PROOF ONLY 1 2 SDMT-Based Numerical Analyses of 3 Deep Excavation in Soft Soil 4 Alessandra Di Mariano 1 ; Sara Amoroso 2 ; Marcos Arroyo 3 ; Paola Monaco 4 ; and Antonio Gens 5 1 5 Abstract: The paper explores the application of conventional (DMT) and seismic (SDMT) dilatometer tests to an important case of deep 6 excavation design. The work presents finite-element analyses simulating a deep excavation close to Barcelona (Spain). A thick layer of soft 7 interbedded sandy and silty soils made characterization based on laboratory testing very difficult. SDMT offered an alternative for estimating 8 the soil stiffness and its stress-strain dependency. Numerical results and high-quality monitoring data show quite close agreement for most 9 phases of the construction process, supporting the use of seismic dilatometer tests in numerical analyses of deep excavations. The paper also 10 indicates the importance of incorporating stiffness data at low strains. FE analyses involved some uncertainties derived from the presence of 11 jet-grouting soil treatments. On this point, a parametric study illustrates the effects of different modeling approaches. DOI: 10.1061/(ASCE) 12 GT.1943-5606.0001993. © 2018 American Society of Civil Engineers. 13 Author keywords: Nonlinear soil stiffness; Seismic dilatometer; Finite-element analysis; Deep excavation; Monitoring; Jet-grout. 14 Introduction 15 Deep 8 excavations 9 in urban areas generally induce ground move- 16 ments, which may damage adjacent structures. These movements 17 are quite sensitive to a number of factors, such as soil mechanical 18 properties, excavation geometry, retaining wall characteristics, con- 19 struction sequence, and construction methods. Accurate evaluation 20 of soil movements is an important aspect of managing third-party 21 risks (Arroyo et al. 2007). Numerical analyses are presently the tool 22 of choice for estimating deep-excavation-induced displacements 23 (Yoo et al. 2014; Dias and Bezuijen 2013). They are particularly 24 necessary when design includes special features (e.g., ground 25 improvement) that are poorly represented in empirical databases 26 (Ou 2016). 27 It is generally acknowledged that soil stiffness nonlinear 28 dependency on strains 10 should be properly taken into account in 29 the analysis of geotechnical structures (Burland 1989). This is es- 30 pecially true for the specific case of deep-excavation-induced 31 displacements (St John et al. 1993; Hashash and Whittle 1996; 32 Jardine et al. 2005; Brinkgreve et al. 2006; Finno 2010; Ou 2016). 33 Several constitutive models can capture 11 the dependence of stiffness 34 on stress-strain levels in a realistic manner, and many of them are 35 readily available in commercial software. However, calibration of 36 a significant number of parameters is a necessary step for their 37 application. Generally, model calibrations can be based on labora- 38 tory testing, but obtaining good stiffness data from it 12 requires high- 39 quality samples and careful testing procedures (e.g., Cho and Finno 40 2009). For gravels, sands, and silts, obtaining unaltered samples 41 for testing in laboratories is a very difficult task. Inverse analysis 42 of monitoring measurements is a rational alternative when data is 43 scarce (Ledesma et al. 1996; Calvello and Finno 2004; Hashash 44 et al. 2006). However, inverse analysis and soil testing work best 45 together (Finno 2010). Trial sections (Arroyo et al. 2007) are ideal 46 for model calibration, but when they are not available the model has 47 to be adjusted as the excavation proceeds. 48 When laboratory testing is difficult, one possible alternative is to 49 rely on in situ tests. Not all in situ tests are equally suitable for this 50 purpose. The relatively cheap standard penetration test (SPT) is a 51 strength-related test with poor repeatability. Khoiri and Ou (2013) 52 state that stiffness parameters obtained from excavation back- 53 analyses may not be correlated to SPT. Self-boring pressuremeter 54 tests (SBPM) can be used to fit a whole stiffness degradation curve, 55 particularly when including unload-reload loops (Jardine 1992; 56 Fahey and Carter 1993). Nevertheless, they are quite sensitive to 57 operational details and are not frequently available. Two in situ tests 58 that are repeatable, easily available 13 , and allow the derivation of soil 59 stiffness decay curves are the seismic cone penetration test (SCPT) 60 and the seismic dilatometer test (SDMT). Indeed, both of these tests 61 allow the obtaining 14 of small-strain modulus measurements from 62 seismic shear wave velocities. While it is very difficult to obtain 63 reliable operative stiffness values from cone penetration tests 64 (CPTs 15 )(Been et al. 2010), values deduced from dilatometer tests 65 (DMTs) can actually give good settlement predictions (Monaco 66 et al. 2007, 2014). 67 Several researchers (Mayne et al. 1999; Lehane and Fahey 2004; 68 Marchetti et al. 2008; Amoroso et al. 2013, 2014; Cox and Mayne 69 2015; Pepe et al. 2015; Rodrigues et al. 2016; Bosco and Monaco 70 2016) have presented procedures to calibrate stiffness degradation 71 curves using seismic dilatometer tests. However, relatively little 72 work has investigated the performance of SDMT-calibrated stiff- 73 ness values in numerical analyses. Arroyo et al. (2008) presented 74 initial results for a trial section in a cut-and-cover railway tunnel in 75 Barcelona, but both monitoring and SDMT results were incom- 76 plete. Later, Sau et al. (2012) reanalyzed the case with improved 1 2 Staff Scientist 3 , International Center for Numerical Methods in Engineering, Barcelona, Spain. 2 Researcher, Istituto Nazionale di Geofisica e Vulcanologia, L ’Aquila, Italy (corresponding author). ORCID: https://orcid.org/0000-0001-5835 -079X. Email: sara.amoroso@ingv.it 3 4 Professor, Dept. of Civil and Environmental Engineering, Geosciences Division, 5 Universitat Politècnica de Catalunya (UPC), Barcelona, Spain. 4 6 Associate Professor, Dept. of Civil, Architectural, and Environmental Engineering, Univ. of L ’Aquila, L ’Aquila, Italy. 5 7 Professor, Dept. of Civil and Environmental Engineering, Geosciences Division, Universitat Politècnica de Catalunya (UPC), Barcelona, Spain. Note. This manuscript was submitted on September 11, 2017; approved on July 16, 2018 No Epub Date. Discussion period open until 0, 0; se- parate discussions must be submitted for individual papers. This paper is part of the Journal of Geotechnical and Geoenvironmental Engineering, © ASCE, ISSN 1090-0241. © ASCE 1 J. Geotech. Geoenviron. Eng.