Abstract
The time-dependent behaviour of high rockfill dams is complex and not easy to accurately predict. Many discrepancies were revealed by the comparison of the observed deformation histories of different dams, and the deformation of some high rockfill dams did not correspond to the general deformation law. Field monitoring is therefore an effective method for understanding complex dam deformation behaviour. In this paper, actual measured deformation data resulting from continuous monitoring of the Maoergai and Qiaoqi dams are analysed. These two dams have similar heights, crest lengths, and alluvium overburden thicknesses. Our aim is to explain the actual deformation histories on the basis of the mechanical behaviours of these dams in order to warn engineers about potential problems that cannot be predicted. The results indicate that the deformation patterns of the two dams are completely different. The dam construction and water impoundment schedule is the major reason for the different horizontal displacement patterns. The reservoir filling rates and rainfall are the main reasons for the different settlement patterns. The case histories are useful for understanding the wide range of possible postconstruction deformation in a dam.
1. Introduction
Rockfill dams are one of the most common dam types around the world. Over the past two decades, many rockfill dams over 100 m have been built in southwestern China, where almost 80% of the national total hydropower resources are concentrated. With the social demand and improvements in compaction techniques, the height of this type of dam is increasing. Thus, the internal stress is high, and the deformation pattern is complex. Moreover, high rockfill dams also show significant postconstruction deformation, which is a major threat to the dam’s serviceability and deserves considerable attention.
The time-dependent breakage of rockfill materials plays a major role in dam creep behaviour [1–4]. In high dams, as the material particles are usually large in size and angular in shape, the rockfill materials are susceptible to particle breakage [5]. In some dams, the rockfill particles may exceed 1200 mm in size. Large particles possess more defects and faults and are subject to high contact forces. Angular particles experience more breakage because stress concentration occurs at their vertices [1]. Zhou and Song [1] reported that the increase in stress levels and particle size will lead to an obvious increase in the creep strain and creep rate of the rockfill. Sohn et al. [6] emphasized the key role of particle shape in breakage by discussing experiments conducted on two materials: glass beads and quartz sand. Liu et al. [7] found that particle breakage mainly occurs in the sharp corners of the particles, and after breakage, the particles become somewhat rounded and smaller. In these studies, the particle size distribution and shape changed obviously during shearing, which can have a great effect on the particle arrangement and rotation and can affect the long-term performance of dams.
To analyse and predict the deformation and stress of rockfill dams, the finite element method has been primarily used based on laboratory-determined mechanical parameters. However, realistic parameters are not easily obtained from laboratory tests [8–10]. The main reason may be the impact of the scale effect on the breakage and creep behaviours of the rockfill material. Scale effects have been shown to significantly influence the representativeness of laboratory results [11–13]. However, the impact of scale effects on strength is still not fully quantified, although they have been studied by many researchers [14–17].
Over the years, the discrete element method (DEM) has been used to simulate laboratory tests of rockfill materials. Zhou and Song [1], Ma et al. [4], and Tapias et al. [18] all found that the creep strain matches well with laboratory creep test results (e.g., the laboratory-determined parameters and DEM-determined parameters are rather similar). However, such models require adopting restrictive assumptions concerning the shape of each particle and the interaction of separate particles. Thus, the numerical results are difficult to directly apply to practical projects.
The values and trends of deformation observed during the construction and operation of several high rockfill dams in southwestern China are complex and may have large differences. In some dams, the dam deformation did not correspond to the general deformation pattern of rockfill dams. For instance, as expected, deformation developed rapidly during the construction phase and the stage of first reservoir impoundment and decreased progressively with time. However, during the second reservoir impoundment, the deformation of the Maoergai dam exhibited a large sudden increase and thus caused longitudinal cracks in the dam crest. In addition to particle breakage, the construction schedule, severe weather conditions, and even topographical and geological conditions have a complexly coupled effect on the dam deformation. In some dams, the initial water impounding proceeds along with the dam construction. This complicated process cannot be reproduced in the laboratory. The actual stress path experienced in the field may not be followed by the calculation procedure. Furthermore, rockfill breakage is sensitive to the humidity of the environment [19]. The effect of the ambient relative humidity on the behaviour of dams is difficult to capture through laboratory tests [20, 21]. Many researchers have placed less emphasis on the environmental conditions than on the mechanical causes of particle breakage.
Interpreting the behaviour of real rockfill dams and their monitoring records is helpful for understanding the time-dependent deformation characteristics of a rockfill dam. The analysis of monitoring data from practical engineering projects is essential for warning engineers about potential problems that cannot be predicted or analysed by numerical simulations. The Maoergai and Qiaoqi dams have similar heights, crest lengths, and alluvium overburden thicknesses but have different deformation patterns. In this paper, the continuous deformation records covering part of the construction, the whole impoundment, and several years of dam operation of the two dams are presented. The interpretation of the differences between the deformation patterns of the two dams can provide a reference for other dams for more reliable predictions.
2. Problems for Some High Core Rockfill Dams
Some deformation phenomena in high rockfill dams are difficult to explain reasonably in terms of laboratory experiments and numerical simulations because of the complexly coupled effects of particle breakage, construction schedule, severe weather conditions, and even topographical and geological conditions on dam deformation. Therefore, in the rockfill dam, continuous monitoring data, which can reflect the state of the dam at every stage, are important and necessary for the interpretation of the deformation behaviour of rockfill dams.
Over the past two decades, 7 core wall rockfill dams with heights exceeding 100 m have been constructed in southwestern China (Table 1). Among these dams, six have been built on deep compressive alluvial layers in valleys with steep abutments. The field observation data from some of these dams show that, in different dams, the deformation patterns are rather different and that the complex pattern of deformation is not consistent with the general deformation behaviour of rockfill dams.
Over a period of 2.5 years after construction, postconstruction settlement in several dams in southwestern China has been obtained. In the Pubugou, Qiaoqi, and Shuiniujia dams, the postconstruction settlements near the middle part of the dams are approximately 0.15%, 0.08%, and 0.16% of the maximum dam height, respectively. However, in the Maoergai dam, the postconstruction settlement near the middle portion of the dam reaches 0.54% of the maximum dam height.
Obviously, the postconstruction settlement in the Maoergai dam is much larger than that in the other dams. The Maoergai and Qiaoqi dams have similar heights, crest lengths, and alluvium overburden thicknesses. The layout of the monitoring gauges in the Maoergai and Qiaoqi dams is shown in Figures 1(a) and 2(a), respectively. The postconstruction deformation values in the two dams over a period of 3.3 years are illustrated in Figure 3. As shown in Figure 3(a), in the Maoergai dam, at the elevation (EL.) 2031 m (0.27H1, where H1 is the dam height of the Maoergai dam), 2048 m (0.40H1), and 2078 m (0.60H1), the horizontal displacements in the river flow direction were 265–328, 310–355, and 590–645 mm, respectively. In the Qiaoqi dam, at EL. 2061 m (0.35H2, where H2 is the dam height of the Qiaoqi dam), 2085 m (0.54H2), and 2110 m (0.74H2), the postconstruction horizontal displacements were approximately 40–90, 110–150, and 160–215 mm, respectively.
(a)
(b)
(c)
(a)
(b)
(c)
(a)
(b)
As shown in Figure 3(b), in the Maoergai dam, the postconstruction settlement values were 390–465 mm at 0.3H1, 330–600 mm at 0.40H1, and 545–865 mm at 0.60H1. In the Qiaoqi dam, at 0.35H2, 0.54H2, and 0.74H2, the postconstruction settlement values were 80–180, 150–180, and 335–350 mm, respectively.
Dam performance during the first impoundment is a critical issue. In this period, the upstream shell of a rockfill dam may undergo rapid settlement with relatively large values because of submerging. This event, usually called “collapse deformation,” has been reported in different rockfill dams [22, 23] and always causes discordant deformation. Then, the dam crest has a high possibility of cracking. For example, longitudinal cracks have been observed on the crest of the Cherry Valley dam [24, 25] and the Pubugou dam [26]. However, field measurements show that, in the Maoergai dam, longitudinal cracks on the crest were not observed during the first impounding but were observed during the second. Therefore, the second impounding period may be another important stage in a dam’s operation life.
Problems also exist in the horizontal displacement of dams. Among the dams over 200 m, the Changheba and Nuozhadu dams also have similar heights. However, during the first impounding, the maximum horizontal displacement in the downstream direction was 150 mm at 0.4 dam height in the Changheba dam and reached 890 mm at 0.5 dam height in the Nuozhadu dam.
3. Measured Deformation of the Maoergai and Qiaoqi Dams
3.1. Maoergai Dam and Its Displacement Monitoring System
The Maoergai gravelly soil core rockfill dam is 147 m high. The length and width of the crest are 458.5 m and 12.0 m, respectively. The dam is built on deep alluvium overburden. A cutoff wall with a thickness of 1.4 m and a maximum depth of 52 m is selected as the vertical antiseepage system to control foundation leakage. Figure 1(a) shows a typical cross section of the dam, which is composed of rockfill zones, transition zones, filter zones, and a central gravelly soil core. The rockfill materials of the dam are composed of quartz sandstone. The grain size distribution curves of the materials are illustrated in Figure 4. The material properties obtained through laboratory are summarized in Table 2.
(a)
(b)
(c)
(d)
In southwest China, high mountains and deep valleys are two vital factors underpinning the major hydro energy projects. The Maoergai canyon is characterized by a wide, V-shaped valley with bank slopes around 40°–50° (Figure 1(b)). The commonly used valley shape factor in practice is the ratio of the dam crest length between the valley abutments and dam height. In the Maoergai dam, the ratio of length to height is about 2.3. Here, the height means the height of the dam plus the thickness of alluvium layers. The dam’s base and abutments consist mainly of metasandstone and phyllite.
The geological condition along the longitudinal section of the dam is illustrated in Figures 1(a) and 1(b). According to geological age, grain size, density, and geological origin, the alluvium overburden (unconsolidated Quaternary deposits) from the top to the bottom is classified into the following four groups: alQ4, al + PlQ4, alQ3, and al + PlQ3. The grain size distribution, dry density, permeability, and thickness of the deposits are listed in Table 3.
Construction of the Maoergai dam started in December 2009 and finished in May 2011. The initial reservoir impounding started in March 2011, when the dam reached a height of 130 m. The highest water level during the initial impounding was 2104.4 m. In October 2012 (during the second reservoir impounding), the water level reached the normal storage level of 2133 m for the first time. The dam construction and reservoir impounding process of the Maoergai dam are shown in Figure 1(c).
A detailed displacement monitoring system is established in the Maoergai dam. In this study, only the monitoring results in the maximum cross section 0 + 244 (Figure 1(a)) are analysed. Tension wire alignment transducers (H1–H21) and hydraulic overflow settlement gauges (V1–V21) are distributed in the downstream zones to measure the internal horizontal displacement and settlement, respectively.
3.2. Qiaoqi Dam and Its Displacement Monitoring System
The Qiaoqi gravelly soil core rockfill dam is 125.5 m high, with a crest that is 439.8 m long and 10.0 m wide. The dam is also built on deep alluvium overburden, and the maximum depth of the cutoff wall in the foundation is 70.5 m. Figure 2(a) shows a typical cross section of the dam, which is composed of rockfill zones, transition zones, filter zones, and a gravelly soil core. The rockfill materials of the Qiaoqi dam are composed of quarried basalt. The grain size distribution curves of the materials are also depicted in Figure 4. The material properties obtained in laboratory are summarized in Table 4.
The Qiaoqi canyon is characterized by a U-shaped valley (Figure 2(b)). Above the elevation of 2095 m, the valley sides gently widen with slopes of around 25°. Lying at below the elevation, the Qiaoqi canyon is incised with steep slopes of around 45°–70°. The ratio of length to height is about 2.3. The height means the dam height plus the thickness of alluvium layers. The dam’s base and abutments are composed primarily of metasandstone and phyllite.
The geological condition along the longitudinal section of the dam is illustrated in Figures 2(a) and 2(b). According to geological age, grain size, density, and geological origin, the deep overburden (unconsolidated Quaternary deposits) from the top to the bottom is classified into the following four groups: , , , and fglQ3. The grain size distribution, dry density, permeability, and thickness of the deposits are listed in Table 5.
The construction of the Qiaoqi dam started in January 2006 and finished in November 2007. The initial reservoir impounding started in December 2006 when the dam was filled to EL. 2090 m and finished in November 2007. The highest water level during the initial impounding was 2125.1 m. In November 2008 (during the second reservoir impounding), the water level reached the normal storage level of 2140.0 m for the first time. The dam construction and reservoir impounding process of the dam are shown in Figure 2(c).
In this study, the results of the monitoring gauges in the maximum cross-section 0 + 215 are used. As shown in Figure 2(a), tension wire alignment transducers (H9 to H23) and hydraulic overflow settlement gauges (V9 to V23) are installed in the downstream side of the dam to measure the horizontal displacement and settlement, respectively. These monitoring devices are installed at EL. 2110 (0.74H2), 2085 (0.54H2), and 2061 m (0.35H2). Meanwhile, reference stations TP12, TP06, and TP01 lie on the downstream slope surface to monitor the exterior displacements of the dam.
3.3. Deformation of the Maoergai Dam
3.3.1. Horizontal Displacement
The starting times of monitoring points on a dam differ because the gauges are installed only after the dam is filled to a certain elevation. In the Maoergai dam, the displacement monitoring at EL. 2031 and 2048 m started in January 2011, and the monitoring at EL. 2078 and 2108 m began in March 2011 and September 2012, respectively. Through August 2014, the dam experienced a total of 3.5 cycles of rising and falling water levels. The horizontal displacements in the river flow direction in the Maoergai dam, as a function of time, fill height, reservoir level, and precipitation, are illustrated in Figure 5. In the figure, positive values indicate deformation towards the downstream direction.
(a)
(b)
(c)
(d)
The figure shows that the horizontal displacements are in the downstream direction and that the total displacement values decrease with height. In the Maoergai dam, the evolution of the horizontal displacement was complex. According to the slopes of the displacement curves, four time intervals were identified (marked by dotted lines in Figure 5): Stage I. In the first reservoir impounding period (March 23, 2011–November 17, 2011), the horizontal displacements increased with the rising water level. The displacements at the lower elevations (EL. 2031 and 2048 m) were much larger than those at the higher elevations (EL. 2078 and 2108 m). Stage II. Before the water level reached 2104.4 m during the second impounding period (November 17, 2011–June 29, 2012), corresponding to the highest water level during the first impounding, the displacement rate decreased, especially at lower levels. Stage III. As the reservoir level rose from 2104.4 m to 2133.2 m (normal storage level) during the second impounding (June 29, 2012–October 18, 2012), the displacement experienced a sudden increase, and the maximum increase reached 610 mm at point H21 (EL. 2108 m). The displacement rates in this period were the highest among all the stages. Stage IV. During the period of operation (October 18, 2012–August 15, 2014), the displacement rate became very low. The displacement curves at lower levels became smooth, which indicates that the displacement in the lower part of the dam gradually stabilized after the second full filling of the reservoir.
3.3.2. Settlement
Generally, during dam construction, significant settlement occurs due to the successively increasing dam weight. The settlement pattern of the Maoergai dam is illustrated in Figure 6. In the Maoergai dam, settlement observations at EL. 2031, 2048, and 2078 m began in January 2011, when the dam was filled to an elevation of 2094 m. At EL. 2108 m, observations started at the beginning of the second reservoir impounding. Though not complete, monitoring data can also provide some useful information.
(a)
(b)
(c)
(d)
In the Maoergai dam, corresponding to the evolution of horizontal displacements, the settlement evolution was complex, and according to the slopes of the settlement curves, four time intervals were identified (marked by dotted lines in Figure 6): Stage I. During the period of construction (before May 20, 2011), the settlement increased with the continual increase in the dam height Stage II. Before water was impounded to the level of 2104.4 m during the second impounding (May 20, 2011–June 29, 2012), the settlement rate decreased Stage III. As the reservoir level rose from 2104.4 m to 2133.2 m (normal storage level) during the second impounding (June 29, 2012–October 18, 2012), the settlement values experienced a sudden increase, and the maximum increase reached 539 mm at EL. 2108 m Stage IV. During the period of operation (October 18, 2012–August 15, 2014), the settlement rate decreased gradually, especially in the lower part of the dam, which indicates that the settlement gradually stabilized after the second complete filling of the reservoir
3.4. Deformation of the Qiaoqi Dam
3.4.1. Horizontal Displacement
In the Qiaoqi dam, monitoring at EL. 2061 and 2085 m started in December 2006, and monitoring at EL. 2110 m started in July 2007. Through October 2012, the dam experienced a total of 5.5 cycles of rising and falling water levels. The horizontal displacements in the river flow direction in the Qiaoqi dam are illustrated in Figure 7.
(a)
(b)
(c)
The figure shows that the horizontal displacements are in the downstream direction and that the total displacement values decrease with height, and the horizontal displacements mainly occurred during the first reservoir impounding (or the dam construction), especially for the lower portion of the dam. As shown in Figure 7, before November 25, 2007, with the water level rising, the horizontal displacements at EL. 2061, 2085, and 2110 m increased 186–355, 388–470, and 480–560 mm, respectively. The monitoring at EL. 2110 m started in July 2007, 4 months after the dam was filled to an elevation of 2110.0 m. Thus, the first reservoir impounding displacements at EL. 2110.0 m during the first months are not included, and the displacements may actually be much greater. After the first reservoir impounding, the displacements were very small. During the second water level cycle (from May 9, 2008 to April 28, 2009), the displacements at EL. 2061, 2085, and 2110 m increased only 70–75, 74–79, and 155–178 mm, respectively. After the second water level cycle, the horizontal displacements changed very little.
3.4.2. Settlement
The settlement patterns of the Qiaoqi dam is illustrated in Figure 8. The settlement values at EL. 2061, 2085, and 2110 m were 560–1220, 620–700, and 440 mm, respectively, during the dam construction phase. During the second water impounding cycle (from May 9, 2008 to April 28, 2009), the settlement values at EL. 2061, 2085, and 2110 m increased by 80–190, 26–56, and 89–93 mm, respectively. After the second water cycle, the dam settlement values at EL. 2061 and 2085 m were negligible. At EL. 2110 m, the settlement values gradually stabilized after 3 years of operation.
(a)
(b)
(c)
4. Analysis of the Deformation Patterns of the Two Dams
4.1. Different Deformation Patterns
Based on the aforementioned analysis, the deformation patterns of the Maoergai and Qiaoqi dams were obviously completely different. By comparing the deformation in the two dams, four conclusions can be drawn:
First, in the upper part of the Maoergai and Qiaoqi dams, most of the horizontal displacement occurred in the second impounding period and the initial impounding period, respectively. Table 6 illustrates the horizontal displacement of the Maoergai dam during every water cycle and the ratio of the displacement to the total displacement that occurred during the first three cycles. In the upper part of the Maoergai dam (0.6H1), where most postconstruction displacement is concentrated, the displacement ratio during the second water level cycle was 72–77%. However, at 0.54H2 and 0.74H2 in the Qiaoqi dam, the displacement percentages were 75–82% and 70–84%, respectively, during the first water level cycle.
Second, the horizontal displacements in the Maoergai dam experienced a large sudden increase after the water level reached 2104.4 m (the highest water level during the initial impounding) during the second reservoir impounding. The maximum increase at 0.80H1 reached 610 mm. In the Qiaoqi dam, horizontal displacements also had a sudden increase when the water level reached 2125.1 m (the highest water level during the initial impounding) during the second reservoir impounding, but the increase was very small.
Third, despite the different deformation patterns, the total horizontal displacements in the two dams are similar at different heights in the two dams. In the Maoergai dam, the total horizontal displacements at 0.40H1 and 0.60H1 were 330–410 mm and 580–650 mm, respectively, whereas in the Qiaoqi dam, the horizontal displacements at 0.35H2 and 0.54H2 were 200–360 mm and 410–490 mm, respectively. The horizontal displacements are mainly caused by the upstream water load. The Maoergai and Qiaoqi dams with similar upstream water loads therefore have similar horizontal displacements.
Fourth, in July 2012 (during the second impounding), the settlement of the Maoergai dam had a very large sudden increase of 410 mm, and among all phases (construction, first impoundment, and operation) in the lifetime of the dam, the maximum deformation rate in the upper part of the dam occurred in this period. In the Qiaoqi dam, the postconstruction settlement was very small, and the maximum deformation rate in the upper part of the dam occurred in the dam construction phase.
Overall, different deformation patterns in the two dams resulted in quite different postconstruction deformation values. The horizontal displacement in the Maoergai dam is considerably larger than that in the Qiaoqi dam. For instance, the horizontal displacement values at 0.40H1 and 0.60H1 in the Maoergai dam are approximately 2.5 and 4.8 times that at 0.54H2 in the Qiaoqi dam. Additionally, the settlement values at 0.40H1 and 0.60H1 in the Maoergai dam are approximately 2.8 and 4.2 times that at 0.54H2 in the Qiaoqi dam. Obviously, deformation differences between the two dams became larger with the increasing dam height, which means that different deformation patterns are mainly caused by the deformation of the dam body instead of overburden layers, otherwise the largest differences should occur in the lower part of the dam. Moreover, the overburden in both dams is composed of gravel with stone and sand layers and has similar thicknesses.
4.2. Analysis
Similar to the case of the Maoergai and Qiaoqi dams, the Changheba and Nuozhadu gravelly soil core rockfill dams also have similar different displacement trends. In the upper part of the Nuozhadu dam, horizontal displacements mainly occurred during the initial impounding, whereas in the upper part of the Changheba dam, the displacements were quite small. As shown in Table 1, these two dams also have similar heights. During the first impounding, the measured maximum horizontal displacement in the Nuozhadu dam was 890 mm towards the downstream side, while in the Changheba dam, the measured maximum horizontal displacement was only 150 mm.
The small horizontal displacement in the Changheba and Maoergai dams during the initial impounding may be caused by collapse deformation. Figure 9 shows the horizontal displacements in the downstream direction of the Changheba core. Due to the collapse deformation, the horizontal displacement of the upstream rockfill and core will develop in the upstream direction, especially for the top part of the dam. Then, the enormous water load acting on the core will have little effect on the downstream displacement. Therefore, collapse settlement in the upstream dam shell can greatly affect the top part of the core and the downstream rockfill zones, leading to small horizontal displacements. In some other dams, such as the Pubugou dam, the influence of collapse settlement is also obvious. Figure 10 shows the horizontal displacement of the downstream side of the Pubugou crest. Figures 9 and 10 show that the displacement in the top part of the dam developed in the upstream direction at the beginning of the initial impounding period and then began to develop in the downstream direction after a certain water level was reached.
In the Qiaoqi and Nuozhadu dams, as shown in Table 7, the initial water impounding occurred simultaneously with the dam construction. Thus, collapse in the upstream shell and core was compensated by new material replacement; thus, the impact of the collapse on the deformation downstream was reduced. Therefore, during the initial impounding, the horizontal displacements of the two dams were large.
In some dams, such as the Maoergai and Qiaoqi dams, the reservoir was not filled to its normal storage level during the initial impounding because of the power plant requirements, dam height, and the seasonal variations in precipitation. Here, we call the highest water level during the initial impounding the first highest water level. The first highest water levels in the Maoergai and Qiaoqi dams were 2104.4 m (28.6 m below the normal water level) and 2125.1 m (14.9 m below its normal water level), respectively. Because of the large difference between the first highest water level and the normal water level in the Maoergai dam, a large increase in dam deformation occurred after the water level reached the first highest water level during the second impounding.
In the Maoergai dam, during the period from June 30, 2012 (the first highest water level) to July 7, 2012, the reservoir level suddenly increased at a very high average rate of 2.6 m/day, with a maximum rate of 4.4 m/day. Due to the high impounding rate, after July 7, 2012, the horizontal displacement increased dramatically, and the maximum increase reached 360 mm in the following month. In the Qiaoqi dam, during the period from September 7, 2008 (the first highest water level) to November 3, 2008 (normal water level), the reservoir level increased at a very low average rate of 0.41 m/day, causing small displacement.
Note that at the Maoergai dam crest, longitudinal cracks appeared just on July 7, 2012. This phenomenon must be caused by the overly large postconstruction horizontal deformation. Although the collapse settlement was smaller during the second impounding than during the first impounding, the sudden large increase in the horizontal displacement in the downstream direction during the second impounding could also have caused cracks.
The settlement difference between the two dams may arise from the difference in the breakage characteristics of the particles in the two dams, as particle breakage has a significant influence on the settlement of rockfill materials. Particle breakage is the result of microcrack propagation within particles and mainly affected by the particle characteristics, stress level, and presence or absence of water [27]. As shown in Figure 4, except for the gravelly core, the grain size distributions of the other materials were similar.
The crack propagation velocity is closely related to the relative humidity. The higher the material humidity is, the greater the amount of particle crushing is. Partial wetting, induced by rainfall, has been shown to have the same relevance as full flooding [28]. Downstream dam shells also collapse after rainfall [29]. The simplified precipitation history based on monthly amounts is illustrated in Figures 5 and 6 for the Maoergai dam site and in Figures 7 and 8 for the Qiaoqi dam site. Both dam site regions are characterized by abundant rainfall, with major rainfall events occurring in the period from May to October every year. At the Maoergai dam site, the average annual rainfall is 620 mm. At the Qiaoqi dam site, the annual rainfall reaches 1000–1400 mm, and the average number of rainy days over the years is 146.7 days.
The annual rainfall at the Qiaoqi dam site was approximately twice that at the Maoergai dam site. Moreover, during dam construction, the Maoergai dam experienced only one rainy season, whereas the Qiaoqi dam experienced two. Therefore, rainfall, which can accelerate the crack growth rate and creep strain evolution, had a much greater impact on the Qiaoqi dam during dam construction. After the completion of construction, particles in the Qiaoqi dam may have had higher local roundness, and the particle size may have decreased; therefore, the contact particle forces were smaller at the same stress level, and the material was hence less compressible. At the same time, crushing generates more fines, and large pores are expected to be occupied by these small particles. Densification reduces the subsequent rearrangement and thus further reduces the compressibility [30]. Therefore, a dam that experiences more rainfall during construction is less sensitive to rainfall after the completion of the construction. Thus, the creep settlements in the Qiaoqi dam were large during the dam construction but small after the completion of the dam construction. However, in the Maoergai dam, the reverse was true. The postconstruction settlement values of the Maoergai dam were very large.
For the Maoergai dam, Figure 6 shows a distinct correlation between measured settlement rates and periods of extreme rainfall intensity during the second impounding. Settlement at EL. 2108 m mainly occurred in the period from June 29 to October 19, 2012. From July to October, when the rainfall was high, the settlement at EL. 2108 m increased by 280 mm at V19, 400 mm at V20, and 550 mm at V21. At V21, approximately 400 mm of settlement occurred in July 2012, coinciding with the heaviest rainfall. Figure 6 also shows that, in the upper part of the dam, the settlement near the surface of the dam (V21) was especially high. However, in the Qiaoqi dam, the increases in the postconstruction settlement at the same elevation were nearly the same. The fact that the outer part of the Maoergai dam is more sensitive to rainfall than that of the Qiaoqi dam further confirms that the settlement rates are closely related to rainfall.
5. Conclusion
Monitoring data analysis of practical engineering projects is helpful in understanding the time-dependent behaviour of rockfill dams and is essential for warning engineers about potential problems that cannot be reproduced by models.
The dam construction and reservoir schedule always differ among different dams because of project cost, makespan, and quality. In dams where the reservoir impounding begins after the completion of the dam construction, the horizontal displacements are small during the first impounding due to the effect of collapse settlements, but significant postconstruction displacement occurs during the second impounding. In dams where the first impounding occurs during the filling of the embankment, the collapse settlement is compensated by new material addition; hence, the horizontal displacements are always large during the first impounding, and the postconstruction displacements are small.
In some dams, the reservoir is not filled to its normal storage level during the first impounding. Therefore, a large increase in the postconstruction horizontal displacement may occur during the second impounding. When the impounding rate is overly high, the displacement will increase dramatically, and longitudinal cracks may occur at the dam crest during the second impounding.
Careful consideration must be given to rainfall during both construction and subsequent phases. A dam that experiences less rainfall during construction is more sensitive to rainfall after the completion of dam construction. In this situation, significant postconstruction settlement is expected to accompany heavy rainfall.
Data Availability
The data used to support the findings of this study are included within the article.
Conflicts of Interest
The authors declare that they have no conflicts of interest.
Acknowledgments
The authors would like to thank the support provided by the National Key Technology R&D Program (no. 2013BAB06B02).