Accurate and timely assessment of the potential risks caused by natural hazards is critical for effective risk mitigation. 

This research topic focuses on developing a generalized framework for real-time hurricane, tsunami, and flooding risk assessment for coastal regions. Various challenges need to be resolved such as addressing the time variation of the responses or extending the approach to loss estimation for infrastructure systems. Also, relevant data collected in real events either through sensor networks or the public/crowdsourcing can also be explored through soft computing and data mining techniques to help establish predictive models and to facilitate further risk assessment.

In the end, real-time risk assessment tools will be developed to enable emergency response manager for risk mitigation planning and decision making.

For systems exposed to multiple hazards, analysis and design must take into account the effects of the competing demands different hazards have to ultimately offer enhanced resilience and robustness. Within this setting, performance-based design, allowing design for specified levels of performance under particular hazard scenarios, is well suited for this task.

​

This research topic will focus on developing a simulation-based framework that can incorporate various uncertainties in hazard load modeling and can address multi-hazard scenarios (including interaction of different hazards) and support multi-criteria decision making (since different hazards can result in different performance requirements). This research will particularly look into developing computationally efficient methods for assessing different performance quantifications in multi-hazard scenarios that ultimately represent different robustness aspects, such as damage (economic robustness), downtime costs (societal robustness) or material use (sustainability robustness).

Key problems that will be investigate are: life-cycle cost analysis of bridges in seismic and hurricane/flood prone zones; risk assessment of offshore wind turbines and oil platforms under wind loads and wave loads; resilience of lifeline structures and other critical facilities.

Tsunami evacuation is an effective way to save lives from tsunami. Realistic evacuation simulation can provide valuable information for accurate evacuation risk assessment and effective evacuation planning. This research investigates the use of agent-based modeling (ABM) for tsunami evacuation simulation of coastal communities, due to its capability of capturing the emergent phenomena, providing a natural description of a system, and being flexible.

Use Machine Learning (ML), especially deep learning (such as Generative Adversarial Network or GANs) for efficient inverse design of topology of metamaterials (periodic structures) with desirable properties, without the need to run expensive optimization.

​

ML can also be used for statistical down-scaling of low-resolution global models (e.g., climate models) to high-resolution regional models.

​

This topic examines layout optimization of buoy arrays (wave farm) in a random sea (described through JONSWAP spectrum), considering uncertainties in the characteristics of sea states. A multiple-scattering based numerical model is adopted to address interaction between buoys and predict the energy output of the array. A stochastic-simulation framework is utilized to propagate the uncertainties and predict the expected power output of the array. Then, the layout is optimized to maximize the power output using a global optimization algorithm, whereas to improve computational efficiency a kriging metamodel approach is established. 

Engineering systems operate under various sources of uncertainties, which must be properly incorporated in the design stage to ensure optimal/high system performance typically expressed through some probabilistic quantity such as optimal life-cycle cost, maximum reliability, or minimal downtime. Traditional design approaches frequently encounter difficulties in dealing with complex systems. Methods relying on stochastic simulation provide a powerful alternative; however, they still face significant computational challenges.

​

This research will investigate the integration of soft and high-performance computing and advanced simulation techniques for establishing versatile stochastic optimization algorithms. Applications include risk-informed multi-criteria design of floor-isolation systems for protection of non-structural components against seismic hazard.

Civil infrastructures suffer from aging and deterioration over time due to either regular operation, or extreme loading conditions, or environmental conditions. Aging and deterioration may considerably reduce the service life and reliability of infrastructures. Therefore, it is critical to consider and incorporate aging and deterioration in the analysis and design of infrastructures, especially regarding reliability, maintenance, life-cycle analysis, remaining service life etc. where the time varying properties and performances of infrastructure play an important role.

​

In general, two types of deterioration mechanisms can be identified, shock deterioration (hazard-related, e.g. due to extreme events such as earthquakes, hurricanes, floods) and gradual deterioration (environment-related). A system may be subject to multiple deterioration processes and also there might be interactions between them. While existing researches have focused on modeling individual deterioration mechanisms and assumed independence between them, this research effort focuses on developing a general stochastic framework to model the impacts of the different deterioration processes and their interactions, as well their impacts on the various (time-varying) performances of engineering systems.

Applications include modeling impact of deterioration due to earthquakes and corrosion on the time-varying fragility, reliability, and life-cycle performances of reinforced concrete (RC) bridges; Bayesian calibration of stochastic deterioration models based on experimental data (e.g., corrosion and ASR).

In the future, this research will also assess the synergistic effects between climate change and aging and deterioration of civil infrastructure (e.g., how climate change impacts the hazards, both in frequency and intensity, and further hazard-related shock deteriorations, and also on how it impacts the environmental variables that influence gradual deteriorations).

Condition assessment and maintenance of the aging infrastructure network have emerged as one of the most important challenges in civil engineering research (acknowledged by both ASCE and NAE as critical challenges of the 21st century). Bayesian updating provides a rational framework for supporting these tasks, with one of the biggest challenges being the computational difficulty in sampling from posterior distributions. The potential complexity of infrastructure models increases this challenge and hinders automated updating of condition and guidance of maintenance actions. To realize real-time, automated condition assessment, the Bayesian updating efficiency needs to be improved.

​

This research will explore integration of soft computing and data mining techniques (to build predictive models) and efficient stochastic sampling techniques to improve computational efficiency. The application will focus on developing real-time tools for rapid condition assessment and optimal maintenance of bridges, because there is an increasing need in the US for timely damage detection and optimal maintenance scheduling for bridge infrastructure. Though there has been a lot of attention in this field on hardware (wireless sensors) and damage detection algorithm developments, more efforts are needed to provide efficient tools that using such information can automatically update the assessment of the bridge condition and support decision making regarding maintenance in real-time. For example, these tools can facilitate bridge inspection/retrofitting scheduling, especially after extreme events/loading (earthquake, blast etc.).