Dissertation Defence: AI-Assisted Framework for Advanced Creep Modeling of Embankments on Soft and Organic Soils with Application to Microbially Induced Calcite Precipitation (MICP)
August 6 at 9:00 am - 1:00 pm
Ahmed Moslem, supervised by Dr. Sumi Siddiqua and Dr. Ramy Saadeldin, will defend their dissertation titled “AI-Assisted Framework for Advanced Creep Modelling of Embankments on Soft and Organic Soils with Application to Microbially Induced Calcite Precipitation (MICP)” in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Civil Engineering.
An abstract for Ahmed Moslem’s dissertation is included below.
Examinations are open to all members of the campus community as well as the general public. Registration is not required for in-person exams.
Abstract
Soft and organic soils present persistent challenges in geotechnical engineering due to their high compressibility, low shear strength, and pronounced time-dependent deformation. With growing interest in sustainable and efficient ground improvement techniques, microbially induced carbonate precipitation (MICP) has emerged as a promising bio-cementation method to enhance soil strength and reduce long-term settlement.
This study proposes an integrated data-driven framework that combines field monitoring, laboratory testing, AI-assisted parameter estimation, and numerical modelling to evaluate embankment performance over organic soils. A comparative analysis of seven constitutive models—Mohr-Coulomb (MC), Hardening Soil (HS), HS with Small-Strain Stiffness (HSsmall), Soft Soil (SS), Soft Soil Creep (SSC), CreepSCLAY1S, and Modified Cam-Clay (MCC)— was performed using field-monitored data. The SSC model was identified as the most suitable for capturing long-term settlement and excess pore water pressure behaviour.
To automate SSC parameter calibration, a Random Forest Regression (RFR) model was trained on 54 organic soil samples to predict modified compression (λ*), swelling (κ*), and creep (μ*) indices from basic soil properties. One-dimensional consolidation simulations validated the model’s predictive accuracy, with feature importance analysis highlighting the dominant role of the creep index (Cα).
To enhance creep modelling, a stress-dependent formulation for Cα was developed based on a stage-wise interpretation that unifies Hypotheses A and B, distinguishing early, intermediate, and late creep phases. This formulation was implemented into the SSC model and validated using case studies from Ballina, Australia, and British Columbia, Canada. Simulations showed significantly improved agreement with observed settlement and excess pore water pressure trends, capturing both stress-sensitive and time-dependent behaviour.
PLAXIS 3D simulations were conducted using SSC parameters calibrated from oedometer tests on untreated and MICP-treated soils from Bolivar Park, British Columbia. Embankments ranging from 1.5 to 8.0 m in height over 1.5 m and 12.0 m thick organic layers were modeled, with regression-based interpolation used to generate predictive design charts. Results indicated that MICP treatment, particularly when combined with unloading strategies, can reduce long-term settlement by up to 43%. The proposed framework is computationally scalable and field-applicable, offering a robust tool for predictive geotechnical design over soft and organic soils.
