In the case of foreign attacks, identifying the most sensitive regions in the American power grid while predicting cascade failure is a vital task with significant national interest due to the energy distribution and security concerns. Such large-scale networks are influenced dynamically by a substantial number of stochastic inputs, such as weather, aging, external attacks, composing a high-dimensional uncertainty space that propagates. Our stochastic models and sensitivity analysis we obtain reliable predictions of the network's failure uncertainty evolution and identify sensitive regions, facilitating decision making and preventive actions during catastrophic events. This study will break a new ground on identifying the most vulnerable regions also enhancing the reliability and resilience of power grid as a whole. This is of critical importance, where the environment and public safety are at risk. the novelty of our approach is that we carry out a 'detailed stochastic failure modeling' for a small yet representing component, e.g., electric wire/cable and a mechanical part subject to thermo-electro-mechanical load, and then, to develop a 'physics and math informed lumped element modeling' of failure and aging for the expanded and global system. Therefore, in our approach, the underlying physics and data-infused modeling would minimally be compromised as it is in the common practice. One graduate student will be supported in each year of this 3 year grant.
The power grid network in the United States is an engineering wonder of complexity, interconnectivity and robustness. However, the high connectivity can potentially lead to cascading failures with the loss of single components. The majority of disturbances is caused by natural events, such as storms, hurricanes, tornadoes, earthquakes. Still, one third of major blackouts have non-natural causes, including human errors, machine failure and intentional attacks. We formulate a new computational-mathematical phase-field based model for thermo-electro- mechanical failure analysis of generating points (sites) on the power grid and the power-transmission lines. Moreover, we develop a reduced-order model which can be in real-time fashioned examined subject to different working/weather conditions across the US. Subsequently, we will approximate the 'probability failure' of the prominent source/generating points also the aging grid network links. We study the propagation of failure uncertainty front through the power grid and identify regions of 'critically high risk' which would be affected by the malfunction of power grid. The model is a robust tool for faster decision making in case of blackouts, prevention of malicious attacks and in prudent, secure expansion of the already complex power networks. With data obtained through smart grid used in monitoring power grid operations, the probabilistic framework serves as a real-time asset in the prediction of rare catastrophic events. In terms of deliverables, in addition to published papers and organization of mini-symposiums in conferences and workshops, we will also provide properly documented open source codes accompanied with benchmark power-grid simulation examples.
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
