Structural performance assessment of electrical transmission networks for hurricane resilience enhancement

Abstract

Hurricanes are one of the main causes for blackouts and related infrastructure damage in the United States. Electrical transmission towers, which are key parts of the electrical transmission networks, are vulnerable to high wind speeds during storms. Collapse of transmission towers may lead to a loss of functionality of transmission lines. This research focuses on regional analysis of electrical transmission networks under hurricane hazards through developing beam elements for analyzing transmission towers, selection of hurricane wind records that incorporate uncertainty quantification, generating collapse fragility curves for transmission towers, and regional damage assessment of transmission networks.Electrical transmission towers are usually made of steel angles. To address the collapse limit state that these structures encounter due to hurricanes, it is important to capture their geometric and material nonlinear response. In the large deformation regime, the steel angle members have coupled axial-flexural-torsional deformations due to the Wagner effect and a non-coincident shear center and centroid. Three-dimensional corotational total Lagrangian beam elements are formulated and implemented in the OpenSees software to consider these coupling effects by invoking Green-Lagrange strains in the basic system that continuously translates and rotates with the element. The formulations try to minimize the number of elements needed through remedying membrane locking, representing nonlinear curvature within an element, and utilizing a kinematic model in the basic system that decouples axial, flexural, and torsional deformations for the first order effect. The fiber section method with uniaxial constitutive laws is adopted to account for material nonlinearity. The elements are validated against a set of experimental and numerical examples. To estimate the response of electrical transmission towers with considering uncertainties of hurricanes, a suite of wind records representing the hurricane hazards for a given location is of great interest. Such a suite consists of many hurricane wind records, which may lead to a high computational cost for the resulting structural analyses. To overcome this issue, this research proposed a machine learning technique to select a representative subset of all collected wind records for a location. First, hurricane wind records, which are expressed as time series with information on both wind speeds and directions, are collected from a synthetic hurricane catalog. Then using an artificial neural network designated as an autoencoder, the high-dimensional wind records are compressed into a set of low-dimensional latent feature vectors, to which a k-means algorithm is applied for clustering. To form the suite of records, a subset of records is selected from each cluster. To address that hurricane wind records are site-specific, a region can be discretized into a set of grids with the proposed hurricane selection procedure applied to the centroid of each grid. This idea is demonstrated using Massachusetts as a testbed. This research then introduces a framework for developing event-based fragility curves that describe the collapse probabilities of transmission towers after a hurricane. The storm maximum gust speed is selected as the intensity measure and incremental dynamic analysis (IDA) is adopted to model collapse using the newly developed beam elements. Uncertainties in wind speeds, directions and durations are accounted for by running IDAs with the selected hurricane wind records. The fragility curve is assumed to be the cumulative distribution function of the collapse capacity, which is the intensity measure at the onset of collapse. As is done in the hurricane selection, a set of fragility curves are also developed grid by grid for transmission towers in the eastern Massachusetts region considering their different locations, orientations, and voltage levels. The final stage of this research involves performing a regional damage assessment of an electrical transmission network in eastern Massachusetts, in which the simulated hurricane wind fields are combined with the developed collapse fragility curves to determine the failure probabilities of geographically distributed transmission towers. The failure probability of each transmission line is then obtained by assuming a series system of multiple towers. Damage assessments of this transmission network are conducted for several hurricane events as examples. These results thus produce a strategy for predicting the damage to the electrical power grid under hurricanes in coastal regions.--Author's abstract