Developing a Framework for Estimating Spatial Geotechnical Properties with a case of Metro Manila, Philippines

• Main Author: Galupino, Joenel
• Format: text
• Language: English
• Published: Animo Repository 2022
• Online Access: https://animorepository.dlsu.edu.ph/etdd_civ/2
• Institution: De La Salle University

As each construction project requires to conduct a geotechnical site investigation, reports are also generated consequently. These reports were just usually stored unused after the construction of the project is completed. It is in this premise that this study was conceptualized, to collect and maximize these data, and to generate a guideline for collecting, storing and estimating geotechnical parameters, which may be used for future projects. Collected geotechnical investigation data and empirical data can be repurposed for generating models that estimate spatial geotechnical properties at other sites with no data. Therefore, it is also significant to have a methodology that provides a procedure for creating models that can assist in the future design of structures. To address this, a new framework was developed for estimating the spatial geotechnical properties of nearby undisturbed areas by incorporating machine learning algorithms to create models. Through a code/program, models for ground elevation, groundwater elevation, SPT N-value, fines content, soil type, and atterberg limits have been developed. Moreover, empirical data were collected and unified to come up with calibrated empirical models that would further process the generated site-specific geotechnical data into important spatial geotechnical properties such as soil strength and soil behavior. Maps and profiles were delineated by deploying the models in multiple locations throughout a case region. Nomographs were also created as a summary of the calibrated models. To test the developed framework, it was applied to Metro Manila, Philippines as a case study with overall accuracy rate is 82%, indicating a very strong relationship between the models and the estimated parameters.