Research Projects

Earthquakes are among the most destructive natural hazards, disproportionately impacting low- and middle-income countries through large-scale fatalities and economic disruption. Effective risk mitigation in seismically active regions hinges on the development of robust seismic risk and damage assessment frameworks that integrate hazard, exposure, and vulnerability/fragility. While tools such as OpenQuake and the Global Earthquake Model (GEM) have advanced the modeling of seismic hazards, fragility functions tailored to regional construction practices remain scarce. This limitation is especially critical in countries like India, where informal and non-engineered construction practices are widespread, often deviating significantly from code-compliant standards. Indian seismic hazard modelling has evolved considerably, but there is still a significant void in the availability of fragility functions for the country’s diverse and regionally distinct building typologies. Most available fragility functions have been derived using disparate methodologies and are not representative of ground realities, especially in seismically vulnerable hill states like Himachal Pradesh. The current practice of adopting global fragility models (e.g., FEMA-HAZUS, GEM) without calibration introduces further uncertainties in risk estimates. Moreover, existing exposure datasets are outdated, and rapid urban growth has further complicated risk assessment. This project aims to address these challenges by developing a comprehensive, region-specific seismic risk quantification tool for Himachal Pradesh. The work is structured across three integrated phases: Phase I: Field Data Collection and Documentation Using a stratified cluster-based approach, the project will collect data on informal construction practices across urban, semi-urban, rural, and remote clusters in Himachal Pradesh. Through interviews with masons, contractors, and engineers, the project will document material choices, reinforcement practices, and structural detailing. Non-destructive testing will also be conducted to quantify material strength. Historical damage data from the 2015 Mw7.8 Nepal and 2011 Mw6.9 Sikkim earthquakes will be compiled to support model validation. Phase II: Multivariate Fragility Function Development The collected data will inform the development of building typologies and the identification of critical structural parameters. A building index will be computed based on statistical weighting of these parameters. Representative building models will undergo nonlinear dynamic analysis (e.g., IDA using SAP2000 or OpenSees), validated with past damage data. Fragility surfaces conditioned on both ground motion intensity and the building index will be generated and benchmarked against global standards. Phase III: Computer Vision and Risk Assessment Tool A machine learning-based classifier (e.g., CNN or EfficientNet) will be trained to identify building typologies from street-view images. These classifications will be mapped to corresponding fragility curves within a user-friendly GUI application. The application will allow end-users to input location-specific imagery or use automated Google Street View downloads. Integrated with hazard datasets, this app will perform real-time seismic risk assessments. The project hypothesizes that incorporating region-specific data into multivariate fragility models, coupled with automated image-based classification, will significantly enhance the accuracy and usability of seismic risk models. This framework is expected to enable data-driven vulnerability mapping at scale, supporting applications in urban planning, retrofitting prioritization, and disaster preparedness. By standardizing fragility modeling for informal RC structures and leveraging AI for exposure mapping, the proposed research promises substantial advancements in seismic risk assessment for Himachal Pradesh and other similarly vulnerable regions across India.