Experimental investigation of the cementation and tensile properties of cement to formation interface considering surface irregularity and drilling mud residue

Cement sheath is an annular structure situated between casing and formation, which is designed to seal the formation and reinforce the borehole. However, owing to the surface irregularity and drilling mud residue, the interface between the cement sheath and the formation is one of the weakest parts for fluid leakage causing well integrity issues, which would also affect the reliability of the cement bond logs. To evaluate the cementation state and tensile properties of the cement-formation interface, rock-cement composite disc specimens are prepared, considering key factors such as lithology, interface irregularity, treatment of oil-based drilling mud contamination, and flushing fluid washing. Flushing efficiency of the drilling mud contaminated rock surface, interface cementation characteristic, referenced tensile strength, and tensile fracture morphology are acquired and analysed comparatively. Results show that (1) the flushing efficiencies of shale and sandstone are 74.1 ~ 61.9% and 52.4 ~ 28.4%, respectively, decreasing with the increase in surface irregularity. (2) The rock-cement interface gradually becomes poorly cemented as the surface cleanliness decreasing and the irregularity increasing. (3) Under the condition of drilling fluid contamination, there exists a transitional area at the cemented interface, where microcracks and pores are developed. (4) With the influence of drilling mud contamination, the referenced tensile strengths of shale-cement and sandstone-cement interfaces drop from 2.716 to 0.586 MPa and from 2.840 to 0.007 MPa, respectively. The effects of underground temperature and pressure environment are not particularly considered in this study and would be investigated systematically in future research.


List of symbols BTS Brazilian tensile strength (MPa) CV
Coefficient of variation (dimensionless) D Diameter of the test specimen (mm) E Young's modulus (GPa) F tc Critical force at tensile fracture initiation (kN) m 0 Initial mass of the half rock disc before drilling mud contamination and flushing (g) m 1 Mass of the half rock disc after depositing and pouring out the surface flowing drilling mud (g) m 2 Mass of the half rock disc after flushing fluid washing (g) P Load corresponding to the tensile fracture initiation at the centre of the specimen (N) T Thickness of the test specimen (mm) UCS Uniaxial compression strength (MPa) η Flushing efficiency (%) μ Poisson's ratio (dimensionless) σ ft Referenced indirect tensile strength of the interface (MPa)

Introduction
During the process of primary cementing, cement slurry is pumped into the well to fill the annular space between casings, or between casing and formation. After solidification, the annular cement sheath has the function of mechanically stabilizing the wellbore and preventing pressurized formation fluids outside the casing from entering the well or flowing between different subsurface zones (Lavrov and Torsaeter 2016). The persistent integrity of the casing-cement-formation system is critical for the long-term safe operation of the wellbore in the fields of hydrocarbon exploitation (Gray et al. 2009;Yousuf et al. 2021) and underground energy storage (Raju and Kumar Khaitan 2012;Liu et al. 2020).
From the viewpoint of casing-cement-formation system ( Fig. 1a), loss of zonal isolation from cement sheath could emerge at three potential locations according to field cement bond logs (Ashena et al. 2014) and laboratory investigation (Xu et al. 2018): (1) the body of cement sheath, (2) the first interface (casing-cement interface), and (3) the second interface (cement-formation) interface. Compared with the former two locations, the cement-formation interface is in a relatively more complex state and has the potential to provide a leakage path, which is a key link for the integrity of the whole underground energy storage or reservoir system. In the stage of well drilling, the formation is exposed to drilling fluid. Then spacers and chemical washes are pumped ahead of the cement slurry for fluid separation and hole cleaning. Ideally, the drilling process leaves a formation surface to which the cement can bond and provide zonal isolation (Ladva et al. 2005). But the actual situation is that the wellbore wall surface is usually undulating, and the drilling fluid is very difficult to be totally removed. The residual filter cake in the grooves hinders the contact and cementation between cement slurry and formation (Fig. 1b). Finally, a natural weak interface is formed. The subsequent operation, such as hydraulic fracturing, periodic injection and production, et al., would inevitably further degrade the cement-formation bonding and eventually lead to a continuous leakage path.
Previous studies mainly focused on the physical and mechanical properties of cement (Quercia et al. 2016;Li et al. 2017;Wang et al. 2018;Xu et al. 2022;He et al. 2022), researches relating of the cement-formation interface is relatively few and limited. Ladva et al. (2005) considered the presence of filter cakes on the permeable sandstone, and they found that the cementing to rocks of pristine conditions gave shear bond strength in the order of 120 psi, whereas the presence of filter cake decreased the value down to 0.1-1 psi. Opedal et al. (2014) quantified the effect of various rock formation types on the cement-formation bonding, and the effect of altering the rock surface with various drilling fluid films before cementing. It was shown that the presence of drilling fluid, and drilling fluid type, was more crucial parameters for cement-formation bonding. Plank et al. (2014) evaluated the properties of the mudcakecement and mudcake-formation interfaces and their effects Fig. 1 (a) The composition of casing-cement-formation system and (b) the irregular cement-formation interface with residual drilling mud 1 3 on the bonding strengths and sealing quality of the cement. Synthetic ceramic core samples were employed to simulate reservoir rock. Liu et al. (2015) evaluated the shear bond strength of the cement-shale interface using a new testing method, without considering the interface irregularity and drilling fluid residue. Lian et al. (2020) performed numerical simulation of cement-to-formation interface debonding. Li et al. (2022) investigated the evolution of permeability of cement-salt rock specimen under cyclic unloading-loading confining pressure. Taghipour et al. (2022) established a novel laboratory setup for realistic wellbore cement and formation integrity by implementing in-situ X-ray Micro-CT imaging. Li et al. (2016) tested the contamination of oil-based mud and its components on the performances of cement slurries, such as fluidity, thickening time, compressive strength, bonding strength, porosity, and permeability. High performance additives were also developed to improve the mechanical properties and bonding strengths of cements (Mohamadian et al. 2020;Mansour et al. 2022).
The above literatures mainly reported the shear bond strength of cement-formation interface, few discussed the aspect of interface tensile debonding (Garcia Fernandez et al. 2019), which was a typical failure mode induced by cement contraction and temperature/pressure fluctuation (Shenold and Teodoriu 2016;Bois et al. 2011). Besides, the surface irregularity of formation was also rarely considered. In the field of concrete research, tensile failure of rock-concrete bi-material discs has been investigated experimentally Qiu et al. 2020a) and numerically Qiu et al. 2020b), which could provide reference to the study of tensile debonding of the second interface.
The literature review showed that the cementation and tensile properties of the cement-formation interface under the influence of surface irregularity and drilling mud contamination have not been well studied. In this study, the shale/sandstone-cement composite specimens are prepared considering the interface irregularity and cleanliness to simulate the cementation state of the second interface in wellbore system. Flushing efficiency of rock surface after drilling mud contamination and flushing fluid cleaning is calculated. The interface cementation state is characterized in mesoscale. The referenced values of indirect tensile strength and fracture morphology under different conditions are acquired and analysed comparatively. Results would be valuable for the understanding of cementation state and tensile capacity of the real underground second interface.

Methodology
This study provides a laboratory method to evaluate the cementation and tensile property of the cement-formation interface, which could simulate the effect of surface irregularity, drilling mud contamination, and flushing fluid washing. The procedures developed to successfully execute the experimental part of the study are summarized as follows ( Fig. 2): (1) Preparation of half disc rock specimen. The irregularity of the rock interface could be considered and customized.
(2) Drilling mud contamination of the rock interface. Pouring drilling mud on the rock interface. Temperature, pressure, and deposited time could be considered during drilling mud contamination. (3) Flushing fluid washing of the rock interface. The residual drilling mud on the rock interface is removed by flushing fluid, and the final flushing effect could be evaluated by flushing efficiency. (4) Preparation of rock-cement disc specimen. Pouring the mixed cement slurry into the mould, which contains the treated half disc rock specimen. Set the temperature, pressure, and time of the curing conditions. (5) Observation of the meso-structure of the rock-cement interface. The initial state of the cement-rock interface is characterized by the industrial microscope. (6) Testing the indirect tensile strength of the rock-cement interface. The interface tensile strength and fracture morphology could be acquired.

Sample preparation
The Brazilian disc specimens, which contain the rockcement interface, are designed to investigate the cementation strength of the interface by implementing indirect tensile test. Factors, such as lithology, interface irregularity, and interface contamination and flushing, are the major considerations. Shale and sandstone outcrops are selected as the rock part in the preparation of rock-cement composite specimens. Because they are commonly encountered lithology during well drilling. Specifically, the shale is derived from outcrops of a shale gas reservoir in Sichuan Basin, China. And the sandstone is collected from outcrops of a tight sandstone gas formation in Ordos Basin, China. According to the result of X-ray diffraction analysis (Fig. 3a), the selected shale mainly contains quartz (69.78%), albite (11.40%), and illite (10.93%). Because of the high quartz content, its basic mechanical properties show a feature of high strength and high Young's modulus, which are 98.5 MPa (UCS), 13.0 (BTS) and 24.10 GPa (E), respectively (Table 1). In comparison, the sandstone is dominantly composited of calcite (47.76%), quartz (20.19%), and albite (15.65%) (Fig. 3b). And the corresponding mechanical parameters are relatively low, with the UCS of 41.2 MPa, BTS of 4.4 MPa, and Young's modulus of 6.23 GPa ( Table 1).
The other half of the composite specimen is cement stone. Class G oil well cement, which is widely used in well cementation, is chosen as the main raw material. The cement slurry formulation is: Class G oil well cement (100%) + Water (40%) + Fluid loss additive (4%) + Defoaming agents (0.25%). Percentages in brackets are mass fractions. The above cement slurry formulation is commonly used in the cementation of shale gas wells in southwestern China. All raw materials are industrially produced and collected from the cementing site. The basic mechanical parameters of cement stone after curing of 28 days are also listed in Table 1.
The preparation procedure of rock-cement composite specimens could be divided into 4 steps.
Firstly, the rock disc specimens, with diameter of 50 mm and thickness of 25 mm, are cut along the centreline using a computer numerical control diamond wire saw (Fig. 4a). The cutting path is designed as three types: (1) straight line, (2) zigzag line with amplitude of 1 mm, and (3) zigzag line with amplitude of 2 mm ( Fig. 5a and b). The straight-line cutting path would manufacture half circular discs with flat and smooth interface, which could simulate the ideal condition of the wall of a wellbore. Interface with zigzag trace of amplitude of 1 mm is an approximate simulation of the irregularity state of real wellbore wall. Zigzag trace with larger amplitude of 2 mm could be regarded as the wall of broken stratum, which would usually exhibit greater irregularity. The prepared half circular rock discs with different surface irregularity could be seen in Fig. 5c and d. In fact, we could not directly observe the real internal wall of underground open hole. So, the complete simulation of the irregularity of the real formation wall is very difficult. But the  observation of collected downhole full diameter cores could help us infer the surface irregularity of the wall. For the formation with good integrity, the collected cores are usually intact, and the surface of cylinder side are relatively smooth with small roughness. For the formation with poor integrity, the collected cores are broken, and the surface becomes rougher. The dimensions of hole collapse blocks could range mm ~ cm. Based on the above observation and inference, the simplification is applied as a compromise method. Secondly, the cutting interface is contaminated with oil-based drilling mud and washed with flushing fluid (Fig. 4b). The drilling mud and flushing fluid are taken from the rig site of a shale gas well in southwestern China. The oil-based drilling fluid is mainly composed of diesel oil, emulsifying agent, wetting agent, oleophilic colloid, and lime. Oil film and oil-based mudcake would form on the surface of casing and wellbore during the drilling process. Correspondingly, the flushing fluid mainly contains wetting reversal flushing agent, suspending agent, stabilizing agent, anti-swelling agent, and weighting material. Specifically, oil-based drilling mud (10 mL) is poured on the interface of the half rock disc. After deposition of 6 h, the drilling mud is poured out, with some of mud adhering on the interface. Then, the interface with residual drilling mud is cleaned with flushing fluid (10 mL) to sufficiently dissolve and wash the adhered mud. The flushing efficiency could be used to quantitatively evaluate the cleaning effect and expressed by where m 0 is the initial mass of the half rock disc before drilling mud contamination and flushing; m 1 is the mass of the half rock disc after depositing and pouring out the surface flowing drilling mud; m 2 is the mass of the half rock disc after flushing fluid washing.
Thirdly, the half circular rock disc is placed in the cylindrical steel mould, which is applied a layer of mineral oil on the inner wall for easy demoulding. Then, mixed cement slurry is poured into the mould and compacted (Fig. 4c). Finally, put the mould into the water bath (constant temperature of 25 °C) under atmospheric pressure and curing for 28 days, which could guarantee the relatively steady micro-structure and mechanical properties of all specimens. It should be noted that the lower surface of the disc specimen is flat and smooth because it is in contact with the bottom face of the mould. For the upper surface, after pouring cement slurry, a circular plastic plate is applied on it to make it level, and excessive cement slurry is timely removed. After demoulding, the standard rock-cement composite disc specimens (φ 25 × 50 mm) are ready for Brazilian indirect tensile test (Fig. 4d), and the cutting and smoothening processes are not further applied on the upper surface of the specimens to avoid disturbance on the relatively weak cement-rock interface. In addition, all specimens are still placed in the water bath environment after curing, and mechanical tests would be implemented as soon as possible.
One important thing need to note is that the interface should along the central line of the disc specimen as close as possible. Even small offset would induce large error in the measurement of interface tensile strength. Because the tensile fracture might initiate and propagate in the cement or rock part, instead of along the interface. Therefore, high machining accuracy is required in the cutting of half disc rock specimens. In addition, after putting the half disc rock specimen in the mould, its cylinder side should be in close contact with the inner wall of the mould.

Experimental design
Three factors are particularly considered in the Brazilian indirect tensile test of rock-cement composite disc specimens. The first factor is the combination type, in which shale and sandstone are selected as the rock part. The second one is the interface irregularity, which is set to 0, 1, 2 mm based on the description in Sample Preparation. The last factor to be evaluated is the interface contamination and flushing. Here three states were designed: (1) untreated state or fresh interface, with no mud contamination and fluid flushing; (2) oil-based drilling mud contamination and flushing fluid washing; (3) oil-based drilling mud contamination without flushing. The effect of pressure and temperature in the process of drilling mud contamination is also considered. A group of shale-cement and sandstone-cement composite specimens are place in a high temperature and high pressure curing kettle and applied the condition of 20 MPa and 80 °C (approximately simulating the pressure and temperature at the depth of 2000 m) during the drilling mud contamination. The other groups of specimens are all put in the environment of atmospheric pressure and 25 °C. All parameters setting of rock-cement interface is presented in Table 2. Four composite specimens are prepared for each testing condition.

Experimental equipment
The Brazilian splitting experiment is conducted on the RMT-150C rock mechanics test system (Fig. 6a). It is a digital controlled electric servo testing machine, which could perform various rock mechanics tests, such as indirect tension, uniaxial compression, triaxial compression, direct shear, and so on. Under the mode of Brazilian splitting test, the maximum value of vertical force is 100 kN, with a precision of 0.001 kN. The displacement of vertical piston is 0-50 mm. The control range of deformation rate is 0.0001-1.0 mm/s. A rate of 0.002 mm/s is adopted to fulfill quasi static loading. The ISRM (International Society for Rock Mechanics) suggested loading type (IRSM 1978) is applied by using upper and lower jaws to transfer vertical load, which could better reduce stress concentration at the contact areas ( Fig. 6b and c).  The referenced indirect tensile strength of the interface could be calculated by where P is the load corresponding to the tensile fracture initiation at the centre of the specimen, N; D is the diameter of the test specimen, mm; t is the thickness of the test specimen, mm. It should be noted that formula (2) is valid for homogeneous and intact material. Strictly speaking, it could not be used for composite disc with a weak interface in this study. Some researchers choose the peak force to indicate the bearing capacity of the half rock and half mortar specimens , while others still use the above equation to calculate the interface tensile strength (Qiu et al. 2020a;Liu et al. 2015). Here, we provide both the critical force at tensile fracture initiation and the calculated tensile stress, which is called the referenced tensile strength and is not the actual value of interface tensile strength.

Analysis of flushing efficiency
Although the conditions of drilling mud contamination and flushing fluid washing are same for each specimen, the final calculated values of flushing efficiency are different for the discrepancy in lithology and surface irregularity. As collected in Table 3, the flushing efficiency decreases with the increase in surface irregularity. Specifically, as the irregularity increases from 0 to 2 mm, the flushing efficiency of shale decreases from 74.1 to 61.9%. Larger grooves would arrest more residual drilling mud and impede fluid washing, reducing the flushing efficiency. Compared with that of shale, sandstone surface shows a further reduction in flushing efficiency. At the condition of flat surface, the flushing efficiency is just 52.4%. At the irregularity of 2 mm, the flushing efficiency drops sharply to 28.4%. The flushing effect is also reflected in the surface state variation of half disc rock specimens during drilling mud contamination and flushing fluid washing. As recorded in Fig. 7, the rock surfaces are covered with residual mud after drilling mud contamination. Residual mud converges on the concave locations of the surface. The greater the irregularity, the more residual drilling mud on the surface. Besides, the mud in the sandstone surface is obviously thicker than that of shale surface. After flushing fluid cleaning, the fresh surface of shale reappears, with most of residual mud washed out. Only a small amount of mud and flushing fluid is left on the surface. While things are different for sandstone. The flushing fluid only washed away the upper surface layer of drilling mud, an appreciable amount is still adhered on the surface. In comparison with shale, oil-based drilling mud is more prone to be attached on the sandstone surface and is very difficult to be totally removed, leading to a lower flushing efficiency. The reason is that the permeability of the selected shale and sandstone is remarkably different. Previous laboratory measurements show that the permeability of the shale is 0.208 × 10 -3 mD , while the permeability of the sandstone is relatively much higher and reaches 0.76 mD ). The extremely low permeability of shale reflects that its microstructure is also dense and compact, which could prevent the deposition and adhesion of drilling mud particles. On the contrary, the permeable sandstone has relatively large pore structure, which is convenient for the adsorption of drilling mud.
The flushing efficiency decreases after the exertion of pressure and temperature (20 MPa-80 °C), which could

Interface cementation in meso-scale
After curing and before Brazilian indirect tensile test, the interface cementation state of rock-cement composite discs is observed and recorded under the industrial microscope. The factors of interface cleanliness, irregularity, and lithology could remarkably influence the final cementation quality, which would be discussed in detail.

Interface cleanliness
The comparison of Fig. 8a, b, and c presents the changes in surface cleanliness on the shale-cement interface cementation. When the shale surface is in a fresh and untreated state, the cement slurry could be in fully contact with shale during setting and hardening period. The cement glues tightly with shale with no obvious micro pores or cracks along their interface (Fig. 8a). This is an ideal cementation state for cement-rock interface. When the shale surface is treated with oil-based drilling mud contamination and flushing fluid washing, the cemented interface is not strictly a curved line, but a striped area with a width of approximately 440 μm (Fig. 8b). Within this area, some defects, such as micro cracks (width of 10 μm) and pores (diameter of 175 μm), are observed. Although flushing fluid could remove most of the residual drilling fluid, there is still a thin layer adhering to the surface. Besides, some flushing fluid would inevitably leave on the surface. These two parts would participate in the cement hydration and hardening process, leading to a striped interface cementation area with some micro flaws. For the surface with drilling fluid contamination and no flushing, the residual drilling mud is relatively thick and distributed unevenly, more gathering in the depressions and less in the protrusions. Consequently, the cement-shale interface is poorly cemented. As shown in Fig. 8c, residual drilling mud mixes with cement and piles up at the low-lying position with a thickness of 855 μm. Above this is a loose cemented area with thickness of 710 μm.

Interface irregularity
As shown in Fig. 8d, b, and e, the surface irregularity increases progressively from 0 to 2 mm. At the irregularity of 0 mm, the drilling mud is not easy to remain on the flat surface and would be sufficiently flushed away. Therefore, the interface is tightly cemented with a thickness of about 150 μm. When the irregularity is set to 1 mm, the thickness of interface cemented area increased to 440 μm, accompanied with the occurrence of microcracks and pores. With the irregularity increases to 2 mm, the interface is not fully cemented, and microcracks with thickness of 100 μm are distributed continuously. This more pronounced concavity is prone to the adhesion of residual drilling mud and adverse to the fluid flushing.

Lithology
As illustrated in Fig. 8b and f, the interface cementation states of shale-cement and sandstone-cement are provided. At the same conditions of surface irregularity and drilling mud contamination and fluid flushing treatment, the sandstone-cement interface cementation state is even worse. The cracks are wider (thickness of 683 μm) and almost interconnected, leading to very limited contact or cemented area. For the reason of surface wettability and larger porosity, oil-based drilling mud could adhere tightly to the sandstone and is quite difficult to clean. The residual drilling mud is massive and almost completely block the establishment of interface cementation.

Indirect tensile strength
The referenced tensile strength of shale-cement interface is controlled by both interface cleanliness and irregularity. As shown in Fig. 9a, under the same condition of interface irregularity, the interface tensile strength decreases sharply with the increase in residual mud. When the shale surface is fresh and untreated before cement pouring, the interface tensile strength is 2.716 MPa, which is close to the value of cement matrix. After treatment of drilling mud contamination and flushing fluid washing, the tensile strength of shale-cement interface declines dramatically by 78.4% and is 0.586 MPa. If the residual drilling mud is unwashed, the interface tensile is only 0.007 MPa, almost completely losing its tensile capacity. For specimens with same interface contamination and flushing treatment but different irregularity, the interface tensile strength is distributed in the range of 0.387-0.745 MPa and increases slightly with the rise of surface irregularity. The increase in surface irregularity brings larger contact area, which could improve the interface tensile property. With the exertion of pressure and temperature (20 MPa-80 °C), the tensile strength decreases from 0.586 to 0.521 MPa. The performance of sandstone-cement interface in tensile strength is different from that of shale-cement interface, which is mainly manifested in specimens treated by drilling mud. For composite specimens with untreated sandstone surface, the interface tensile strength is 2.840 MPa (Fig. 10) tensile strength decreases from 0.007 to 0.006 MPa. It could be inferred that, whether wash or not, the tensile strengths of sandstone-cement interface are both very weak after oilbased drilling mud contamination. The reason is that the residual drilling mud is firmly adhere to sandstone surface and flushing fluid could not effectively remove it. The critical forces at tensile fracture initiation and the referenced tensile strengths are collected in Table 4.

Tensile fracture morphology at the interface
The tensile fracture morphology is closely related to the lithology, interface cementation and irregularity state. Tensile fractures mainly initiate and propagate along the interface. But the specific features are different for specimens at each condition.
For shale-cement specimens with same surface contamination and flushing treatment but different surface irregularity (Fig. 11a, b, and c), tensile fractures extend strictly along the interface. Further observation on the two complementary fracture surfaces shows that the shale part remains intact, with some adherent cement spreading on the surface, which implies that effective cementation has been established at local limited areas of shale-cement interface. Most other areas are poorly cemented for the drilling mud contamination and incomplete flushing, where a thin layer of dried drilling mud could still be spotted. When the shale surface is fresh and untreated, the tensile fracture is mainly in the cement next to the interface (Fig. 11d). Because the cement is firmly glued with shale and has relatively lower tensile strength. At the condition of surface contamination with no flushing, the tensile fracture is along the interface, and the grooves of shale and cement are full of unconsolidated drilling mud (Fig. 11e). Residual drilling mud stands between shale and cement, hindering the formation of interface cementation.
The tensile crack paths of sandstone-cement discs are similar with that of shale-cement. But the state of fracture surface is different. In the range of sandstone-cement discs with the treatment of mud contamination and flushing, the volume of residual mud grows with the increase in surface irregularity. At the irregularity of 0 mm, although the residual mud is relatively very few, but it spreads all over the sandstone surface, preventing the effective interface cementation (Fig. 11f). At the irregularity of 1 mm, the fluctuation of tensile fracture surface becomes gentle, because the grooves of sandstone fracture surface are filled with residual drilling mud (Fig. 11g). This phenomenon also exists in condition of irregularity of 2 mm, where more drilling mud deposits in the larger depressions (Fig. 11h). At the condition of fresh and untreated surface, the interface is firmly bonded, and the tensile fracture extends in the cement part closely adjacent to the interface, which is very similar to that of the shale-cement specimen (Fig. 11i). When the sandstone surface is treated with mud contamination and no flushing, the tensile rupture surface is a nearly flat plane,  0.008 4.07 × 10 -3 ST-2-3-3 0.010 5.09 × 10 -3 ST-2-3-4 0.006 3.06 × 10 -3 ST-P-2-2-1 0.011 5.60 × 10 -3 0.006 0.228 ST-P-2-2-2 0.014 7.13 × 10 -3 ST-P-2-2-3 0.015 7.64 × 10 -3 ST-P-2-2-4 0.008 4.07 × 10 -3 as the residual drilling mud almost completely covers the irregularity surface of sandstone (Fig. 11j).

Conclusions
By considering the interface irregularity and cleanliness in preparation of the rock-cement disc specimens, the Brazilian indirect tensile tests are implemented. The flushing efficiency of rock surface, interface cementation state, indirect tensile strength, and tensile fracture morphology are recorded and analysed. The conclusions are as follows: (1) The irregular surface is prone to trapping residual drilling mud and hindering fluid washing. The flushing efficiency of shale and sandstone are 74.1-61.9% and 52.4-28.4%, respectively, decreasing with the increase in surface irregularity.
(2) Compared with that of shale, oil-based drilling mud could adhere more tightly to the sandstone and is quite difficult to remove, which is attributed to the relatively higher permeability of sandstones. (3) The rock-cement interface gradually becomes poorly cemented as the surface cleanliness decreasing and the irregularity increasing. There exits an interface cemented transitional area with thickness of 100 μm-1 mm, where micro cracks and pores are developed. (4) After drilling mud contamination, the referenced tensile strengths of shale-cement and sandstone-cement interfaces drop from 2.716 to 0.586 MPa and from 2.840 to 0.007 MPa, respectively. (5) The environment of high temperature and high pressure could aggregate the contamination of drilling mud and reduce the flushing efficiency. (6) Without surface contamination, the tensile fracture is mainly in the cement next to the interface. While Tensile fracture morphology at the rock-cement interface: (a) shale-cement, irregularity 0 mm, drilling mud + flushing; (b) shalecement, irregularity 1 mm, drilling mud + flushing; (c) shale-cement, irregularity 2 mm, drilling mud + flushing; (d) shale-cement, irregularity 1 mm, fresh and untreated interface; (e) shale-cement, irregularity 1 mm, drilling mud; (f) sandstone-cement, irregularity 0 mm, drilling mud + flushing; (g) sandstone-cement, irregularity 1 mm, drilling mud + flushing; (h) sandstone-cement, irregularity 2 mm, drilling mud + flushing; (i) sandstone-cement, irregularity 1 mm, fresh and untreated interface; and (j) sandstone-cement, irregularity 1 mm, drilling mud tensile fractures mainly initiate and propagate along the interface for specimens treated with drilling mud contamination.