Introduction

During the intensive excavation of deeply buried tunnels, the joint network inside a rock mass often tends to be loose or open due to the relief of stress, which in turn can often lead to weaknesses in the entire jointed rock mass. The so-called excavation damage zone (EDZ) is often used to describe this excavation-induced deterioration (Maxwell et al. 1998; Brady and Brown 1985; Millard et al. 2009; Zhu et al. 2014) in the form of joint opening and crack propagation (Weinberger et al. 2010; Li et al. 2010; Zhang et al. 2015; Bonilla-Sierra et al. 2015; Pei et al. 2016). The damage to the jointed rock mass during the unloading process often causes serious problems. The largely reduced modulus of a damaged rock mass leads to inhomogeneous settlement and subsequent deformation of the installed upper structures (Ozcan et al. 2013; Xiang et al. 2015; Kofichko 1996; Justo et al. 2014). Additionally, unloading of the confining stress generally leads to shearing deformation and slipping failure of the whole joint, resulting in collapse or sliding of the jointed rock mass (Sharifzadeh et al. 2010; Maleki 2011; Satıcıo and Unver 2015; Jin et al. 2015).

The actual damage to the jointed rock mass is a process of dynamic evolution that starts under a certain initial geo-stress state (σ1, σ2, σ3) during excavation-associated unloading. It develops with time and slows down at a certain redistributed stress state after excavation. The in situ test provides a key tool to understand the real damage behavior of jointed rock masses, because no laboratory experiments can reproduce the actual geo-stress conditions and the complex in situ engineering boundaries (Watters et al. 2000; Gage et al. 2014; Yarahmadi et al. 2015; Zhan et al. 2016). Many in situ testing methods have been proposed to evaluate damaged rock masses, such as ultrasonic wave testing, borehole camera/television, and micro-seismic monitoring. Meaningful findings on rock mass damage have been reported, such as the opening of existing joints, the generation of new cracks, and the intersections of fractures (Li et al. 2013; Rodriguez et al. 2014; Armand et al. 2014; Scholtès and Donzé 2015). Field and laboratory evidence has shown that damage of a jointed rock mass is a time-dependent behavior due to joint opening, closing and extension (Malan 2002; Fahimifar and Soroush 2005; Shao et al. 2006; Li et al. 2011). However, the available field test data obtained so far about the time-dependent performance of the jointed rock mass EDZ are insufficient. Thus, insufficient data restrict the engineering estimations of jointed rock mass damage in practice and leads to many uncertainties in long-term safety engineering of dam foundations (Nabil and Ala 2001; El-Naqaand Kuisi, 2002; Lin et al. 2015; Fan et al. 2015).

As a typical example of an embedded joint rock mass, the columnar jointed rock mass presents a “special” geological structure. The joint network formed in an ordered colonnade pattern consists of roughly hexagonal units, and is transversely cut by other joint sets (Schultz 1995; Lore et al. 2000; Jiang et al. 2014). Some typical columnar jointed rock sites, such as the Chinese Tengchong, Hong Kong’s High Island Reservoir, and the US Banks Lake, have now become popular landscape attractions for their unique beauty (Lucas and Stephen 2008; Fang et al. 2011), as shown in Fig. 1. It is widely accepted that columnar jointed rock masses originate from lava flows and are formed by thermally induced contraction–cooling (Lore et al. 2000; Budkewitsch and Robin 1994; Müller 1998; Goehring et al. 2006). However, their geotechnical engineering response to unloading excavation has rarely been discussed due to limited engineering practice. The exposed columnar jointed basalt (CJB) at the Baihetan hydropower station in southwest China is a typical columnar jointed rock mass. Due to its undesirable geotechnical properties, such as the jointed structure, brings challenges to the construction of the concrete dam and underground tunnels. Prior investigations and tests have shown that the CJB has a low deformation modulus and strength and thus results in heterogeneous settlement and slope instability for the dam (Yan et al. 2011; Shan and Di 2013; Feng et al. 2016; Zhang et al. 2016). According to the corresponding Chinese national standards, engineering experiences and the Baihetan hydraulic design standards, the classification of the rock mass quality on the dam foundation after excavation must satisfy the “III” rank, and its ideal deformation modulus should be larger than 6 GPa (El-Naqa and Kuisi 2002; PSCC 2003; PSCC 2005; Shi et al. 2008; Wu et al. 2009a; Luo et al. 2015). Therefore, detailed field tests on the depth and corresponding time-dependent evolution of the EDZ are essential for the safety assessment of the concrete dam. In addition, understanding the time-dependent mechanical properties of the CJB after excavation is of great importance for making a timely decision to reinforce the CJB in the dam foundation, such as the installation of rock bolts, anchor cables, grouting, etc.

Fig. 1
figure 1

Typical scenic spot of the columnar jointed rock mass at the Washington Banks lake

Full size image

In the present work, we carried out a detailed investigation of the geometric characteristics and the unloading-induced failure performance of the CJB on the left side of the Baihetan dam base. The field test data were obtained from the ultrasonic wave detection of 30 boreholes in the columnar jointed rock mass. The evolution of the entire damaged depth after excavation, the spatial distribution and time-dependent evolution of the damage degree in the EDZ were also studied using back statistical analyses. The CJB damage mechanism was investigated with the aid of additional field observations by using a digital borehole camera. Reinforcement measurements for the damaged CJB on the Baihetan dam base are then discussed according to the reported field test results.

Background of the Baihetan project

The Baihetan hydropower station is located on the lower reaches of the Jinsha River, China, with its left bank on the side of Ningnan County in Sichuan Province and its right bank on the side of Qiaojia County in Yunan Province. The station, with a total of 1600 MW of installed capacity, is the second largest hydraulic project in China (only smaller than the Three Gorges station). Its concrete dam, designed as a double-curvature arch format, is situated in an unsymmetrical V-shaped valley. The arch dam is 289 m in height, with the crest at 834 m asl, and the normal reservoir water level is at 825 m, as shown in Fig. 2a. Igneous basalt in the Baihetan region is the result of historic magmatic and volcanic eruptions. The rock stratum is typically a monoclinal orientation, with the geological attitude of NE30°–50° in trend and SE∠15°–25° in dip angle. The surface weathering of the rock mass in the dam’s valley is evident. The horizontal weathering depth on the left bank is approximately 67–89 m, and pm the right bank it is relatively smaller, possibly approximately 29–45 m due to the non-uniform unloading effects on the left and right banks. The field measurement of the initial geo-stress ranges from 4.89 to 10.0 MPa in the valley region, whereas under the river bottom, the maximum principal stress can reach approximately 40.32 MPa.

Fig. 2
figure 2

The Baihetan dam foundation: (a) typical geologic transverse profile of the valley; (b) exposed columnar jointed basalt on the left dam site

Full size image

The basalt on the dam site belongs to the upper Permian system (P2β42–P2β33), and the typical CJB exposed on the dam base ranges from 575 to 670 m in altitude on the left bank and from 585 to 545 m on the right bank (Fig. 2b). The CJB, characterized by small columnar blocks, formed the cooling process of the lava from a volcanic eruption in the Permian epoch. The CJB prisms on the left bank are irregular polygons in cross-section, and the axis trends towards the valley with a dip angle of approximately 70°. Prior tentative field tests, such as the in situ triaxial compression tests and the bearing plate test in some experimental tunnels at other positions (Jin et al. 2015), showed that the deformation modulus of the undisturbed CJB approximately ranged from 7 to 11 GPa along the horizontal direction, and from 5 to 9 GPa in the vertical direction. However, after the excavation, the deformation modulus test by bearing plate experiment sharply decreased to a lower range of 3.0–5.2 GPa along the unloading direction due to the induced excavation damage (Shi et al. 2008;Justo et al. 2006; Xu et al. 2011, Cui et al. 2018).

Overall, both the structural strength and the deformation modulus investigation for the CJB indicated the seriously excavation-induced deterioration on the dam base. Hence, the structural characteristics and the time-dependent damage of CJB are key factors for the general stability of the concrete dam and the long-term safety of the reservoir. Therefore, an experimental zone was selected at an altitude of 650–660 m, as shown in Fig. 2b, for a more detailed investigation of the CJB structure and its time-dependent damage behavior after excavation.

Investigation for filed CJB on the dam base

Along with the excavation of the rock mass above the dam base, an in situ investigation was carried out at the altitude of 620–670 m where the CJB was exposed. The geological analysis showed that the CJB mass was cut by three types of joint sets with different physical properties and geometric scales, i.e., a columnar joint set between prisms at the meter scale, an upright internal joint set inside the prism at the decimeter scale, and a horizontal internal joint set inside the prism at the centimeter scale, as shown in Fig. 3. These joint sets consisted of a complicated joint network that deeply affected the mechanical properties of the CJB. The equivalent cylindrical diameter of the columnar blocks and the average space of the horizontal joint sets were key parameters for understanding the whole geometric structure of the CJB.

Fig. 3
figure 3

CJB’s joint network including three types of joint sets

Full size image

Geometric statistical analysis for field CJB

It has been shown that the exposed prism sections of the CJB on the dam foundation were not only of hexagonal shape but also of other kinds of irregular polygons because of the unsatisfactory historical heat-exchange environment (Saliba and Jagla 2003; Bosshard et al. 2012; Hetenyi et al. 2012). Moreover, the prismoid axes were not vertical due to the inclined stratum. The scanline mapping technique and the stereographic projection method were therefore utilized to investigate the geometric characteristics of the CJB.

The excavated surface of the dam foundation and the walls of the surveying tunnels inside the CJB were the available measuring windows. Considering the complicated close joint structure, a reasonable choice of measuring lines was very important. Three groups of measuring lines in trends 10°, 65°, and 126° were selected, intersecting roughly at 60° to each other, as shown in Fig. 4a. Moreover, this layout can also avoid the possible orientation dependency of the measuring lines. Therefore, the edge lengths of the exposed prisms along the measuring lines and the orientations of columnar joints on the surface can be measured. Along these measuring lines, the columnar joint orientations can be obtained. To ensure an objective and reasonable statistical result, a total of 150 sets of altitude parameters of the joints in three groups (50 sets for each group) were measured randomly. All these joint orientation data were analyzed with Schmidt pole plots and contour diagrams (see Fig. 4a). In the Schmidt net diagram, the pole plots were shown to be nonsymmetrical, which directly proved that the CJB cross-section was an irregular polygon. This result was in good agreement with the field observations, as shown in Fig. 4b. In addition, the pole plots in the diagram were not lumped in certain localized zones, and were not similar to those of a regular stratified rock mass. This implied that the columnar joints had a special closed intersection network pattern as well as the same dip angles to a large degree.

Fig. 4
figure 4

Poles contour diagram of columnar joints (a) and field CJB exposed on the slope (b)

Full size image

The investigation of the transverse sections of the CJB exposed on the surface showed that most of the block sections were quadrangular and pentagonal due to the intensive incisions from the steep joint network, as shown in Fig. 5. Measurements on the diameter of the columnar blocks in the exposed CJB on the dam foundation showed that the block edges were 15–35 cm in length, with an average of 29 cm (Table 1), based on 190 samples. Moreover, the logarithmic normal distribution reasonably described the variability of the CJB edge lengths.

Fig. 5
figure 5

Statistical polygon ratio of the CJB transverse section

Full size image
Table 1 Statistical data of edge length for the columnar jointed blocks
Full size table

Figure 6 showed the frequency histograms of the intervals and the fitted lognormal frequency function of the horizontal joint set inside the columnar blocks. The lognormal distribution fitted the observed results reasonably well, showing the mean value of 1.36 cm. This kind of joint set originated from the delayed cooling of the lava as mentioned above and was initially closed before the excavation due to the surrounding high geo-stress. When the upper rock mass was excavated, the normal stress on the joint plane experienced an unloading process. Hence, this intensively horizontal joint set exposed on the dam foundation was loosened and partially opened step by step, thereby leading to the decrease in the deformation modulus with time. This process should also cause time-dependent deformation and local failure of the rock mass due to the reduced frictional forces on the joint plane.

Fig. 6
figure 6

Statistical interval distribution of the horizontal joint set

Full size image

Field failures of the CJB on the dam foundation

The field investigation showed that this exposed columnar jointed rock mass exhibited several types of failure modes, including opening of columnar prisms, cracking of concrete, loosening of internal joints, and shearing deformation of horizontal joints (Fig. 7). These kinds of failure may all be related to the CJB joint sets.

  1. 1)

    The opening of the columnar and vertical internal joint sets induced in the separation of columnar prisms, as well as the disintegration of the columnar block itself, as shown in Fig. 7a. These local deformations of columnar prisms may easily lead to the cracking of sprayed concrete on the dam base (see Fig. 7b).

  2. 2)

    The opening of horizontal internal joint sets easily resulted in the shearing tendency of the columnar blocks, as shown in Fig. 7c. This type of opening may also cause extra shearing deformation and even local collapse due to the reduced frictional resistance if no appropriate supporting measurements were applied, as shown in Fig. 7d.

Fig. 7
figure 7

Field failure performances of columnar jointed basalt exposed on the dam foundation: (a) opening of columnar prisms, (b) cracking of concrete, (c) loosening of internal joint, (d) shearing deformation of horizontal joint

Full size image

Time-dependent damage tests for the CJB

To understand the unloading damage response of the CJB on the left dam foundation after excavation, we selected a rectangular experimental zone of 90 m in length and 14 m in height on the dam base to consider the field testing condition, as shown in Figs. 2a and 8. A series of tests were carried out to study the time-dependent CJB damage behavior using ultrasonic P wave methods.

Fig. 8
figure 8

In situ experiment site and corresponding subzones

Full size image

Testing scheme

The experimental zone was entirely in the columnar jointed rock mass, and its surface was inclined to the Jinsha River with a dip angle of approximately 45°. A shearing belt with a small dip angle (approximately 15°–25°), called LS3319, ran through the testing zone and introduced differences of rock mass quality. The quality classification of the CJB mass below the shearing belt was III-1 rank, whereas the CJB quality classification above was slightly weaker, i.e., III-2, according to the Chinese national standard for engineering classification of rock masses (DB 50218 1995; Chen and Xu 2017). The experimental zone was divided into three subzones, i.e., the downside region of LS3319, subzone A; the upper region of LS3319 at EL 660–655 m, subzone B and subzone C, as shown in Fig. 8. Thus, these geological differences between subzones might affect the initial damage depth induced by the unloading excavation for the dam foundation.

Thirty boreholes in total were drilled in the experimental zone before the excavation, with each experimental subzone including 10 boreholes. All these testing holes were normal to the face of the experimental zone and were 10 m in length. The test used the ultrasonic P wave detection method in a single borehole (Aydin 2014; Meglis et al. 2005), and the detection probe had one emission component and two receiving components (Fig. 9). For every 0.1 m as the probe moved in the borehole along the axes, a P wave velocity data would be obtained. And, by assembling all the data, a wave–velocity curve along the borehole depth could be obtained (Aydin 2014; Malmgren et al. 2007; Wu et al. 2009b). By evaluating the tested curves, the damage depth of the Baihetan CJB was defined as the point at the obvious turning point on each curve. If there was no obvious turning point, a point with a 10% reduction in the wave velocity in the undamaged segment would be used.

Fig. 9
figure 9

The integrated ultrasonic wave detection equipment for rock damage

Full size image

The wave velocity of the undisturbed columnar jointed rock mass was obtained from Eq. (1). The average P wave velocity of the undamaged rock mass in testing holes was defined as:

$$ \overline{v^o}=\frac{1}{n}\sum \limits_{i=1}^n{v_i}^o $$
(1)

where \( \overline{v^o} \) is the average P wave velocity of the undisturbed columnar jointed rock mass, vio is the measured P wave velocity of the non-damaged parts along the borehole, and n is the number of measured velocities.

Temporal evolution of the CJB damage depth

The field testing of the CJB experimental zone lasted for 7 months. All the detected curves of P wave velocity versus depth for the boreholes showed that the lower P wave velocity appeared at the shallower part and the higher P wave velocity appeared at the deeper part (i.e., undisturbed rock mass) in the borehole, as shown in Fig. 10. These results indicated that both the final stable damage depth and the EDZ stable time were different in different subzones; for example, the average CJB damage depth at subzone A was smaller than subzones B and C, as shown in Fig. 11.

Fig. 10
figure 10

Typical tested P wave velocity vs. depth cures in the boreholes (tested at 15 days after excavation)

Full size image
Fig. 11
figure 11

Total tested depths of EDZ in the CJB experimental zone (results at 220 days after excavation)

Full size image

All the testing work began in Novrmbrt 2014, the day after the upper rock mass above the experimental zone was excavated by blast. In situ tests by the ultrasonic P wave apparatus showed that:

  1. 1)

    The average P wave velocity of the undisturbed CJB was approximately 5132 m/s, but that of the damaged columnar jointed rock mass was in the range of 4800–3200 m/s in subzone A. The final depth of the EDZ stabilized within less than 15 days; whereas the adjacent blasting opening could induce a sharp depth increase that could be observed in boreholes A1–5 and A2–1, as shown in Fig. 12.

  2. 2)

    In subzone B, the average P wave velocity of the undisturbed CJB was approximately 5072 m/s, but that of the damaged columnar jointed rock mass was in the range of 4700–2960 m/s. It was evident in Fig. 13 that the CJB in subzone B had time-dependent damage behavior, as the damage depth increased with time continuously, until approximately 30 to 45 days after the excavation. Then, the EDZ depth finally stabilized.

  3. 3)

    The average P wave velocity of the undisturbed CJB was approximately 5085 m/s in subzone C, but that of the damaged columnar jointed rock mass was in the range of 4600–2890 m/s. The CJB at subzone C also obviously had time-dependent damage behavior, as shown in Fig. 14. Consistent with the findings in subzone B, it took approximately 3045 days for the EDZ depth evolution to stop.

Fig. 12
figure 12

Temporal evolution of the CJB damage depth gained in subzone A

Full size image
Fig. 13
figure 13

Temporal evolution of the CJB damage depth gained in subzone B

Full size image
Fig. 14
figure 14

Temporal evolution of the CJB damage depth gained in subzone C

Full size image

By comparing Figs. 12, 13 and 14, it can be found that the time-dependent CJB damage in subzones B and C was more serious than in subzone A, although all the temporal damage ceased after a long time, as recorded in Table 2. The differences of the tested results between the subzones maybe related to the rock quality and influenced by the shearing belt, i.e., LS3319, in the experimental zone.

Table 2 Summarization of damage characters for the CJB in different zones
Full size table

To obtain the temporal evolution pattern for this III-2 rank columnar jointed rock mass, regressive analyses were performed based on the damage depth data obtained in subzones B and C. The results showed that the temporal evolution of the CJB damage depth was generally in accordance with the Curtis logic equation (Eq. 2),

$$ L=\frac{1}{1+3.6064\ {e}^{-0.1144T}}\operatorname{} $$
(2)

where L is the normalized EDZ damage depth ratio, i.e., the current/final damage depth, and T is the time of the CJB after excavation. Figure 15 shows the obtained data and the fitted Curtis curve, and the results match each other reasonably well. It is also worth noting that as T became sufficiently long, the damage depth ratio L would converge to a constant, given that there was no external disturbance.

Fig. 15
figure 15

Fitting curve of Curtis logic equation for the damage depth data gained in the subzones B and C

Full size image

Spatial damage degree in the EDZ of the CJB

The in situ detections shown above revealed that the P wave velocities in the EDZ of the CJB varied at different depths of the borehole. This was manifested by the fact that low velocity was detected at the shallower part and high velocity was detected at the interior part along the borehole axis (Fig. 16). The result indicated that the damage degree of the unloaded CJB EDZ was spatially inhomogeneous, which may also leaded to spatial anisotropy of the CJB deformation modulus.

Fig. 16
figure 16

Typically time-dependent reduction of P wave velocity curve and increase of damage depth in EDZ

Full size image

According to Rabtonov’s 1963 definition of the damage index for a destructive test (Rabtonov 1963), the damage of the rock mass can be defined by the wave velocity,

$$ {D}_f=1-\raisebox{1ex}{${v}_i$}\!\left/ \!\raisebox{-1ex}{$\overline{v^o}$}\right. $$
(3)

where Df refers to the damage degree of the rock mass. The damage degrees of the CJB’s EDZ in subzones B and C were thereby calculated, and the results clearly showed heterogeneous damage after excavation. As shown in Fig. 17, serious damage occurred at the opening of the hole, whereas no or little damage was found in the deeply undisturbed rock mass. Regression analyses were carried out based on the damage data of the B and C subzones, and the results indicated the Sigmoid function can ideally describe the spatial distribution of spatial damage degree for the Baihetan columnar jointed rock mass from the surface to the interior. It is shown in terms of Eqs. 4 and 5,

$$ {\displaystyle \begin{array}{l}{D}_B^{mid}=1-\frac{1}{1+2.7197{e}^{\left(1-3.3456P\right)}}\\ {}{D}_B^{down}=1-\frac{1}{1+1.0879{e}^{\left(1-3.3456P\right)}}\\ {}{D}_B^{up}=1-\frac{1}{1+6.7993{e}^{\left(1-3.3456P\right)}}\end{array}} $$
(4)
$$ {\displaystyle \begin{array}{l}{D}_C^{mid}=1-\frac{1}{1+3.1197{e}^{\left(1-3.5658P\right)}}\\ {}{D}_C^{down}=1-\frac{1}{1+1.2479{e}^{\left(1-3.5658P\right)}}\\ {}{D}_C^{up}=1-\frac{1}{1+7.7993{e}^{\left(1-3.5658P\right)}}\end{array}} $$
(5)

where P is the ratio between the current testing position and the final EDZ depth; \( {D}_B^{mid} \), \( {D}_B^{up} \), and \( {D}_B^{down} \) are the middle fitting, up boundary, and down boundary equations for testing in subzone B, respectively; and \( {D}_C^{mid} \), \( {D}_C^{up} \), and \( {D}_C^{down} \) are the middle fitting, up boundary, and down boundary equations for testing in subzone C, respectively.

Fig. 17
figure 17

Regression analysis respectively for the damage data of subzone B (a) and subzone C (b)

Full size image

Time-dependent damage evolutionof CJB’s EDZ

Since the disturbed CJB showed spatially inhomogeneous damage degree characteristics, the time-dependent evolution of the CJB at different positions was further analyzed. P wave velocity data with a final damage depth of more than 2 m in subzones B and C were selected and analyzed. The velocity data in the EDZ were divided uniformly into 5 segments along the borehole axis, i.e., 0–1/5, 1/5–2/5, 2/5–3/5, 3/5–4/5, and 4/5–1 of EDZ depth. The average velocity for the testing data of a given depth at different testing times was calculated. Figure 18 summarizes the computed results for each segment, as well as the fitting functions. It was shown that the average velocity of all the testing data had a tendency of time-dependent reduction, although the fitting function and final stable values were different. It was also shown that the time-dependent decrease of the CJB velocity in the EDZ was self-convergent under no disturbance (i.e., converging to a constant value).

Fig. 18
figure 18

Time-dependent evolution of the CJB P wave velocity at different depth and corresponding fitting curves in the EDZ: (a) 0–1/5 EDZ depth, (b) 1/5–2/5 EDZ depth, (c) 2/5–3/5 EDZ depth, (d) 3/5–4/5 EDZ depth, (e) 4/5–1 EDZ depth

Full size image

Regression analyses for the above velocity data in different depths showed that the negative exponent function, as described in Eq. 6, can reasonably describe the time-dependent evolution of CJB velocity,

$$ D(t)=a\cdot {t}^{-b} $$
(6)

where D is the damage degree, t is the time, and a and b are the fitting parameters corresponding to different positions. It was interesting to note that none of the fitting curves cross into each other, as shown in Fig. 18f. By converting the weakening degree in the P wave velocity into the damage degree according to Eq. 3, it can be found that the time-dependent evolution of the damage degree in the CJB seems to have a similar deteriorating pattern in different positions.

In summary, the field observations from the ultrasonic P wave velocity tests indicate that the EDZ of the CJB on the dam foundation experiences not only spatially inhomogeneous damage in degree but also temporal damage in depth and degree. Therefore, the EDZ of the Baihetan columnar jointed basalt can be considered as a 4D damage process, i.e., having spatially 3D inhomogeneous deformation and time-dependent evolution.

Discussion

Generally, the jointed rock mass is composed of intact rock and joint sets (Barton et al. 1974; Hoek and Brown 1997; Sakurai 2010), and, hence, its damage mechanism should be distinguished from normal homogeneous material. To grasp the damage principle of the columnar jointed rock mass, a digital borehole camera (DBC) was used to observe the unloading damage behavior of the CJB in the testing holes. The results showed that many open joints existed in the testing borehole, with widths of 0.1–3 mm, as shown in Fig. 19. Through successive surveys by the DBC, it was also shown that joint widths gradually changed with time. Due to the excavation, the stress within the surrounding rock mass was unloaded, initiating the opening of the originally closed joints. However, until a new mechanical equilibrium state was reached, the induced open of joints developed continuously. Meanwhile, the mechanical properties of the damaged CJB, such as the deformation module, shearing strength, wave velocity, etc., became weakened.

Fig. 19
figure 19

Time-dependent cracking development of the borehole jointed rock mass observed by digital borehole camera

Full size image

Since opening of the joints inside the CJB was the intrinsic mechanism of time-dependent damage and a key factor for the stability of dam construction, the necessary technical means must be carried out to reduce or close up the joints, and improve CJB mechanical properties. Assuming the opening of the original joints was following the principal CJB damage mechanism, the damage degree can also be defined by its geometric definition (Kachanov 1958), i.e.,

$$ D=1-\raisebox{1ex}{$\overline{A}$}\!\left/ \!\raisebox{-1ex}{${A}_0$}\right. $$
(7)

where D is the material damage degree, A0 is the initial cross-sectional area of the undamaged material, and \( \overline{A} \) is the effective loading area of the damaged material. By comparing with Eq. (3), where the damage degree is defined by the wave, the rock wave velocity can be considered directly proportional to the effective loading area in the excavated damage zone, i.e.,

$$ {v}_i\propto \overline{A} $$
(8)

Therefore, the rehabilitation of the damaged CJB can be achieved by increasing the effective contacting and loading area of the rock mass.

In addition, as a typical friction-type material, the rock mass strength generally satisfies the Mohr–Coulomb criterion, i.e.,

$$ \tau ={C}_0+{\sigma}_n\tan \varphi $$
(9)

where τ is the material shearing strength, C0 is the cohesion, σn is the normal stress, and φ is the friction angle. To improve the effective loading strength of the damaged jointed rock mass, one way is to amend the physical properties of the material (C0 and φ), and the other way is to enhance the exterior normal stress (σn). Therefore, two technical methods can be employed:

  1. 1)

    Perform the injection of liquid cement inside the damaged rock mass, which can solidify the joints and fill their lacunas, i.e. to increase the effective loading area, and also improve the C0 and φ of the CJB;

  2. 2)

    Install prestressed rock bolts to enhance the normal stress on joints and restrain further cracking of the CJB.

During the actual reinforcement of the columnar jointed basalt on the dam foundation, wet-ground cement with 0.5 MPa pressure was grouted inside the columnar jointed rock mass and rock bolts with 120 kN prestress were also installed to limit the time-dependent deformation and damage (Fig. 20). The corresponding DBC test indicated that these methods can improve the integrity of the columnar jointed basalt.

Fig. 20
figure 20

Actual reinforcement for the jointed basalt: (a) grouted cement observed by digital borehole camera; (b) pre-stress rock bolt on the dam foundation

Full size image

Conclusions

The columnar jointed rock mass is a special geo-material with its natural self-organized network normally in a hexagonal prismatic shape. However, in situ investigations of the CJB on the Baihetan dam foundation showed that most of the transverse sections were quadrangular and pentagonal, and a few were trilateral and hexagonal. The edge lengths were in a logarithmic normal distribution instead of being uniformly distributed, with a mean length of 29 cm. The joint network inside the CJB included not only the columnar joints but also the horizontally and vertically distributed internal joints with different historical origins and mechanical properties.

Field ultrasonic P wave tests for the damaged columnar jointed rock mass revealed that the Baihetan CJB had a specific time-dependent damage property. By using backward regression analysis, it was shown that: (1) the EDZ depth increased with time after excavation and converges in a certain time depending on the rock quality, which was in accordance with the Curtis logic equation; (2) the CJB damage degree in the EDZ varied in position in accordance with the Sigmoid function; and (3) the EDZ damage degree also changed with time, and its wave velocity fitted a negative exponential function. Overall, the damage evolution of the CJB’s EDZ on the Baihetan dam foundation can be considered as a 4D process, i.e., spatially inhomogeneous distribution and time-dependent evolution.

Successive field observations using a digital borehole camera indicated that the unloading damage mechanism of the Baihetan CJB was mainly related to the time-dependent opening of the initially blank and closed joints after excavation. For improving the mechanical properties of the damaged CJB, two reinforcing methods, i.e., the injection of liquid cement and the installation of prestressed rock bolts inside this damaged rock mass, have been suggested.