PRACTICAL DISCUSSION ON THE INTERPRETATION OF GEOMONITORING MEASUREMENTS AND THEIR INFLUENCE ON TUNNEL SUPPORT DESIGN.

Transcription

1 PRACTICAL DISCUSSION ON THE INTERPRETATION OF GEOMONITORING MEASUREMENTS AND THEIR INFLUENCE ON TUNNEL SUPPORT DESIGN. A CASE STUDY G. Angistalis, I. Papadatos, I. Rentzeperis, E. Saridou Egnatia Odos A.E. 6 th klm Thessaloniki-Thermi, Thessaloniki, Greece, GR 57 1 Keywords: convergence, rate, countermeasures INTRODUCTION This is a paper that discusses, in a practical/empirical manner, the movements and their impact on the support measures- which have been measured during the excavation of the top heading of the Driskos motorway tunnel (now in operation), during the years 2 to 22. The total length of the tunnel is almost 4.5km and consists of two parallel 12m span tunnels. It belongs to the Egnatia Motorway in Northern Greece. The measurements of 37 (3 point) monitoring stations have been analyzed and evaluated. The evaluation has been focused on the measured maximum vertical and horizontal movements, their rate of development (mm/day) and their relation with the rock mass quality and support behavior. The driving force behind this discussion has been the observation that, in general, high early (measured close to the excavation front) rates of convergences (mm/day) have resulted later in failures of the installed support. The discussion is structured as follows: a) Description of the rock mass and support measures, b) presentation of the movements measured, c) correlation of the movements with the rock mass and the support behavior, d) description of support performance, failures, back analysis and counter measures, e) testing the results against existing models, and finally, f) the description of an attempt to obtain an empirical model for the prediction of the support behavior. ROCK MASS, SUPPORT MEASURES The rock mass Driskos tunnel has been excavated in the so-called Ionian flysch formation, which mainly comprises alternations of siltstones and sandstones in various proportions, with shear and fault zones. Figure 1 Fault and shear zones in the flysch formation of Driskos tunnel. Hard rock pieces represent Sandstone and Siltstone. In fault/ shear zones the geomaterial is soft and clayey. 1

2 The thickness of these zones varied from some centimeters to generally less than one meter. Ground water has been almost everywhere present, in the form of drops or small inflows through faults and sheared/fractured zones. The strength parameters of the rock mass are given in a following paragraph. Figure 1 shows a fault/shear zone in the flysch formation of Driskos tunnel in a borehole core. In the fault and shear zones, the formation has been similar to soft soil, which could be easily molded with the fingers. Figure 2 shows a geological mapping data sheet of a similar a rock mass. Figure 2 - Geological mapping of faults, crossing alternations of siltstone-sandstone layers The excavation and support measures For the excavation and support of the 4.5 klm tunnel, five support classes have been initially designed and applied in almost its 2/3. At the section under consideration the tunnel has been excavated sequentially in two stages, that is top-heading and bench. Additional measures have been implemented later, to tackle the higher displacements observed in some areas of the tunnel. Table 1 describes support classes III and IV. The strength of the shotcrete has had a compressive strength equal to 4, MPa in 8 hours, and 28,5 MPa in 28 days (2x2x2cm cubes). The mix design of the shotcrete has complied with specification ACI 214. These support classes (III and IV) have been applied in most of the tunnel s length and provided a basis for the correlation of the geo-monitoring data prior to the implementation of additional measures. Table 1 - Excavation and support measures for support classes III and IV (top heading) Support measures per excavation step III IV Shotcrete thickness at the top heading [fiber reinforced 15cm 2cm, (3Kgr/m 3 )] Fully grouted rock bolts: pattern, length, diameter 1,3m x 1,3m, l=4m and 3m, Ø 25 1,2m x 1,m, l=4m and 6m, Ø 25 Steel ribs - Lattice girders 95/D26 per 1m Excavation step A phase (mainly drill and blast) 1,5m/(day) 1m/(day) GEOMONITORING RESULTS General The movements observed, during the excavation of the top heading of the tunnel, have been measured at three point convergence stations. One measuring point has been installed on the tunnel roof and the two others have been installed at either side of the excavation profile. The monitoring stations evaluated, belong to a discrete part of the tunnel (~2,klm). In this tunnel part, the geological 2

3 conditions and the excavation and support methods used, have been relatively similar. The width of the excavation is 12m and the height is 6m. Results For a length of almost 2, km of the excavated tunnel, the convergence results from thirty seven (37) stations have been evaluated together with the support class applied, rock mass conditions, geology and ground water, overburden and observed behavior of the shotcrete shell. a/a Table 2. Data of Monitoring stations in the section of Driskos tunnel under consideration. Description of Water Maximum horizontal Maximum Over geologic formation movement (mm) vertical burden movement (mm) Support class 1 See comment1 Dry to damp 3,4<3 1 23,1 IV 2 -//- Dry to damp 8,9<3 15<3 35,81 IV 3 -//- 13,7<3 2<3 42 IV 4 -//- Dry to damp 6,2<3 18,5<3 56,87 IV 5 -//- Dry to damp 7,7<2 21>2 57,56 III 6 -//- Dry to Damp 9,9<2 24>2 57,96 III 7 -//- Wet to dripping 6,3<3 49>3 58,41 IV 8 -//- Wet to dripping 1,9<3 37>3 58,9 IV 9 -//- Wet to dripping 22,8<3 56>3 59,4 IV 1 -//- Damp 7,5<3 64>3 59,4 IV 11 -//- Dripping 7,5<3 27<3 63 IV 12 -//- Dry 17,9<2 26>2 82,35 III 13 -//- Dry 5,1<2 34>2 115 III 14 -//- Dry 6,<2 32>2 111,49 III 15 See comment 2 Water drops 24,2<3 133>3 125,82 IV 16 -//- Dry to Damp 37,4>3 25>3 12,82 IV 17 -//- Damp to wet 8,7<3 43>3 89,3 IV 18 See comment 1 No data 7,7 <3 14<3 25 IV 19 -//- No data 8,9<3 15<3 35 IV 2 -//- No data 13,7<3 2<3 43,5 IV 21 -//- Dry 4,4<3 14<3 6,29 IV 22 -//- Dry to damp 8,3<3 11<2 64 III 23 -//- Dry 1,4<3 31>2 66,5 III 24 -//- dripping 6,4<3 4>3 67,8 IV 25 -//- Wet 8,2<3 64>3 68,83 IV 26 See comment 3 Wet 11,2<3 44>3 69,1 IV 27 -//- Dry -Damp 16,1<3 55>3 72,14 IV 28 -//- Damp 3,4<3 44>3 84 IV 29 -//- Wet to dripping 2,4<3 5>3 11,21 IV 3 -//- Dry 2,4<2 21>2 117,8 III 31 -//- Dry to damp 2,4<2 21>2 125,9 III 32 -//- No data 1,2<2 93>2 116,19 III 33 See comments1&3 3lit/h from 23,6<3 67>3 124,52 IV fault 34 -//- 23,2<3 8>3 126 IV 35 -//- 1-2m 3 /h 15,3<3 61>3 97,74 IV 36 -//- Light flow 4,2<3 5>3 18 IV 37 -//- Light flow 7,4<3 8>3 89,3 IV Comment 1: Alternations of thin to medium bedded sandstone and siltstone. Comment 2: Major fault zones were identified (more than 1 meter thick). Comment 3: Significant fault zones. 3

4 The results are shown in table 2. The maximum vertical and horizontal movements measured are compared with the maximum anticipated ones (2mm and 3mm) that derived from the numerical calculations of the tunnel design. The first occurred during the excavation of the top heading of the tunnel, while the numerically predicted ones, correspond to the excavation of the entire tunnel cross section (top heading and bench). Both the vertical and horizontal movements of table 2 are the maximum reported. The vertical movements measured have been proved to be higher than the horizontal ones. Figure 3 shows the ratio of vertical to horizontal movements per measuring station. The ratio for station a/a No32 has not been included in the chart because its value (7) is well outside the cluster of the majority of the ratio values. The average ratio of vertical to horizontal movements is 8. However, most of the ratio values are between 2 and 6. Ratio of vertical/ horizontal movemen Measuring stations Figure 3 - Ratio of vertical to horizontal movements h(mm) /3/ 15/7/ 15/11/ 15/3/1 15/7/1 15/11/1 15/3/2 15/7/2 15/11/2 Figure 4 - Pattern of vertical movements. The lines represent the movements of the three measuring points. One on the roof and two on the sidewalls of the tunnel h(mm) /3/ 16/3/ 17/3/ 4

5 These values may be firstly attributed to some degree to practical construction reasons, such as the foundation conditions of the shotcrete shell. The areas of foundation of the shotcrete shell generally suffer from over-excavation, especially when steel sets are used. Also, the water accumulation in these areas may cause deterioration and softening of the siltstone or the soil-like fault geomaterial. This may result in some ground consolidation in the foundation areas. Figure 4 shows a typical pattern of the vertical movements from station a/a No26. The three curves represent the vertical movements measured in three points (one at the top and two at either side of the shell) of the shotcrete shell. Discussion and evaluation The highest values of movements have been reported at convergence stations a/a No15 and a/a No16 in areas of fault zones, of a thickness of approximately one meter. Water has been present in the form of drops or small inflows. h maximum (mm R = RMR Figure 5 - RMR vs maximum vertical movements [ h maximum (mm)] To evaluate the rock mass conditions during construction, at this particular part of the tunnel, we have used the RMR classification system. The average RMR value for support class IV was 32 (poor rock), while the highest value was 37 and the lowest 23. For support class III, the average RMR value was 46 (sound rock), while the highest value was 54 and the lowest 39. The relatively higher movement values were obtained in the areas of support class IV (see table 2). Figure 5 shows a correlation between the maximum vertical movements and the RMR values, while in figure 6 the RMR values have been replaced by the overburden. In the preparation of these charts, the values of the two monitoring stations mentioned in the first paragraph have been considered as outliers and have not been included in the chart. The extreme values of movements observed at these stations may be probably explained by the presence of thick faults: the behavior of the ground/shotcrete shell system in this area has been mostly controlled through the poor soil-like strength characteristics of the fault gouge, instead of those of the surrounding rock mass. Despite the big scatter of data, we generally observe an increase of movements when the RMR values decrease. The upper left cluster of points represents mainly the values related to support class IV, while the lower right cluster of points represents the values related to support class III. Considering the uncertainties involved in the classification of the rock mass and the development of 5

6 movements, i.e. mainly the timing of the support measures completion, the timing of installation of the measuring station, and the rate of excavation, this tendency seems to be significant. The correlation coefficient of the trend line is R= Overburden (m) R = Vertical movements (mm) Figure 6 - Maximum vertical movement vs overburden The chart in figure 6 correlates the magnitude of the movements with the overburden. The correlation coefficient for is R=.34. Despite the fact that this correlation is not high, the movements generally tend to increase with the increase of the overburden. The rate of the early movements has proved to correlate with the history of the measuring station and the behavior of the ground/support system. Figure 7 correlates the average rate of vertical movements observed during the first days with the maximum movements measured. Cracks in the shotcrete shell start to develop when the vertical movements exceed the value of 4 to 5mm or ~1% of the 6m excavation height. h maximum (mm R = Rate (α) of early movement (mm/day) Figure 7 - Rate of movement (a) vs maximum vertical movement [ h maximum (mm)] 6

7 This behavior is associated with a rate of an early movement of almost.3cm to.4cm per day. Higher rates resulted in more extensive damage of the shotcrete shell. These high rates generally correlate well with the rock mass conditions and especially with the areas of the fault zones filled with soft clayey material. SHOTCRETE SHELL PERFORMANCE AND BACK ANALYSIS Shotcrete shell failures and countermeasures The deformations resulted in cracking of the shotcrete shell for almost 22m of the tunnel s length (figure 8). Apart from the cracks, in this part of the tunnel, the anticipated development of movements, during the excavation of the tunnel bench, has been also a challenge. The top heading had to be strengthened by means of additional measures, prior to the commencement of the bench excavation. Based on the monitoring data, a detailed back analysis design has been carried out. (Koronakis, 21). Figure 8 - Crack development in the shotcrete shell The aim has been to obtain numerical results of the actual movements and compare the calculated stress of the shotcrete shell with the maximum available. Plain strain finite element analyses have been carried out. The design simulated the rock mass as an elastoplastic continuum medium and applied different k ratios (horizontal/vertical stress) to produce the actual pattern of movements. The design parameters have been calculated using the Hoek and Brown failure criterion. The input data used for support class IV are: GSI=2, σ ci = 15, m i = 8, Em(GPa) =.689, C(Kpa)=255, φ =2.9, γ(kn/m 3 )=27, ν=.3 and K The back analysis results have proved that the stress values developed in the shotcrete shell exceed the allowable one, thus resulting in the development of cracks. The idea behind the measures (grout improvement of the rock mass) designed to strengthen the damaged areas, was to keep the increases in the stresses and movements to the minimum possible 7

8 during the excavation of the bench. To estimate the strength properties of the improved ground around the excavation, it has been assumed that 5% of the area, to be grouted, would be filled with grout. Therefore the grouted rock mass strength would have a σ cm improved =.95σ cm +.5σ grout (σ cm =.74MPa, for the rock mass, σ grout = 3 MPa, for the grout). The calculated input data (for the subsequent analyses) for the improved rock mass (of the support class IV) have been: GSI=31, σ ci = 25 m i = 8, E m (GPa) =1.7, C(Kpa)=74, φ =24.7, γ(kn/m 3 )=27, ν=.3 and K Figure 9 shows the magnitude of the yield zone around the tunnel bores: a) after bench excavation (with no ground improvement), b) after bench excavation (with ground improvement). In the first case yield zone extends more than 9 meters, while in the second case yield zone is limited to 6 meters. The saving, in the total maximum vertical displacement, between the two cases is 67mm. (a) (b) Figure 9 Extend of yield zone around the two tunnel bores:(a) more than 9 meters after bench excavation (no improvement), (b) less than 6 meters after bench excavation (with improvement+additional rock bolts)) Table 3 shows the stress increases in the shotcrete shell for the abovementioned cases. The additional strengthening measures have been resulted to the minimum possible stress increase (from 49.81MPa to 59.64MPa) in the partially failed shotcrete shell. They have been selected as cost effective in the context of the construction contract, including improvement of the rock mass by means of grouting, installation of additional anchors and excavation of the bench with a small (1 meter) excavation step. The measures taken to strengthen/improve the damaged shotcrete shell (to ensure safe working conditions) have not been included in the design analyses and are not discussed herewith. Table 3 - Back analysis results (maximum stresses) in the Driskos tunnel section under consideration. Support section IV Maximum stress in the shotcrete shell (top heading). Maximum allowable, 28,5MPa Excavation of phase A (top heading). Actual situation. 49,81 > 28,5 Excavation of phase B without additional measures 73,2 > 28,5 Excavation of phase B: Additional measures: Rock bolts 1,3mX1,3m 8m long in top heading, grouting of the rock mass at the sidewalls of phases A and B of the tunnel of a depth reaching 5m, plus grouting below the tunnel invert to a depth reaching 3m. 59,64> 28,5 8

9 Grouting Figure 1 shows the pattern of the holes (per 1,5m of tunnel length) drilled for the grout injection around the tunnel. The pressures and volumes of grout injected are also shown (Koronakis, 21). Figure 1 Pattern of grouting. Pressures and volumes of grout injection in a section The length of the drill holes has been 4,5m and 6m, Ø55mm in diameter. Grouting has been generally carried out in two stages at 5m and 2m respectively (no casing, packers only). The pressures of grout injection (w/c < 1 and 3%-5% of bentonite) have been ranging from 1bar to 3bars and the volume of the grout injected has been generally less than 4 liters per hole. The maximum values, of grout volume injected have been recorded near and around the excavation perimeter. For the completion of the tunnel cross section, the bench has been excavated at an excavation and support step equal to 1 meter. The bench excavation has not triggered any further movements or development of cracks of the support shell. ASPECTS OF EMPIRISM Magnitude of movements Α brief analysis of the real absolute magnitudes of the movements followed the general principle of the methodology for estimating the potential tunnel squeezing, proposed by Hoek & Marinos (Hoek & Marinos, 2) using closed form solutions for a circular tunnel. According to this methodology, the strain percentage of an unsupported tunnel (tunnel closure/tunnel diameter x 1) may be predicted by application of the following equation: ε =.2( σ cm po ) ( 2 ) (1) Where σcm = rock mass strength and po = in situ stress. For the current data, the actual percentage strains (for the supported tunnel) have been plotted against the ratio of the rock mass strength σcm /in situ stress po. The rock mass strength has been 9

10 calculated with application of the Hoek & Brown failure criterion. The values of in situ stress p o are the overburden x unit weight of the rock mass. This is shown in the figure No Percentage strain ε ε =,7(σ cm /p o )-,8 R =, Rock mass strength σcm /in situ stress po Figure 11 Percentage strain (ε=1 x tunnel closure/tunnel diameter) of the supported tunnel vs rock mass strength σ cm /in situ stress p o The correlation (power curve) is high; however, the number of points is relatively limited. Similar data have not been included in the charts, as they could influence in the wrong way the correlation. Figure No 12 shows the curve proposed by Hoek & Marinos for unsupported tunnels and the curve for the supported Driskos tunnel. Due to the installed support, the latter is shifted towards limited strain values. 14 Percentage strain ε ε =,7(σ cm /p o )-,8 curve for Supported Driskos Tunnel ε =,2(σ cm /p o )-2 curve for Unsupported tunnel (Hoek & Marinos 2) Rock mass strength σ cm /in situ stress p o Figure 12 Percentage strain (ε=1 x tunnel closure/tunnel diameter) for the supported and unsupported tunnel vs rock mass strength σ cm /in situ stress p o 1

11 Rate of movements The early rate of the shotcrete shell movements development depends mainly on the rock mass strength, its speed of failure, its creep potential and the strength/stiffness development of the shotcrete shell. Generally speaking, the question associated with the behavior of the support is α) whether the young shotcrete will be strong enough to withstand the early loads imposed on it by the rock mass and b) what its post failure behavior is going to be. The early and final shotcrete strength, - and finally its performance and efficiency - will depend on the rate of its loading and on its properties/mix design. Depending on the early loading and the early rate of deformation, the final strength/modulus of elasticity of the shotcrete may be significantly reduced. Moussa, reports (Moussa, 1993) that early loading of the shotcrete shell up to 7% of its strength at early stages (almost one week) may result in significant losses of strength at later stages. In the context of this discussion, the rates of the strains development have been plotted against the ratio σ cm /p o. The quantity σ cm /p o is the one that correlates better with the early rate of movement. The result is shown in figure 13. It is obvious that, the lower the rock mass strength σ cm /in situ stress p o is, the higher the early rate of strain development becomes. The higher correlation (R=.81) is described by the logarithmic equation (2). The area above the dotted line on the plot includes the points that correspond to the measuring stations, where shotcrete failures have occurred. σ α = -2,13ln p cm o -,7 (2) The equation (2) of this curve is an example that predicts for a 12 m span tunnel and for the set of support measures described in table 1 - the early rate of strain development and the subsequent potential of shotcrete failure, from the ratio σ cm /p o. Of course, different support measures/method of excavation and related parameters will result in different curves and this is an issue that is related to the evaluation of more data. 8 α, rate of movements (mm/day σ cm/p o α= -2,13ln(σ cm /p o )-,7 R=,81 Figure 13 Rate of early movements (a=mm/day) in the shotcrete shell vs rock mass strength σ cm /in situ stress p o 11

12 Generally speaking, this equation proves that the poorer the rock mass is and the higher the in-situ stress, the faster its failure becomes (through the development of creep strains) and the faster the subsequent load application on the shotcrete shell will be. It illustrates the importance - when selecting/designing a supporting system for a tunnel in poor ground - of comparing in terms of time, the strength/stiffness development of the shotcrete and the stresses induced by the rock mass. CONCLUSIONS The movements which have been measured during the excavation of the top heading of a significant section of Driskos tunnel have been found to correlate with the rock mass quality and to a lesser degree with the overburden. The principles of the methodology proposed by Hoek & Marinos (Hoek & Marinos, 2), for strains relating to rock mass strength and in situ stress, apply well in the evaluated data of Driskos tunnel; the aforementioned parameters have been found to correlate with the measured strains in a similar manner. The rate of the early movements of the shotcrete shell was found to correlate well with the maximum observed ones, and the damages induced to it. It also correlates with the ratio rock mass strength σ cm /in situ stress p o. This rate seems to be a good indicator for predicting the shotcrete shell performance, testing the design, and proactively counteract. However, additional data are required, in order to determine significant values of early strain development rates, which may be used as an engineering tool. REFERENCES Hoek, E., Marinos, P. (2), Predicting Squeeze, Tunnels and Tunnelling International, December 2, pp Koronakis, N. (21), Definitive Design of Driskos Tunnel, Omikron Kappa Design Consultants Ltd, Athens, Greece, EDR GmbH, Munchen, Germany (Unpublished). Moussa, A. (1993), Finite Element Modelling for Shotcrete in Tunnelling, Ph.D. Thesis, Innsburg University, Austria. 12