Abstract:Biological soil stabilization through Microbial-Induced Calcite Precipitation ( MICP ) demonstrates the potential to be an eco-friendly and sustainable approach to enhance the strength properties of cohesionless granular soils. No systematic study has been performed so far in the literature on predicting the strength of bio-mediated sands prior to testing. In this study, to overcome this deficiency, based on a comprehensive set of experimental data compiled from literature, the amounts of Precipitated Calcium Carbonate ( PCC ) and Unconfined Compressive Strength ( UCS ) of MICP -treated sands are estimated through hybridizing Artificial Neural Network ( ANN ) and Adaptive Neuro-Fuzzy Inference System ( ANFIS ) models with metaheuristic algorithms. During the data extraction procedure, a total of 70 and 115 datasets are used to predict the PCC and UCS , respectively. To this end, five key features of urea concentration ( M u ), calcium chloride ( M cc ), optical density of bacterial suspension ( OD 600 ), pH and total volume of injection per volume of sample ( V t ), are first selected as the inputs of the hybrid models to estimate the amount of calcium carbonate precipitation within the porous structure of MICP -treated granular medium. The experimental data on the PCC along with the basic characteristics of the host sand including mean grain size ( D 50 ), uniformity coefficient ( C u ) and void ratio ( e ), are then adopted as the inputs of the numerical models in the second phase whilst the UCS of bio-mediated sands is set as the output. The performance and accuracy of implemented models are rigorously assessed through analyzing various statistical indices and performance criteria. According to the numerical analysis, the ANN -Particle Swarm Optimization ( PSO ) and ANFIS - PSO hybrid models show the best performance in predicting PCC and UCS of bio-mediated sands, respectively. Using the well-established Gene Expression Programming ( GEP ) algorithm, practical correlations are developed to predict the PCC and UCS values of MICP -mediated soils based on the characteristics of biological binding agents as well as the host material. The Machine Learning ( ML )-based models presented in this study provide engineers with a fruitful framework for the rough and preliminary estimation of the amount of calcite precipitation as well as the UCS of bio-mediated sands in the field prior to stabilization and performing complementary tests. Nevertheless, it is important to consider the limitations of extrapolation in the models presented in this research.

Prediction of hydraulic conductivity of sand with multivariate-index properties using optimal machine-learning-based regression models

Prediction of Soil-Water Characteristic Curve Using Artificial Neural Network Approach

Predicting soil hydraulic conductivity using random forest, SVM, and LSSVM models

Novel Insights in Soil Mechanics: Integrating Experimental Investigation with Machine Learning for Unconfined Compression Parameter Prediction of Expansive Soil

Comparing machine learning approaches for estimating soil saturated hydraulic conductivity

Machine Learning Prediction Model for Clay Electrical Conductivity and Its Application in Electroosmosis Consolidation

Hierarchical Bayesian Model for Predicting Small-Strain Stiffness of Sand

Predicting soil water retention curves using machine learning: A study of model architecture and input variables

Prediction of hydraulic conductivity of sodium bentonite GCLs by machine learning approaches

Predicting saturated and near-saturated hydraulic conductivity using artificial neural networks and multiple linear regression in calcareous soils

A simple model for predicting the hydraulic conductivity of MICP-treated sand

Estimating soil–water characteristic curve (SWCC) using machine learning and soil micro-porosity analysis

A modified model predicting the hydraulic properties of variably saturated soil accounting for film and capillary flow

Prediction of UCS of fine-grained soil based on machine learning part 1: multivariable regression analysis, gaussian process regression, and gene expression programming

Suggestion for a new deterministic model coupled with machine learning techniques for landslide susceptibility mapping

Prediction of UCS of fine-grained soil based on machine learning part 2: comparison between hybrid relevance vector machine and Gaussian process regression

Novel hybrid machine learning models including support vector machine with meta-heuristic algorithms in predicting unconfined compressive strength of organic soils stabilised with cement and lime

Modeling of electrical resistivity of soil based on geotechnical properties

Machine learning–based uncertainty modelling of mechanical properties of soft clays relating to time‐dependent behavior and its application

Predicting the precipitated calcium carbonate and unconfined compressive strength of bio-mediated sands through robust hybrid optimization algorithms

Novel Genetic Algorithm (GA) based hybrid machine learning-pedotransfer Function (ML-PTF) for prediction of spatial pattern of saturated hydraulic conductivity