Reassessing the Berlin M1 Pit Excavation
- deepexcavation
- Jul 14
- 4 min read
MQ4 section Reanalysed Using DeepEX FEM Analysis
Introduction
Urban deep excavations present unique geotechnical challenges, especially when undertaken in heterogeneous soils like Berlin sand. A landmark study by Nikolinakou et al. (2011) examined the performance of the 20-meter-deep M1 excavation pit adjacent to Berlin's Hauptbahnhof station, using the advanced MIT-S1 soil model within a finite element framework. This project stands out as a benchmark for performance-based design and soil-structure interaction modelling in granular soils.
At DeepExcavation LLC, we revisit this iconic case using DeepEX with our integrated Finite Element Method (FEM) capabilities. Our goal is to re-create the excavation model (section MQ4) within DeepEX, simulate the staged construction process, and compare the computed wall deflections against both the field measurements and those predicted in the original study.
Project Overview
The M1 pit was excavated underwater to a depth of around 20 meters (18 meters in the MQ4 section). The support system consisted of a perimeter diaphragm wall and a single row of prestressed tiebacks. The site is underlain by medium to dense Berlin sands, and a significant portion of the original analysis was dedicated to calibrating advanced soil parameters based on extensive lab and in-situ testing.
The original study emphasized the impact of two critical parameters on model accuracy, namely, the initial earth pressure ratio (K₀), and the in-situ void ratio (e₀ - 0.6 for S0, 0.53 for S1, and 0.4 for S2), as shown in Figure 1.
 |
Figure 1 – Typical cross-section with soil profile. |
The soil is characterized by the following stratigraphy: the friction angle (ϕ'), 31º for S0, 34º for S1, and 37º for S2.
These were estimated using field dynamic penetration tests and local geological history. Instrumentation data from multiple sections revealed mostly consistent deflection trends, with one outlier due to local variability in soil properties.
Description of the analysis
The MQ4 section is composed by a wall 1.20 meters thickness and 24.80 meters height. The tieback is characterized by the properties presented in Table 1.
Table 1 - Tieback anchors' properties.
Lfree (m) | Lfixed (m) | sh (m) | inclination (º) | tendons | As (cm2/m) | P kN/m | D (mm) | E (GPa) |
30.5 | 8 | 1.5 | 19 | 8 | 9.68 | 213 | 15.2 | 210 |
DeepEX FEM Model
The MQ4 section excavation is recreated within DeepEX using the FEM module. The key modelling steps include (Figure 2):
§ Geometry & Construction Staging: The 18-meter-deep excavation, diaphragm wall configuration, and tieback placement were modelled based on published details.
§ Soil Profile Input: The Berlin sand layers were defined using representative stratigraphy, and material parameters were set according to those reported in the original study. The Fill soil is admitted having a friction angle of 30º.
§ Soil Model: We used the built-in effective stress model in DeepEX FEM to approximate the behaviour of Berlin sand.
§ Groundwater Conditions: A hydrostatic water profile was applied to reflect the submerged excavation environment.
 |
Figure 2 – Numerical model with geometry and the last step of the analysis (DeepEX). |
Methodology
The analysis was conducted in staged excavation steps to closely simulate the actual construction sequence. Prestressed tiebacks were activated at the appropriate depths, P = 213 kN/m. Key assumptions were:
- Drained behaviour due to the granular nature of Berlin sands.
- Plane strain condition.
- Initial stresses computed using the K₀ method.
The FEM solver in DeepEX provided wall displacements (Figure 3), ground deformations, and support loads at each stage (e.g., Figure 4, for the last excavation stage).
Results and Discussion
Our results revealed:
Wall Deflections: Predicted maximum horizontal displacements closely matched the field measurements for most sections, with deviations within 10–15%, echoing the accuracy reported in the 2011 study.
§ Maximum wall deflection at first prestressing (in cm): measured = 0.9, prediction = 0.2, present study = 0.77. The results from DeepEX presents a close agreement with the measured by instrumentation, for a ratio of 0.77/0.9 = 0.86.
§ Maximum wall deflection at excavation (in cm): measured = 3.4, prediction = 2.6, present study = 2.34. The results from DeepEX presents a close agreement with the measured by instrumentation, for a ratio of 2.34/3.4 = 0.69.
Support Loads: Tieback forces were reasonably predicted, validating both the structural modelling and soil-structure interaction captured by DeepEX.
§ Maximum axial force (in kN/m): measured = 270, prediction = 300, present study = 278. The results from DeepEX are considerably close to the measured by instrumentation, for a ratio of 278/270 = 1.03.
Critical Sensitivities: As in the original study, small variations in the initial K₀ and e₀ values significantly impacted deflection magnitudes, highlighting the need for precise parameter estimation.
 |
Figure 3 – FEM mesh with the representation of the wall deflection in colour grading (DeepEX). |
Notably, the section MQ4 exhibited higher wall deflections in the field and could not be replicated with more than a close agreement, reinforcing the original authors' (Nikolinakou et al., 2011) assertion about spatial variability effects.
 |
Figure 4 – Wall deflection and maximum anchor reaction (DeepEX). |
 |
Figure 5 – Wall deflections, surficial settlements for measured and predicted by numerical analysis and prestressing force (Nikolinakou et al., 2011). |
Conclusion
This revisitation of the M1 Berlin pit case study demonstrates that DeepEX, with its integrated FEM capabilities, can effectively replicate complex excavation behaviours when supported by well-characterized soil parameters. The results reaffirm the importance of detailed geotechnical investigations and the sensitivity of FEM predictions to basic state parameters like K₀ and e₀.
By leveraging DeepEX, engineers can confidently undertake deep excavation modelling with a high degree of fidelity and insight into critical risk areas. This case also highlights the potential of re-analysing classic case studies using modern, commercial tools to refine understanding and improve best practices.
References
Nikolinakou, Maria, Whitttle, Andrew, Savidis, Stavros and Schran, Ute. (2011). Prediction and Interpretation of the Performance of a Deep Excavation in Berlin Sand. Journal of Geotechnical and Geoenvironmental Engineering. 137. 1047-1061. 10.1061/(ASCE)GT.1943-5606.0000518. DOI: https://doi.org/10.1061/(ASCE)GT.1943-5606.0000518
Let us show you how to reduce your design time by up to 90%!
