top of page

Multiscale Modeling of Flood Protection Infrastructure


Resilience-Based Design of Flood Protection Infrastructure (FPI) is a topic of vital importance for out communities for several reasons: 1) infrastructure systems are vital lifelines for the Country, 2) the general state of the Country’s FPI demands for immediate action and has dramatically proven to grant insufficient protection to communities (see Figure), 3) the expected increased frequency of natural disturbances acting on these systems requires an overall improvement in design, construction and maintenance practices. The importance of this topic is testified by the fact that the all the major Science and Engineering Academies in the world recognize this problem as one of the most critical challenges to be addressed in the next years by our society.


Recent flooding events striking Edgecumbe (left) and Wellington (right) communities

In our research, we aim to bring disruptive improvements in the design and maintenance of Flood Protection Infrastructure. These systems are generally designed with a local approach that involves the evaluation of the geometrical and mechanical parameters of the system in the most critical region, usually defined as a 2-Dimensional cross section. The so-obtained design is then simply repeated throughout the structure under the assumption that the Factor of Safety (FoS) of the designed section represents a lower bound for its value throughout the system. This approach, therefore, neglects all 3-dimensional and structural effects that may amplify the system response to natural hazards and it is therefore against safety. Our numerical tools allow to perform large full-scale simulations by means of a Machine Learning Based Multiscale Modeling technique. The results obtained from the numerical models are embedded in a resilience analysis toolset specific for FPI that integrates physical failures from erosion or over topping with societal impacts. It also connects FPI with other community lifelines and infrastructure to enable the prediction of cascading failures and community impacts. The framework can allow stakeholders to test different failure scenarios and act as a planning tool to provide a reference for determining where remedial action is most needed.


Graphical interpretation of the proposed multiscale approach


Comparison between the detailed numerical model and the proposed surrogate Machine Learning based approach 

bottom of page