Some new pre-warning criteria for creep slope failure

SCIENCE CHINA Technological Sciences

© Science China Press and Springer-Verlag Berlin Heidelberg 2011 tech.scichina.com www.springerlink.com

*Corresponding author (email: [email protected])

• RESEARCH PAPER • December 2011 Vol.54 Suppl. 1: 210–220

doi: 10.1007/s11431-011-4640-5

Some new pre-warning criteria for creep slope failure

XU Qiang1*, YUAN Yong1, ZENG YuPing1 & HACK Robert2

1 State Key Laboratory of Geo-hazard Prevention and Geo-environment Protection, Chengdu University of Technology, Chengdu 610059, China;

2 The International Institute for Geo-information Science and Earth Observation (ITC), Enschede, the Netherlands

Received July 19, 2011; accepted October 11, 2011; published online November 18, 2011

The forecasting of the failure time of a slope remains a worldwide problem because of many different possibilities of geologi-cal conditions in combination with many varying external factors such as climate and vegetation, and not well-defined or un-known time effects in deformation and failure models. The aim of this paper is to suggest a new method to carry out the phase division and to explore tangential angular features of the displacement-time curve of creep slopes as well as the acceleration characteristics in the process of slope deformation and the pre-warning criteria for critical failure. An imminent failure is pro-ceeded with usually three basic deformation phases, namely primary creep (“decelerated”), secondary creep (“steady- state”) and tertiary creep (“accelerated”). Mostly, only during the accelerated phase the imminence of a possible slope failure is rec-ognized. The analysis of displacement data from a series of landslides allowed to recognize different evolutionary patterns of displacement. A quantitative approach was proposed to describe the tangential angle of the displacement-time curve and a new criterion based on the angle was put forward to divide the accelerated phase into three sub-phases: initial acceleration, medium acceleration, and the critical failure. A pre-warning criterion for critical failure is also proposed consequently. Changes of acceleration showed completely different characteristics from those of cumulative displacement and displacement in the process of slope deformation. The values of acceleration usually oscillate around 0 prior to the critical failure phase, whereas the acceleration increases abruptly when the deformation moves into the critical failure phase. This allows, therefore, for a method to forecast the time of the failure. So it is possible to define different alert acceleration threshold values to be used for emergency management.

creep slope, displacement-time curve, tangential angle, alert acceleration threshold, pre-warning of critical failure

Citation: Xu Q, Yuan Y, Zeng Y P, et al. Some new pre-warning criteria for creep slope failure. Sci China Tech Sci, 2011, 54(Suppl. 1): 210220, doi: 10.1007/s11431-011-4640-5

1 Introduction

The forecasting of the failure time of a slope remains a worldwide difficulty at present because of the variations of geological conditions, slope structure, deformation failure models, and many other external factors, even though the prediction of the failure time of landslides has been recently discussed extensively, similar to the discussions on earth-

quake prediction. This discussion is more important because of the severe losses of lives and economy caused by the very frequent occurrence of landslides in recent time.

Generally, research on forecasting the time of a slope failure has gone through four stages or approaches, namely empirical equation prediction [1–3], prediction by dis-placement-time statistical analysis [4–6], non-linear predic-tion [7–13] , and comprehensive prediction [14–20]. How-ever, this has not led to a solution because of the following reasons:

(1) At present, the majority of scholars focus on the study

Xu Q, et al. Sci China Tech Sci December (2011) Vol.54 Suppl. 1 211

of developing mathematical models for the forecasting of the failure time of slope. Real slopes, however, differ from each other widely depending on their individual geological and geomorphologic development and the failure time of slope is influenced by many factors. Many mathematical models can not include all the factors and as a result are rarely applicable to the majority of landslides. This is the reason why successful prediction using mathematical fore-casting models is rare as shown in the research [17].

(2) The evolution of an instable slope shows individual-ity, such as oscillations or developing step transitions, and the vast majority of landslides, especially certain types of landslides, show some common evident features during their deformation evolution, such as characteristics of slow creep phenomena in creep slopes. These features and char-acteristics have not been included in detail in most studies.

(3) The further development and miniaturization of monitoring tools and technologies and the possibilities of remote real-time monitoring can all provide important tech-nical options for the forecasting of slope failure. At present, the monitoring work itself is overweighted, but at the same time, research on data utilization and early warning is not in-depth.

For these reasons, research should be confined to slopes that have certain characteristics in common. This paper, therefore, focuses on the study of deformation feature with time and pre-warning criteria for creep slopes.

Deformation monitoring results of a considerable number of landslides show that slopes composed of loose soil (such as ancient landslides, artificially filled slopes, or other soil slopes formed by other causes) and rock slopes with a time-dependent failure mechanism (such as toppling-creep slopes, slopes with sliding-bending deformation and the creep-crack deformation slopes in highly weathering zones) deform generally over a long time from the beginning to the failure and that the deformation has the characteristics of slow creep, a phenomenon known as “slope creep” [21, 22]. Such slopes with characteristics of long-term creep defor-mation are commonly denoted by “creep slopes”. Since creep slopes have a regular pattern of deformation in space and time, it may be possible to develop an early-warning criterion and to forecast the time of failure.

Early in the 1960s, it was found that the deformation of certain slopes is directly related to the creep of rock and soil [1]. A well-known empirical formula for forecasting the failure time was proposed on the basis of rock and soil creep tests and of analogy between the test result and the deformation curve of the slope [2]. This initiated forecasting the failure time of a slope. This is the slow deformation or flow of slopes involving soil, rock and ice, taking place un-der gravity and external loadings [23]. Different models of slope creep have been proposed, depending on the adopted function (linear, exponential, power law) and definition of “failure” [24]. Saito and Uezawa [1] and Saito [2, 3] as-sumed a non-linear visco-plastic bulk behaviour of slope

materials and showed that the logarithm of the time of fail-ure is proportional to the logarithm of the strain rate. Fuku-zono [4] and Voight [25, 26] brought important improve-ments to the slope creep theory. Crosta and Agliardi [27] discussed how to set velocity thresholds to achieve different levels for forecasting landslides, and proposed criteria for pre-alert (30 d), alert (15 d) and emergency (7 d) especially for the Ruinon rockslide, based on the “accelerating creep” model proposed by Voight. However, the method is rarely applicable in practice because mostly an expert is needed to determine the alert velocity threshold of each slope and even of different benchmarks on a slope. Because the speed of slope deformation prior to the failure changes considera-bly from a few millimeters per day to meters per day [17], and the threshold speed difference can be more than an or-der of magnitude even on the same slope but measured at different benchmarks [27]. In this paper, therefore, two more practical pre-warning indexes, called the tangential angle index and the acceleration threshold index for slope failure are proposed.

2 Three-phase law of creep slopes

A large number of landslides show that a creep slope, under gravity, usually goes through three phases from beginning to the final failure. Namely, primary creep (“decelerated”), secondary creep (“steady-state”) and tertiary creep (“accel-erated”) [3, 18], similar to the creep curve of rock and soil masses [3, 18] (see Figure 1).

But, the displacement-time curve of a slope shown in Figure 1 is of only an “ideal” curve and applies only to a slope under its constant self weight. The development and evolution of a slope is inevitably subject to the influence of a variety of external factors (such as rainfall, human engi-

Figure 1 Sketch of three deformation phases of a creep slope. S, cumula-tive displacement; t, time; AB, primary creep; BC, secondary creep; CF, tertiary creep; CD, initial accelerated sub-phase; DE, medium accelerated sub-phase; EF, critical failure sub-phase.

212 Xu Q, et al. Sci China Tech Sci December (2011) Vol.54 Suppl. 1

neering activities, etc.). The displacement-time curve for real slopes shows oscillations and fluctuations because the slope is in a complex and open system on the surface of the earth's crust and subject to many processes. Therefore, the common displacement-time curve is often varying or fluc-tuating and/or contains steps (so-called ladder type), but overall the trend is in compliance with the three-phase law above. The former is often featured with fluctuations if the development of the slope deformation is affected by non-periodic external factors (such as rainfall, artificial ac-tivities) or as a result of man-made errors in monitoring. The latter is usually step-shaped if the deformation evolu-tion of the slope is subject to the influences of periodic ex-ternal factors, such as the annual rainfall in the flood season, the schedule of the normal annual water level of a reservoir, such as in the Three Gorges Reservoir Area. The rainfall in the yearly flood season causes a rapid increase in deforma-tion and after the flood season the deformation gradually returns to steadiness. The overall displacement curve shows several steps corresponding to the annual rainfall in the an-nual flood season.The displacement-time curve of a slope under gravity can be obtained through a proper filter (such as smoothed filter, least squares filtering, etc.) filtering the “noise” arising from the external factors.

Another point to note is the start time of monitoring. In Figure 1, the curve of slope deformation shows the defor-mation from the beginning to the final failure of the slope. However, in practice, the monitoring apparatus are usually set up only when the deformation of the slope has devel-oped to a degree that is recognized as a potential hazard. Moreover, the initial deformation phase of a slope is often very short. Therefore, the displacement-time curve of many real slopes may only contain two phases, the secondary creep and tertiary creep phase, with the primary phase being absent. Sometimes only the curve of the tertiary creep phase is obtained and the first two phases are absent.

3 Pre-warning criterion of tangential angle

Figure 1 shows that the displacement-time curve tends to increase continuously especially after the slope deformation enters the accelerated deformation phase. The displace-ment-time curve becomes almost vertical and the angle with respect to the horizontal axis is close to 90° in the final critical failure phase. Some scholars [8, 28] proposed the tangential angle of displacement-time curve and used it to forecast the failure time of slope. The deformation tangen tial angle refers to the angle between the tangent of the displacement-time curve and the horizontal axis at a given time. Wang and Zhang [28], based on a statistical analysis of a considerable number of landslide cases, found that the tan-gential angle just before the failure of the slope is generally 89°–89.5°. Therefore, it was proposed to use it as one of the criteria for landslide forecasting, named the “deformation

tangential angle criterion”. However, the definition of the criterion is problematic

because of units of the coordinates of the displacement-time curve may have different dimensions and scales. Although the displacement-time curve maintains the characteristics of its three phases if any of the two coordinate scales is ex-tended or compressed, the tangential angle of the displace-ment-time curve will change. For example, if the x-axis is fixed whereas the y-axis scale is stretched, the tangential angle at a point of time during the accelerated deformation will increase. x-coordinate scale transformation will result oppositely in the same transformation of the y-coordinate. In other words, in terms of the displacement data monitored for the same slope, the tangential angle measured at the same point of time would vary if different coordinate scales are adopted to map the displacement-time curve. Thus, the criterion of tangential angle of displacement-time is not so unambiguously. Here, we try to propose a coordinate trans-formation to unify the dimensions of the x and y coordinates, which is employed to ensure the unambiguous definition of the tangential angle and establish the criterion for landslide forecasting accordingly.

We note that there is a manifest steady deformation phase in the slope evolution reflected in the displace-ment-time curve (Figure 1). In this deformation phase, the deformation rate is basically a constant value, and the cu-mulative displacement s and time t are in a linear relation, that is, s = vt, in which v is the displacement rate in the sec-ondary deformation phase. s and t have a non-linear rela-tionship in the other two phases.

Since the displacement rate v during the secondary creep phase is nearly a constant for the slope, the displacement reported on the y-axis can be reset with respect to this con-stant displacement rate, thus obtaining values with the same dimension (i.e. time) and unit as the x-axis. Definition can be made as follows:

( )

( ) .s i

T iv

(1)

Here, s(i) is the cumulative displacement of time ti; v is the average velocity of the secondary creep phase; T(i) is the relative time corresponding to monitoring time ti, with same dimension and unit as time.

Figure 2 is the curve of relative time T with respect to the monitoring time t (hereafter referred to as T-t curve ), cor-responding to the displacement-time curve (Figure 1). Ac-cording to the T-t curve, an improved expression of the tan-gential angle i can be obtained:

1

( ) ( 1)arctan arctan ,i

i i

T i T i T

t t t

(2)

where i is the tangential angle of T-t curve; ti is the ith of monitoring time; t is the monitoring period, such as 1 day, 1 week, etc.; t is the change in volume during a unit of


Figure 2 T-t curve converted from the displacement-time curve. T, The value of longitudinal coordinate after transformation; t, time; AB, primary creep; BC, secondary creep; CF, tertiary creep; CD, initial accelerated sub-phase; DE, medium accelerated sub-phase; EF, critical failure sub- phase.

time period. By defining i the following is valid.

When i <45°, the slope deformation lies in the primary

creep phase; When i ≈45°, the slope deformation lies in the secon-

dary creep phase; When i >45°, the slope deformation goes into the terti-

ary creep phase. It is worth nothing that cumulative displacement should

be concerned in calculating the tangential angle. In addition, equalization of intervals is needed to harmonize the moni-toring period, in other words, to maintain the consistency of different deformation phases if the monitoring period ( t ) of different deformation phases varies.

As stated above, the correct determination of the displacement rate v during the secondary creep phase is the key to obtaining the improved tangential angle. Due to the interference of external factors and measurement errors, the displacement rates at different times cannot be absolutely the same but tend to fluctuate within a certain range, even if the slope deformation is in the secondary creep phase. Therefore, the average rate of the slope deformation can be calculated and be regarded as the velocity of secondary creep. The procedure to obtain the specific rate is as fol-lows:

(1) The division of the slope deformation phase: deter-mine the slope deformation phase and distinguish the sec-ondary creep phase according to the deformation monitor-ing curve.

(2) The determination of the rate v in the secondary creep phase: the rate v is equal to the arithmetic mean of the dis-placement rates within various time intervals in the secon-

dary creep phase.

1

1,

m

ii

v vm

(3)

where vi refers to the deformation rate of the ith time inter-vals in the secondary creep phase, and m refers to the num-ber of time intervals.

4 Pre-warning criterion of acceleration

Besides the deformation rate as the basic indicator reflecting the slope deformation characteristics, acceleration change should also be an important feature in the analysis. Some typical cases were studied in order to identify the changes of the acceleration of the slope deformation. Firstly, the Jiminsi landside (Figure 3) in China was used to illustrate time-variation of the acceleration. There was complete se-ries data of monitoring for this landslide. And the cumula-tive displacement-time curve and deformation rate-time curve are shown in Figures 3(a) and (b); the acceleration- time curve is shown in Figure 3(c).

Figure 3(b) shows that the slope has a complete curve starting from the initial deformation to the slope failure. More concretely, the slope deformation rate has the follow-ing characteristics:

(1) In the initial phase it increases from 0 and then de-creases, and remains nearly a constant after the entry of the secondary creep phase;

(2) It increases gradually at a pronounced speed after the entry of the accelerated phase of deformation. The rate in-creases abruptly in the critical failure phase and continues to increase until the slope fails. The cumulative displacement- time curve (Figure 3(a)) has an increasing trend, only with a different rate of increase in each phase. In contrast, the ac-celeration-time curve in Figure 3(c) shows a completely different pattern of changes in contrast to Figures 3(a) and (b). In the initial deformation phase, the acceleration a in-creases from 0 to a certain maximum value and thereafter decreases rapidly to zero or even becomes negative. This indicates that the deformation starts rather instantly but slows down rapidly. However, in the secondary creep phase, the acceleration a varies around 0, and the mean is about 0. After the entry of the tertiary creep phase, the acceleration a is mostly positive, but it is still fluctuating and the fluctuat-ing amplitude is larger than that in the former phase. Once the deformation enters the critical failure phase, the accel-eration a shows a rapid growth.

Another information significant to slope failure forecast can be found also in the acceleration-time curve in Figure 3(c). Before the slope enters the critical failure phase, the acceleration, in general, varies around 0. The fluctuating amplitude is the smallest in the secondary creep phase and


Figure 3 Deformation curves of Jiminsi landslide. AB, primary creep; BC, secondary creep; CF, tertiary creep; EF, critical failure sub-phase. (a) Cumulative displacement-time curve; (b) deformation rate-time curve; (c) acceleration-time curve.

relatively large at the initial and the accelerated deformation phases, but the overall fluctuating amplitude does not ex-ceed threshold (Figure 3(c)). From the Jiminsi landslide curve, the can be determined approximately as 0.5 mm/d2. Once the acceleration exceeds , it signifies the entry of the critical failure phase. An alert for the imminent failure of a slope should be given at this moment. An early but not too early warning will be of great significance to the protection of people’s lives and property as well as to the emergency services and maintenance engineering. In other words, the real automatic pre-warning of the imminent failure of slopes can be achieved by means of automatic monitoring, real- time transmission and automatic processing of computer data if the pre-warning criterion for critical failure is ob-

tained through the analysis of the acceleration-time curve of a slope.

To test and verify the laws postulated above, the research further studied the acceleration characteristics of other land-slides, such as an open pit landslide in Chile, as well as the Daye landslide in China (Figure 4). For these landslides, the monitoring data of deformation are relatively complete (Figures 4 and 5).

As shown in Figures 4 and 5, the characteristics and pat-terns of deformation are found to be completely similar among the three landslides above. In particular, all the ac-celeration curves, the focus of concern in this paper, appear very manifest with sudden jumps before and after the entry of the critical failure phase. The pre-warning criterion for critical failure is not difficult to be determined according to the acceleration curve (mainly referring to the peak value and the mean of the accelerations in the accelerated defor-mation phase before the entry of the critical failure phase).

Figure 4 Deformation curves of the Daye landslide in China. (a) Cumu-lative displacement-time curve; (b) deformation rate-time curve; (c) ac-celeration-time curve.


Figure 5 Deformation curves of the open pit landslide in Chile. (a) Cu-mulative displacement-time curve; (b) deformation rate-time curve; (c) acceleration-time curve.

According to Figure 4, the pre-warning criterion for criti-cal failure of the Daye landslide is identified at about 0.2 mm/d2, and to Figure 5 it is identified at about 6 mm/d2.

5 Case study

Taking the recent Baishi landslide (Figure 6) as an illustra-tion, the following is to test and verify the proposed method. Baishi landslide is located in Beichuan County of Mianyang City in Sichuan, China. The slope was approximately 300 m long, 260 m wide and 25 m thick on average, and about 2×106 m3 in total volume. Manifest evidence of deformation has been found since December 2006 (Figure 6). Lives and property of 695 people living in the Baishi village below the

slope would have been endangered if the slope had failed on a large scale, because the height difference from the toe of the landslide to the Baishui River is more than 500 m. On July 28, 2007, the overall slope collapsed in a large scale, blocking off the Baishui River to form a barrier lake (Figure 7). Fortunately, the slope failure did not result in life and property losses to the local residents due to geotechnical monitoring and forecasting in advance. This helped the timely evacuation and the construction of a drainage tunnel through the mountain on the opposite side of the slope. In order to give accurate forecast to the deformation of the Baishi landslide, comprehensive deformation monitoring on the landslide has been implemented since January 2007 (Figure 8).

Figure 6 Full view of Baishi landslide on December 28, 2006.

Figure 7 Full view of the Baishi landslide after sliding on July 28, 2007.


Figure 8 Distribution of benchmarks of Baishi landslide.

The continuous monitoring revealed that all the bench-marks within sections H2 and H3 started to enter the accel-erated deformation phase in the second half of April in 2007, whereas the displacement rate at the benchmark located in section H1 was always varying within a certain range. Therefore, sections H2 and H3 were identified as the major deformation areas, whereas section H1 was identified as relatively stable. This was confirmed by the subsequent failure of the slope (Figure 7). Accordingly, the focus of concern was put on sections H2 and H3 in the late period of the slope monitoring and forecasting. An analysis of the monitoring data from 8 benchmarks in sections H2 and H3 found similar displacement rates and cumulative displace-ment magnitudes as well as the evolution trend at each benchmark within sections H2 and H3, indicating the syn-chronic deformation of the overall sections H2 and H3. Figure 13 shows the cumulative displacement-time moni-toring curves at the four representative benchmarks in sec-tions H2 and H3. There were 199 iso-interval pieces of monitoring data at each benchmark from January 10, 2007 to the occurrence of the slope failure. Figure 10 shows the corresponding T-t curves at the four benchmarks. Table 1 is about the newly defined tangential angles obtained before the failure of the Baishi landslide.

It is shown in Figure 10 and Table 1 that the newly de-

Table 1 Tangential angles calculating from the final series of data (prior to the critical failure)

Benchmark 2 Benchmark 6 Benchmark 10 Benchmark 14

79.11 78.43 75.27 79.09

80.55 76.94 76.95 79.27

80.95 79.06 77.8 80.95

81.06 79.57 77.54 81.60

82.62 79.89 80.44 81.66

83.96 81.19 81.5 83.32

84.76 82.99 82.54 84.55

85.86 83.88 84.54 85.43

87.79 85.22 86.93 86.62

89.47 87.39 89.25 88.29

Figure 9 Displacement-time curves of four benchmarks of Baishi land-slide. (a) Benchmark 2; (b) benchmark 6; (c) benchmark 10; (d) benchmark 14.

fined tangential angles are obviously greater than 45° when the slope deformation enters the accelerated deformation phase. The slope deformation rate increases evidently when the tangential angle of the slope exceeds 80°; the slope shows prominent signs of imminent failure when the tan-gential angle of the slope exceeds 85°. After the tangential angle surpasses 85°, the deformation rate and the tangential angle tend to increase sharply with time. The tangential an-gle reaches about 89° right before the slope failure.


Figure 10 T-t curves of four benchmarks of Baishi landslide.

Therefore, referring to Figure 1, we can work out the relatively quantitative criteria for the division of the three sub-phases of the acceleration deformation phase using the improved tangential angle, namely, when the improved tangential angle>45°, the slope deformation enters the pri-mary creep phase; when the improved tangential angle>80°, the slope deformation enters the accelerated deformation phase; when the improved tangential angle>85°, the slope deformation enters the critical failure phase.

Benchmark 2 is chosen as an example in analyzing the acceleration characteristics. Figure 11 is the deformation rate-time curve of benchmark 2 on the Baishi landslide. In the end of 2006, the deformation rate of the Baishi landslide, was very high, up to 20–100 mm/d, or even higher. As is shown in Figure 9, the slope deformation was still at the secondary creep phase by the end of March, 2007. In order to reflect the subtle characteristics of the slope deformation more clearly, the monitoring data after the entry of critical failure phase (after July 22, 2007) were truncated because the magnitude of the data was too large in contrast to that of the previous phase and the overall display accuracy might be otherwise affected, as shown in Figure 12. And the Fig-ure indicates that prior to March 26, 2007, the deformation rate of the slope was varying around 80 mm/d, showing manifest features of steady deformation. Based on the anal-ysis of monitoring data and accurate judgment on the slope evolution phase, in January 2007 the relevant experts, made a decisive plan to excavate the drainage tunnel inside the mountain opposite to the landslide in case of the river block. Figure 12 also shows that after March 26, 2007, the defor-mation rate of the slope started to climb and showed itself with evident characteristics of accelerated deformation, in-dicating the slope’s entry of the accelerated deformation

Figure 11 Deformation rate-time curve of benchmark 2 in 2007.

Figure 12 Deformation rate-time curve of benchmark 2 prior to critical failure phase in 2007.

phase. Figure 13 is the acceleration-time curve of the bench-

mark 2 of the Baishi landslide. The figure shows that before the slope deformation entered the critical failure phase (July 22, 2007), the acceleration was varying around 0 even if it was at the accelerated deformation phase, whereas the cu-


Figure 13 Acceleration-time curve of the 2nd benchmark in 2007.

mulative displacement and the deformation rate tended to increase gradually after the entry of the accelerated defor-mation phase. Figure 13 indicates that the acceleration rises sharply and abruptly right after the entry of the critical fail-ure sub-phase. This shows that in the slope deformation evolution, the acceleration has completely different charac-teristics before and after the entry of the critical failure phase, showing significant catastrophic characteristics and the catastrophic point being the threshold for entering the critical failure phase.

In order to reflect the subtle characteristics of the changes of the acceleration at utmost, the monitoring results in the last two days( July 27 and 28) were also truncated because of the large magnitude of the acceleration, as shown in Figure 14 which indicates that the acceleration is always varying around 0 and its mean is approximately 0 both at the secondary and tertiary creep phases. In contrast, the acceleration surges up abruptly after July 23, and the acceleration-time curve shows noticeable “sudden jumps”.

More importantly, Figure 14 also indicates that although the acceleration keeps varying prior to the slope’s entry of the critical failure phase, the varying amplitude, through the entire deformation process, has always been confined to a certain range, i.e. the threshold value . As Baishi landslide is concerned, the acceleration pre-warning criterion for critical failure can be set at 100 mm/d2.

In practical landslide forecasting, the following aspects are notable when using the acceleration pre-warning crite-rion to predict the critical failure of a slope.

(1) As each landslide has its own manifest characteristics of individuality apart from the universal features, such as the three-phase law, the acceleration pre-warning crite-rion of each landslide varies and there is no uniform

pre-warning criterion for critical failure, hence the indica-tor needs to be specified through the analysis of the charac-teristics of the acceleration curve.

(2) In determining the specific value of , the utmost ac-curacy is unnecessary, as long as the usual sudden jump is

Figure 14 Acceleration-time curve of 2nd benchmark prior to July 26, 2007.

correctly identified. For instance, the pre-warning criterion for critical failure of the Baishi landslide can be set at 100 mm/d2, but can also be set slightly larger, for example at 150 or 200 mm/d2. In fact, the given value of the pre- warning criterion for critical failure during the actual forecast of Baishi landslide was 200 mm/d2, and an alert on critical failure was sent on July 24, making it successful as early warning. This is due to the fact that the acceleration value will by far exceed this value when the slope enters the critical failure state. Figure 14 also indicates that the actual acceleration of Baishi landslide reached up to 2500 mm/d2 prior to the failure.

(3) The premise of the slope failure is that the slope de-formation has entered into the accelerated deformation phase. Therefore, the overall judgment on the slope evolu-tion phase needs to be noted when the acceleration pre- warning criterion for critical failure is used to forecast a slope failure. The indicator can be used for the critical fail-ure forecast only when the slope deformation has entered the mid and last periods of the accelerated deformation sub- phase and the value of the acceleration has exceeded the set value of .

(4) In stock transaction, the expected transaction price can be set in advance and automatic transaction happens when the set conditions are satisfied. The acceleration pre- warning criterion for critical failure, similar to stock trans-action, can also be set in advance on condition of automatic monitoring of the landslide, and automatic pre-warning for failure will be achieved in accordance with the monitoring results.

6 Conclusions

We have applied a variety of monitoring and modeling me-


methods to better understanding of the nature of early warning, and have outlined

(1) Like any development and evolution, the evolution of a landslide also goes through a process of emergence (the occurrence of deformation), development (the continuous deformation), thriving (the accelerated deformation), and demise (the slope failure), in which a creep slope, in general, experiences three basic phases of primary creep, secondary creep and tertiary creep. Correct judgment on the deforma-tion phase of a slope has played a central and determining role in early warning. The entry of the accelerated deforma-tion phase is the premise of the slope failure. At the former two phases of the slope, failure will not happen normally, regardless of the high deformation rate or the large cumula-tive displacement magnitude and the manifest evidence of the macroscopic deformation failure. On the contrary, once the slope deformation enters the accelerated deformation phase, especially the critical failure phase, the slope failure will be inevitable if urgent engineering measures are not taken. This law is of great significance in guiding emer-gency engineering for landslides.

(2) A new method is proposed to obtain the tangential angle after transforming the cumulative displacement-time curve of a slope to achieve the same dimension between the x and y coordinates. The improved tangential angle will not vary with the stretch or compression of the coordinates. The quantitative criteria are given for further dividing the terti-ary creep phase of a slope into three sub-phases, assisted by the improved tangential angle. That is, when the improved tangential angle> 45°, the slope deformation enters the ini-tial deformation sub-phase; when the improved tangential angle> 80°, the slope deformation enters the medium accel-erated deformation sub-phase; when the improved tangen-tial angle> 85°, the slope deformation enters the highly ac-celerated (the critical failure ) phase. Results from the test of some landslide cases that have relatively complete de-formation monitoring curves show that the criteria above for the division and the forecast of critical failure are universal to some extent.

We have furthermore shown the changes of acceleration have completely different characteristics from those of cumulative displacement and deformation rate in the whole process from the beginning to the failure of the slope. In the whole period before the slope deformation enters the critical failure phase, the acceleration value basically keeps fluctu-ating around 0, whereas once it enters the critical failure phase, the acceleration rises abruptly and sharply. That is to say, the acceleration shows completely different character-istics before and after the slope enters the critical failure phase, showing significant catastrophic characteristics at the entry of the critical failure phase. Based on this feature, the way to forecast critical failure by acceleration and the fore-cast for critical failure has been proposed.

This work was supported by Ya-lung River Joint Research Foundation of National Natural Science Foundation of China and ErTan Hydropower Development Company Ltd. (Grants No. 50579099), the National Basic Research Program “973”, Project of the Ministry of Science and Tech-nology of the People’s Republic of China (Grant No. 2008CB425801), the Creative Team Program of the Ministry of Education in China (Grant No. IRT0812), and the Major Cultivation Project in University Scientific and Technological Innovation Program of the Ministry of Education, the De-velopment of Monitoring and Forecast System for Geohazards Based on WEB.

1 Saito M, Uezawa H. Failure of soil due to creep. In: Proceedings of the 5th International Conference on Soil Mechanics and Foundation Engineering, Paris, 1961. 315–318

2 Saito M. Forecasting the time of occurrence of a slope failure. In: Proceedings of the 6th International Conference on Soil Mechanics and Foundation Engineering. Canada: University of Toronto Press, 1965. 537–541

3 Saito M. Forecasting time of slope failure by tertiary creep. In: Pro-ceedings of the 7th International Conference on Soil Mechanics and Foundation Engineering. Mexico: Sociedad Mexicana de Mecanica de Suelos, A.C., 1969. 677–683

4 Fukuzono T. A new method for predicting the failure time of a slope. Proc IV International Conference and Field Workshop on Landslide. Tokyo: Tokyo University Press, 1985. 12–20

5 Yan T, Wu F Q, Ying K L. The static and dynamic laws of landslide system and quantitative space-time prediction of the instability of slopes. Earth Sci-J Geology Univ China, 1989, 14(2): 117–133

6 Zhang Z Y, Wang S T, Wang L S. Principle of engineering geology analysis (in Chinese). Beijing: Geologic Publishing House, 1992. 374–377

7 Huang R Q, Xu Q. Synergetic prediction model of slope instability (in Chinese). Mountain Res,1997, 15: 7–12

8 Li T B, Chen M D, Wang L S. Landslide Real-Time Tracking Predic-tion (in Chinese). Chengdu: Chengdu Scientific and Technical Uni-versity Publishing, 1999. 16–69

9 Qin S Q, Jiao J J, Wang S J. The predictable timescale of landslides. Bull Eng Geol Environ, 2001, 59: 307–312

10 Qin S Q , Jiao J J, Wang S J. A nonlinear dynamical model of land-slide evolution. Geomorphology, 2002, 43: 77–85

11 Gao W. Study on displacement prediction of landslide based on grey system and evolutionary neural network. Computational Methods in Engineering & Science. Berlin, Heidelberg: Springer, 2007. 275

12 Zheng M X. A new predicting method of landslide by fractal theory. Geotechnical Engineering for Disaster Mitigation and Rehabilitation. Berlin Heidelberg: Springer, 2008. 975–980

13 Huang Z Q, Law T K, Liu H D, et al. The chaotic characteristics of landslide evolution: A case study of Xintan landslide (in Chinese). Environ Geol, 2009, 56: 1585–1591

14 Zhong Y Q. Study of prediction method by synthetic information for Huanglashi landslide (in Chinese). Chin J Geological Hazard Control, 1995, 6(4): 67–74

15 Wang N Q, Liu S H, Zeng S W. A study on forecasting method of the comprehensive analysis of macroscopic signs of landslide (in Chi-nese). Chin J Rock Mech Eng, 1999, 11(1): 34–38

16 Wang S Q, Xu J J. New method to monitor and forecast landslide disaster in short-term (in Chinese). J China Three Gorges Univ (Natural Sciences), 2006, 28(5): 385–388

17 Xu Q, Huang R Q, Li X Z. Research progress in time forecast and prediction of landslides (in Chinese). Adv Earth Sci, 2004, 193: 478–483

18 Xu Q, Tang M G, Xu K X. Research on space-time evolution laws


and early warning-prediction of landslides (in Chinese). Chin J Rock Mech Eng, 2008, 27(6): 1104–1112

19 Chen Z Y, Sun P, Wang Y J, et al. 3-dimensional slope stability analyses using non-associative stress-strain relationships. Sci China Ser E-Tech Sci, 2009, 52(9): 2517–2527

20 Zheng H, Zhou C B. Global analysis on slope stability and its engi-neering application. Sci China Ser E-Tech Sci, 2009, 52(2): 507– 512

21 Terzaghi K. Mechanism of landslides. In: Paige S, ed. Application of Geology to Engineering Practice (Berkey Volume). Washington, DC: Geological Society of America, 1950. 83–123

22 Haefeli R. Creep problems in soils, snow and ice. In: Proc. 3rd Int. Conf. on Soil Mechanics and Foundation Engineering. 1953, 3: 238–251

23 Emery J J. Simulation of slope creep. In: Voight, ed. Rockslides & Avalanches Developments in Geotechnical Engineering, 1979. 669– 691

24 Cruden D M, Masoumzadeh S. Accelerating creep of the slopes of a coal mine. Rock Mech Rock Eng, 1987, 20: 123–135

25 Voight B. A method for prediction of volcanic eruption. Nature, 1988, 332: 125–130

26 Voight B. A relation to describe rate-dependent material failure. Sci-ence, 1989, 243: 200–203

27 Crosta G B, Agliardi F. How to obtain alert velocity thresholds for large rockslides. Physics Chem Earth, 2002, 27: 1557–1565

28 Wang J D, Zhang Z Y. Systematic project research on typical high speed loess landslide (in Chinese). Chengdu: Sichuan Science and Technology Publishing House, 1999. 140–153

Some new pre-warning criteria for creep slope failure

Documents

Transcript of Some new pre-warning criteria for creep slope failure