Title

Plaxis Bulletin
Issue 33 / Spring 2013 Modelling Swelling Rock Behaviour in Tunnelling South Toulon Tube: Numerical Back-analysis of In-situ Measurements Stability Analysis of the Red River Dike: The Past to the Present

Ed

Table of contents
Page 5

The Plaxis Bulletin is the combined magazine of Plaxis bv and the Plaxis users association (NL). The bulletin focuses on the use of the finite element method in geotechnical engineering practise and includes articles on the practical application of the PLAXIS programs, case studies and backgrounds on the models implemented in PLAXIS. The bulletin offers a platform where users of PLAXIS can share ideas and experiences with each other. The editors welcome submission of papers for the Plaxis bulletin that fall in any of these categories. The manuscript should preferably be submitted in an electronic format, formatted as plain text without formatting. It should include the title of the paper, the name(s) of the authors and contact information (preferably e-mail) for the corresponding author(s). The main body of the article should be divided into appropriate sections and, if necessary, subsections. If any references are used, they should be listed at the end of the article. The author should ensure that the article is written clearly for ease of reading. In case figures are used in the text, it should be indicated where they should be placed approximately in the text. The figures themselves have to be supplied separately from the text in a vector based format (eps,ai). If photographs or scanned figures are used the author should ensure that they have a resolution of at least 300 dpi or a minimum of 3 mega pixels. The use of colour in figures and photographs is encouraged, as the Plaxis bulletin is printed in full-colour.

03 04 05 10
Page 10

Editorial New developments Modelling Swelling Rock Behaviour in Tunnelling South Toulon Tube: Numerical Back-analysis of In-situ Measurements Stability Analysis of the Red River Dike: The Past to the Present 3D Modelling of Frozen Ground and Associated Volume Expansion Recent activities

14

20
Page 14

22

Colophon
The Plaxis Bulletin is a publication of Plaxis bv and is distributed worldwide among Plaxis subscribers Editorial board: Ronald Brinkgreve Erwin Beernink Arny Lengkeek Design: Jori van den Munckhof Any correspondence regarding the Plaxis bulletin can be sent by e-mail to: bulletin@plaxis.nl or by regular mail to: Plaxis Bulletin c/o Annelies Vogelezang PO Box 572 2600 AN Delft The Netherlands For information about PLAXIS software contact your local agent or Plaxis main office: Plaxis bv P.O. Box 572 2600 AN Delft The Netherlands info@plaxis.nl www.plaxis.nl Tel: +31 (0)15 251 7720 Fax: +31 (0)15 257 3107

Page 22

Page 20

Editorial

The Plaxis head office has moved and from the begining of this new year we have started from our new location in Delft. We are very exited about the coming period for Plaxis, and expect to be able to provide the valuable products and services as we have been doing since the founding of our company in 1993. With this move we look to the future for 20 more years of success. In this issue of the Plaxis bulletin we have again tried to collect interesting articles and useful information for you. In the New Developments column we take a look at PLAXIS facilities to deal with pore pressure in saturated and unsaturated conditions. These facilities will also be available in 3D with the relesase of the new 3D PlaxFlow module. The new module is scheduled for release together with PLAXIS 3D 2013 (expected summer 2013). The first users article involves the modelling of swelling rock behaviour in tunnelling. The article compares the results of a numerical back analysis of a constitutive swelling model to in-situ measurements, using swelling parameters derived from laboratory swelling tests. The second users article is about the South Toulon Tube, and involves a numerical backanalysis on in situ measurements. The numerical model is described and the simulation is validated by comparing it to in situ measurements. The good fitting with the different measurements recorded in situ shows that the three-dimensional numerical modeling, with discretization of the inclusions, is a reliable tool to simulate the complex phenomenon of interaction between the excavation process, the reinforcements and the ground reaction.

The third users article covers a stability analysis of the red river dike. The authors compare the different levels of safety in the past with those of the present situation by using PLAXIS code that includes several material models. In stability analysis, PLAXIS demonstrated successfully the dike safety factor in flood waves with different construction stages. In addition to the contributions by PLAXIS users, there is an article on 3D modelling of frozen ground and associated volume expansion where PLAXIS Expert Services provided assistence with setting-up the models for analysis. We wish you an interesting reading experience and look forward to receive your comments on this 33rd Plaxis bulletin. The Editors

www.plaxis.nl l Spring issue 2013 l Plaxis Bulletin

Title

New developments
Ronald Brinkgreve, Plaxis bv

An important principle in geotechnical engineering is the division of total stress into effective stress and pore pressure. PLAXIS fully supports this effective stress principle. It provides several facilities to deal with pore pressure in saturated as well as unsaturated conditions. Pore pressure, in this respect, is generally pore water pressure, since air pore pressure is ignored. With the release of the new 3D PlaxFlow Module (expected summer 2013), all facilities to deal with pore pressure are available in 2D

as well as 3D.
The new 3D PlaxFlow Module allows for transient groundwater flow as well as fully coupled flow-deformation analysis. In the latter case, pore pressure can change in time as a result of (undrained) loading as well as a (simultaneous) change in hydraulic conditions. This involves the modelling of partially saturated soil behaviour in the unsaturated zone above the phreatic level. The behaviour in the unsaturated zone is defined by the soil-water retention curve, which is included in the Flow tab of material data sets for soil & interfaces. Since in most geotechnical projects unsaturated soil data is not available, PLAXIS includes predefined curves for different types of soil. In principle, the unsaturated zone includes suction (positive pore water stress as a result of capillary action). The amount of suction included in the active pore pressure (= effective suction) depends on the degree of saturation, Sr. The latter depends on the selected soil-water retention curve (Fig. 1). For very permeable soils, like coarse sand and gravel, the degree of saturation decreases rapidly with the distance above the phreatic level, whereas in low permeable soils, like silt and clay, the saturation above the phreatic level remains high. Therefore, especially in low permeable soils the influence of suction can be significant. In frictional soils, effective suction provides a kind of artificial cohesion. In the new 3D PlaxFlow Module, suction can be taken into account when needed, but in many cases it might be desired to avoid the influence of suction on the results. Therefore, the option Ignore suction is available for any type of calculation (Plastic, Safety, Dynamic, Consolidation and Fully coupled flowdeformation) to avoid the generation of suction in the unsaturated zone, equivalent with classical calculations based on Terzaghis effective stress. The Ignore suction option removes the need for different calculation modes. It simplifies the use of calculation features and the interpretation of results. Backgrounds and further details about the various possibilities in the new 3D PlaxFlow Module are described in the corresponding manuals. With the new implementation we have tried to make unsaturated soil behaviour, groundwater flow and fully coupled analysis accessible and understandable for all geotechnical engineers. We are very keen to know from you if we have succeeded in reaching this challenging goal.

Figure 1: Example of a soil-water retention curve y is the suction height above the phreatic level; Sr is the degree of saturation

Plaxis Bulletin l Spring issue 2013 l www.plaxis.nl

Title

Foto by Christian Ammering

Modelling Swelling Rock Behaviour in Tunnelling

Bert Schdlich & Helmut F. Schweiger, Institute for Soil Mechanics and Foundation Engineering, Graz University of Technology, Graz, Austria Thomas Marcher, ILF Consulting Engineers, Innsbruck, Austria

Although a great amount of practical experience has been gained in the last decades, tunnel design in swelling rock is still a very challenging task, as the recent examples of the Engelbergtunnel in southern Germany and the Chienbergtunnel in Switzerland demonstrate. Reliable prediction of swelling pressures and swelling deformations especially in anhydritic rock is extremely difficult due to the heterogeneity of the material and the complexity of the involved transport mechanisms. However, modern design codes and engineering practice demand capacity checks for tunnel linings, which usually can only be provided by numerical analysis with an appropriate constitutive model. Such a constitutive swelling model, which adds swelling strains in dependence on the stress level and accounts for the time dependent evolution of swelling, has been implemented for Plaxis. This article compares the results of a numerical back analysis with this model to in-situ measurements, using swelling parameters derived from laboratory swelling tests.

Swelling rocks are geomaterials which increase in volume if water is allowed to infiltrate. The most prominent rock types exhibiting swelling behaviour are certain types of claystone and anhydrite-bearing rocks, which can be commonly found in northern Switzerland and southern Germany. Tunnelling in such materials is notoriously difficult: If a flexible invert lining is installed, large invert heave evolves after tunnel excavation. In case these deformations are prevented by a rigid support concept, large swelling pressures may develop at the tunnel lining. It is well known that swelling deformations at least in claystone reduce with the logarithm of stress, and that swelling deformations can be completely suppressed by sufficiently high pressure. The chemical processes in anhydrite swelling, on the other hand, are completely different, and the semi-logarithmic relationship between swelling strains and stress level (Grob 1972) is not universally accepted for these materials. Evolution of swelling with time in both claystone and anhydrite depends on the availability of water, which is governed by the permeability of the material, layering of the subsoil and the amount of water recharge. As some of these factors relate to characteristics of the specific boundary value problem rather than

the material itself, parameters determining the time-swell behaviour cannot be transferred from laboratory tests to large-scale problems. Constitutive model The constitutive model used in this paper has been implemented by T. Benz (NTNU Norway) as a user-defined soil model for PLAXIS. The model employs four parameters for strength and stiffness and three parameters for swelling. f and c are the well-known Mohr-Coulomb friction angle and cohesion, E and n are the isotropic elastic Youngs modulus and Poissons ratio, respectively. Crossanisotropic elasticity can also be considered but is not used in this study. The meaning of the swelling parameters is shown in Figure 1 and Figure 2. The maximum swelling pressure sq0 is the axial stress beyond which no swelling occurs, the swelling potential kq gives the inclination of the swelling curve in semi-logarithmic scale and the parameter hq is related to the time until the final swelling strain has developed (Wittke & Wittke 2005). Cross-anisotropic swelling can be considered in the model, but again this feature is not used here. (1)

If necessary the time-swell behaviour can be related to elastic and plastic volumetric strains, evel and evpl, by using parameters A el and A pl to define the time swelling parameter hq: (2)

Positive volumetric strains (loosening of the material) result in faster approach of the final swelling strain, while negative volumetric strains delay or may even stop the evolution of the swelling strains. This approach accounts for the dependency of the swelling rate on the penetration rate of water. Construction of the Pfndertunnel The 6.7 km long first tube of the Pfaendertunnel near Bregenz (Austria) was constructed in 1976-1980 according to the principles of the New Austrian Tunnelling method (NATM). While top heading and bench excavation were carried out without major difficulties, significant invert heave of up to 30 cm was observed after about 75% of the tunnel length was excavated. These observations lead to detailed laboratory investigations of the swelling characteristics of the Pfaenderstock material, an extensive monitoring program and to the installation of additional anchors in the tunnel invert.

www.plaxis.nl l Spring issue 2013 l Plaxis Bulletin

Laboratory swelling tests The Pfaenderstock consists of various layers of sandstone, conglomerate, claystone and marl, which are summarized as upper freshwater molasse. The marl (claystone) layers were identified as the rock type causing the swelling due to their high content of Montmorillonite (Weiss et al. 1980). Czurda & Ginther (1983) distinguished between undisturbed molasse marl (series A, Figure 4) and the fault zone material (series B, Figure 5). Series A samples showed

higher swelling potential, but lower maximum swelling pressures than the samples of series B. This notable difference was attributed to relaxation and swelling of the series B samples before the samples could be tested. For the back analysis two swelling parameter sets are considered, which represent the upper and lower boundary of the test results. The time swelling parameters A 0, A el and A pl are calibrated to match the in situ time-swelling curve. Numerical model and material parameters The 2D finite element model used in this study is shown in Figure 6. Tunnel geometry and basic material parameters of the marl layer (E = 2.5 GPa,

f = 34, c = 1000 kPa) have been taken from John et al. (2009). Tunnel overburden is ~200 m above the tunnel crown, which is representative of the cross section at km 5+373. Linear elastic plate elements are used for the shotcrete lining, with E = 7.5 GPa for the young and E = 15 GPa for the cured shotcrete. The final concrete lining is modelled with volume elements assuming linear elastic behaviour and a stiffness of E = 30 GPa. The final lining thickness varies between 50 cm at the invert and 25 cm at the crown. Swelling parameters are listed in Table 1. Sets 1a, 1b and 2a only employ A 0 for the time dependency of swelling, while in set 2b evolution of swelling with time is entirely governed by elastic volumetric strains.

Fig. 1: Semi-logarithmic swelling law (Grob 1972)

Figure 2: Influence of hq on evolution of swelling strains

Figure 3: Pfaendertunnel cross section 1st tube (after John & Pilser 2011)

Figure 4: Swelling test results, series A

Figure 5: Swelling test results, series B

Plaxis Bulletin l Spring issue 2013 l www.plaxis.nl

Figure 7: Development of invert heave with time Figure 6: Finite element model (dimensions in m)

Figure 8: Profile of vertical displacements, a) numerical analysis at t = 7180 d, b) measurements km 5+820 (after John 1982)

Figure 9: Development of pressure on the lining (set 1b)

Parameter Swelling potential kq [%] Max. swelling stress sq0 [kPa] A0 Ael Apl Table 1: Swelling parameters

Set 1a 3.0 1000 5.0e 0.0 0.0

-3

Set 1b 3.0 1500 2.5e 0.0 0.0

-3

Set 2a 0.75 4000 3.0e 0.0 0.0

-3

Set 2b 0.75 4000 0.0 9.0 0.0

Calculation phases After top heading / invert excavation (assuming pre-relaxation factors of 75% and 37.5%, respectively), the concrete invert arch is installed. Swelling is confined in the model to an area of 15 m x 15 m below the tunnel invert. After a swelling phase of 65 days, the final lining is activated, followed by another swelling phase of 115 days. John (1982) reported that the decision on invert anchoring and pre-stressing was based on the swell heave deformations observed up to this point. In the cross section considered here this resulted in an anchor pattern of 2.2 m spacing.

Evolution of invert heave with time Figure 9 compares the time-swelling curves calculated with the different parameter sets with the measured invert heave in km 5+373. The measurements plot close to a straight line in logarithmic time scale, which cannot be reproduced exactly by the exponential approach employed in the model. The match with the measured invert heave is, however, sufficient from a practical point of view. Set 1a delivers too little invert heave (10mm), and the development of deformations completely

stops after activating the prestressed anchors. Increasing the maximum swelling stress by 50% (set 1b) yields ~50% more deformation and a better match with the measurements. While such a significant influence may be expected, it should be noted that experimental results for these two sets plot so close to each other that either of the two parameter sets appears justified (Figure 4). Surprisingly, sets 2a and 2b which represent much smaller free-swell deformations deliver more invert heave than sets 1a and 1b. This is a result of the higher maximum swelling stress in sets 2a and 2b, which activates swelling in deeper rock layers, yet with a small swelling potential. Swelling deformations are thus more widely distributed with set 2a and 2b. Modelling the evolution of swelling with time entirely in dependence on elastic volumetric strains (set 2b) results in a slightly more prolonged time-swell-curve than using a constant value of A 0 (set 2a). In set 2b the rate of swelling does not only decrease due to the convergence with the final swelling strain, but also due to negative elastic volumetric strains. The large positive volumetric strains after tunnel excavation are gradually reduced in the swelling phases by the increasing swelling pressure.

www.plaxis.nl l Spring issue 2013 l Plaxis Bulletin

Modelling Swelling Rock Behaviour in Tunnelling

Figure 10: Variation of maximum swelling stress

Figure 11: Variation of time swelling parameters (set 1b)

Figure 12: Variation of rock stiffness (set 1b)

Figure 13: Variation of stress pre-relaxation

Distribution of swelling strains over depth The proportion of the rock mass which is affected by swelling depends primarily on the maximum swelling stress. For set 1b ( sq0 = 1500 kPa) the swelling zone is confined to about 2 m below the tunnel invert, which matches well with the sliding micrometer measurements in the neighbouring cross section km 5+820 (Figure 8). The swelling zone with set 2a ( sq0 = 4000 kPa) is much deeper due to the higher maximum swelling pressure, even though similar invert heave is obtained with both parameter sets. These results indicate that the maximum in situ swelling pressure is rather in the range of 1000-2000 kPa than close to the in-situ stresses. Swelling pressure Figure 9 shows the distribution of swelling pressure on the tunnel invert lining for different stages in time for parameter set 1b. The circumferential distance L is measured from the tunnel invert, such that L = 0 m is directly at the invert and L = 5 m is the end of the swelling area. No pressure measurements are available. Due to the stiffer support provided to the tunnel lining at the sides of the tunnel, the maximum

swelling pressure does not occur at the tunnel invert but at a distance of ~3.8 m. Anchor prestressing increases the normal stress on the lining by about 90 kPa. The difference to the distributed prestressing force of (0.8*640 kN / 2.2 m / 2.2. m) = 106 kN/m2 is a result of the already closed final lining, which distributes part of the applied load in circumferential direction. Comparing the increase in pressure to the swelling line of set 1b at 200-300 kPa (Figure 4) explains the limited influence of prestressing in the numerical calculations. Even though anchor prestressing increases the pressure by ~45%, reduction of final swelling strain is only about 18% due to the semilogarithmic swelling law. Additionally, the effect of prestressing diminishes rapidly with increasing distance to the tunnel, and the deeper rock layers remain virtually unaffected. Variation of maximum swelling pressure The calculated invert heave is notably sensitive to the maximum swelling pressure sq0 assumed in the numerical analysis. As the variation of this parameter in the laboratory swelling tests is rather large albeit concealed by the logarithmic stress scale sq0 has been varied from 500 kPa

to 2000 kPa ( A 0 = 2.5e -3). Results indicate a linear increase of invert heave with sq0 (Figure 10). This is primarily the result of the increasing depth of the swelling zone below the tunnel invert (Figure 8), and not so much due to higher swelling strains directly underneath the tunnel invert. Influence of other material parameters The influence of other material parameters on swelling deformations is limited. As expected, the time swelling parameter A 0 has a notable influence on the evolution of swelling deformations, but not on final deformations (Figure 11). Varying the elastic rock stiffness had virtually no effect on swelling deformations after tunnel excavation (Figure 12), but naturally changed deformations during tunnel excavation. Variation of the 2D pre-relaxation factors which in most practical cases are an educated guess rather than a thoroughly derived parameter also had no notably influence on swelling deformation (Figure 13). As no temporary invert lining was installed after top heading excavation, stresses in the rock mass at the tunnel invert drop to ~0 during tunnel excavation, independent of the pre-relaxation factors applied in the excavation phases.

Plaxis Bulletin l Spring issue 2013 l www.plaxis.nl

Concluding remarks This article presented the results of a back analysis of measured swelling deformations in the Pfaendertunnel (Austria). A constitutive model based on Grobs swelling law and exponential convergence with final swelling strains over time was used for the numerical calculations. Input swelling parameters were derived from laboratory swelling tests. Due to the large variation of laboratory test results, the sensitivity of model predictions on the input swelling parameters was investigated. Different sets of swelling potential kq and maximum swelling stress sq0 delivered very similar swelling deformations at the tunnel lining, as increasing sq0 is roughly equivalent to increasing kq. However, good match with the measured displacement profile below the tunnel invert was only obtained with sq0 = 1500 kPa, which represents the upper edge of the experimental results on undisturbed molasse marl. Using higher values of sq0 (and lower values of kq) delivers too large swelling zones. The invert heave measurements plot close to a straight line in logarithmic time scale, which cannot be exactly reproduced by the exponential approach of the constitutive model. The match with the measured evolution of swelling, however, is sufficient from a practical point of view.

References Czurda, K. A., and Ginther, G. (1983) Quellverhalten der Molassemergel im Pfnderstock bei Bregenz, sterreich, Mitt. sterr. geolog. Ges., 76, pp. 141-160. Grob, H. (1972) Schwelldruck im Belchentunnel, Proc. Int. Symp. fr Untertagebau, Luzern, pp. 99-119. John, M. (1982) Anwendung der neuen sterreichischen Tunnelbauweise bei quellendem Gebirge im Pfndertunnel Proc. of the 31st Geomechanik Kolloquium, Salzburg, Austria. John, M., Marcher, T., Pilser, G., and Alber, O. (2009) Considerations of swelling for the 2nd bore of the Pfndertunnel, Proc. of the World Tunnel Congress 2009, Budapest, Hungary, pp. 50-61. John, M., and Pilser, G. (2011) Criteria for selecting a tunnelling method using the first and the second tube of the Pfnder tunnel as example, Geomechanics and Tunnelling, 4(11), pp. 527-533. Weiss, E. H., Mller, H. M., Riedmller, G., and Schwaighofer, B. (1980) Zum Problem quellfhiger Gesteine im Tunnelbau, Geolog. Palont. Mitt. Innsbruck, 10(5), pp. 207-210. Wittke, W., and Wittke, M. (2005) Design, construction and supervision of tunnels in swelling rock, Proc. 31st ITA World Tunnelling Congress 2005, pp. 1173-1178.

www.plaxis.nl l Spring issue 2013 l Plaxis Bulletin

South Toulon Tube: Numerical Back-analysis of In-situ Measurements

J.P. Janin, H. Le Bissonnais, A. Guilloux, TERRASOL, Paris, France, D. Dias, University Joseph-Fourier, Grenoble, France, R. Kastner, LGCIE, INSA de Lyon, Lyon, France, F. Emeriault, INP, Grenoble, France

Full face excavation with ground reinforcement has become a common technique to build large tunnels in soft rock or hard soil. Nevertheless, at the design phase, it remains difficult to assess the effect of the different construction and reinforcement elements on the control of the ground movements and settlements. In order to improve the understanding of ground response to this tunnelling method, a monitoring section has been installed during the construction of the south Toulon tunnel (France). An important database was obtained and subsequently used for numerical back-analysis. A 3D FE calculation (PLAXIS 3D v.2010), modeling the real pre-reinforcements system and workflow steps, permitted to simulate the in situ measurements.

The south Toulon tunnel will connect motorways A50 and A57 from Nice to Marseille. It is an urban shallow tunnel, 12 m in diameter and 1820 m in length, excavated through very difficult heterogeneous soils and with a limited overburden (maximum of about 35 m). The construction method used is full face excavation and ground reinforcement ahead of the tunnel face by pipe umbrellas (long forepoling) and face bolting. The construction sequences and the amount of reinforcement were continuously adapted to the overburden, the soil conditions and the measured settlements (Janin et al [2011]). In addition to the regular settlement measures, a specific monitoring zone was set up to improve the understanding of ground response and to collect precise data for validating a 3D numerical model. The complex phenomenon of interaction between the excavation process, the reinforcements and the ground reaction demand in fact a 3D approach. In this paper, the numerical model is described and the simulation is validated by comparing it to in situ measurements. 2 Presentation on the Analised Section The monitoring section is situated in Alexandre Ier garden on the west side of Toulon at the excavation progress PM 882. In this zone the cover depth was about 25 m. 2.1 Instrumentation The instruments set up in the analyzed section are presented in Figure 1. The instrumentation from the surface is composed of 2 inclinometers on both sides of the tunnel, one vertical

extensometer on the tunnel axis and 3 surface target prisms (X, Y, Z) close to the previous instruments. In addition, 4 radial extensometers, 6 vibrating wire strain gauges placed on the steel rib, 5 pressure cells and convergence targets were installed from the tunnel. 2.2 Geology The geological profile of the section has been drawn (see Fig. 1) based on the borehole investigations. They showed that the geological stratigraphy is generally horizontal and the degree of alteration of the phyllitic bedrock is considerably high. Despite the variations of materials characteristics (especially in the

bedrock), average ground properties of the different strata were proposed at the detailed design phase (see Table 1). 2.3 Excavation method The south Toulon tunnel was excavated on the basis of the so-called ADECO-RS method developed by Lunardi [2008]. According to this method, pipe umbrellas (6 or 14 of inclination) and horizontal face bolts were installed. The excavation process progressed by 1.5 m steps and after each step one HEB 180 steel rib was installed. In this zone, the tunnel invert (HEB 220 + concrete) was realized with a distance to tunnel face of about 39 m.

Figure 1: Three-dimensional model

10

Plaxis Bulletin l Spring issue 2013 l www.plaxis.nl

Depth (m) 0 to 3.5 3.5 to 5.9 Below 5.9

Soils Fill Colluviums Bedrock

g(kN/m3) 19 20.8 24.2

Eref50 = Erefoed (MPa) 1.6 40 240

Erefur = 3.Eref50 (MPa) 4.8 120 720

c (kPa) 2 10 40

j () 20 30 25

y () 0 0 0

vur 0.2 0.2 0.2

K0 0.5 0.5 1

Table 1: Main geotechnical parameters

Support type Walls tunnel Tunnel invert Tunnel face

Support description HEB 180 (spacing 1.5 m) + 25 cm shotcrete HEB 220 (spacing 1.5 m) + 30 cm shotcrete 15 cm shotcrete

E (GPa) 13.5 14 10

Thickness (m) 0.25 0.3 0.3

25 m. A three-dimensional non uniform mesh with smaller elements around the excavation is used. Finally, the model contains 158000 tetrahedral elements and 247000 nodes. All movements are fixed at the bottom of the model and horizontal displacements are blocked in models lateral faces. 3.2 Mechanical parameters and simulation of the excavation process The ground is represented by the non linear elasto-plastic HS model (Hardening Soil Model) implemented in the PLAXIS code. Hejazi [2008] showed, in a tunnel excavation study, that this model might produce ground deformations that better fit with in situ measurements than using the linear elastic/Mohr-Coulomb model. The geotechnical parameters considered in the simulation are listed in Table 1. The initial stress field is considered as isotropic (in conformity with the studies previously realized on Toulon soils by Constantin [1988] and Dias [1999]). In the numerical model, the shotcrete at tunnel face, the excavation support and the tunnel invert are modeled by plate elements with a linear elastic behavior. The mechanical parameters are defined in Table 2. The parameters of support and tunnel invert are determined by homogenization based on steel and shotcrete characteristics and ribs spacing. A rigid interface is considered between the support and the ground.

Table 2: Mechanical parameters of tunnel support

3 Three-dimentional Back-analisys A three-dimensional model was realized to simulate the tunnel excavation process of the area where the monitoring zone had been placed. The analysis was carried out by means of the three-dimensional finite element code PLAXIS 3D (version 2010). 3.1 Geometry Considering the geometry symmetry, only half of the entire domain is taken into account in the analysis as shown in Fig. 2. The real shape of Toulon tunnel with a cross section area of 120 m2 is imported from CAD. The tunnel is then modeled by extruding the surface in X direction. The tunnel temporary support and the pre-reinforcement are modeled and the process is explained in the following paragraph. In order to avoid boundary effects, the extension of the mesh is equal to 150 m in X and Y directions and 70 m in Z direction. The cover depth is about
Figure 2: Geological section and instruments

www.plaxis.nl l Spring issue 2013 l Plaxis Bulletin

11

South Toulon Tube: Numerical Back-analysis of In-situ Measurements

The calculation is carried out in drained conditions. The tunnel excavation is simulated in a first stage by 10 steps, 3 meters long, followed by 60 steps with an excavation length of 1.5 m as done in situ. In each phase, the tunnel lining is installed 1.5 m behind the tunnel face, on which the shotcrete application is simulated as well. The tunnel invert is activated 39 m behind the tunnel face progress. The pre-reinforcements, i.e. the pipes constituting the long forepoling and the face bolts, are simulated using embedded pile structures (see Fig. 3). Thanks to in situ pullout tests, a realistic limit skin resistance could be introduced between the piles and the soil. The maximum friction resistance measured along the soil/bolt interface is equal to 135 kN/m. Figure 4 shows the main characteristics of the pre-

reinforcement system considered in the numerical simulation, based on works realized in the studied zone. As far as the umbrella pre-support is concerned, 13 autodrilling pipes (51/33 mm) were taken into account and renewed every 9 m. The same inclination of 6 has been kept all along the model in order to simplify the meshing. As for the face bolts, a constant length of 18 m is considered. Besides, Dias and Kastner [2005] proved that the bolting system is characterized by the global stiffness (ES). Therefore, the same number of bolts (20) is kept all along the model in order to simplify the meshing. The real bolting density, installed in situ, is simulated varying proportionally the bolts modulus and the friction resistance based on the material characteristics listed in Table 3.

4 Comparisons Between Numerical Simulation and Measurements Figure 5a shows the settlements of different surface points, placed directly above the tunnel axis in the analyzed zone, against their distance from the tunnel face. The excavation started to influence settlements more or less 30 m ahead of the tunnel face. Afterwards, settlements accelerated and finally reach a threshold 50 m behind the tunnel face with a settlement of around 20 mm. The progression of settlements obtained with the numerical calculation fits with the measurements evolution. Similarly, threedimensional modeling seems to well represent the shape of the transverse settlement trough both in terms of maximum settlement and trough width (see Fig. 5b). The final measurements of inclinometer movements in the monitoring section plane

Steel bolts Modulus E (GPa) Cross section A (m2) Moment of inertia I (m4) Table 3: Face bolts characteristics 210 0.488 10-3 0.0327 10-6

Fibreglass bolts 40 0.8 10-3 ~0

Figure 3: Umbrella forepoling and face bolts modeled with embedded piles

Figure 4: Simulation of pre-reinforcements in the numerical model

Figure 5: Comparison between in situ measurements and 3D simulation - Settlement evolution against tunnel face advance (a) and transverse settlement trough (b)

12

Plaxis Bulletin l Spring issue 2013 l www.plaxis.nl

South Toulon Tube: Numerical Back-analysis of In-situ Measurements

showed two important phenomena (see Fig. 6). On one hand, the first few meters of inclinometers converged towards the tunnel and, on the other hand, a local convergence (belly) is observed at the tunnel level. The numerical approaches correctly estimate the two phenomena. A comparison between the tunnel support deformation obtained with numerical calculation and that inferred from in situ measurements is also performed. The 3D model proves to be able to represent the support deformation with an acceptable accuracy. The difference between the measurements and the numerical results, especially in the bottom part of the rib, is probably related to the simplifications related to the way of modeling (homogenization of steel and shotcrete characteristics, excavation of the stross carried out at the same time as that of the tunnel face). 5 Conclusions The monitoring section, installed during the construction of Toulon tunnels south tube, allows analyzing the evolution of soil deformation during the excavation progress. It also permits to create an important database, used later to validate a numerical simulation. A 3D numerical model was performed with PLAXIS 3D, taking into account the real excavation process and the pre-reinforcements installed in situ. The good fitting with the different measurements recorded in situ shows that the three-dimensional numerical modeling, with discretization of the inclusions, is a reliable tool to simulate the complex phenomenon of interaction between the excavation process, the reinforcements and the ground reaction. Finally, PLAXIS 3D seems to be a useful and efficient tool to predict the movements caused by the excavation of a tunnel, realized with ground pre-reinforcements.

References Constantin, B, Durand, J.P., Thone M. (1998), Progrs technologiques dans le cadre de lutilisation de la mthode du prdcoupage mcanique Toulon. In : Journes dtudes Internationales de Chambry, Association Franaise des Travaux en Souterrain, Paris, pp. 171180. Dias, D. (1999), Renforcement du front de taille des tunnels par boulonnage - Etude numrique et application un cas rel en site urbain. Thse Doc. Lyon : INSA de Lyon, p. 320. Dias, D, Kastner (2005), Modlisation numrique de lapport du renforcement par boulonnage du front de taille des tunnels. Canadian Geotechnical Journal, Vol. 42, pp. 16561674. Hejazi Y, Dias D, Kastner R, (2008), Impact of constitutive models on the numerical analysis of underground constructions. Acta Geotechnica, Vol. 3, n 4, pp. 251258. Janin, J.P., Dias, D, Emeriault, F, Kastner, R, Le Bissonnais, H, Guilloux, A, (2011), Settlement monitoring and tunneling process adaptation case of South Toulon Tunnel. The seventh International Symposium on Geotechnical Aspects of Underground Construction in Soft Ground, tc28 Rome. Rome. Lunardi, P. (2008), Design and construction of tunnels Analysis of controlled deformation in rocks and soils (ADECO-RS). Berlin: Springer.

Figure 7: Comparison between in situ measurements and 3D simulation Deformation of the tunnel support Figure 6: Comparison between in situ measurements and 3D simulation - Inclinometer movements in monitoring section plane

www.plaxis.nl l Spring issue 2013 l Plaxis Bulletin

13

Stability Analysis of the Red River Dike: The Past to the Present
Pham Quang Tu, P.H.A.J.M van Gelder, Technical University of Delft, The Netherlands Trinh Minh Thub, Hanoi Water Resources University

The Red River delta is located in Northern Vietnam with an area of around 15,500 km2 and a population of nearly 20 million people. The delta is protected by approximately 3000 km of dikes which are classified into four grades from III, II, I to special; and each dike type attains a different level of safety. In the past, most of the dikes, which were not high enough, failed mainly due to overflowing; therefore the area was usually threatened with flooding. Nowadays, as a result of many measures that have been imposed in the delta such as strengthening the dike systems, constructing the dams in the upstream river for multiple purposes, the flood risk in the area is reduced significantly. In this paper, the authors compare the different levels of safety in the past with those of the present situation by using Plaxis code that includes several material models.

The Red River (known as Song Hong in Vietnamese) flows from the mountain areas of Southern China to the Gulf of Tonkin (East Sea) with a total length of over 1150km (the length in Vietnams area is about 510km), as seen in Figure 1. The River is the biggest one in the northern part of Vietnam with the total dike length of nearly 1700km that take up more than a half of the length of the dike systems of the delta. The dikes are divided into different parts due to their specifically protective functions, and each part will follow a different level of safety. In the Hanoi area, the length of dikes of grade 1 and special*1 is around 250km in a total of 470km (DDMFC, 2009), (see Figure 2). In this part, four periods of dike development will be discussed to give readers an overview of dike rehabilitation for hundreds of years ago. The first period of the stability analysis is the empirical period of Vietnam which started in the 11th century, when the first dikes were built, till 1890s. However, the significant progress in flood protection was only made from the beginning of the 19th century in the Nguyen dynasties. At that time, people knew how to measure the height of flood water level and strengthen the dike system after each flood. As a result, the dike crests were increased each year after dike breaches and floods, without any prediction for the next levels of

Figure 1: Red river basin (from Khoi [2010])

*1 special grade is used in Vietnam dike system as the top grade which is more important than the grade 1(for instance, the dike system in Hanoi area, the capital of Vietnam)

14

Plaxis Bulletin l Spring issue 2013 l www.plaxis.nl

safety. In this period the last geometry of dike and the severity of the typical flood in the year 1893 were opted for the stability analysis. There were 38 years with dike failures during this period causing serious damage to the economy and fatalities (GDM, 1995), meanwhile in the 20th century, only 8 years with dike failure were observed. The second period of the analysis is from 1890s to 1945 when France fully established their ruling in Vietnam. After assisting the Nguyen dynasties in governing the whole country, the French widely exploited the natural resources; as a result, infrastructures were developed for their purposes. At that time, French engineers were requested to transfer strategically their experiences of flood protection in Hanoi, for instance, the 1st and the 2nd dike program, under the supervision of the Dikes Commission ( see GDM, 1995; Phan, 1985). Consequently, there were some noticeable progresses in flood mitigation during this time; for instance, the increase of the dike crest from 10.5m to 13.0m and the construction of Day Dam for flood diversion to lower the water level in Hanoi area, etc.
Figure 2: Red river around Hanoi area

Year 1893 1915 1945 1971 1996 Future*2 Figure 3: Probablility distribution of maximum water level

Flood water level (m) 10.5 (13)* 12 13.5 14.6 12.98 14.9

Flood duration (days) 7 12 5 5 23 16

Note* 9 10.5 10.5 11.5 11.5 11.5

Table 1: Flood water level

*2 Flood duration is calculated following a threshold level (here, alarm level 3~11,5m at Hanoi)

www.plaxis.nl l Spring issue 2013 l Plaxis Bulletin

15

Stability Analysis of the Red River Dike: The Past to the Present

The third period, the Vietnam war, spread from 1945 till 1975 which resulted in the anomalies of river dikes. At that time, the dikes could be bomb targets or were excavated to slow down enemy assaults. Although much effort spent on the rehabilitation of the whole dike system of the Red River Delta, they were still facing a high flood risk, and even the rehabilitation works had been undertaken for the whole dike system in the Red River Delta. Additionally, the attempts to heightening and strengthening the dike system such as millions of man-days, and cubic meter of earthworks, have been done in this period, which made it to the top of amount of dike rehabilitation work (see FPD, 2000). In terms of dike engineering, the anomalies of dike embankment did not only come from its very construction, but also from such social activities. The last period for the analysis is the Renovationof-Economy time (known as Doi Moi in Vietnam) from 1986 till now. Economically, private-owned enterprises were established and equally treated as state-owned ones. As a result, new technologies were applied to dike safety in the projects funded by the Asian Development Bank (ADB) or by local governments. The safety level of the flood defence system has been improved significantly due to the application of effective measures such as planting and protecting forests in upper basins, constructing dams and reservoirs to store water, strengthening dikes, and raising awareness of flood protection in the public. In short, the Red River dikes are now able to withstand a flood frequency from 1/250 year to 1/500 year comparing to 1/50 year in the third period (Ha, 2010). During the past hundred years, the overflowing and the resulting erosion of inner slope, that came from the limited height of dike system, caused many dike failures. In times from the empirical period until now the dike embankment compaction ratio, K*3, has varied considerably. In addition, soil properties and their distribution also changed underneath the dikes. These two aspects play an important role in dike safety assessment, particularly for flood wave durations. In this paper, dike stability is studied following a deterministic approach taking into account first the historical data to illustrate different stress states and water levels, and second different soil models. Input Data The selected location in this case study is 10km upstream of Hanoi Old quarter, namely Lien Mac (in Vietnamese). In 1915, there was a serious dike breach at this dike section caused by overflowingthe main failure mechanism (see Local Dike Board No.3, 1971-2010). Firstly, the hydraulic boundary condition is considered with the statistical data of over 100 years. After that, the maximum water level and flood duration are fitted to the theoretical distributions. The water levels in the calculations have been chosen following the historical data of the typical flood waves in the years 1893, 1915, 1945 and 1971. It could be seen from Figure 3 that the water level was maximum at 14.6m in 1971, and equalled a flood return period of 1/50 year. Additionally, flood mitigating measures have been

Figure 4: Typical dike strengthen strategies

Ground level

Ground level

Ground level

Ground level

Ground level

GL + 7.9m

Figure 5: Typical geotechnical profiles

carried out, such as planting trees in the upstream basin, constructing reservoirs for many multiple purposes, strengthening dike systems, and applying flood diversion. A rising level of around 14.9m*4, equivalent to a flood frequency of 1/500 year, will be considered for the near future (for more details, see Table 1). Secondly, dike geometry was collected from the local dike boards and historical documents. In the 19th century-empirical period-there were no drawings of dike cross sections due to the limitations of both scientific knowledge and poor archives of works. In the period dominated by the French, the dike dimensions were formulated based on historical documents and updated with research works (Phan, 1985; Gauthier, 1930; Monsieur and Dominique, 1991). The data including documents of dikes have been properly protected, so all typical sections relating to each dike program are quite clear and available now. The strategies of dike rehabilitation are described geometrically in Figure 4. Thirdly, from geotechnical investigations, the profiles of soil are illustrated with different typical combinations. Nguyen (1999) summarized the four possible cross sections for analysis in the project of Red River dike rehabilitation. Tran (2012) also recommended five types of ground conditions for the assessment of the Red River dikes. In fact, both the studies are nearly similar, although the latter described more clearly the influence of the silty fine sand layer to the dike stability. In this paper, the authors take in to account the 1st profile for

calculation; the 4th and the 5th types seem to be more dangerous for dike embankments, but these profiles are vanished from the study area and they will be discussed in another report (see Figure 5). Finally, soil properties were selected from the report of Geotechnical investigation, which were carried out in 1994 for the ADB project (HEC1, 1994). The dike embankment is represented by different compaction ratios which is assumed varying from K < 0.80 (in 1890s) to K~ 0.95 (currently). The soils of embankment have been remoulded in the laboratory with the mentioned compaction ratios, then the strength and the compression were determined by direct shear test and oedometer test. In fact, the soil parameters (of layers 2b, 7, 8 following HEC1 (1994)) could not be determined at the construction period, for instance during 1890s, 1915, 1945, etc. It is assumed that the changes of soil parameters are disregarded in this case study. Strength and stiffness of soils may be influenced by different stress states that resulted in actual loading progresses. In this case, both Mohr Coulomb (MC) and Hardening Soil (HS) models were applied, and parameters were selected following Table 2 and 3. Calculation Phases, Results, and Discussion Following the manual of Plaxis 2D-2011 (Brinkgreve, 2011) these problems could be analysed by the Plaxis finite element code which includes multiple choices for material models, construction stages, and calculation types, etc. However, it is hard to model stress states from

*3 Compaction ratio K = gdry/gmax where, gdry is gdry density of soil; gmax is the maximum density of soil, which is found by plotting a density - moisture curve in compaction test in laboratory (ASTM D698 & ASTM D1557). *4 Predicted results of flood frequency of 1/500 year, (Khoi, 2010)

16

Plaxis Bulletin l Spring issue 2013 l www.plaxis.nl

Stability Analysis of the Red River Dike: The Past to the Present

a hundred years ago; here we ignored the time interval in each calculation phase, and we opted for undrained behaviour as the drainage type. Time dependent conditions in each flooding season are considered by transient analysis. The calculation phases are illustrated below. Initial: Gravity loading, the initial conditions have been calculated and the classical mode is chosen for calculation; Phase 1: The 1st dike stage of construction follows the dike geometry of 1893; the new dike is built up and the ground water level is approximately set on the surface level because the earthwork had been carried out in dry seasons; Phase 2: The high water level in 1893, which is set at 10.5m and transient flow is modelled for the analysis with the flood duration of 7 days; Phase 3: The analysis of dike safety has been used to determine the factor of safety in the flood in 1893.

The procedures are repeated from phase 1 to 3 for each flood in 1915, 1945, and 1971 with water level and flood duration given in Table 1. . Figure 6 shows the results of the safety calculation phases, and the plastic zones enlarge depending on both the dike geometry and the flood water levels. The phreatic levels reach the critical situation in most cases, and the dike embankments were saturated. The level of dike crest has increased many times with different construction methods (Phan, 1985). In this paper, only four different geometry of dikes are selected because they were dealing with the typical flood waves and the major project of dike rehabilitation. Dike failure was observed twice in the year 1915 at the study location, comparing to eight cases of dike failure during the 20th century (Ha, 2010), and where the overflowing was the dominant mechanism of failure. Table 4 and Figure 7 show that we have the maximum of dike safety in 1893

with the level of dike crest of 10.5m, and the safety factor (Msf) minimizes at the predicted situation with a flood frequency of 1/500 year. However, in terms of flood management, the dike in the year 1893 might be safe at the water level frequency of 1/1 year (see Figure 3). It means that if there is a flood with a higher water level of 10.5m, dikes could failed by the overflowing and erosion of the inner slope, even if its safety factor reaches 1.54. In 1971, the dike could be able to withstand in the critical states, e.g., Msf ~ 1. It showed reasonable evidence for dike assessments around Hanoi at that time with some dike sections at the birth of collapse (see FPD, 2000). At the predicted flood level of 1/500 year, the water level is not raised freely because the Red River discharge is redistributed by the reservoir systems at the upstream rivers. However, the flood duration increases dramatically, from 5 days (in 1893) to 16 days (in the predicted flood of 1/500 year) and dikes are also at the risk of critical situations. If

Identification Drainage type gunsat gsat E n(nu) G Eoed cref f (phi) Flow model Soil type kx ky

Unit

Dikes1880 Undrained (A)

Clay1 Undrained (A) 18.5 19 4500 0.3 1731 6058 12 9 Van Genuchten Medium fine 0.014 0.014

Clay2 Undrained (A) 19.5 20 7000 0.28 2734 8949 20 12 Van Genuchten Fine 0.0003 0.0003

Sand Drained 19 19 12000 0.33 4511 17780 0.1 28 Van Genuchten Sand 1.27 1.27

Dikes1945 Undrained (A) 18 18.7 3800 0.32 1439 5438 12 10 Van Genuchten Silty clay loam 0.04 0.04

Dikes 2000 Undrained (A) 18.7 19.3 4500 0.27 1772 5623 18 12 Van Genuchten Silty clay loam 0.03 0.03

kN/m3 kN/m kN/m

3

17 17.5 3000 0.33

kN/m kN/m

1128 4445 9 7 Van Genuchten Silty clay loam

kN/m2 deg

m/day m/day

0.05 0.05

Table 2: Soil parameters for Mohr - Coulomb model

Identification Drainage type gunsat gsat Eref50 Erefoed E

ref ur

Unit

Dikes1880 Undrained (A)

Clay1 Undrained (A) 18.5 19 3150 5500 16500 0.5 12 9 Van Genuchten 0.014 0.014

Clay2 Undrained (A) 19.5 20 5500 7500 23500 0.5 20 12 Van Genuchten 0.0003 0.0003

Sand Drained 19 19 7500 7800 20000 0.5 0.1 28 Van Genuchten 1.27 1.27

Dikes1945 Undrained 18 18.7 4400 4300 13000 0.5 12 10 Van Genuchten 0.04 0.04

Dikes 2000 Undrained (A) 18.7 19.3 5500 6000 17000 0.5 15 12 Van Genuchten 0.03 0.03

kN/m3 kN/m
3

17 17.5 3000 3400 10000 0.5

kN/m2 kN/m2 kN/m

2

power (m) cref f(phi) Flow model kx ky m/day m/day kN/m deg
2

9 7 Van Genuchten 0.05 0.05

Table 3: Soil parameters for Hardening Soil model Notes: The data used in tables 2 & 3 are adapted from HEC1 [1994]

www.plaxis.nl l Spring issue 2013 l Plaxis Bulletin

17

Stability Analysis of the Red River Dike: The Past to the Present

Figure 6: Stability of dikes vs construction stages

Figure 7: Msf vs Material Models and flood frequency

Figure 8: Msf vs steps calculation of MC model

Msf Year MC 1893 1915 1945 1971 1/500 1.538 1.282 1.32 1.032 1.031 HS 1.549 1.226 1.279 0.987 0.989

Flood wave Frequency (%) 100 91 20 2.2 0.2 Return period (1/year) 1.0 1.1 5.0 45.5 500.0

Table 4: Summary of calculation results

18

Plaxis Bulletin l Spring issue 2013 l www.plaxis.nl

Stability Analysis of the Red River Dike: The Past to the Present

other failure mechanisms presented, the dike could have failed. Comparing the material models of MC and HS, we can see from Figure 7 that the former has a higher value of Msf than that of the latter. A possible explanation is that the HS model mobilize soils strength lower than that of the MC model (Brinkgreve, 2009); in this case, HS model is preferable to application. Conclusions An adequate assessment of dike safety is a complex issue that should include not only the instability but also piping and other failure mechanisms. In stability analysis, PLAXIS 2D 2011 demonstrated successfully the dike safety factor in flood waves with different construction stages. A dike location in Red River in Hanoi with historical data of dike failure is selected and the calculation results demonstrated reasonably comparing to the past observations, and dikes stability has been significantly improved from the past until now. The dikes stability was calculated in critical states of both the flood in 1971 (with return period of 1/50 year) and the flood frequency of 1/500 year and was found to be unsafe. In contrast, it was estimated to be safe in the flood with a return period of 1/1 year in 1893, with regard to stability only. Further research should be carried out to investigate the influences of the heterogeneity of dike embankment and subsoil data, and other related issues such partially saturated soil behaviour and advanced soil models. Acknowledgements This paper is funded by the project of Upgrading the Training Capacity in Coastal Engineering of the Water Resources University of Vietnam from the Royal Netherlands Embassy in Hanoi of Vietnam, and CICAT and Hydraulic Engineering Department of TU Delft; the authors wish to thank for their support.

References Hanoi DDMFC. Report on current condition of river dike systems in Hanoi area before flood season 2009. Technical report, 2009. General Department of Meteorology GDM. History of Vietnam Meteorology, Vol 1, Ministry of Natural Resources and Environment of Vietnam. Hanoi Meteorology publishing, 1995. Phan, K. (1985). A brief history of water management in Vietnam (in Vietnamese). Hanoi: Hanoi Social Science Publishing. Ha Van Khoi. Study the scientific background for un-flooded some lower area in the red river basin and Hoanglong river basin. Technical report, 2010. Hanoi Local Dike Board No3. General background of red river dike systems in Hanoi area, board no3. Technical report, 1971. J Gauthier. Digues du Tonkin (Hanoi, Imprimerie Dextreme-orient). M6409 National library of Vietnam, 1930. M.D Hemery. Histoire du euve rouge, gestion et amenagement dun systeme hydraulique au tonkin des annees 1890 jusqua la seconde guerre mondiale, tome 1 History of the Red River, management and Layouts of a system of hydraulic tonkin 1890s until the Second World War, Volume 1. Universite de Paris 7, 1991. Nguyen, C. M. Some geotechnical engineering aspects of dike in red river delta, calculation revitalization. In Proceeding of Geology engineering and geo-environment in Vietnam, 1999. Tran, V. T. Study on the geological conditions of the red river dikes in Hanoi area. Technical report, 2012. HEC1. Geology investigation reports, Hanoi dikes improvement project. Technical report, 1994. R.B.J Brinkgreve. Plaxis 2D-2011 manual, vols 1:5. Plaxis BV, 2011. Hanoi. FPD. Hanoi - 50 years with natural hazard protection work, Flood Protection Department of Hanoi. Hanoi publishing, 2000. R.B.J Brinkgreve. Material models for rock and soil - Lecture note CT4360. Technical University of Delft, 2009.

www.plaxis.nl l Spring issue 2013 l Plaxis Bulletin

19

3D Modelling of Frozen Ground and Associated Volume Expansion

Joseph Sopko, Ground Freezing Services, Moretrench

Plaxis was contracted by Moretrench to provide assistance in setting-up 3D finite element models for the analysis of how ground freezing process and associated volume expansion would impact existing slurry wall panels. Thanks to PLAXIS Expert Services, valuable results have been obtained in terms of displacement and integrity assessment for the existing slurry walls.
What is ground freezing? The ground freezing process converts pore water into ice by the continuous circulation of a cryogenic fluid within a system of small diameter, closed-end pipes installed in a pattern consistent with the shape of the area to be treated. The frozen pore water acts as a bonding agent, fusing together particles of soil or rock to significantly increase compressive strength and impart impermeability. heat from the surrounding soil. When the soil temperature reaches 32F (0C), ice begins to form around the pipes in the shape of vertical, elliptical cylinders. As the cylinders gradually enlarge with time, they interconnect to form a continuous wall. Once the frozen wall reaches design thickness, the freeze plant is typically operated at a reduced rate to maintain the condition. ground and propagates thermally, rather than by displacement. The ground remains largely undisturbed during freeze pipe installation and the freezing process and, in most instances, reverts to its original state once freezing is discontinued. Ground freezing provides both excavation support and groundwater control. While the use of ground freezing for the construction of deep shafts through water-bearing soils is the most common application, the technique is also appropriate for mining of cross passages and connector tunnels, stabilization of mixed-face tunnels, frozen-arch canopies, and containment of contamination. Finite element modelling The dimension of the FE model is 100m*50m*60m. Only half of the geometry has been modelled due to symmetry conditions. The model contains a total of 64,000 elements and 90,000 nodes (i.e

PLAXIS professionals assisted us tremendously in moving forward our analytical tools and ability to understand the mechanics of our project...
Freeze pipes are typically installed vertically in the soil and connected in series-parallel. The coolant is pumped down a drop tube to the bottom of the freeze pipe and flows up the annulus, withdrawing Unlike other groundwater cut-off and excavation support techniques, ground freezing is a minimally invasive technique that requires only limited physical penetration into the

FE mesh

Calculated displacement after ground frozen

20

Plaxis Bulletin l Spring issue 2013 l www.plaxis.nl

270,000 dof). Each finite element has 10 nodes. The slurry wall panels have been defined around a 11m diameter shaft. The frozen pipes have been placed along a series of concentrated circles. In PLAXIS the frozen process is modelled in sequence by applying an isotropic volumetric prescribed strain of 1.5 % (so 0.5% in X-direction, 0.5% in Y-direction, 0.5% in Z-direction ) followed by replacing the properties of the swollen clusters by those of frozen soil after each expansion phase.

The model consists of the following construction sequence: Vacancy: Advisor PLAXIS Expert Services Initial stress definition and initial water level Slurry wall construction Excavation Ground freezing part I at diameter D1 Ground freezing part II at diameter D2>D1 Ground freezing part III at diameter D3>D2 Ground freezing part IV at diameter D4>D3 Plaxis bv is currently looking for an advisor to strengthen our expert services team. The advisor is expected to plan, execute and report projects and results on his/her own. He/she also acts as project leader and maintains contacts with clients, universities and other external parties. For more information please visit: www.plaxis.nl/jobs

The implementation of PLAXIS to applications of artificial ground freezing has significantly advanced the analysis of complex problems...

PLAXIS Expert Services added value: Quick start on the job Set-up of fully optimized and ready to run 3D FE models Regular model review Next business-day advanced technical assistance Customer quotes The implementation of PLAXIS to applications of artificial ground freezing has significantly advanced the analysis of complex problems, particularly the mechanics associated with ground expansion as well as the decrease in frozen soil strength with time under load. Applying PLAXIS to this problems was a major undertaking for our engineers. PLAXIS professionals assisted us tremendously in moving forward our analytical tools and ability to understand the mechanics of our project as well as being able to present the results to our client with incredible output graphics.

The PLAXIS consultants eagerly accepted the challenge, fully understood our concerns and rapidly and efficiently assisted us in generating the models that accurately simulated the complex issues with stress and deformation on existing structures. Their assistance has enabled us to make PLAXIS a readily available tool to analyse a wide arrange of ground freezing challenges. Their knowledge, accuracy and efficiency are unsurpassed in this industry. About Moretrench Moretrench is geotechnical contractor delivering geotechnical solutions to the underground, industrial and environmental remediation industries. Moretrench offers a wide range of specialized services with skilled and experienced engineering teams in the field of dewatering and groundwater control, ground freezing, earth retention and anchors, deep foundations, underpinning, grouting and ground improvement, environmental remediation.

Stress development in the slurry panel

www.plaxis.nl l Spring issue 2013 l Plaxis Bulletin

21

Recent Activities

Seminar Offshore Foundations Houston On January 28, 2013 Plaxis organised a short seminar in Houston on the use of PLAXIS 3D for offshore foundations in geotechnical engineering. PLAXIS 3D is rapidly gaining popularity in modelling subsea behaviour of soil and structures in the US and around the globe. This free event was well attended by professionals coming from major oil companies, engineering consulting firms, and universities. A selection of offshore relevant features of PLAXIS 3D where presented. On the subject of soil

material modelling, special attention was given to the total stress approach, anisotropic undrained shear strength, and the NGI-ADP model in PLAXIS 2D and PLAXIS 3D. Also discussed were the importance of load vs. displacement controlled analysis, side-swipe technique, and the use of interface elements in PLAXIS for modelling skin friction, gapping and slipping. Additionally consideration was given to computational efficiency by optimising the calculation mesh, utilising a multicore solver, and using the new option called Gradual Error Reduction in PLAXIS 3D 2012.

The seminar was concluded with a live demonstration of building and analysing an offshore foundation in PLAXIS 3D, this clearly illustrated the ease and speed of work in PLAXIS 3D. A description of this case study on a combined loading of a suction anchor can be found in our online Knowledge Base. A similar seminar was also organised in Singapore end of 2012, and due to the popularity of both events plans for others around the world are already underway, with the first coming up in Australia later this year. So keep an eye out for this and more upcoming events.

22

Plaxis Bulletin l Spring issue 2013 l www.plaxis.nl

New Head Office in Delft The construction of our new head office is now finalised, so as of the 2nd of January we have moved to our new location. Our new visiting address will be: Computerlaan 14 2628 XK Delft The Netherlands Further contact details remain unchanged. Our new office offers many new facilities, including space for hosting PLAXIS Academy trainings. For 2013 we have planned to host a series of workshops. Each month a different subject will be addressed for either PLAXIS 2D or PLAXIS 3D, for instance modelling tunnels, dynamics or groundwater. Whereas computational geotechnics courses focus on the background and application of the Finite Element Method in geotechnical engineering, PLAXIS trainings focus directly on learning and using the features of the PLAXIS software. Hence, the main goal of these workshops is to teach the use and basic concepts of different features in the PLAXIS software and to train the participants in using the software themselves. Besides that, the workshops will, to an extent, be touching upon the new functionality and improved workflow of PLAXIS software. Furthermore, the participants will get acquainted with the proper use of PLAXIS 2D and/or PLAXIS 3D by doing hands-on exercises. During these exercises experienced Plaxis tutors will provide individual assistance. Visit our site at www.plaxis.nl/events for a full list of the workshops and topics included. We wish you all a prosperous 2013, and would like to welcome you to our new office!

Product update: PLAXIS 2D PLAXIS 2D 2012 is now available! The new software is equipped with, amongst others, the new embedded pile row facility. With the success of the embedded pile element in PLAXIS 3D, users started to ask for a similar facility in PLAXIS 2D. However, the stress state and deformation pattern around piles is fully three-dimensional, so at first sight it seemed very difficult to develop an element in 2D that can realistically model piles. Nevertheless, we have succeeded to do so.

Furthermore the new software has many new and improved features such as; Structural forces in volumes (When soil is used to model structures it is now possible to directly visualize the structural forces in the Output program), Inverse analysis of material parameters (parameter optimisation), and Implementation of free field boundaries to enhance the domain of the dynamics calculations (Dynamics module, available upon request for PLAXIS VIP).

Scematic representation of embedded pile

www.plaxis.nl l Spring issue 2013 l Plaxis Bulletin

23

Title

Activities 2013
March 3 6, 2013 Geo-Congress 2013 San Diego CA, U.S.A. March 4, 2013 Workshop on the use of PLAXIS 2D for Earthquake Geotechnical Analysis Milan, Italy March 7, 2013 Grundlggningsdagen 2013 Stockholm, Sweden March 11 14, 2013 Advanced Course on Computational Geotechnics Schiphol, The Netherlands March 18, 2013 Belgium Plaxis Users Meeting Brussels, Belgium March 26 - 27, 2013 Training on PLAXIS 2D for Earthquake Geotechnical Analysis Istanbul, Turkey March 27 - 28, 2013 Training on the use of PLAXIS 2D Kuala Lumpur, Malaysia April 3, 2013 Workshop: Introduction to PLAXIS 3D Delft, The Netherlands April 15, 2013 Workshop: Utilisation de PLAXIS 3D en gotechnique Paris, France April 17 - 19,2013 EURO:TUN 2013 Bochum, Germany April 23, 2013 Workshop on Soil Parameter Optimisation and Sensitivity in PLAXIS Delft, The Netherlands May 1 - 4, 2013 Seventh International Conference on Case Histories in Geotechnical Engineering Chicago IL, U.S.A. May 15, 2013 French Plaxis Users Meeting Paris, France May 16, 2013 Workshop: Utilisation de PLAXIS 2D dans lingnierie gotechnique sismique Paris, France May 21 24, 2013 Advanced Course on Computational Geotechnics Berkeley CA, U.S.A. June 4, 2013 Workshop on Modelling Dynamics in PLAXIS Delft, The Netherlands June 17 20, 2013 Standard Course on Computational Geotechnics Manchester, United Kingdom June 23 - 26, 2013 Rapid Excavation and Tunneling Conference 2013 Washington DC, U.S.A. June 25 - 26, 2013 Russian Plaxis Users Meeting St. Petersburg, Russia July 3, 2013 Workshop: Introduction to PLAXIS 2D Delft, The Netherlands July 18, 2013 UK Plaxis Users Meeting London, United Kingdom August 12 15, 2013 Standard Course on Computational Geotechnics Gothenburg, Sweden September 2 - 6, 2013 International Conference on Soil Mechanics and Geotechnical Engineering Paris, France September 2 - 4, 2013 Standard Course on Computational Geotechnics Gdansk, Poland September 4, 2013 Workshop on Modelling Tunnels in PLAXIS 3D Delft, The Netherlands September 10 13, 2013 Advanced Course on Computational Geotechnics Manchester, United Kingdom September 12, 2013 Workshop: Modlisation de leau dans les sols sous Paris, France September 18 20, 2013 Advanced Course on Computational Geotechnics Chennai, India September 29 October 03, 2013 GeoMontral 2013 Montral, Canada October 2, 2013 Workshop: Introduction to PLAXIS 3D Delft, The Netherlands November 6, 2013 Workshop on Soil Parameter Optimisation and Sensitivity in PLAXIS Delft, The Netherlands November 6 - 8, 2013 European Plaxis Users Meeting Karlsruhe, Germany December 04, 2013 Workshop on Modelling Groundwater in PLAXIS 3D Delft, The Netherlands

Plaxis bv Computerlaan 14 2628 XK Delft

P.O. Box 572 2600 AN Delft The Netherlands

www.plaxis.nl Tel +31 (0)15 2517 720 Fax +31 (0)15 2573 107

Plaxis Asia Singapore Tel +65 6325 4191

16 Jalan Kilang Timor #05-08 Redhill Forum 159308 Singapore