Three-dimensional effects in slope stability for shallow excavations Analyses with the finite element program PLAXIS
Niclas Lindberg
Master of Science Program in civil engineering Luleå University of Technology Department of Civil, Environmental and Natural resources engineering
PREFACE
This master thesis is the final part of my five year education in civil engineering at Luleå university of technology. The investigation has been done on behalf of Luleå university with inspiration from Trafikverket.
I would like to thank my supervisor Hans Mattsson from Luleå University of technology for all the guidance and inspiration during both the courses and the time doing this investigation. I also want to thank for the possibility to work and learn more about numerical modelling. A special thanks to Per Gunnvard at Luleå university for all the guidance and patience during the time of making this investigation.
Finally, thanks to all my friends and family during the years at Luleå university of technology.
Luleå, Mars 2018 Niclas Lindberg
i ii ABSTRACT
The purpose with this study was to investigate the impact of three-dimensional effects in slope stability for three-dimensional excavations and slopes with cohesive soils and compare the results with the method provided by the Swedish commission of slope stability in 1995 regarding three-dimensional effects. Both the factor of safety and the shape of the slip surface was compared between the methods but also the results from their equivalent two- dimensional geometry.
The investigation was performed with models created in the finite element software PLAXIS 3D and the limit equilibrium software GeoStudio SLOPE/W. Three-dimensional excavations with varying slope angles, external loads and slope lengths were tested for three different geometry groups in PLAXIS 3D. The equivalent two-dimensional geometries were modeled with SLOPE/W and recalculated with the three-dimensional effect method provided from the Swedish commission of slope stability.
The results show that the methods match well for slopes with inclinations 1:2 and 1:1 when an external load is present on the slope edge, and the factor of safety is greater and not close to 1,0. For an excavation with vertical walls or when no external load is present, the methods match poorly. The results also show that for a long and unloaded slope, the factor of safety approaches the value obtained from a simplified two-dimensional analysis.
The results imply that the recommendations from the Swedish commission of slope stability are reliable for simple calculations of standard cohesive slopes.
Keywords: Slope stability; 3D-effects; FEM
iii iv SAMMANFATTNING
Syftet med detta examensarbete var att undersöka tredimensionella effekters inverkan vid släntstabilitet hos tredimensionella schakter och slänter av kohesiva jordar och jämföra resultatet med den metod som svenska skredkommissionen rekommenderat år 1995 gällande tredimensionella-effekter. Både säkerhetsfaktorn och formen hos den utbildade glidytan jämfördes mellan metoderna samt resultatet från dess ekvivalenta tvådimensionella geometri.
Undersökningen utfördes med hjälp av modellering i det finita elementprogrammet PLAXIS 3D och gränslastanalysprogrammet GeoStudio SLOPE/W. Tredimensionella schakter med varierande släntlutningar, externa laster och släntlängder testades hos tre olika geometrigrupper i PLAXIS 3D. De ekvivalenta tvådimensionella geometrierna modellerades i SLOPE/W och räknades sedan om tredimensionellt enligt den metod som svenska skredkommissionen rekommenderat.
Resultatet visar att metoderna överensstämmer väl för schakter med släntlutningen 1:2 och 1:1 där en extern last finns närvarande på släntkrönet och säkerhetsfaktorn är större än och inte nära 1,0. För schakter med vertikala schaktväggar eller schakter där ingen extern last närvarar överensstämmer metoderna inte väl. Resultatet visar också att en långsträckt obelastad slänt har en säkerhetsfaktor som stämmer väl överens med en simplifierad tvådimensionell analys.
Resultatet föreslår att rekommendationerna från svenska skredkommissionen är tillförlitliga för enklare beräkningar av normala släntstabilitetsproblem i kohesiva jordar.
Nyckelord: Släntstabilitet; 3D-effekter; FEM
v vi TABLE OF CONTENTS
1. INTRODUCTION ...... 1
1.1 Background ...... 1 1.2 Purpose and objective ...... 2 1.3 Limitations ...... 3 2. THEORETICAL BACKGROUND ...... 5
2.1 Finite element method ...... 5 2.2 Limit equilibrium method ...... 10 2.3 Three-dimensional slope stability ...... 13 3. SOFTWARE ...... 19
3.1 PLAXIS ...... 19 3.2 GeoStudio, SLOPE/W ...... 23 4. NUMERICAL MODELLING WITH PLAXIS 3D...... 25
4.1 Geometries and cases ...... 25 4.2 Model specification ...... 27 4.3 Phases ...... 29 4.4 Material parameters ...... 29 4.5 Mesh and boundaries ...... 30 5. SLOPE/W MODELS AND 3D-EFFECT CALCULATIONS ...... 31
5.1 SLOPE/W models ...... 31 5.2 Calculation of three-dimensional effects ...... 32 6. RESULTS AND ANALYSIS...... 33
6.1 Two-dimensional comparison ...... 35 6.2 Three-dimensional analysis ...... 36 6.3 3D-effects method compared with PLAXIS 3D models ...... 39 6.4 Safety analysis in PLAXIS ...... 43 7. DISCUSSION ...... 45
7.1 Suggestions on further studies ...... 46 REFERENCES ...... 47
APPENDIX A1 –SFR/DEFORMATION-CURVES, MODEL A ...... 51 APPENDIX A2 –SFR/DEFORMATION-CURVES, MODEL B ...... 55
vii APPENDIX A3 –SFR/DEFORMATION-CURVES, MODEL C ...... 58 APPENDIX B1 – TOTAL DEFORMATIONS, MODEL A ...... 59 APPENDIX B2 – TOTAL DEFORMATIONS, MODEL B ...... 67 APPENDIX B3 – TOTAL DEFORMATIONS, MODEL C ...... 74 APPENDIX C1 –SLOPE/W SLIP SURFACE GROUP A ...... 78 APPENDIX C2 –SLOPE/W SLIP SURFACE GROUP B ...... 79 APPENDIX C3 –SLOPE/W SLIP SURFACE GROUP C ...... 80
viii 1 INTRODUCTION
1.1 Background
Slope stability analysis is a branch of geotechnical engineering and concerns the stability for both natural and constructed soil slopes. Many construction projects require excavations in soil to construct pipes and cables, and also foundations for buildings and bridges. Roads and railways are usually built with soil embankments, where stability analysis must be performed to satisfy the safety regulations. Several methods have been developed throughout history to calculate the stability of slopes, which results in a factor of safety. The factor is defined as the ratio of the shear strength available of the soil compared to the necessary strength to maintain equilibrium (Bishop, 1955). The most common approach to calculate the stability of a slope is with the limit equilibrium method (LEM), with an assumption of plane-strain conditions. (Zhang, Guangqi, Zheng, Li, & Zhuang, 2013) With the software and computer power available today, this method can obtain multiple failure surfaces with their factors of safety in a very short time. The finite element method (FEM) is also a popular technique which is a numerical method that discretizes a problem into elements and solves partial differential equations numerically.
To establish safe and economical solutions for slopes and excavation stability, high accuracy in the calculation models are required, even though several assumptions to simplify the real- life problems are inevitable in most engineering models. A very common assumption in slope stability analysis is the plane-strain condition, which is a simplifying assumption that means that the value of the strain component perpendicular to the plane of interest is equal to zero. This is usually valid when structures are very long in one dimension in comparison to the others. With this assumption, no curvatures, corners or change of geometry in one dimension can be accounted for at all. This means that the failure surface must exist perpendicular to the plane of interest. (Zhang, Guangqi, Zheng, Li, & Zhuang, 2013) In the past, this assumption was almost necessary because of the limitations in the three-dimensional slope stability analysis methods. Location, shape and direction of the slip surface are usually unknown, which makes the problem very complex to solve. (Cheng, Liu, Wei, & Au, 2005)
A three-dimensional slope stability analysis is important because the factor of safety is known to be higher than in a two-dimensional analysis. This means that a more economic design is possible. (Cheng, Liu, Wei, & Au, 2005) The Swedish commission of slope stability proposed
1 a calculation method for the treatment of three-dimensional effects (Skredkommissionen 3:95, 1995). The calculations are valid for simple cases with an idealized slip surface in cohesive soils and is based on the limit equilibrium approach. The treatment of three-dimensional effects in slopes made of frictional soils were described as unknown. The Swedish transport administration often perform shallow excavations for smaller construction works. If three- dimensional effects are considered for slope stability, the total excavation cost could be smaller. They are currently using the recommendations from 1995 but is interested in a modern FE comparison to confirm or add knowledge to the subject.
1.2 Purpose and objective
All real-life slopes are three-dimensional problems, and because the factor of safety is known to be higher for slope calculations where three-dimensional effects are accounted for, more knowledge is required to make more economic designs that are still considered safe. The purpose of this master thesis is therefore to compare the difference in factor of safety and slip surface shape between three-dimensional and two-dimensional slope stability analysis for shallow excavations using the commercial software PLAXIS 3D and SLOPE/W. The SLOPE/W results will then be extended to three-dimensional surfaces using the 3D-effects method described by the Swedish commission of slope stability. Different slope geometries and soil materials will be tested in the models along with external loads and excavations. The objective is to compare the factor of safety and slip surface shapes for shallow excavations obtained from the three-dimensional-effects method (3D-effects) and PLAXIS 3D models.
The following questions will be investigated:
· What difference in factor of safety is obtained from a stability analysis of a three- dimensional excavation problem modeled in three-dimensions than in its two- dimensional equivalence? · Does the 3D-effects calculation method recommended by the Swedish commission of slope stability match the results obtained from finite element 3D calculations?
2 1.3 Limitations
This report is limited to only study undrained analyzes for one type of cohesive soil material. The varied parameters are the slope angles, the excavation lengths and the magnitude of the external load. The groundwater level, excavation depth and load area are kept the same in all models.
3 4 2 THEORETICAL BACKGROUND
2.1 Finite element method
Almost every physical phenomenon can be described mathematically using differential equations. The problem with differential equations is that most of them are very hard or even impossible to solve with analytical methods. The alternative way of solving them is with numerical methods which gives approximate solutions, e.g. the finite element method. The main feature with the finite element method is that the body or region is discretized into smaller elements, on which the differential equations describe its behavior. Every element has nodal points where the variables are assumed to be known. The nodal points are usually located at the element boundaries. The elements are attached together to form an element mesh, which can be seen in Figure 1. The more elements a body is discretized into, the more nodal points it has. With more nodal points comes more unknowns, which in general produce a solution with higher accuracy. The method is then carried out by solving for the unknowns in the nodal points, and shape functions describe the behavior of the element in between nodes. (Ottosen & Petersson, 1992)
Figure 1 Element mesh of a soil body in PLAXIS 3D The finite element method was primarily developed to solve structural problems, but can be used for several types of problems, such as heat transfer, electromagnetic problems and groundwater flow, and is applicable to problems in one, two or three dimensions. (Ottosen & Petersson, 1992).
5 2.1.1 The mathematical background to the finite element method
For structural analysis, the differential equations describing the element behavior must be derived from equilibrium equations. These equations can be obtained from an infinitesimal stress cube. On the surface of the cube that can be seen in Figure 2, stress components are acting, where indicates the plane on which the stress acts, and the direction of the stress.
Figure 2 An infinitesimal stress cube A three-dimensional body have 9 stress components, where only 6 of them are independent because of moment equilibrium. If there is force equilibrium in every cartesian direction, the following three equilibrium equations can be obtained and expressed as:
+ + + =0 (2.1) + + + =0 (2.2) + + + =0 (2.3) In equation (1) – (3), b denotes the body forces in the three cartesian directions. These equations written in matrix form, together with the body force vector gives the expression:
00 0 0 0 0 + =0 (2.4) ⎡ ⎤ ⎡ ⎤ ⎢ ⎥ ⎢ 0 ⎥ 00 ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎣ ⎦ This is called the strong formulation, which also can⎣ be written:⎦
+ =0 (2.5) ∇ 6 The operator is written in the transposed form because of its use in the non-transposed form later. The finite element approach is based on the weak formulation of the differential ∇ equations, which requires some mathematical manipulations from the strong formulation. The equilibrium equations must be multiplied with an arbitrary weight function, in this case a three-dimensional weight vector:
= (2.6) The stresses acting on the surface of the body acts as boundary conditions, and can be expressed with a traction vector:
= (2.7) The cartesian components of must fulfill the boundary conditions: