution should be captured with a high-speed camera and analyzed via DIC software during all tests.Hence,one surface of the specimen was selected,and a speckle pattern was applied to this surface using a speckle-making tool,as shown in Fig.4.Compared with the cement mortar specimen,there may be noticeable texture or speckle features on the rock specimen surface.The DIC strain analysis need not be based on the fabricated speckle in this condition.However,speckle fabrication can optimize strain analysis results.

Table 1Mechanical properties of the specimen.

3.2.Monitoring equipment and scheme

In addition to stress and strain data,AE signals,thermal changes,and crack propagation inferred from high-speed camera images were captured.During testing,the high-speed camera(FR-Stream 4Coaxp high-speed camera including a long-time recording system of 560 frames per second)with a resolution of 2336×1728 pixels was used.The strain evolution of the rock surfaces based on DIC was analyzed using the TEMA 2D DIC software,and the strain calculation principle can be found in Pan et al.(2009).The AE signals were monitored with a PAC Express-8 AE instrument.The three types of tests were conducted in the same laboratory environment and the same AE threshold of 40 dB was set to eliminate the effects of background noise in the three types of tests.An A615 thermal infrared camera from FLIR Systems,USA,was used for thermal infrared signal monitoring,for which the resolution was up to 640×480 pixels and the thermal sensitivity was less than 0.05°C.The analysis of the temperature and strain was limited to an area of one-third of the surface around the center of the specimen.The monitoring equipment and monitoring scheme are shown in Fig.5.

3.3.Test process and results analysis

3.3.1.Direct compressive-shear test

The direct compressive-shear test was performed under constant normal stress of 5 MPa.After the specimen and the shear box were placed in the frame,the normal load was first applied at a loading rate of 0.5 kN/s.After the predetermined normal load value,the lateral load was applied under displacement control at a loading rate of 1 mm/min until the rock exhibited failure.

The recorded AE counts,AE amplitudes,temperature changes are shown in Fig.6 against the testing time.The shear stress-time curve shows a strong nonlinearity at low shear stresses(i.e.below 6 MPa).At higher shear stresses,the curve was approximately linear with respect to time approaching the peak shear strength.At 280 s,the shear stress reached 11.16 MPa,and the specimen failed indicated by a sudden shear stress drop.Contemporarily with failure,the AE count reached the peak value of approximately 2800.With further shearing in the post-failure range,the shear stresstime curve dropped gradually to an ultimate value accompanied by a small number of AE events.The temporal distribution of AE signals was reasonably consistent with the shear failure process of the specimens.The amplitude was mainly in the range of 40-55 dB,and a small number of higher amplitude AE signals in the range of 70-75 dB were generated near specimen failure.There was a quiet period of AE signals near the failure of the specimen.

The incremental maximum principal strain was selected for the analysis.Its evolution on the specimen surface in the direct compressive-shear test is shown in Fig.7.Due to a large number of images,the analysis of the strain started from 200 s and was limited to an area of one-third of the surface around the center of the specimen.At 200 s,the shear stress was approximately 50%of the peak shear strength without any obvious deformation on the specimen surface,while the strain was approximately 0.002-0.009.At 232 s,the shear stress was approximately 70%of the peak shear strength,and the strain appeared localized along with a band on the left side of the specimen.Until 235 s,the thickness of the deformation band expanded further without further penetration towards the right specimen side.The strain magnitudes reached 0.03.At 245 s,there was an obvious deformation band on the right side of the specimen,which continued to expand laterally until 270.59 s.The shear stress reached 90%of the peak shear strength at that time.The left and right bands approach each other,associated with the formation of many cracks on the specimen surface.The maximum strain reached 0.13 at this time.At 280 s,the stress reached the peak shear strength.Afterward,shearing continued until an ultimate stage.At this time,there were three obvious cracks on the right side of the specimen,causing some spalling at the rock surface.The results of DIC analysis suggest three stages of crack formation:(1)At low shear stresses,stress concentrations occurred on the left side of the specimen,causing the formation of tensile and shear micro-cracks;(2)With increasing load,the tensile cracks initiated on the right side of the specimen;(3)Before failure,these cracks extended further until the shear coalescence leading to the ultimate failure of the specimen.

Fig.8 shows the temperature evolution on the surface of specimens in the direct compressive-shear test.Combined with theanalysis of the temperature-time curve in Fig.6,the results show that the surface temperature of the specimens increased continuously with increasing lateral load.At 200 s,a higher-temperature area appeared on the left side of the specimen.At 240 s,the higher-temperature area on the left side expanded further towards the specimen’s center,and another higher-temperature area appeared on the right side of the specimen.The right side temperature was about 0.3°C higher than that on the left side.The different temperatures may be associated with energy dissipation due to crack growth at the left side of the specimens,whereas the energy on the right side kept accumulating.At 260 s,the highertemperature areas on the left and right sides began to coalesce.At 280 s,the specimen failed.In the post-peak range,frictional slip continued between the upper and lower crack surfaces,and the temperature continued to increase by 0.2°C until 410 s.The spatial evolution of temperature and strain on the surface of specimens were basically consistent during the test,suggesting that temperature measurements can be utilized to detect local strain/stress concentrations and failure processes.

Fig.4.Fabrication process of the rock specimen with a speckle pattern on one surface.

Fig.5.Monitoring equipment and monitoring scheme of tests.

Fig.6.Relationships among shear stress,AE counts,AE amplitude,temperature,and time in the direct compressive-shear test.

Fig.7.Strain evolution on the specimen surface in the direct compressive-shear test:(a)200 s;(b)230 s;(c)232 s;(d)235 s;(e)245 s;(f)270.59 s;(g)270.6 s;and(h)280 s.

3.3.2.Direct tensile test

The cubic specimen and the auxiliary loading equipment were installed as shown in Fig.3c,and the specimen was placed in the designated position.A normal tensile load was applied to the specimen,and the loading rate was set to-0.5 kN/s until the specimen failed.

The relations between tensile stress,AE counts,AE amplitude,temperature,and time are shown in Fig.9.Before 22 s,it is the pre-tightening stage between the several connecting parts and the specimen and not the real tensile test.Hence,Fig.9 just shows the real test stage wherein the applied loading ramp was stable,and the tensile stress exhibited a linear increase with respect to time.When the stress reached 0.77 MPa,the specimen failed followed by a substantial tensile stress drop,which emanated an audible sound and a large number of AE signals generated about 0.8 s prior to the failure.It mainly indicated that a large number of cracks were initiated in the specimen before failure.There were two obvious AE signal peaks in this tensile test at 28 s and 38 s,which nearly corresponded to the two peaks in the tensile stress-time curve,indicating that micro-crack formed in the specimen before the tensile failure.The AE amplitudes were mainly in the range of 50-75 dB,and a small number of AE signals with higher amplitudes of 80-100 dB were generated at the peak tensile stress.The amplitudes range of the AE signals in the direct tensile test was higher than that in the compressive-shear test.Further,there was no quiet period prior to failure in this direct tensile test.

Fig.8.Temperature evolution on the specimen surface in the direct compressive-shear test:(a)0 s;(b)200 s;(c)240 s;(d)260 s;(e)280 s and(f)410 s.

Fig.9.Relationships among tensile stress,AE counts,AE amplitude,temperature,and time in the direct tensile test.

The strain evolution on the specimen surface in the direct tensile test is shown in Fig.10.The strain analysis started from 20 s and was limited to an area of one-third of the surface around the center of the specimen.At 23 s,the tensile stress was approximately 15%of the peak tensile strength.At 30.5 s,the tensile stress reached 80%of the peak tensile stress,and tensile strain concentrations appeared at the specimen surface(areas A,B,and C in Fig.10).The strain reached a magnitude range of 0.05-0.09.However,there were no macroscopic cracks visible on the rock surface.According to the AE count-time curve in Fig.9,a certain number of micro-cracks occurred inside the specimen at this time.At 34.4 s,strain localized on the left side of area A where the strain was approximately 0.02-0.04.With increasing tensile stress,the strain localization in the center and right side of the specimen(areas B and C)increased further to values of approximately 0.08-0.12.Then,the strain localizations expanded rapidly.At 34.49 s,areas A and B coalesced,and the maximum strain reached 0.14.Shortly,0.01 s later,the specimen failed along a tensile crack as shown in the image at 34.5 s.The DIC image shows that significant strain localizations illuminating the ultimate crack surface only appeared when the tensile stress was close to 95% of the peak tensile strength.Before that,it was impossible to effectively infer the position of the tensile crack from spatially distributed strain localizations.

The temperature evolution at the specimen surface in the direct tensile test is shown in Fig.11.In combination with the temperature-time curve analysis in Fig.9,the results show that the surface temperature increased with the increasing vertical load.However,the temperature change was only 0.05°C compared to the initial temperature.After failure,the surface temperature abruptly dropped by 0.2°C.As observed during compressive shear tests,no temperature band with elevated temperatures was observed during the tensile test.

Fig.10.Strain evolution on the specimen surface in the direct tensile test:(a)20 s;(b)23 s;(c)30.5 s;(d)32 s;(e)34.4 s;(f)34.48 s;(g)34.49 s and(h)34.5 s.

A valid direct tensile test should result in failure at the midpoint of the specimen(Perras and Diederichs,2014),therefore,this direct tensile test can be thought as valid.However,the cracking path is worth noticing since it has undulating surfaces and is inclined to the loading direction.Based on the previous studies(Shang et al.,2016;Komurlu et al.,2017;Kourkoulis et al.,2018;Savino et al.,2018),we selected several typical failure profiles of specimens under the direct tensile tests,as shown in Fig.12a-d.The failure profiles have different features,not only with relatively flat surfaces,but also with undulating surfaces.It is believed that the most reasonable explanation is the heterogeneity of the specimen and the fact that its inner micro-cracks are not mathematical planes normal to the tensile loading axis(Kourkoulis et al.,2018).To further validate the effectiveness of the apparatus as for the direct tensile test on the specimen with macro-crack,we also conducted a direct tensile test on a jointed specimen,and its failure characteristic is shown in Fig.12e,which also meets the requirement of the valid test(Perras and Diederichs,2014).Compared with the intact specimen,the jointed specimen also created inclined cracks near the joint tip,and the cracks run through the rock bridge in an approximately horizontal direction.The jointed specimen’s tensile strength is 0.483 MPa,slightly less than 75%of the tensile strength of intact specimen.The test results indicated that the new test apparatus could perform a direct tensile test for cubic rock specimens successfully.

3.3.3.Direct tensile-shear test

Based on the direct tensile test results,the vertical tensile stress in the direct tensile-shear test was set to approximately 20%of the tensile strength of the specimen.The direct tensile-shear test was performed under a constant normal tensile force of-3 kN.The cubic specimen and the auxiliary loading equipment were installed as shown in Fig.3b,and the specimen was placed in the frame.Afterward,the tensile normal load was applied at a loading rate of-0.2 kN/s.After the predetermined normal load was reached,the lateral load was applied under displacement control at a rate of 1 mm/min until the specimen failed.

The recorded AE counts,AE amplitude,and temperature changes versus time are shown in Fig.13.At the beginning of loading,the shear stress and time exhibited a certain nonlinear relation associated with the adjustment of the system.After 22 s,the shear stress increased linearly with respect to time,accompanied by few AE signals.When the shear stress reached 2.76 MPa,the specimen exhibited shear failure.The shear strength of the specimen was significantly lower than that under the compressiveshear stress.When the shear stress approached the peak strength,a large number of cracks occurred,associated with a large number of AE count(i.e.9500).Prior to failure,the stress-time curve remained almost linear and failure occurred instantaneously,which was different from the compressive-shear test.After the specimen failure,the shear stress and AE signals decreased significantly.The AE amplitudes were mainly in the range of 45-70 dB.The highest AE amplitude did not appear at failure and AE count peak,but at the shear stress reaching 88% of the peak strength.At that time,the AE amplitude was approximately 80-95 dB.Compared with the AE responses in the direct compressiveshear and direct tensile tests,the amplitude of the AE signals in the tensile-shear test was between the two aforementioned tests.There was a certain quiet period before tensile-shear failure,but it was far less obvious than that in the compressive-shear test.

The strain evolution on the specimen surface in the tensileshear test is shown in Fig.14.The strain analysis started from 54.5 s and was limited to an area of one-third of the surface around the center of the specimen.At 54.5 s,the shear stress was approximately 85% of the peak strength.At this time,the surface strain was relatively low and within the range of 0.0004-0.003.At 59.31 s,there was an obvious strain concentration on the left side of the specimen near the shear box.After an additional 0.05 s,spalls occurred on the left side of the specimen near the shear box,which decreased the DIC analysis area(i.e,the speckles in that area could no longer be captured).The strain increased further and was approximately 0.005-0.02.At 62.26 s,deformation bands developed at the right side of the specimen,and several cracks appeared on the specimen surface.At this time,the shear stress was approximately 97%of the peak strength,and the strain was around 0.02-0.04.Then,0.01 s later,deformation bands appeared on the left side of the specimen with strains up to 0.03-0.05.At the same time,the deformation bands on the right side expanded further with strain up to 0.08-0.15.Visible cracks appeared on the specimen surface,followed by tensile-shear failure.Compared with the response in the compressive-shear test,the deformation bands appeared later in the tensile-shear test,similar to the response in the direct tensile test.However,the size of the bands was significantly larger in the tensile-shear test than those in the other two tests.The AE signal analysis showed that large-scale cracks might have developed at the lower shear stress level because a certain amount of AE signals with higher amplitudes(＞60 dB)were captured before 25 s.In the previous research,the AE amplitudes were also usually used to estimate the crack scale(Christl et al.,1989;Li et al.,2020).Compared with the tensile-shear tests,the AE amplitudes were mainly below 60 dB at the early loading stage in the compressive-shear and direct tensile test.As the number of cracks continued to increase with increasing shear stress,shearing coalescence gradually formed a potential crack surface,approximately normal to the applied tensile stress.After that,the ultimate tensile crack formed,which may explain that the highest AE amplitude did not appear at the AE count peak.

The temperature evolution on the specimen surface in the direct tensile-shear test is shown in Fig.15.Combining the analysis of the temperature-time curve in Fig.13,the results show that the initial temperature of the specimen surface slightly increased as the shear load increased.After 30 s,the temperature of the specimen surface decreased continuously as cracks continued to appear on the specimen surface.When the specimen failed,the temperature abruptly dropped.However,the temperature did not change significantly during the whole test,and the range of change remained about 0.1°C.The tensile-shear failure process was similar to that in the direct tensile-test,i.e.there was no obvious temperature response.

Fig.11.Temperature evolution on the specimen surface in the direct tensile test:(a)20 s;(b)30 s;(c)34 s and(d)34.9 s.

Fig.12.Several direct tensile failure profiles from(a)Shang et al.(2016),(b)Savino et al.(2018),(c)Komurlu et al.(2017),and(d)Kourkoulis et al.(2018);and(e)The test on the jointed specimen using the new apparatus.

Fig.13.Relationships among shear stress,AE counts,AE amplitude,temperature,and time in the direct tensile-shear test.

Fig.14.Strain evolution on the specimen surface in the direct tensile-shear test:(a)54.5 s;(b)59.31 s;(c)59.36 s;(d)59.5 s;(e)62.25 s;(f)62.26 s;(g)62.27 s;and(h)62.28 s.

Fig.15.Temperature evolution on the specimen surface in the direct tensile-shear test:(a)0 s;(b)30 s;(c)60 s;(d)62 s and(d)62.5 s.

4.Discussion

In this paper,a new multifunctional rock testing system with'tensile-compressive-shear'loading functions was developed.To verify the effectiveness of the system,a series of tests was carried out on cement mortar.Three tests were conducted,including a direct compressive-shear test,a direct tensile test,and a direct tensile-shear test,and the test results verified the feasibility and effectiveness of the developed testing system.For the direct tensile test,the most challenging issue was the insufficient adherence between the mortar specimen and the loading system.Compared with the cement mortar specimens,the surface of the real rock is smoother and can more tightly fit the auxiliary loading system.

The tensile strength of the specimen was estimated to be 0.77 MPa,and the compressive strength was 18.29 MPa.The peak shear strength of the specimen from the compressive-shear test was 11.16 MPa at a normal stress of 5 MPa,whereas the peak shear strength of the specimen from the tensile-shear test was 2.76 MPa at a normal tensile stress of 0.15 MPa.The vertical compressive stress enhanced the shear strength of the specimen,whereas the tensile stress provided the opposite effect.With increasing shear load,the cracks increased,accompanied by shearing coalescence of small-scale tensile cracks.Compared with the tensile-shear tests,the propagation of shear cracks was suppressed in the compressive-shear tests due to the effect of vertical compressive stress.For normal tensile stress,shear cracks propagated in a direction approximately normal to the applied tensile stress.

Regarding the response of the temperature during testing,the range of temperature change in the direct compressive-shear test was significantly higher than that in the direct tensile and the tensile-shear tests.When no visible cracks occurred on the specimen surface,the energy accumulated in the specimens and the temperature exhibited an increasing trend.For the vertical compressive stress tests,the crack opening was significantly smaller compared to the direct tensile-shear and the direct tensile tests.Hence,the energy dissipation was smaller and the temperature continuously increased before the specimen failed.Based on the monitoring results and observed phenomena,the spatial temperature evolution on the specimen surface in the direct compressive-shear test may serve as a precursor signal where cracks eventually occur.The results suggest that temperature monitoring may be an effective precursor for damage initiation under compressive-shear conditions.Large-scale tests and field tests will be performed in subsequent studies to further explore the effectiveness of thermal infrared technology for monitoring rock mass failure processes in underground engineering.

A comparison of the AE signals from the three tests showed apparent differences.There was a clear quiet period in the direct compressive-shear test before the specimen failed,which was consistent with the previous studies(Zhou et al.,2018).However,this phenomenon did not appear in the direct tensile test,although a certain degree of AE signal reduction before failure was observed.Further,AE signals in the direct tensile test primarily accumulated before failure.Based on the test results of the three specimens in this research,the peak AE counts were the highest in the direct tensile test and lowest in the compressive-shear test.The AE amplitude distribution in the direct tensile test(50-75 dB)was higher than that in the compressive-shear test(40-55 dB).During testing,tensile failure was accompanied by an audible sound.The AE amplitude in the direct tensile-shear failure was between those in the two aforementioned tests.In addition,the highest AE counts and amplitudes appeared both near the peak strength in the direct compressive-shear and the direct tensile tests.In contrast,in the tensile-shear test,the highest value of AE counts appeared near failure and the highest amplitude value at approximately 80%of the peak shear strength.This finding indicates significant tensile breakages in the later loading stage of the tensile-shear test process.

We also observed differences in the spatial and temporal strain distribution on the specimen surface under the three different stress states.In the direct compressive-shear test,a conspicuous deformation band appeared on the specimen surface when the shear stress was approximately 70% of the peak shear strength.Regarding the other two tests,the phenomenon of the shear band appeared near 95%of the peak strength.For all tests,the maximum local strain before failure was approximately 0.15.

5.Conclusions

This research developed a new rock testing system with“tensile-compressive-shear”loading functions to conduct laboratory tests on specimens under different stress states.Preliminary tests,including the direct compressive-shear tests,the direct tensile tests,and the direct tensile-shear tests,were performed to verify the effectiveness of the system and the evolution of AE signals,DIC strain data,and thermal infrared images were captured during the testing process.Some conclusions drawn from this study are summarized hereafter:

(1)The developed rock testing system adopted various new equipment to realize highly accurate simulations of tensile,compressive,and shear stresses,and enabled direct testing of cubic specimens with multiple dimensions.The newly designed auxiliary loading equipment with an open-type design ensured a larger specimen surface exposed during testing,which enabled it to be visually observed using a high-speed camera,a thermal infrared camera,and other optical monitoring equipment.

(2)Before failure,only the compressive-shear test showed a gradual failure process with the onset of nonlinear behavior.The direct tensile and the direct tensile-shear tests showed abrupt failure with a distinct post-failure stress drop.There were obvious differences in the shape of the shear stresstime curves from the direct tensile-shear and the direct compressive-shear tests.The former showed strong brittle failure characteristics and tensile cracks produced little debris in the test process.

(3)There was an obvious quiet period when the failure was almost reached in the direct compressive-shear test,while the phenomena were unobvious in the direct tensile-shear test and did not occur in the direct tensile test.The peak AE amplitude of the direct tensile-shear test was obtained when the shear stress reached 80% of the peak shear strength,distinguished from the nearly failure time in the other two tests,indicating that the tensile cracks contribute much to failure during the later loading period in the direct tensile-shear test.

(4)The deformation concentration bands in the compressiveshear test formed earlier than those in the other two tests in terms of stress level.The maximum local strain prior to failure was approximately 0.15 in each test.The temperature response on the specimen surface was obvious only in the compressive-shear test and temperature anomaly occurred earlier than the AE signals,thereby providing obvious precursory characteristics.In-depth research is still needed on the utilization of thermal infrared imaging technology to monitor rock mass stability in actual engineering applications.

The performance of this testing system has been successfully assessed on the rock-like specimens and further experiments on the real rock specimens will be conducted in the next stage.In addition,the investigation on the scale effect on the failure behaviors of specimens with different sizes and the assessments on rock behaviors for the stress conditions at deep underground excavations will be our research focus by using this new test system.

Declaration of competing interest

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

Acknowledgments

We acknowledge the funding support from the National Natural Science Foundation of China(Grant Nos.U1806226 and 51979154).The experiments were conducted in the Research Institute of New Material and Intelligent Equipment(Co-funded by Shandong University and Qihe County).Much appreciation to the engineers who provide much convenience to us in the experiment process.