Case Study: Real-Time PLAXIS Calibration for a Deep Excavation in High-Speed Rail with DAARWIN

This case study presents the application of DAARWIN, a cloud-based geotechnical platform, to perform real-time calibration of a PLAXIS numerical model during the excavation of a deep retained cutting for a high-speed rail project. The goal was to enhance confidence in design parameters, optimise temporary works, and implement the Observational Method at full operational scale.

Project Overview

The case focuses on a deep retained excavation forming part of a high-speed rail corridor. The structure included diaphragm walls, ground anchors, and temporary props supporting excavation depths up to 20 metres.

The geological profile consisted of alternating layers of stiff overconsolidated clays and dense sands, with a fluctuating groundwater table influencing both stability and deformation.

Given the variable ground conditions and sensitivity of nearby infrastructure, instrumentation and monitoring (I&M) was installed to capture wall movements, surface settlements, anchor loads, and pore pressures throughout construction.

Objectives of the DAARWIN Deployment

The main objectives were to:

• Establish a digital link between the PLAXIS numerical model and site monitoring data.

• Use DAARWIN to perform automated real-time model calibration (backanalysis).

• Quantify ground stiffness and strength parameters based on actual site performance.

• Apply the Observational Method to make data-driven adjustments during construction.

Methodology

Model Setup and Data Integration

The finite element model, developed in PLAXIS 2D, simulated the staged excavation and retaining wall system. The model included advanced soil constitutive behaviour to represent the stiff clay and sand strata.

Monitoring data—comprising inclinometer profiles, piezometric levels, settlement points, and strain gauges—were automatically uploaded into DAARWIN’s cloud platform, synchronised directly from the site database.

Observation points within DAARWIN were then matched to the corresponding PLAXIS calculation nodes, allowing direct comparison between predicted and measured responses.

Automated Back analysis and PLAXIS Calibration

Using machine learning algorithms, DAARWIN performed iterative calibration of the PLAXIS model. The system automatically adjusted key parameters, including:

• Undrained shear strength for cohesive strata.

• Young’s modulus and stiffness degradation relationships.

• Interface friction and wall-soil interaction properties.

  
  

Each iteration compared the simulated PLAXIS output against observed field data

until convergence was achieved between measured and predicted responses. The resulting calibrated soil parameters represented a more realistic reflection of in-situ conditions.

Real-Time Feedback and Decision Support

DAARWIN provided interactive dashboards visualizing wall deflections, surface settlements, and pore pressure evolution in real time.By continuously comparing the PLAXIS model predictions with measured performance, the engineering team could confirm that the excavation behaviour remained within design limits and make informed adjustments to the construction sequence when needed.

This closed feedback loop allowed real-time validation of design assumptions, reducing reliance on conservative estimates and supporting agile, evidence-based decision-making on site.

Results and Insights

Improved Understanding of Ground Behaviour

The calibration process revealed that the in-situ stiffness of one clay layer was significantly higher than initially assumed. Updating the PLAXIS model with this refined modulus improved the match between predicted and observed wall movements, reducing horizontal displacement discrepancies by over 30%.

Optimisation of Temporary Works

With enhanced confidence in the calibrated PLAXIS model, the design team safely reduced reliance on certain temporary props, enabling faster excavation sequences and lower material usage. This optimisation led to measurable cost and carbon savings, while maintaining full compliance with stability and deformation criteria.

Enhanced Implementation of the Observational Method

Through the integration of real-time monitoring and model calibration, DAARWIN provided a structured, auditable implementation of the Observational Method.

Stakeholders could visually track performance, validate assumptions, and agree on modifications transparently—transforming the method from a reactive philosophy into a proactive digital process.

Discussion

This case demonstrates the power of combining numerical modeling (PLAXIS) with digital backanalysis (DAARWIN) to reduce geotechnical uncertainty during construction.

By linking predictive analysis with real-world monitoring data, engineers can continuously update their understanding of ground behaviour, leading to smarter and leaner designs. The approach bridges the gap between theory and reality—turning monitoring data into actionable intelligence.

Furthermore, the ability to run real-time calibration in the cloud makes the Observational Method feasible for large, data-intensive projects, supporting better collaboration across design, construction, and monitoring teams.

Conclusions

The application of real-time PLAXIS calibration through DAARWIN in this high-speed rail excavation demonstrated that:

• Continuous backanalysis enhances understanding of actual ground behaviour.

• Model calibration allows data-driven optimisation of temporary works.

• The Observational Method can be executed effectively through digital workflows.

• The combined use of PLAXIS and DAARWIN provides a robust, scalable framework for geotechnical risk management in complex infrastructure projects.

  
  

This study highlights the transition from static geotechnical models to adaptive, data-informed design, where numerical models evolve with the ground’s real response—resulting in safer, faster, and more sustainable construction.

 Book your free demo here Back Analysis | DAARWIN