Ioakim, Andreas
(2025)
Model updating frameworks for the estimation of lateral soil-pile interaction parameters.
PhD thesis, University of Nottingham.
Abstract
Resilient infrastructure systems, particularly in safety-critical applications such as offshore wind turbines (OWTs), demand advanced monitoring frameworks that can accommodate complex dynamics and inherent uncertainties. Structural Health Monitoring (SHM) is central to this goal, ensuring the operational integrity of OWT foundations while supporting safety, efficiency, and sustainability. This thesis advances the field of SHM by integrating physics-based modelling with data-driven techniques, combining the reliability of numerical simulations and the adaptability of data-oriented methods. This hybrid approach overcomes the limitations of purely data-driven methods, which can produce accurate but physically inconsistent predictions, and purely physics-based models, which can be computationally expensive and sensitive to parameter uncertainties.
Focusing on soil-pile interaction (SPI), this work addresses key challenges in estimating critical system parameters that influence the dynamic response and structural integrity of pile foundations. The thesis develops and evaluates deterministic and stochastic model updating frameworks, each designed to handle varying data availability and changing environmental or operational conditions. The deterministic framework employs frequency-response functions (FRFs) using input-output data, while the stochastic frameworks utilise output-only data and Modal Assurance Criterion (MAC)-based updating. These frameworks are validated using numerical simulations, with field tests conducted in selected scenarios.
Throughout the thesis, numerous challenges are systematically addressed. These include achieving both physical interpretability and computational feasibility, maintaining accuracy under measurement uncertainties, ensuring robustness against varying operational conditions, and promoting scalability to real-world OWT systems. Addressing these challenges requires careful selection of an appropriate modelling approach, including model reduction strategies to balance accuracy and efficiency. The choice of parameters to be estimated, objective function formulation for robust performance, and selection or development of suitable optimisation algorithms are also critical factors. Additionally, the frameworks are designed to be applicable under operational conditions, where environmental and loading uncertainties must be accounted for.
This thesis demonstrates that hybrid SHM frameworks, integrating data-driven and physics-based methods, enhance the predictive accuracy and reliability of OWT foundation monitoring. The findings emphasise the role of vibration data in model updating for estimating operating parameters while reducing uncertainty. These frameworks offer a structured approach to quantifying uncertain parameters and assessing the condition of pile foundations under dynamic uncertainties. By bridging theoretical insights with practical frameworks, this research contributes to the advancement of SHM for pile foundations.
Actions (Archive Staff Only)
 |
Edit View |
Tools
Tools
