On the effects of unloading and reloading of excavations in clay and silt A Case Study of the Varberg Tunnel Project Master’s thesis in Infrastructure and Environmental Engineering ALICE HULTIN AXEL RECKMAN DEPARTMENT OF ARCHITECTURE AND CIVIL ENGINEERING CHALMERS UNIVERSITY OF TECHNOLOGY Gothenburg, Sweden 2024 www.chalmers.se www.chalmers.se Master’s thesis 2024 On the effects of unloading and reloading of excavations in clay and silt A case study of the Varberg Tunnel Project ALICE HULTIN AXEL RECKMAN Department of Architecture and Civil Engineering Division of Geology and Geotechnics Geotechnics Research group Chalmers University of Technology Gothenburg, Sweden 2024 On the effects of unloading and reloading of excavations in clay and silt A case study of the Varberg Tunnel project ALICE HULTIN AXEL RECKMAN © ALICE HULTIN, 2024. © AXEL RECKMAN, 2024. Supervisor: Mats Karlsson, Department of Architecture and Civil Engineering Supervisor: Daniel Baltrock, Implenia Sverige AB Examiner: Jelke Dijkstra, Department of Architecture and Civil Engineering Master’s Thesis 2024 Department of Architecture and Civil Engineering Division of Geology and Geotechnics Geotechnics Research group Chalmers University of Technology SE-412 96 Gothenburg Telephone +46 31 772 1000 Cover: A visualization showing the ground profile and geotechnical structure on the site. Typeset in LATEX, template by Kyriaki Antoniadou-Plytaria Printed by Chalmers Reproservice Gothenburg, Sweden 2024 iv On the effects of unloading and reloading of excavations in clay and silt A Case Study from the Varberg Tunnel Project ALICE HULTIN AXEL RECKMAN Department of Architecture and Civil Engineering Chalmers University of Technology Abstract In connection with pipe laying in an excavated trench supported by steel sheet piles in clay with underlying layered silty soil, large settlements were measured on a newly laid pipe in a project in Varberg, Sweden. This work aims to investigate possible causes of the settlement by numerical modelling of the geotechnical design. The work also aims to draw conclusions that can help in future excavation work in similar soil profiles. The project initially identified relevant sounding points with in-situ geotechnical tests and laboratory tests. A soil profile was developed from the empirical interpretation of CPTu logs using soil behaviour type (SBT) classification systems. Soil parameters were determined from field and laboratory tests. A numer- ical modelling was performed of the unloading and reloading effects of a supported excavation. The numerical analysis shows that the excavation bottom heaves during unloading and mechanical swelling. During reloading and pipe laying, there is an immediate settlement and a consolidation settlement. The settlement of the excava- tion bottom (and the pipe) during reloading in the numerical model may be due to an elastic rebound effect of the bottom under the pre-consolidation pressure in the soil. A difference arises between the total heave and total settlement corresponding to the magnitude of the mechanical swelling. The work also identifies other potential causes of the actual settlement of the excavation bottom (and the pipe) but which were omitted in the numerical modelling. These factors are briefly addressed and include hydraulic uplift, water flows along the soil-sheet pile interface due to pene- tration of permeable water-bearing subsoil layers, installation and removal effects of sheet piles, and the influence on the strength properties of the soil from vibrations during sheet pile driving in layered silty soils and compaction works in the shafts. Keywords: excavation, unloading effects, reloading effects, layered soil, vertical dis- placement, heave, mechanical swelling, settlement, embedded retaining structures, sheet piling. Om effekterna av avlastning och återbelastning av schakter i lera och silt En fallstudie från projekt Varbergstunneln ALICE HULTIN AXEL RECKMAN Institutionen för arkitektur och samhällsbyggnadsteknik Chalmers tekniska högskola Sammanfattning I samband med rörläggning i en schaktad ledningsgrav inom stålspont i lera med underliggande skiktad siltig jord uppmättes stora sättningar på en nylagd ledning i ett projekt i Varberg, Sverige. Detta arbetet syftar till utreda möjliga orsaker till sättningen genom numerisk modellering av den geotekniska konstruktionen. Ar- betet strävar också efter att dra slutsatser som kan vara behjälpliga i framtida schaktarbeten i liknande jordprofiler. Projektet identifierade inledningsvis relevanta sonderingspunkter med utförda in-situ geotekniska tester samt laboratorietester. En markprofil utvecklades från empirisk tolkning av CPTu-loggar genom diagram för jordklassificering. Jordparmeterar bestämdes utifrån fält- och laboratorieförsök. En numeriska modellering utfördes av av- och återbelastningseffekter av en spontad schakt. Den numeriska analysen visar att schaktbotten häver sig vid avlastning samt på mekanisk svällning. Vid återbelastning och ledningsläggning sker en omedelbar sättning samt en konsolideringssättning. Sättningen av schaktbotten (och lednin- gen) vid återbelastning i den numeriska modellen kan bero på en elastisk återfjädring av botten under förkonsolideringstrycket i jorden. En differens uppstår mellan den totala hävningen och totala sättningen som motsvarar storleken på den mekaniska svällningen. I arbetet identifieras också andra potentiella orsaker till den verkliga sättningen av schaktbotten (och ledningen) men som utelämnades i den numeriska modelleringen. Dessa faktorer behandlas kortfattat och innefattar hydraulisk bot- tenupptryckning, vattenflöden längs kontaktytan mellan jord och spont på grund av penetration av permeabla vattenförande underliggande skikt, installations- och dragningseffekter av spontar och påverkan på jordens hållfasthetsegenskaper från vibrationer under spontdrivning i skiktad siltig jord samt packningsarbeten i schak- ten. Nyckelord: schaktning, avlastningseffekter, återbelastningseffekter, skiktad jord, vertikala rörelser, hävning, mekanisk svällning, sättning, nedsänkt stödkonstruk- tion, spontning. Acknowledgements This master thesis was carried out during the spring of 2024 as part of the master’s programme in Infrastructure and Environmental Engineering at Chalmers University of Technology, in collaboration with Implenia Sverige AB and the Varberg Tunnel project. We would like to thank Daniel Baltrock at Implenia Sverige AB for proposing this thesis topic, providing supervision, and sharing valuable project insights. We also thank the project team for their assistance in gathering data. At Chalmers University of Technology, we thank our supervisor, Mats Karlsson, for his excellent feedback and sharing of insights from engineering practice. We also thank Jelke Dijkstra for accepting the responsibility of being our examiner and for his valuable feedback. We extend our heartfelt thanks to our families and friends for their support during the work. Alice Hultin, Gothenburg, June 2024 Axel Reckman, Gothenburg, June 2024 vii List of Acronyms Below is the list of acronyms that have been used throughout this thesis listed in alphabetical order: B Billion CAL Calibrated CAUC Consolidated Anisotropic Undrained Compression CPT Cone Penetration Test CPTu Cone Penetration Test with Piezocone CRS Constant Rate of Strain DSS Direct Simple Shear ECI Early Contractor Involvement FCT Fall Cone Test FEM Finite Element Method FVT Field Vane Test IL Incremental Loading M Million MC Mohr Coulomb MUR Soil Survey Report (Markteknisk undersökningsrapport) NC Normally Consolidated OC Over Consolidated OCR Over Consolidation Ratio POP Pre-overburden Pressure PS Parametric Study SBT Soil Behaviour Type SGU Geological Survey of Sweden (Sveriges geologiska undersökning) SQD Specimen Quality Designation SS Soft Soil SSC Soft Soil Creep ix x Nomenclature Below is the nomenclature of indices, sets, parameters, and variables that have been used throughout this thesis. Latin letters A Area as Swelling index Bq Normalized Pore Pressure Ratio b Load factor (swelling) c′ Effective cohesion cu Undrained shear strength cv Consolidation coefficient (vertical) Cαe Secondary consolidation index (void ratio) e Void ratio f ′ c Yield surface fs Sleeve resistance Fr Normalized Friction Ratio g Gravity H Height i Hydraulic gradient Ir Soil behavior type index ki Permeability K0 Lateral earth pressure ratio at rest KNC 0 Lateral earth pressure at rest in the NC-region M0 Constant constrained modulus below the effective vertical precon- solidation pressure, Swedish method Mc Stress ratio at critical state in triaxial compression xi Me Stress ratio at critical state in triaxial extension ML Constant constrained modulus between the stresses σ′ c and σL, Swedish method Mul Unloading modulus M ′ Modulus number mv Volumetric compressibility p′ Mean effective stress p′ 0 Magnitude of the yield surface q Deviatoric stress qc Tip resistance qn Net cone resistance qt Corrected Cone Resistance Qt Normalized Cone Resistance St Sensitivity Su Undrained shear strength t Time u Pore water pressure u0 Equilibrium pore pressure based on water table depth u2 Measured pore pressure vur Poisson’s ratio for unloading-reloading wL Liquid limit wn Natural water content Greek letters ϵ1 Axial strain σ′ c Apparent preconsolidation pressure γ Unit weight ρ Density σL Stress limit σvo Total Overburden Stress σ′ vo Effective Overburden Stress σ′ cv Ratio between the preconsolidation stress and the vertical effective stress σh Horizontal effective stress xii σ′ n Normal effective stress τ Strength index τf Shear stress µ Correction factor ϕ′ Friction angle ϕ′ c Critical state friction angle κ∗ Modified swelling index λ∗ Modified compression index µ∗ Modified creep index xiii xiv Contents List of Acronyms ix Nomenclature x List of Figures xvii List of Tables xxi 1 Introduction 1 1.1 Problem statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.2 Aim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.3 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.4 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.5 Project outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 2 Site description 5 2.1 History of the area . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.2 Geology and hydrogeology overview . . . . . . . . . . . . . . . . . . . 6 2.3 Construction works . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 2.4 Conceptual site model . . . . . . . . . . . . . . . . . . . . . . . . . . 10 3 Theory 11 3.1 Groundwater flow in soils . . . . . . . . . . . . . . . . . . . . . . . . 11 3.2 Unloading/reloading response . . . . . . . . . . . . . . . . . . . . . . 11 3.3 Settlements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 3.4 Excavations with embedded retaining walls . . . . . . . . . . . . . . . 13 3.4.1 Basal heave failure . . . . . . . . . . . . . . . . . . . . . . . . 13 3.4.2 Hydraulic uplift . . . . . . . . . . . . . . . . . . . . . . . . . . 13 3.4.3 Ground movements . . . . . . . . . . . . . . . . . . . . . . . . 14 3.5 Reference projects on anisotropic permeability . . . . . . . . . . . . . 15 3.6 Reference projects on excavation heave . . . . . . . . . . . . . . . . . 16 3.7 Soil classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 3.8 Evaluation of soil properties . . . . . . . . . . . . . . . . . . . . . . . 21 3.8.1 Piston sampler sampling . . . . . . . . . . . . . . . . . . . . . 21 3.8.2 Triaxial shear test . . . . . . . . . . . . . . . . . . . . . . . . . 21 xv Contents 3.8.3 Direct simple shear test (DSS) . . . . . . . . . . . . . . . . . . 22 3.8.4 The constant rate of strain (CRS) consolidation test . . . . . . 22 3.8.5 Sample quality . . . . . . . . . . . . . . . . . . . . . . . . . . 25 3.9 Numerical modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 4 Methods 33 4.1 Development of a soil profile . . . . . . . . . . . . . . . . . . . . . . . 35 4.2 Evaluation of soil properties . . . . . . . . . . . . . . . . . . . . . . . 35 4.3 Calibration of soil parameters . . . . . . . . . . . . . . . . . . . . . . 36 4.3.1 Numerical modelling of structural elements . . . . . . . . . . . 38 4.4 Method for assessing economic and environmental impacts . . . . . . 38 5 Results form geotechnical site characterisation 41 5.1 The geotechnical site investigation . . . . . . . . . . . . . . . . . . . . 41 5.2 Sample quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 5.3 Soil profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 6 Results from the numerical analysis 49 7 Results for economical and environmental impact 63 8 Discussion 65 8.1 Ground profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 8.2 Unloading effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 8.3 Reloading effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 8.4 Wall removal effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 8.5 Additional factors contributing to vertical displacements of the exca- vation bottom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 8.6 Soil test and sensitivity . . . . . . . . . . . . . . . . . . . . . . . . . . 69 8.7 Economical and environmental impact . . . . . . . . . . . . . . . . . 71 9 Conclusions 73 9.1 Recommendations for further studies . . . . . . . . . . . . . . . . . . 74 Bibliography 75 A Borehole I B Soil properties III C Soil test Clay XV D Soil test Silt XIX E Numerical Results XXIII xvi List of Figures 2.1 Overview of building foundations in the surrounding area (Golder Associates AB, 2020) . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 2.2 Soil map (Geological Survey of Sweden, 2023) . . . . . . . . . . . . . 7 2.3 Geotechnical structure for the new pipeline with sheet pile walls and bracing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 2.4 Plan over the settlements gauges that was used during the pipe con- struction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2.5 Monitoring of pipe settlement through displacement gauges. . . . . . 9 2.6 Conceptual Site Model . . . . . . . . . . . . . . . . . . . . . . . . . . 10 3.1 The soil behaviour type classification charts represented in the semilog Qt − Fr space and the semilog Qt −Bq space (Robertson, 1990). . . . 19 3.2 log-log Q−∆u2/σ ′ vo SBTn chart as proposed by Schneider et al. (2008). 20 3.3 Evaluation of compressibility parameters from the CRS-test (Sällfors, 1975) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 3.4 The reality and definition for the relationship between the modules for oedometer tests (Amundsen et al., 2015) . . . . . . . . . . . . . . 26 3.5 The definition for the relationship between the natural water content wN and volumetric strain ϵv0 for triaxial and oedometer tests (R. Larsson et al., 2007) . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 3.6 The modified stiffness parameters that are used in the SS/SSC model represented in the oedometer space (Karstunen & Amavasi, 2017). . . 28 3.7 The relationship between κ∗, λ∗, and M in the overconsolidated (OC) region and the normal consolidated (NC) region (Olsson, 2010). . . . 29 3.8 Coefficient of secondary compression αs(max) at the apparent precon- solidation pressure for different water content (P.-E. Larsson et al., 1997) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 3.9 Effect from delayed consolidation and groundwater movements of ge- ological history on the pre-consolidation pressure (Parry & Wroth, 1981) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 4.1 The Methodology employed for soil profile and soil properties. . . . . 34 4.2 The inSAR measurement of the vertical displacement of the site for the excavation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 4.3 Buildings LCA stages according to EN 15978 . . . . . . . . . . . . . . 40 xvii List of Figures 5.1 Results for borehole U34G06, U05G16 and U17G05 according to the (SBTn) classification diagram based on Robertson’s (1990) normal- ized parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 5.2 results of Soil classification diagrams in two different plotting formats from(Schneider et al., 2008) . . . . . . . . . . . . . . . . . . . . . . . 43 5.3 Result for sample quality with R. Larsson et al., 2007 method . . . . 44 5.4 The picture shows the developed soil profile based on the method proposed by Schneider et al. (2008) and the evaluated soil properties is shown. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 5.5 The illustrations show the developed soil profile in 3D from CPTu soundings, interpreted by empirical relations according to Schneider et al. (2008), and the geotechnical structure. . . . . . . . . . . . . . . 47 6.1 The soil geometry that was used in the numerical analysis. . . . . . . 51 6.2 The figure shows the influence of the mesh quality on the vertical dis- placement in of the excavation bed. The factor written together with the name represents the coarseness factor for the respective element distribution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 6.3 The results from the sensitivity analysis of the main material pa- rameters are compiled in the figure. The material parameters were varied by the factors indicated. The settlement values represent the settlement occurring from the onset of reloading until the end of con- solidation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 6.4 Presentation of the staged construction steps along with the prevalent water levels throughout the analysis. The water level outside the excavation is kept constant while the level inside is lowered 0.2 m below the excavation level. . . . . . . . . . . . . . . . . . . . . . . . . 58 6.5 The vertical displacements that occurred during the construction works are presented here for the validated (VAL) model and the cal- ibrated (CAL) model from the parametric study. The displacement measured from a node at -1.9 meters is used as the benchmark for comparison against pipe settlements. . . . . . . . . . . . . . . . . . . 59 6.6 The excess pore pressure development that arose during the staged construction are shown at multiple depth (nodes). . . . . . . . . . . . 60 6.7 The pipe settlements in the case study are compared with the excava- tion bottom (and pipe) settlement measured from node 5802 at -1.9 m in the numerical model. Settlements are recorded from the onset of reloading and end on day 34, when site monitoring stops. . . . . . 61 B.1 Soil unit weight plotted against level. . . . . . . . . . . . . . . . . . . IV B.2 Undrained shear strength plotted against level. . . . . . . . . . . . . . V B.3 Liquid limit plotted against level. . . . . . . . . . . . . . . . . . . . . VI B.4 Oedometer modulus (M0) plotted against level. The red line repre- sents the selected values for the initial analysis in PLAXIS before the soil test. Due to the limited number of measurement points in the silt layer, the same slope of the trend line for the clay was also applied to the silt and then verified through the soil test. . . . . . . . . . . . VII xviii List of Figures B.5 Oedometer modulus (ML) against level. The red line represents the selected values for the initial analysis in PLAXIS before the soil test. Due to the limited number of measurement points in the silt layer, the same slope of the trend line for the clay was also applied to the silt and then verified through the soil test. . . . . . . . . . . . . . . . VIII B.6 Preconsolidation pressure against level. . . . . . . . . . . . . . . . . . IX B.7 Limit pressure plotted against level. . . . . . . . . . . . . . . . . . . . X B.8 Consolidation coefficient plotted against level. . . . . . . . . . . . . . XI B.9 Permeability plotted against level. The red line shows the selected values for the initial analysis in PLAXIS before the soil test. Due to the limited number of measurement points in the silt layer, the slope for the trend line was determined by averaging the lowest values and applied to the entire silt layer, and then verified through the soil test. XII B.10 Water ratio plotted against level. . . . . . . . . . . . . . . . . . . . . XIII B.11 Sensitivity plotted against level. . . . . . . . . . . . . . . . . . . . . . XIV C.1 p′ − q plot showing triaxial (CAUC) test stress paths for clay. . . . . XV C.2 Stress paths for clay for the triaxial test, the original values in PLAXIS soft soil model, the values from the soil test in PLAXIS soft soil model, and the values from the PLAXIS parametric study for soft soil model for vertical effective stress, horizontal effective stress, shear stress, and pore pressure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . XVI C.3 The plot shows the results for clay for the CRS test, the original values in PLAXIS soft soil model, the values from the soil test in PLAXIS soft soil model, and the values from the PLAXIS parametric study for soft soil model for the plot of vertical effective stress against strain. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XVI C.4 The results of the oedometer modulus for clay are shown for the CRS test, the original values in PLAXIS soft soil model, the values from the soil test in PLAXIS soft soil model, and the values from the PLAXIS parametric study for soft soil model. . . . . . . . . . . . . . . . . . . XVII D.1 p′ − q plot showing triaxial (CAUC) test stress paths for silt. . . . . . XIX D.2 The plot shows the results for silt for the triaxial test, the original values in PLAXIS soft soil model, the values from the soil test in PLAXIS soft soil model, and the values from the PLAXIS paramet- ric study for soft soil model for vertical effective stress, horizontal effective stress, shear stress, and pore pressure. . . . . . . . . . . . . XX D.3 The results for silt are shown for the CRS test, the original values in PLAXIS soft soil model, the values from the soil test in PLAXIS soft soil model, and the values from the PLAXIS parametric study for soft soil model for the plot of vertical effective stress against strain.XXI D.4 The results of the oedometer modulus for silt for the CRS test, the original values in PLAXIS soft soil model, the values from the soil test in PLAXIS soft soil model, and the values from the PLAXIS parametric study for soft soil model. . . . . . . . . . . . . . . . . . . XXI xix List of Figures E.1 The vertical displacement of the excavation bottom is shown at vari- ous depths (nodes) plotted against the total calculation time. . . . . . XXIII E.2 The evolution of excess pore pressure plotted against the total calcu- lation time. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XXIV E.3 Illustration of the deformed mesh after unloading of the excavation. . XXV E.4 Illustration of the deformed mesh after the final consolidation step. . XXV E.5 Illustration of plastic points in the numerical model during reloading (backfilling). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XXVI xx List of Tables 3.1 The soil behaviour type classes presented by Robertson (1990). . . . . 19 3.2 The soil behaviour type classes according to Schneider et al. (2008). . 20 3.3 Sample quality estimation according to (Amundsen et al., 2015) com- pilation of different calculation methods for the CRS (oedometer) and triaxial test methods . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 4.1 Material properties used for numerical modelling of the embedded walls. 38 4.2 Bracing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 5.1 Sample quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 6.1 Input soil properties for validation with SOIL TEST in PLAXIS . . . 52 6.2 Construction procedure of the excavation . . . . . . . . . . . . . . . . 53 6.3 Validated soil properties in the numerical model. . . . . . . . . . . . . 54 6.4 The adopted soil properties from the parametric study. . . . . . . . . 57 7.1 Results for environmental impact presented in tonnes of carbon diox- ide equivalent (ton CO2eq) . . . . . . . . . . . . . . . . . . . . . . . . 64 7.2 Results for economical impact presented in million . . . . . . . . . . 64 A.1 Compilation of selected boreholes and analysed tests. . . . . . . . . . II xxi List of Tables xxii 1 Introduction The Swedish systems for stormwater, drinking water and wastewater management will need large investments following an increased urbanisation, a growing need for maintenance, and climate adaption for increased resilience against growing precip- itation (Svenskt Vatten AB, 2023). The investments in new stormwater, drinking water and wastewater pipes in Sweden amounted in 2021 to SEK 8.363 billion (B) and in existing pipes to SEK 5.090 B. This corresponds to a total yearly expendi- ture of SEK 13.461 B in the pipe network and 60% of the total investment in the systems. Nevertheless, an additional SEK 10 B is needed to cope with the growing demand for maintenance works in the system that was mainly built in the 1950s- 1970s. These drivers are together challenging the systems performance. Solutions that contribute to the development of the pipe systems are essential for people, the environment, and society. The construction industry could play a key role by con- ducting cost-effective projects for laying of new pipes and replacement of existing pipes approaching the end of their serviceable life. The laying of pipes in soft clay is typically performed in excavated trenches sup- ported by sheet pile walls in urban areas, where open-cut excavations are not suitable (Franzén et al., 2000). This approach often results in significant pipe settlements, impacting serviceability and leading to additional construction works and increased project costs. Moreover, construction of transportation infrastructure can some- times require pipes to be relocated. Consequently, an improved understanding of the ground movements in trench excavations for pipes and more reliable predictions of settlements can increase the profitability in construction projects. Currently, there is a lack of knowledge about the properties of silt and the mecha- nisms linked to settling problems. Therefore, SGI, together with the Swedish Traf- fic Agency and Chalmers, has chosen to delve more deeply into the behavior and properties of silt, according to (Löfroth et al., 2024). The report concludes that hydrogeological and topographic conditions create high pore water pressure in the most permeable areas of layered soils with varying hydraulic conductivity (Löfroth et al., 2024). This leads into this master’s thesis, which will also explore the prob- lem of silt related to settlement, situating this report within the existing body of knowledge on silt behavior. 1 1. Introduction The Swedish Transport Administration (Trafikverket) invests SEK 7.9 B in the de- velopment of Sweden’s main west coast railway line through the the Varberg tunnel project (Swedish Transport Administration, 2022). The railway route is important for passenger and freight traffic in Western Sweden. The project will add double- tracks to the link and thus improved railway capacity and increase the robustness of the system. The execution undertaken by Implenia Construction GmbH and Implenia Sweden AB, under an Early Contractor Involvement (ECI) contract with Trafikverket, comprises 9 km of double-tracks, a 2.8 km drill & blast tunnel, a 300 m cut & cover concrete tunnel, a 900 m concrete trough, and associated civil engineering works (Implenia AG, n.d.). The city of Varberg manages a wastewater treatment plant located in the north of Varberg. In the planning stage of the new tunnel project, the railway track area adjacent to the treatment plant was projected to expand, reducing the accessibility to existing pipes. Consequently, the pipes were relocated to a free corridor to ensure accessibility for future maintenance and renovations. 1.1 Problem statement The contractor laid pipes in excavated trenches supported by sheet pile walls and bracing. The walls were driven to the prescribed level. The trench was excavated to the final excavation level. Pipes and vertical displacement gauges monitoring the position were installed according to the instructions. During backfilling large pipe settlements arose and soon stabilised. The pipe settled more than the tolerances allowed and the construction approach was revised. 1.2 Aim The aim of this thesis is to investigate the mechanisms controlling the unloading and reloading effects in both the short and long term for excavations (3.2 m wide and 4-5 m deep). Additionally, it seeks to explain the pipe settlements and the mechanisms contributing to the observed behaviour in the considered case study. 1.3 Objectives The research questions for this project are as follows: • What are the mechanisms influencing the ground movements typically associ- ated with excavations with a depth of 4-5 meters? • What influences the unloading and reloading response of a deep excavation? • What geotechnical project risks exist given the site investigation? 2 1. Introduction • How can the pipe settlements from the site be explained? • How do alternative approaches for trenched excavations compare against each other? 1.4 Limitations The numerical modelling will only capture the unloading and reloading response of the excavation and the subsequent pipe settlements. The soil geometry employed will be based on the developed ground profile. The ground movements and soil dis- turbance arising from wall installation, wall removal, and workmanship will not be modelled, as the exact mechanisms and modelling procedures are not clear. Addi- tionally, the swelling behaviour of expansive soils will not be included in the analysis. 1.5 Project outline The project will encompass a literature review of embedded wall design and associ- ated ground movements, ground profiles typically associated with geotechnical risks, and numerical modelling of excavations. The site investigation will start with a desk study. Soundings will be selected and assessed, yielding a 3D ground profile for the site. Soil behaviour will be assessed, and soil properties evaluated from in-situ and laboratory tests. The next step is to identify the key factors in the ground profile and at the site that are believed to be important for recreating the pipe settlement in the numerical model. This will lead to the development of soil geometry for the numerical modelling, which will be performed using the geotechnical finite element software PLAXIS 2D. The evaluated properties will be validated by simulating stress paths in the SOIL TEST module in the software. A sensitivity analysis will identify the principal parameters controlling the behaviour in the model. Ultimately, a para- metric study will seek the ultimate soil properties that yield the pipe settlements observed at the site. 3 1. Introduction 4 2 Site description This chapter provides background information of the site, the excavation, and de- scribes the local soil conditions. Additionally, it presents the construction methods used for the new pipe, the geotechnical structure, and the existing structures. 2.1 History of the area The West Coast Railway between Malmö and Gothenburg was inaugurated in 1886 and has since been used to transport people and goods(Varberg Kommun, 2021). The railway station and the station house were built on what was then filled seabed. A goods depot, with railway tracks right outside, was built in 1900 north of the station house on land that resembled a swamp but was filled in. The West Coast Railway became very popular, and in 1920, the track was relaid, and the station area expanded. With an increasing population and trade, the traffic authority decided in 2019 to expand to a double track and a train tunnel through Varberg. The area by the old warehouse, where a former railroad track lay, will now have a wastewater pipeline built on the land previously characterised by filled marshland. To understand the local conditions of the site, it is important to assess the existing constructions in the area and their foundations. The foundations of the structures and facilities adjacent to the new wastewater line, marked in red, are illustrated in Figure (Figure 2.1) below. The existing Getterö Bridge and the new Getterö Road, marked in grey, are foundation-reinforced with piles. The building to the east of the line, where the old warehouse once stood, is now replaced by a new construction with a foundation reinforced with a slab on land, marked in yellow. The buildings west of the line are foundation-reinforced with piles marked in blue. The red area in Figure (Figure 2.1) indicates where the new wastewater line will be built. This area previously had a railroad track, which has been removed and replaced by a low noise barrier. The embankment was removed in connection with the construction of the new wastewater line. Additionally, there is a temporary road where the new line will be built, allowing all construction traffic to pass through. 5 2. Site description Figure 2.1: Overview of building foundations in the surrounding area (Golder Associates AB, 2020) 2.2 Geology and hydrogeology overview The geological historical site of Varberg creates complex conditions as the ground consists of many different deposits, both post-glacial and glacial (Påsse, 1990). The location under investigation is located right along the coastline, providing unique geotechnical conditions. The rock formation under the Varberg tunnel is presumed to mainly be banded gneiss with smaller part of mafic rocks. This rock formation was formed approximately 1420 million years ago and consists of alternating layers of quartz-feldspar and garnet. Most of the soil types and layers have been formed from this bedrock. Other soil types present are embedded within the others, such as flint, which formed in Skåne and was transported here by sea ice. The material above the bedrock is assumed to be moraine/till, which is also classified as a firm material. The soil layer overlying the bedrock is characterised by glacial clay or post- glacial clay. Both glacial clay and post-glacial clay can occur as both homogeneous and laminated layers. In the laminated clay, layers of sand, gravel, shell layer may occur. Above the clay, there are a few meters of wave-washed gravel and postglacial sand mixed with or overlain by fill material. The different soil types are presented in the soil map from SGU in the figure (Figure 2.2) below. 6 2. Site description Figure 2.2: Soil map (Geological Survey of Sweden, 2023) A general cross-section of the soil layer sequence developed by Implenia from surveys and material from the Soil Survey Report (MUR) is presented at the beginning of the process. The soil profile was taken from material from the soil type map from SGU, CPT surveys, interpretation of CPT using the computer program CONRAD and piston sampling. This survey indicates that the ground level is located at a depth of +2m to +3m above sea level and that groundwater levels are located at 1-2 meters below ground level. The top layer was fill with elements of sand, gravel and brick and had a varying thickness from 0m to -1m up to ground level. Under the fill, an area of sand/fine sand with a maximum thickness of 1.8m occurs in some parts, mainly in the south. The sand and fill layers are underlain by a clay layer that varies between 0m and -8m. The clay layer above goes to a silty clay/clayey silt layer down to a depth of 18m. The next layer is a solid layer consisting of moraine that overlays the rock surface, where the rock surface starts at a depth of approximately 16-20 meters. This geological locale and the produced cross-section described above types on a system with two aquifers, partly an unconfined aquifer in the sand layer/fill layer and partly a confined aquifer in the silty clay/clay silt. The clay layer in between will act as a confining bed. The first cross-section from Implenia is reported in (Figure 2.6) below under the conceptual site model chapter. 2.3 Construction works The City of Varberg manages a wastewater treatment plant located to the north of Varberg. In the planning stage of the tunnel, the railway track area adjacent to 7 2. Site description the treatment plant was projected to expand, reducing accessibility to the existing underground utility lines connecting to the plant. Consequently, to enhance acces- sibility to perform future maintenance and renovation works, the pipe was realigned to an open corridor along the east side of the railway tracks. The new wastewater pipe was planned with a length of 580m using precast concrete pipe segments of diameter 1200mm, 1000mm and 800mm installed in an excavated trench. The pipe that starts in the south uses 1000 mm concrete segments with an excavation bottom situated between -1.6 m and -0.2 m. In the middle, a 1200 mm pipe was installed at the bottom of the excavation, between -1.9 m and -2.1 m. At the northern end, an 800 mm pipe was installed with an excavation bottom at a level between -1.4 m and -1.9 m. The ground surface level is at approximately +2 m. The construction approach involved laying the pipe in a braced excavation using 10 m long sheet pile walls. The wall type was VL603 with steel quality S355GP. The bracing was HEB300 with steel quality S355N/M/J. The walls was installed by means of vibration using a side grip vibratory-type sheet pile driver until the top of the walls was at +2,5m. The material properties of the wall and the bracing is presented in Table 4.1 and Table 4.2. The excavation proceeded to the final excava- tion level was at -1,9m. The backfilling began with constructing a reinforced pipe bed of 0,5m thickness. A vibratory plate compactor was used inside. Subsequently, the pipe was installed and settlement gauges connected (Figure 2.4). Lastly, the walls were removed. The gauges recorded significant pipe settlements during the backfilling of the excavation (Figure 2.5). Figure 2.3: Geotechnical structure for the new pipeline with sheet pile walls and bracing 8 2. Site description Figure 2.4: Plan over the settlements gauges that was used during the pipe con- struction. Figure 2.5: Monitoring of pipe settlement through displacement gauges. 9 2. Site description 2.4 Conceptual site model The first step in developing a numerical model that illustrates important aspects of the real world is to create a conceptual site model. This model is a simplification of reality used to understand the problem. The conceptual model developed for the pipeline project in Varberg is presented in (Figure 2.6). This model shows adjacent structures, rail ways and roads. The conceptual site model highlights the adjacent structures, infrastructures, and that soil tests are only available on the west side of the projected pipeline. Figure 2.6: Conceptual Site Model 10 3 Theory The chapter presents key concepts relevant to the considered case and summaries the main findings of the literature review. 3.1 Groundwater flow in soils The groundwater flow through soil can be estimated using Darcy’s Law, as shown in Equation 3.1. The flow depends on the cross-sectional area (A), the permeability of the soil (k), and the hydraulic gradient (i), which can be derived from Equation 3.2 (Brown et al., 2023a). q = Aki (3.1) i = dh dx (3.2) Natural soils formed under varying depositional environments can exhibit anisotropic features in the macro-structure(Brown et al., 2023a). Distinct alternations in par- ticle sizes, as in partings, laminations, and layering influences the soil permeability and ground water flow conditions. Layering on a macroscopic level can result in the horizontal permeability being 100 to 1000 times larger than the vertical. In engi- neering practice, a reliable estimate of a soil’s permeability is derived from in-situ measurements. Laboratory tests use a limited number of samples, and sampling can result in disturbed specimens, affecting the specimen’s permeability. 3.2 Unloading/reloading response The unloading response of soils depends on the unloading stiffness (Persson, 2004). The unloading modulus is thus relevant for estimating heave in excavations. The stress dependency of the unloading modulus was analysed for a soft clay deposit in the Gothenburg region. The study encompassed field monitoring in a deep excava- tion and laboratory testing, including triaxial, IL oedometer, and Bender element tests of specimens. As the overburden pressure was reduced, the unloading modulus decreased. Moreover, the initial large unloading modulus can be explained by the fact that secondary consolidation is acting in the opposite direction of the swelling 11 3. Theory (Persson, 2004). The swelling rate must thus be larger than the secondary consolida- tion rate for heave to develop. A relationship for the unloading modulus in soft clay is presented in Equation 3.3 (Persson, 2007). The showed unloading modulus is a re- fined numerical fitting of the relationships recognised earlier by Persson (2004). The equation depends on the pre-consolidation pressure (σ′ c) and the over-consolidation ratio (OCR), being OCR = σ′ c/σ ′ v. S is the total swelling. Mul = 35σ′ c · e 3.5 OCR (3.3) The unloading behaviour in silt is characterised by swelling, and the unloading modulus can be calculated from Equation 3.4, using empirical values of the swelling index (αs) ranging from 0.2 to 0.6 % (R. Larsson, 1995). Mul = σ′v as (3.4) The reloading behaviour and modulus of soft soils are influenced by previous swelling as seen in Equation 3.5 (R. Larsson, 1986). The load factor (b) corresponds to the stress state at which swelling develops in the soil. The reloading modulus in a stress range σ′ v < b ·σ′ c can be calculated with Equation 3.5 and will be near constant. For σ′ v > b · σ′ c, Equation 3.6. Mrl = b · σ′ c − σ′ v as · ln ( b·σ′ c σ′ v ) = b · σ′ c − σ′ v S (3.5) Mrl ≃ σ′ v as (3.6) 3.3 Settlements Loading saturated soil with low permeability results in undrained (ρu), primary consolidation (ρc) and sometimes secondary/creep settlements (ρs) (Brown et al., 2023a). Equation 3.7 shows the total settlement (ρt). ρt = ρu + ρc + ρs (3.7) Primary consolidation settlements (ρc) in soils with a low permeability arise from the loading response and the subsequent consolidation process (Knappett & Craig, 2019). Loading increases the overburden pressure. The soil particles want to re- arrange to take up less space, but such a movement is counteracted by the incom- pressible nature of water. The water flow rate through the boundaries of the soil is lower than the loading rate, resulting in an increased excess pore water pressure (ue) and an undrained condition. A seepage gradient develops, draining the soil 12 3. Theory and reducing the pore water pressure. The consolidation rate is governed by the soil’s permeability (k). The consolidation continues until ue = 0, and the pore water pressure equals the static pore water pressure. At this stage, the consolidation is completed and the soil is drained. The resulting vertical displacement is the primary consolidation settlement. At the same time as the primary consolidation occurs, the soil can exhibit sec- ondary consolidation/creep settlements that depends on the soil viscosity (Knappett & Craig, 2019). The secondary consolidation index is calculated using Equation 3.8 (Olsson, 2010). The secondary consolidation is dependent on the degree of consoli- dation. Cαe = ∆e ∆log(t) (3.8) 3.4 Excavations with embedded retaining walls Earth retaining structures provide lateral support to the ground (Gaba et al., 2017). Various types exist, encompassing gravity retaining walls, hybrid designs, and em- bedded retaining walls. For an embedded wall, lateral support for the retained ground is achieved by driving the wall below the excavation base. Various types of embedded retaining walls exist, including sheet pile walls, king post walls, contigu- ous bored pile walls, secant bored pile walls, and diaphragm walls. Deep and narrow excavations often use sheet pile walls, with additional support provided by bracing or props installed as the excavation progresses (Knappett & Craig, 2019). 3.4.1 Basal heave failure The earth pressure outside excavation is exerted by the soil weight and a potential surface load. When the overburden pressure above the excavation level is equal to or exceeds the shear strength of the soil, basal heave failure can occur (Knappett & Craig, 2019). Failure occurs along a circular shearing plane below the excavation. Braced excavations in use one or more levels of props to maintain stability as they progress, resulting in larger stress reductions during excavation and, eventually, larger risk for bottom heave. 3.4.2 Hydraulic uplift In dry excavations in fine-grained soil layers that overly coarser soil with higher permeability, high water pressures in the underlying formation can lead to instability and hydraulic uplift (Fredriksson et al., 2018). The stability of the excavation base against hydraulic uplift can be assessed using Equation 3.9, using the unit weight of water ρw, gravity g, height between ground water table and permeable stratum H, saturated soil density ρm, and the distance between excavation level and permeable stratum d. The partial factor γd depends on the safety class for the geotechnical 13 3. Theory structure and is 0.83 for safety class (SF) 1, 0.91 for SF 2, and 1.00 for SF 3. Consequently, the overburden pressure of soil in the excavation has to be greater or equal to the water pressure in the more permeable coarse soil layer beneath to ensure stability. ρw · g ·H ≤ 0.9ρm · g · d γd · 1.1 (3.9) The high water pressures beneath the excavation can be reduced by installing wells and thus reduce the risk for hydraulic uplift to develop in the excavation (Brown et al., 2023b). The excavation can be cut off from the water-bearing stratum by taking the embedded retaining walls into an underlying soil formation with lower permeability or by grouting and sealing the zone below the wall. A passive relief system for the water pressure can be installed in the excavation. 3.4.3 Ground movements Excavations in front of embedded retaining structures result in ground movements, which depend on various factors and their complex interactions in three dimensions (Gaba et al., 2017). Factors contributing to ground movements within and around a deep excavation encompass wall installation, the excavation effect (changes in the stress state in the soil during unloading, excavation geometry, ground stiffness and strength, the type and stiffness of the wall and support system, changes in the groundwater regime and pore pressures, the construction procedure, and the quality of workmanship), groundwater flows, and project-specific procedures (e.g. sheet pile removal). This section covers soil movements from installing embedded retaining structures, excavation work before the wall, and elastic and time-dependent heave. Wall installation Sheet piles are installed through driving, hydraulic jacking or pressing depending on the ground conditions (Gaba et al., 2017). Smaller ground movements can arise dur- ing the installation procedure. Sheet pile driving also generates ground vibrations, resulting in compaction of loose coarse-grained soils. Excavation works Excavation between the retaining walls reduces lateral passive earth pressures. Mo- bilising shear stresses to the yield limit of the soil in excavation results in shear deformations and local plastic ground movements (Gaba et al., 2017). Elastic and time-dependent heave The overburden pressure exerted on the underlying soil is reduced as the excavation progresses (Gaba et al., 2017). This leads to an elastic displacements upwards of the excavation bottom in the short term. The soil stiffness determines the imme- diate heave for excavations in clay. The unloading effect on an excavation bottom in saturated clay also results in time-dependent swelling, which adds to the total 14 3. Theory excavation heave in the long term (Knappett & Craig, 2019). The reduced over- burden pressure increases soil volume and void space, resulting in decreased pore pressures and negative excess pore pressures. Swelling arises as the seepage restores the pore pressure within the soil to the long-term equilibrium, steady-state seepage and drained conditions. The soil permeability governs the swelling rate. 3.5 Reference projects on anisotropic permeabil- ity The conceptual design of underground construction projects needs to account for the macro-structure of the soil, which is influenced by the depositional environment (Brown et al., 2023a). The characteristic features can also vary within a soil layer, e.g. coarse partings, layering, and lamination, resulting in anisotropic permeability. The conceptual design of the New Palace Yard underground car park adjacent to Westminster Palace in London showed the importance of thorough ground investiga- tions of the soil macro-structure and its implications on the construction procedure (Brown et al., 2023a). The setting with surrounding historical buildings meant small tolerances for ground movements, thus presenting a geotechnical engineering chal- lenge for the project. The underground project involved an 18.5 m deep excavation supported by reinforced concrete diaphragm walls and struts. Geotechnical site in- vestigations revealed 10mm thick fine sand and silt partings with a 50 mm spacing within the London clay between 19 and 30m depth, with a decreasing frequency with depth. A more homogeneous clay was estimated to extend from a depth of 30m. The presence of permeable partings in the clay meant that the horizontal permeabil- ity could be high. Assessments of the groundwater conditions on the site indicated hydrostatic water pressures. Consequently, the soil underlying the excavation could exhibit high pressures and present a geotechnical hazard, encompassing an increased risk for hydraulic uplift and base failure. Various mitigation measures were assessed. A relief system was inappropriate due to concerns about its effectiveness over time. Additionally, it was challenging to perform a representative seepage analysis for the design, as the flow properties can vary greatly in fine-grained soil with coarse part- ings. Ultimately, the approach employed 30m diaphragm walls extending into the underlying homogeneous clay, there by reducing the risk of hydraulic uplift and base failure. A construction project for a sewer in a deep excavation in Southampton encountered difficulties due to a water-bearing stratum underlying the excavation level (Ward, 1957). The contractor used sheet piles to support the trench excavation, carried out in a stiff sandy clay overlaying a laminated soil layer between fine sand and silty. The sewer was constructed on a concrete slab on the excavation bottom. During excavation, hydraulic uplift arose from the high water pressures in the underlying formation and the subsequently constructed precast concrete sewer pipes settled and cracked after backfilling. The solution became a relief system for the water pressure in the underlying formation. A post-hole auger was driven into the excavation bot- 15 3. Theory tom, filling the hole with gravel. Excess water was pumped during the construction. A permeable layer was constructed on the bottom, followed by the concrete slab and the sewer. This approach led to the successful completion of the project. Hydraulic uplift was identified as a significant risk in a 10m deep excavation project in sensitive clay in central Ottawa (McRostie et al., 1996). The excavation would later accommodate the new Police headquarters complex. In 1981, it was the deepest open dry excavation in the city. A geotechnical site characterisation identified the subsurface as silty clay until 13 meters depth, clayey silt until 18 meters, a stratum of sandy silt, and glacial till at 21 meters. High water pressures in the more permeable stratum below the excavation level were recognised as a potential geotechnical hazard that could lead to hydraulic uplift. The risk was mitigated in the planning phase of the contract by developing a control system for the water pressures of wells and electro-osmosis. The contractor awarded the contract was concerned about the overall cost efficiency and the influence that it might have on the workmanship within the excavation and thus employed the observational approach. The unloading effect would result in negative pore pressures, and a groundwater gradient would develop to seek equilibrium. Nevertheless, a slow seepage gradient due to estimated low horizontal permeability would allow the excavation to be completed without additional measures. The excavation started with a test and then progressed in stages. Extensive monitoring of both heave and pore pressures was performed. In case of an uncontrolled rise in water pressure or heave, the contractor would install wick drains. When the bottom was reached, a raft foundation was installed. The contractor successfully employed the observational method and completed the excavation with minimal heave. 3.6 Reference projects on excavation heave Full-scale field studies on bottom heave and deformation in trench excavations sup- ported by sheet piles in Swedish soft clay have been performed by Magnusson (1975) and Franzén et al. (2000). Magnusson (1975) analysed the heave behaviour in deep excavations and whether the unloading response was elastic and plastic. A proposed approach to predict the former was to use a bottom heave safety factor N . A value for N smaller than 4 signifies elastic heave and a greater plastic. Reloading of the excavation base by backfilling results in settlements, and the magnitude depends on the deforma- tion behaviour of the base heave. In clay, the subsequent settlement, developing during the construction time, is equal to the elastic heave if the soil structure re- mains unchanged, and the reloading restores the stress state to the initial condition. Reloading to a restored stress state and with an unchanged clay structure on a base heave composed of both elastic and plastic deformations results in an additional settlement, apart from the elastic rebound, developing during 3 to 6 months. The additional settlement is sometimes associated with the retaining wall extraction and the following remaining cavity. 16 3. Theory Franzén et al. (2000) performed full-scale field trials of pipe laying in trench exca- vation between temporary sheet pile walls in soft clay in the Gothenburg region. Bottom displacements were monitored and analysed during and after construction. Various approaches to minimise the total settlements were analysed. In all attempts, the initial bottom heave was followed by an even greater settlement of the pipe. The study found that longer sheet piles reduced the mobilised shear stresses and the bottom heave, leading to smaller subsequent pipe settlements during backfilling. Longer sheet piles also created a larger cavity in the soil below the trench base, resulting in larger vertical settlements during wall extraction(Franzén et al., 2000). The removed walls contained 0-15 mm of soil, further increasing the cavity in the ground (Franzén & Spetz, 1998). Cutting the walls at the excavation level did not reduce bottom settlements (Franzén et al., 2000). Moreover, walking on the excava- tion bottom disturbed the clay, leading to larger pipe displacement. Ultimately, the study recognised that the total bottom displacement arose form was complex and arose from various sources. 17 3. Theory 3.7 Soil classification A prevalent approach for soil classification is based on physical soil characteristics (e.g. grading) (Robertson, 2016). Soil classification in geotechnical projects could benefit from combining physical and behavioural characteristics, as it better repre- sents the in situ condition of the soils. The in situ soil behaviour can be measured through a cone penetration test, and various approaches for subsequent data analy- sis and soil classification exist (e.g. Robertson (1990), Robertson (2009), Schneider et al. (2008) and Robertson (2016)). Cone penetration testing at a constant rate in situ provides continuous logging of test properties in the ground (Brown et al., 2023a). Subsequent data analysis can accurately identify soil stratification, classification and various properties. The elec- tric static cone penetrometer (CPT) continuously logs the tip resistance (qc) and the sleeve resistance (fs) on the sleeve above the cone. The piezoncone penetrome- ter (CPTU) also measures the pore water pressure during penetration. While this method can yield precise geotechnical property values for sand, additional in situ methods are necessary to complement it for clay and silt (Swedish Institute for Stan- dards, 2023). Piston sampling and subsequent testing can better estimate density, water content, and liquidity index (Knappett & Craig, 2019). Robertson (1990) presented a soil behaviour type classification (SBT) system from CPTU logs (Figure 3.1 that classifies soils according to Table 3.1). Normalised parameters are obtained from the CPTU data with Equation 3.10, 3.11 and 3.12 (Robertson, 2016). The SBT system presented by Robertson (1990) is a refinement of a previous method and considers an increase in effective overburden stress with depth in both penetration and sleeve resistances (Robertson, 2009). The semilog Qt − Fr space has limited accuracy for classifying softer, fine-grained soils, particu- larly within the lower Qt range (< 10 kPa). In such cases, an alternative approach is to plot the CPTU parameters in the semilog Qt − Bq space, as the classification is linked to the penetration pore pressure. Qt = qt − σvo σ′ vo (3.10) Fr = [ fs qt − σvo ] · 100% (3.11) Bq = u2 − u0 qt − σvo = ∆u qt − σvo (3.12) 18 3. Theory Figure 3.1: The soil behaviour type classification charts represented in the semilog Qt − Fr space and the semilog Qt −Bq space (Robertson, 1990). Table 3.1: The soil behaviour type classes presented by Robertson (1990). Zone Soil Type 1 Sensitive fine-grained 2 Organic 3 Clay 4 Silt-mixtures 5 Sand-mixtures 6 Sand 7 Gravelly sand to sand 8 Very stiff sand to clayey sand 9 Very stiff fine-grained Soil structure at the micro and macro scales contributes to the different behaviours observed in situ compared to ideal soils (Robertson, 2016). As a result, a refine- ment of the classification approach for soils with pronounced micro structure was developed. It is also recognised that the model presented by Robertson (1990) is reliable in ideal soils with a homogeneous macrostructure. The model presented by Schneider et al. (2008) is better suited for soils with pronounced macrostructure, as 19 3. Theory the model has adjusted boundaries, especially in areas with higher Fr values. This adjustment creates clearer distinctions between silt, clay, and sand. The approach for soil classification with SBTn charts presented by Robertson (1990) from CPTU logs is less reliable in soils with a considerable macrostructure, such as layering (Robertson, 2016). The prevalent SBTn charts are also less accurate in transitional soils (Schneider et al., 2008). Consequently, new charts were proposed based on the general tip resistance and the pore pressure, encompassing the following spaces: log-log Q− ∆u2/σ ′ vo (Figure 3.2), semilog Q− ∆u2/σ ′ vo and semilog Q−Bq, which uses Bq as the model by Robertson (1990) and new boundaries (Schneider et al., 2008). The soil behaviour in mixtures of clay, silt, and sand with um = 0 during testing is best represented in the log-log Q − ∆u2/σ ′ vo space as it emphasises the key features of these soils. However, all spaces will yield the same soil classification. The approach characterises the soils according to Table 3.2. Figure 3.2: log-log Q−∆u2/σ ′ vo SBTn chart as proposed by Schneider et al. (2008). Table 3.2: The soil behaviour type classes according to Schneider et al. (2008). Zone Soil Type 1a Silts and "Low Ir" clays 1b Clays 1c Sensitive clays 2 Essentially drained sands 3 Transitional soils 20 3. Theory 3.8 Evaluation of soil properties Obtaining an established soil profile allows for the investigation of soil properties using Triaxial shear tests, Consolidation tests with a constant rate of strain (CRS), Direct simple shear tests (DSS), and sampling with piston samplers. 3.8.1 Piston sampler sampling The initial parameters analysed included density ρ, water content wn, liquidity index wL, sensitivities and preconsolidation pressure (σ′ c) obtained from sampling with a piston sampler. Although triaxial shear tests, CRS, and DSS tests provided values for density and water content, samples from the piston sampler were considered more comprehensive and consequently prioritised highest in collecting these parameters. All values for density, moisture content, liquidity index, and sensitivity were com- piled and plotted against depth, with trend lines drawn for each layer to determine the final value for the layer. Effective stress can be calculated using the soil layer’s densities and known water level. The fill, sand and rock densities were evaluated according to TK GEO 13 and (R. Larsson, 1995). From piston sampler sampling, fall cone tests and field vane (FVT) tests were also conducted in situ to determine the value of the strength index, τ . This value of the strength index was then used to calculate the undrained shear strength cu by com- puting a correction factor µ according to Equations 3.13 to adjust the strength index, τ , to the undrained shear strength cu according to Equations 3.14. The correction factor incorporates the soil’s liquidity index to better account for soil behaviour and thus provide a more accurate value for the undrained shear strength. Finally, the preconsolidation pressure (σ′ c) was evaluated using Hansbo’s empirical method for both fall cone tests and field vane (FVT) tests, as described in Equation and the undrained shear strength (uncorrected) was used Equations 3.15, (R. Larsson et al., 2007) and (Swedish Institute for Standards, 2007). µ = (0.43 wL )0.45 (3.13) cu = τ ∗ µ (3.14) τ = σ′ c ∗ 0.45 ∗ wL (3.15) 3.8.2 Triaxial shear test Triaxial compression test is a prevalent testing approach for shearing of soils (Knap- pett & Craig, 2019). The shear strength of clay depends on the history of applied 21 3. Theory stresses and thus the state of consolidation (Azizi, 1999). With increasing consol- idation degree, the linear relationship between shear strength and applied normal effective stresses becomes non-linear. Hence, the relationship between the shear stress at failure for a heavily overconsolidated clay and the applied normal effective stresses follows Equation 3.16, and depends on the peak frictional angle (ϕp) and the apparent cohesion (c′). The parameter c′ is the shear strength exhibited by the clay at zero normal effective stresses. The additional strength and stability within the clay result from increasing effective stresses from negative pore pressure, leading to suction. In practice, c′ is challenging to measure accurately and can thus be set to zero as a conservative approach. τf = c′ + σ′ n tanϕ′ p (3.16) Normally consolidated clay and granular soils do not exhibit any apparent cohesion, and the relationship linking the shear strength to the effective normal stresses follows Equation 3.17, where ϕc is the critical frictional angle. τf = σ′ n tanϕ′ c (3.17) 3.8.3 Direct simple shear test (DSS) The values from DSS tests were utilized to correct the shear strength obtained from fall cone and wing probing tests, complementing the results from Triaxial shear tests. According to Swedish Institute for Standards (1991b), DSS is employed to assess the controlling shear strength as a function of the normal stress acting on the shear plane. Both undrained shear strength and effective strength parameters are determined for DSS. Parameters obtained from the Direct Simple Shear (DSS) test included density (ρ), water content (wn), undrained shear strength (cu), and preconsolidation pressure (σ′ c). Direct shear tests are conducted on undisturbed samples, primarily layered soil with alternating silt and clay TK GEO 13. The shear strength during direct shearing on a horizontal sliding surface is determined through direct shear tests. These tests are also particularly suitable for layered and varved soils, where the shearing is directed towards any potentially weaker layers in the soil (R. Larsson et al., 2007). This indicates that compared to the corrected strength determined by field vane (FVT) and fall cone tests, active triaxial tests typically yield higher undrained shear strengths, while passive triaxial tests yield lower undrained shear strengths. The average undrained shear strength from active and passive triaxial tests and direct shear tests is slightly higher than the corrected strength from field vane (FVT) and fall cone tests, especially in low-plasticity clays (R. Larsson et al., 2007). 3.8.4 The constant rate of strain (CRS) consolidation test From CRS, the calculations of the parameters density (ρ), preconsolidation pressure (σ′ c), limit stress (σ′ L), permeability (k), and oedometer modulus (M0, ML, M ′) were 22 3. Theory performed. An important parameter for obtaining information about the strength and defor- mation properties of the soil was the apparent pre-consolidation pressure (σ′ c). This parameter indicated the degree of soil consolidation and the maximum pressure to which the soil had previously been subjected, including creep deformation (Sällfors, 1975). Given the extensive past construction in the area, this parameter played an important role in understanding the loading history and predicting the soil’s be- haviour under the new loading. Consolidation was caused by two main mechanisms: vertical load exceeding the previous load and creep deformation resulting from par- ticle movement, especially in clay particles that reorganised under constant effective stress (Bjerrum & Aitchison, 1973). To calculate the apparent preconsolidation pressure from CRS, Sällfors (1975) method was used in combination with Swedish Institute for Standards (1991a) to analyse CRS. M0 Constant modulus at stresses below the apparent pre-consolidation pressure, ML Constant modulus for stresses between the apparent preconsolidation pressure and the limit stress σL, and mod- ulus number M’ were also evaluated according to Swedish Institute for Standards (1991a) and Sällfors (1975). The parameters were evaluated according to Figure 3.3. The parameters for the compression modulus are assessed based on the plotted modulus-stress curve, which typically yields an M0 value that tends to be lower than the actual field conditions, as per past experiences. As a common practice, the evaluated M0 from a CRS test is often multiplied by a factor of 3-5 to approximate the field conditions better (Olsson, 2010) Figure 3.3: Evaluation of compressibility parameters from the CRS-test (Sällfors, 1975) According to Karstunen and Amavasi (2017), the apparent preconsolidation pressure is an important parameter for the Soft Soil models in PLAXIS, as the pre-overburden pressure (POP) and the Overconsolidation Ratio (OCR) used in the material models depend on this parameter. POP was of significant importance for the model’s re- sponse and its sensitivity. The apparent pre-consolidation pressure can be evaluated from Triaxial tests, DSS tests, fall cone tests (FC) and field vane test (FVT). 23 3. Theory To calculate the Oedometer module from CRS tests, the definition provided by R. Larsson, 2008 is employed, utilizing the relationship between the vertical effective stress σv and strain ϵ1 according to Equation 3.18. The oedometer module serves to assess the material’s load-bearing capacity, particularly in settlement calculations. This module assumes that horizontal deformation is restrained and that the vertical effective stress increases continuously (R. Larsson, 1995). M = ∆σ′ v ∆ϵv (3.18) 24 3. Theory 3.8.5 Sample quality To ensure a correct and accurate analysis, it is important that the samples are of high quality. Quality assessment is used to investigate possible sample disturbances by classifying whether and how disturbed the samples are. Several methods exist to determine the degree of disturbance in a sample. The methods used in this report are based on the theories of (Terzaghi et al., 1996) (which is based on (Andresen & Kolstad, 1979)), (Karlsrud & Hernandez-Martinez, 2013) and (R. Larsson et al., 2007). The first two theories can be applied to both laboratory tests: the Oedometer (CRS) and triaxial tests. They are based on evaluating the volumetric strain ϵv0 at the preconsolidation pressure σ′ p for the CRS samples and at the horizontal effective pressure σ′ h0 for triaxial tests (Andresen & Kolstad, 1979; Terzaghi et al., 1996). The samples are then classified according to the Specimen Quality Designation (SQD) system, which ranting the disturbance level of the samples from A to E (Terzaghi et al., 1996). The range of the classes and their meanings are specified in Table 3.3. Table 3.3: Sample quality estimation according to (Amundsen et al., 2015) com- pilation of different calculation methods for the CRS (oedometer) and triaxial test methods Specimen Qual- ity Designation (SQD) (Terzaghi et al. 1996) M0/ML Karlsrud and Hernandez- Martinez (2013) Volumetric strain ϵv0 (%) SQD Ratio M0/ML Sample quality <1 A >2 Very good to excellent 1-2 B 1.5-2 Good to fair 2-4 C 1.5-2 Poor 4-8 D 1-1.5 Very poor >8 E <1 The second method is used only for CRS samples and use the oedometer modules to determine a stiffness ratio, called the oedometer stiffness ratio. This relationship is defined between M0 The initial consolidation modulus, which indicates the stiffness of the soil at the start of loading and ML The consolidation modulus at a later stage of loading, which indicates the stiffness of the soil after some consolidation (Karlsrud & Hernandez-Martinez, 2013). 25 3. Theory These values, M0 and ML, are determined according to the procedure shown in Figure 3.4. The classes and their meanings, which categorise the sample quality based on the oedometer stiffness ratio, are given in Table 3.3 . This classification is important for assessing the degree of disturbance in the samples and is essential for the correct interpretation of the results of the CRS tests (Amundsen et al., 2015). Figure 3.4: The reality and definition for the relationship between the modules for oedometer tests (Amundsen et al., 2015) The last theory to evaluate the quality introduced by R. Larsson et al., 2007 is useful for both the Oedometer (CRS) and triaxial tests. R. Larsson et al., 2007 method is a further development of Lunne et al., 1997. The method uses the value for the volumetric strain, ϵv0, but instead of the preconsolidation pressure, σ′ p, the natural water content wN is applied. The classes and their meanings, which categorize the sample quality based on the volumetric strain, are given in Figure 3.5. Figure 3.5: The definition for the relationship between the natural water content wN and volumetric strain ϵv0 for triaxial and oedometer tests (R. Larsson et al., 2007) 26 3. Theory 3.9 Numerical modelling Elasto-plastic hardening models are suitable for numerical modelling of lightly over- consolidated soft soils (Karstunen & Amavasi, 2017). These constitutive models incorporate elastic properties, a yield function, a plastic potential, and a hardening rule. The associated hardening rule links plastic deformation to the progression of the yield surface. The internal hardening parameter allows the model to recognise distinct features of soft soils, such as the stiffness dependency on recent stresses and the strength increase along the normal compression line. Two elasto-plastic harden- ing models are the Soft Soil (SS) model and the Soft Soil Creep (SSC) model. The SS model uses a failure criterion defined by the Mohr-Coulomb model. The SSC model incorporates rate-dependency to recognise secondary consolidation/creep in the analysis. The yield function of the SS model is influenced by the critical state (Karstunen & Amavasi, 2017). The critical state concept in geomechanics is an ultimate state condition, reached when continuous constant shearing results in no more changes in mean effective stress (p′), deviatoric stress (q), or void space (v) of a soil specimen (Equation 3.19) (Wood, 1991). ∂p′ ∂ϵq = ∂q ∂ϵq = ∂v ∂ϵq = 0 (3.19) At this state, the critical state stress ratio (η) evaluated as the ratio between the deviatoric stress (qc) and the vertical effective stress (σ′ c) is equal to the critical state line (M) (Equation 3.20) (Wood, 1991). qc p′ c = η = M (3.20) The frictional angle at critical state (ϕ′ c) can estimated from the critical state ratio in triaxial compression (Mc), typically given by a triaxial compression test (CAUC) (Equation 3.21) (Karstunen & Amavasi, 2017). Ultimately, the SS yield surface is defined. sinϕ′ c = 3Mc 6 +Mc (3.21) The critical state ratio in extension (Me) can be estimated from the ϕc using Equa- tion 3.22 (Karstunen & Amavasi, 2017). This value is typically given by a triaxal tests in extension (CAUE). sinϕ′ e = 3Me 6 −Me (3.22) Moreover, the model parameters for the soft soil (SS) creep (SSC) model encom- passes the modified swelling index (κ∗), the modified compression index (λ∗), and 27 3. Theory Poisson’s ratio at unloading/reloading (νur), being different from the initial Pois- son’s ratio (ν) (Karstunen & Amavasi, 2017). Olsson (2010) has compiled empirical approaches for evaluating parameters for the SS/SSC model. κ∗ can be derived through Equation 3.23 and λ∗ can be evaluated by using Equation 3.24. σ′ vc rep- resents the vertical preconsolidation stress. The oedometer modulus (M) is stress dependent and is M = M0 in the range 0 < σ′ v < σ′ c and M = ML in σ′ c < σ′ v < σ′ L (R. Larsson, 2008). κ∗ ≃ 2 ∗ σ′ v M (3.23) λ∗ ≃ 1.1 ∗ σ′ vc ML (3.24) The slopes corresponding to stiffness parameters κ∗ and λ∗ are shown in the semilog ϵp − p′ space from an oedometer test (Figure 3.6) (Karstunen & Amavasi, 2017). κ∗ equals to the unloading (swelling) slope in the oedometer space. An alternative approach for evaluating κ∗ is to use the initial loading slope with caution. Figure 3.6: The modified stiffness parameters that are used in the SS/SSC model represented in the oedometer space (Karstunen & Amavasi, 2017). A relationship exists between the modified soft soil strength parameters and the oedometer modulus, as shown in Figure 3.7 (Olsson, 2010). 28 3. Theory Figure 3.7: The relationship between κ∗, λ∗, and M in the overconsolidated (OC) region and the normal consolidated (NC) region (Olsson, 2010). The creep parameter µ∗ is typically evaluated from IL oedometer tests. Nevertheless, in engineering practice in Sweden, these test are rare due to their high cost (Olsson, 2010). An alternative approach to evaluate µ∗ is first find the creep index r1 from Equation 3.25, where WN is the water content (Christensen, 1995). µ∗ can be calculated from Equation 3.26 (Olsson, 2010). r1 = ( 75 w1.5 N ) (3.25) µ∗ = 1 r1 (3.26) An second alternative approach is to evaluate the coefficient of secondary compres- sion at the apparent preconsolidation pressure (αs(max)) using the water content, as illustrated in Figure 3.8 (P.-E. Larsson et al., 1997). 29 3. Theory Figure 3.8: Coefficient of secondary compression αs(max) at the apparent precon- solidation pressure for different water content (P.-E. Larsson et al., 1997) αs(max) is used to calculate µ∗ according to Equation 3.27, as proposed by (Olsson, 2010). µ∗ = αs ln10 (3.27) A third alternative approach is based on a constant representing the secondary compression behaviour of soil, denoted as the ratio Cα/Cc (Mesri & Castro, 1987). Essentially, the method follows the same principle as the relationship proposed by (Christensen, 1995), where µ∗ is evaluated based on an r1. Nevertheless, the third method incorporates the ratio between the preconsolidation stress σ′ c and the vertical effective stress σ′ v, along with the oedometer modulus M . This approach aligns more closely with the method used to evaluate the other parameters, κ∗ and λ∗. Equation 3.28 below illustrates this, as adapted by (Olsson, 2010) to match the Swedish parameters. r1 = ML (0.04 ± 0.01) ∗ σ′ vc (3.28) The calculation of soil parameters for the soft soil model in PLAXIS involves deter- mining key factors such as the overconsolidation ratio (OCR), pre-overburden pres- sure (POP), and coefficients related to lateral earth pressure. From the apparent pre-consolidation pressure (σ′ c) and the vertical effective stress (σ′ c), the overcon- solidation ratio (OCR) and the pre-overburden pressure (POP) could be evaluated according to Equation 3.29 and 3.30 (Parry & Wroth, 1981). 30 3. Theory OCR = ( σ′ v σ′ c ) (3.29) POP = σ′ v − σ′ c (3.30) The soft soil models parameters OCR and POP can be assessed by plotting the in- situ effective vertical stress and the final evaluated trendline for (σ′ c) against depth in the same plot (Equation 3.9) (Karstunen & Amavasi, 2017). Figure 3.9: Effect from delayed consolidation and groundwater movements of ge- ological history on the pre-consolidation pressure (Parry & Wroth, 1981) The coefficient of lateral earth pressure at rest in the normally consolidated (NC) region, (Knc 0 ), is calculated using Jaky’s formula (Equation 3.31), as reported by Karstunen and Amavasi (2017). The coefficient of lateral earth pressure at rest for the overconsolidated (OC) state, (K0), is determined using Equation 3.32 (Swedish Institute for Standards, 2007). 31 3. Theory Knc 0 = 1 − sinΦ′ c (3.31) K0 = (1 − sinΦ′ c) √ OCR (3.32) R. Larsson et al. (2007) proposed another approach for calculating K0 that is more appropriate for uniform clays and accounts for the water content. 32 4 Methods This chapter presents the approach employed for developing a ground model for the construction site, evaluating soil properties for the numerical model, and validating the numerical model. The soil profile and its input properties constitute the foundation of the modelling process and are essential for obtaining reliable results. Consequently, significant time and multiple steps are required to establish a detailed, accurate ground profile with reliable characteristics. The methodology employed to achieve maximum precision and reliability is described in the Figure 4.1. The investigation methods utilised in the various stages of the methodology and the derivation of soil properties are outlined in the Figure 4.1. Many investigation methods produced similar parameters but with variations in the results. To ensure the highest possible accuracy, the tests were prioritised according to regulations from The challenge at the site was that only a limited number of borehole investigations and tests had been conducted. 32 boreholes were selected to create a representative 3D soil profile with as accurate soil properties as possible. In order to obtain an accurate analysis with high quality, the samples were evaluated based on three different methods (Terzaghi et al., 1996), (Karlsrud & Hernandez- Martinez, 2013), and (R. Larsson et al., 2007). Sample quality assessment evaluates how disturbed the triaxial and oedometer tests were. This evaluation was conducted to determine which samples should be prioritized and which are of poor quality. The quality of the samples was assessed using ϵv0, σ′ p, σ′ h0, and wN . 33 4. Methods Figure 4.1: The Methodology employed for soil profile and soil properties. 34 4. Methods 4.1 Development of a soil profile The development of a three-dimensional ground profile started with assessing the relevance of available field measurements on the site. A limited number of field tests conducted near the planned pipe were chosen and incorporated into the assessment. In the approach, all Cone Penetration Testing with pore pressure measurements (CPTU) and piston sampling across the entire study area were selected to provide an overall understanding of the ground profile and the soil behaviour. During this phase, the homogeneity of soil behaviour across the area was evaluated, and any variations were examined. The soil classification from CPTU was performed in line with the approaches involv- ing normalized soil behaviour type (SBTn) charts proposed by Robertson (1990), Robertson (2016), and Schneider et al. (2008). The initial assessment was performed by assessing SBTn charts and comparing density, water content, and liquidity index from piston sampler sampling. CPTU test indicated various sudden significant in- creases in qt and fs, as well as decreases in u. The approach recognized the layering of silt in the clay. As a result, a second approach from (Schneider et al., 2008) was employed, where a more detailed profile was developed for the area where settlement issues were observed. This decision was based on the characterisation of this area as having larger volumes of silt than the rest, resulting in differences in density and water content compared to the overall area. The outcome was a general 3D soil profile across the entire area and a more specific 2D profile for the area experiencing settlement issues. The data from boreholes closer to the excavation were weighted higher. 4.2 Evaluation of soil properties Piston sampling provided soil properties encompassing density (ρ), water content (wn), liquid limit (wL), sensitivities, and the preconsolidation pressure (σ′ c), which was calculated using Hansbo’s empirical method for both fall cone tests and field vane (FVT) tests according to Equations 3.15. The CRS tests provided the stiffness properties of the formations in the ground. The tests were evaluated according to Swedish Institute for Standards (1991a) and thereby provided the following properties: ρ, σ′ c, limit stress σ′ L, permeability k, and oedometer modulus (M0, ML, M ′). The pre-consolidation pressure (σ′ c) and compression modulus were important for further analysis and development of input parameters in the subsequent numerical analysis. Permeability k was assessed according to Swedish Institute for Standards (1991a) for the clay, where a trend-line that remained constant with depth was derived from the values obtained from CRS. A trend line was also drawn from the CRS values for the silt, but these values correspond more closely to those of the clay than those of the 35 4. Methods silt. Therefore, that value was considered to represent the clay between the layers of silt and a new empirical value for the silt was derived from (R. Larsson, 1995). The permeability of the silt was then calculated by determining the percentage of silt compared to clay from the three closest boreholes and based on soil classification according to (Schneider et al., 2008). Finally, a weighted value for the silt was determined, taking into account the behaviour of the sieved silt. Permeability for the fill, sand, and rock was evaluated based on TK GEO 13 and (R. Larsson, 1995). The parameters evaluated from the Triaxial shear test included ρ, wn, ϕ′, cu, c′, σ′ c, and σ′ h. The main parameters for further analysis and those whose final results were prioritised were ϕ′, σ′ c, and c′. Parameters calculated from the Direct Simple Shear (DSS) test included ρ, wn, cu, and σ′ c. The next step involved calculating the input parameters OCR, POP, Mc, κ∗, λ∗, µ, νur, Knc 0 , and K0 for Plaxis. To determine which POP and OCR to utilise, our values were evaluated and compared with our results, as shown in Figure 3.9. It suggests a constant POP and a decreasing OCR with depth due to changes in groundwater movements. Furthermore, only POP will be utilized in the PLAXIS analysis as its behaviour best corresponds to the soft clay in our area. Values for OCR and POP were determined for the silt and clay layers at the mid-depth of each layer, along with their corresponding σ′ c and σ′ c. K0 was calculated according to Equation 3.32 from (Swedish Institute for Standards, 2007). Further methods were considered, but they did not meet the site-specific requirements demanded by the soil properties. The value for the friction angle was evaluated using the critical state line (Mc). The critical state line was used to achieve the best-fitted model. Finally, the model-specific parameters for the soft soil model, κ∗, λ∗, µ, νur, were evaluated. An important consideration when evaluating κ∗ is that when assessing CRS oedometer tests within the overconsolidated zone, this often yields a value for κ∗ that is too low compared to field conditions (Sällfors, 1975). It’s crucial to consider that κ∗, λ∗, and µ∗ are based on mean values derived from σ′ v and σ′ c, rendering these parameters highly sensitive and reliant on accurate estimation of the mean value (Olsson, 2010). To mitigate this sensitivity, values are checked in soil tests and through sensitivity analysis. The parameter κ∗ is typically evaluated from the unloading/reloading path in a CRS test (Olsson, 2010). Nevertheless, no such test data were available for the construction site. This could result in an excessively large κ∗. 4.3 Calibration of soil parameters The material properties in the numerical model were validated to verify the cor- respondence between the tests and the numerical interpretation by PLAXIS. The validation procedure involved using the soil test module in PLAXIS for the critical 36 4. Methods layers, encompassing the clay and the silt layers. The numerical representation of the strength of the stiffness properties for a stratum was verified against one triaxial (CAUC) and one CRS test. The strength parameters (ϕ′ and c′ ref ) were calibrated against a triaxial test assumed taken on a specimen from the clay formation on the site. The calibration was performed in the p′ − q plane to match the Mc line of the modelled soil layer with the peak of the stress path from the triaxial test. An acceptable match was achieved by increasing ϕ and the σ′ 3 of the cell. The κ∗ was calibrated against the initial loading in the ϵ1−q plane of the triaxial test. Typically, the parameter is evaluated from the unloading/reloading path in a CRS test (Olsson, 2010). Nevertheless, no such tests were available for the construction site. As a result, this could potentially result in an excessively large κ∗ in the model. The λ in the model was calibrated against the ML from CRS tests taken in the formation. The fit was made using a higher λ∗ and shifting the curve to the right by slightly increasing the initial stress. The µ∗ in the representation of clay formation was verified against inSAR measure- ment from the site. The yearly creep rate is 0.49mm/year and are presented in Figure 4.2. Figure 4.2: The inSAR measurement of the vertical displacement of the site for the excavation. 37 4. Methods 4.3.1 Numerical modelling of structural elements The geotechnical structure encompass the sheet pile walls and the bracing. The walls was modelled as plate elements and material properties from Table 4.1. The bracing uses node-to-node anchors and properties from Table 4.2. Table 4.1: Material properties used for numerical modelling of the embedded walls. Parameters Unit Steal sheet pile wall, VL603, S355GP Material type - Elastic EA (kN/m) 2, 43E6 EI (kNm2/m) 31.20E3 d (m) 0.3924 w (kN/m/m) 1 v(nu) - 0.3 Table 4.2: Bracing Parameters Unit Bracing, HEB300, S355N/M/J2 Material type - Elastic EA (kN/m) 3,13E6 Lspacing (kNm2/m) 1 4.4 Method for assessing economic and environ- mental impacts The economic and environmental impacts were evaluated for four different solutions: Sheet pile wall, trench box (kojapo), Slope, and Steered JT drilling. The method used to calculate the economic effects of the four different solutions was based on Implenia’s own calculations for production costs. These calculations included all costs associated with each solution, such as the purchase of materials, machine costs, transport, and labor costs for the works. The method used to evaluate the environmental impact was based on the standard SS-EN 15978:2011 (Swedish Institute for Standards, 2011). The environmental im- pact of the construction was assessed for the product stage and construction process 38 4. Methods stage (A1-A5) as shown in Figure 4.3. The environmental impact was evaluated for three different solutions: Sheet pile wall, trench box (kojapo), and Slope. For the sheet pile wall, raw material supply by a manufacturer in Holland included the transport within the production and manufacture of the sheet pile according to the manufacturer’s own EPD. In the construction process stage (A4-A5) for the sheet pile, the transport from Holland to the project in Varberg was included, along with transport using excavators and trucks during the execution of the shaft and the backfilling, use of pumping, and finally the manufacture of the rock crusher used in the backfilling. This stage (A4-A5) was calculated according to two different cases defined by the Swedish Transport Administration: one for Earth excavation, case B road, and another for Earth excavation, case A road. In Case B, soil masses are considered contaminated and must be transported to special facilities, with a transport distance of 83 km to a facility in Gothenburg. In Case A, soil masses are classified as clean but cannot be used for any other purpose in the project, resulting in transportation of the masses to a disposal site 3 km away. The installation impact from the driving of the sheet pile is not included due to missing information. The material used in the backfill was evaluated as crushed rock sourced from this project in Varberg. When calculating this, the emissions for all values, except those in parentheses, are based on the Swedish Transport Administration’s climate calculation, which has been developed and adapted for Implenia’s project using their own reference values. For the trench box, two different cases were calculated. The first case used the same evaluated value as for the sheet piling in the product stage (A1-A3), and the second case excluded the influence from the product stage (A1-A3) to resemble a scenario where the trench box is rented out and therefore reused. In the construction process stage (A4-A5) for the trench box, transport from the company that rented out the trench box in Gothenburg to Varberg was calculated. In addition, the emissions for the trench box were included and calculated in the same way as for the sheet pile. For slopes, no product stage (A1-A3) was included because the rock crusher ma- terial used is included in the construction process stage (A4-A5). The difference evaluated for the slope is a gradient of 1:1, which is considered slightly too low. The slope results in twice the volume of the shaft and the backfill. The impact in the construction process stage (A4-A5) was calculated in the same way but with the new larger volume. 39 4. Methods Figure 4.3: Buildings LCA stages according to EN 15978 40 5 Results form geotechnical site characterisation 5.1 The geotechnical site investigation The geotechnical site investigation involved selecting high-quality soundings taken from the construction site and in an area extending 200 meters to develop a 3D ground profile. To obtain an accurate analysis, the quality of the samples was ex- tensively evaluated using three different methods to ensure that our main probes were of high enough quality to give an accurate result. CPTU-tests were princi- pally used to assess the stratification and pressure soundings to assess the level of firm ground/bedrock. The soil characterisation of the CPTU logs where performed according to Robertson (1990) and Schneider et al. (2008). For the three most rele- vant soundings (U05G16, U34G06 and U17G05) the results are presented in Figure 5.1. The CPTU logs losses of pore pressure on various occasions, which indicates a change in the soil macrostructure towards a more permeable deposit. This aligns with the results presented. Given the spread of soil types and the varying soil macrostructure, the approach proposed by Schneider et al. (2008) was implemented for soil characterisation. The result is presented in Figure 5.2. 41 5. Results form geotechnical site characterisation 1 10 100 1000 -0.600 -0.400 -0.200 0.000 0.200 0.400 0.600 0.800 1.000 1.200 Q t Bq Bq diagram Borehole U34G06 Bq 1.00 10.00 100.00 1000.00 0.1 1 10 Q t Fr Fr diagram U34G06 Serie1 1.00 10.00 100.00 1000.00 0.1 1 10 Q t Fr Fr diagram Borehole U05G16 Fr diagram 1.00 10.00 100.00 1000.00 -0.600 -0.300 0.000 0.300 0.600 0.900 1.200 Q t Bq Bq diagram Borehole U05G16 Bq 1.00 10.00 100.00 1000.00 -0.600 -0.400 -0.200 0.000 0.200 0.400 0.600 0.800 1.000 1.200 Q t Bq Bq diagram- Borehole U17G05 Bq 1.00 10.00 100.00 1000.00 0.1 1 10 Q t Fr Fr diagram – Borehole U17G05 Fr diagram Figure 5.1: Results for borehole U34G06, U05G16 and U17G05 according to the (SBTn) classification diagram based on Robertson’s (1990) normalized parameters 42 5. Results form geotechnical site characterisation 1 10 100 1000 Q t Δ𝑢2/𝜎𝑣0 Schneider´s (2008) first classification diagrams - Borehole U34G06 Δ𝑢2/𝜎𝑣0 diagram 0-5 m 5-10 m 10-15 m 15-20 m 1 10 100 1000 Q t Bq Schneider´s (2008) third classification diagrams - Borehole U34G06 Bq diagram 0-5 m 5-10 m 10-15 m 15-20 m 1 10 100 1000 Q t Bq Schneider´s (2008) third classification diagrams - Borehole U05G16 Bq diagram 0-5 m 5-10 m 10-15 m 15-20 m 1 10 100 1000 Q t Δ𝑢2/𝜎𝑣0 Schneider´s (2008) third classification diagrams - Borehole U05G16 Δ𝑢2/𝜎𝑣0 diagram 0-5 m 5-10 m 10-15 m 15-20 m 1 10 100 1000 Q t Bq Schneider´s (2008) third classification diagrams - Borehole U05G16 Bq diagram 0-5 m 5-10 m 10-15 m 15-20 m 1 10 100 1000 Q t Δ𝑢2/𝜎𝑣0 Schneider´s (2008) third classification diagrams - Borehole U05G16 Δ𝑢2/𝜎𝑣0 diagram 0-5 m 5-10 m 10-15 m 15-20 m Figure 5.2: results of Soil classification diagrams in two different plotting formats from(Schneider et al., 2008) 43 5. Results form geotechnical site characterisation 5.2 Sample quality After the soil profile had been determined, characteristic values for the soil layers were evaluated. To determine characteristic values for soil properties with high accuracy, the quality of the samples was examined using three different methods. The results for the quality of the samples are reported in Figure 5.3 and Table 5.1. The results showed that the triaxial samples were of very high quality and will therefore have high importance in further analysis. However, these samples were unfortunately located far from the pipe. According to the SQD, none of the CRS samples were of good quality, in contrast to the methods by (Karlsrud & Hernandez-Martinez, 2013) and (R. Larsson et al., 2007), which indicated samples of higher quality. Based on the results of the quality assessment and the samples’ locations, it was determined that samples U05G16, U34G06, and U17G05 were the most relevant and will be prioritized in further analysis. Figure 5.3: Result for sample quality with R. Larsson et al., 2007 method 44 5. Results form geotechnical site characterisation Table 5.1: Sample quality Sample Type ϵv0 (%) SQD Sample quality Ratio M0/ML Sample quality U05G12-3m CRS 3 C Poor 1.3 Poor U05G12-4m CRS 3 C Poor 2.9 Very good U05G16-5m CRS 3 C Poor 1.6 Good to fair U05G16-6m CRS 3 C Poor 2.4 Very good U05G13-3m CRS 3 C Poor 3.1 Very good U05G13-5m CRS 2.7 C Poor 2.3 Very good U05G13-9m CRS 4 D Very poor 2.1 Very good U05G17-5m CRS 5 D Very poor 1.6 Good to fair U05G17-6m CRS 3 C Poor 2.8 Very good U05G17-7m CRS 3 C Poor 2.0 Good to fair U05G18-5m CRS 5 D Very poor 1.4 Poor U17G05-4m CRS 3 C Poor 1.2 Poor U17G05-8m CRS 3 C Poor 2.3 Very good U05G13-5m Triaxial 1 B Good to fair - - U05G17-5m Triaxial 0.35 A Very good - - 45 5. Results form geotechnical site characterisation 5.3 Soil profile The developed ground profile is principally based on the soil characterisation of U34G06 and is presented in Figure 5.4. The soil layer denoted "Silt" is in fact a stratum containing a significant amount of pronounced variations in the macrostruc- ture. The global ground conditions of the site indicates that these variations are connected layers or larger lenses of coarser deposits in a finer-grained soil as the appear in multiple soundings logs. These layers have a thickness ranging from 0.2 m to several meters of soil type 1a, 2 and 3 according to Schneider et al. (2008). To capture the key feature being the higher permeability of this layer, it is modelled as a continuous silt layer. The layer denoted as "Friction material" has been indi- cated to be firm ground in pressure soundings and is thus thought of as a boundary condition with fixed deformation in the subsequent numerical analysis. The result for the produced 3D model with the final soil layer based on interpretations of all CPT according to the Schneider et al. (2008) method. The global ground profile was developed from the soundings compiled in Table A.1 in Appendix A. The evaluated soil samples are shown in Figure B.1, Figure B.2, Figure B.3, Figure B.4, Figure B.5, Figure B.6, Figure B.7, Figure B.8, Figure B.9, Figure B.10, and Figure B.11 in Appendix B. Silt Fill Sand Clay Bedrock +2m -1.5m -4.5m -16.5m -18.5m -23.5m +0.5m [m.a.s.l] WL+1m ρfill = 1.8 [t/m3] ρs.fill = 2.1[t/m3] ϕfill = 45° ρsand = 1.8 [t/m3] ρs.sand = 1.8[t/m3]ϕsand = 32° ρclay = 1.74 [t/m3] wn =43 [%] wL = 33 [%] St.clay = 15 OCR = 1.33 kx,y = 0.0605E-3[m/day] Selected material parameters kx,y = 864[m/day] kx,y = 8.64[m/day] ρSilt = 1.94 [t/m3] wn =30 [%] wL = 27 [%] St.silt = 15 OCR = 1.13 kx,y = 4.89E-3[m/day] Figure 5.4: The picture shows the developed soil profile based on the method proposed by Schneider et al. (2008) and the evaluated soil properties is shown. 46 5. Results form geotechnical site characterisation U34G06 U05G16 U34G04 U34G03 N122-07 U05G14 U05G13 14T2037 U05G15 U05G17 U05G18 U05G20 Figure 5.5: The illustrations show the developed soil profile in 3D from CPTu soundings, interpreted by empirical relations according to Schneider et al. (2008), and the geotechnical structure. 47 5. Results form geotechnical site characterisation 48 6 Results from the numerical analysis A numerical model was developed to analyse the unloading and reloading effects at the excavation bottom. The aim of the num