Dorothy Reed

Methodology for Assessing the Resilience of Networked Infrastructure

In this paper, we outline a method to characterize the behavior of networked infrastructure for natural hazard events such as hurricanes and earthquakes. Our method includes resilience and interdependency measures. Because most urban infrastructure systems rely on electric power to function properly, we focus on the contribution of power delivery systems to post-event infrastructure recovery. We provide a brief example of our calculations using power delivery and telecommunications data collected post-landfall for Hurricane Katrina. The model is an important component of a scheme to develop design strategies for increased resilience of urban infrastructure for extreme natural hazard scenarios.

ANALYSIS OF UNIFORMLY AND LINEARLY DISTRIBUTED MASS DAMPERS UNDER HARMONIC AND EARTHQUAKE EXCITATION

Abstract The effectiveness of multiple mass dampers has been investigated by Igusa and Xu [Dynamic characteristics of multiple tuned mass substructures with closely spaced frequencies. Earthq. Engng Struct. Dynam. 21 (1992) 1050–70], Yamaguchi and Harnpornchai [Fundamental characteristics of multiple tuned mass dampers for suppressing harmonically forced oscillations. Earthq. Engng Struct. Dynam. 22 (1993) 51–62], Abe and Fujino [Dynamic characterization of multiple tuned mass dampers and some design formulas. Earthq. Engng Struct. Dynam. 23 (1994) 813–35] and Kareem and Kline [Performance of multiple mass dampers under random loading. J. Struct. Engng 121 (1995) 348–61]. In this paper, we extend the results of these previous investigations to examine the performance of uniformly and linearly distributed multiple mass dampers, respectively. These systems were selected to ascertain whether the distribution of masses located close to the central mass damper would influence the performance of the entire system in reducing vibration. We evaluate performance numerically through assessing the effectiveness and robustness of each system, as well as considering the effects of redundancy, under harmonic excitation. In this regard, we evaluate the performance of the system when certain individual dampers do not function. We show that the uniformly distributed mass system is more effective in reducing the peak dynamic magnification factor. The linearly distributed system is more robust under mistuning. It is more robust to damping variation for low damping values but the effectiveness of the two systems converges as damping increases. The uniformly distributed system is slightly more reliable when an individual damper fails. The eleven mass system is optimum for both configurations for harmonic excitation. The 21-mass system is more effective in structural vibration decay in both cases for the El Centro earthquake simulation.

A non-linear numerical model of the tuned liquid damper

The Tuned Liquid Damper (TLD) is modelled numerically as an equivalent tuned mass damper with non-linear stiffness and damping. These parameters are derived from extensive experimental results described in References 1 and 2. This Non-linear Stiffness and Damping (NSD) model captures the behaviour of the TLD system adequately under a variety of loading conditions. In particular, the NSD model incorporates the stiffness hardening property of the TLD under large amplitude excitation.

Autoregressive representation of longitudinal, lateral, and vertical turbulence spectra

Abstract A method for the simulation of fluctuating wind velocity time histories based on a combination of time series models and existing expressions for longitudinal, lateral, and vertical turbulence spectra is outlined. Related expressions for calculating the integral scale of turbulence, x Lu , are presented. Calculated values of x Lu using this method are close to those obtained by conventional methods from various field data.

Energy Supply System Performance for Hurricane Katrina

In 2005, wind-induced structural damage from Hurricane Katrina resulted in extensive energy supply disruptions across the Gulf of Mexico. In this paper, we examine the resilience of these systems with a particular emphasis on the electric power delivery systems. We investigate the correlations of the power outage data with weather parameters such as wind speed, rainfall, and storm surge. We examine the resilience and recovery of the electric delivery system spatially and temporally and compare the results with other natural disasters. We conclude that the behavior of the delivery systems under hurricane loadings is affected by rainfall and storm surge as well as wind speed. Widespread failures may occur at wind speeds less than predicted peak values.

Tuned liquid dampers under large amplitude excitation

Abstract Rectangular tuned liquid dampers were investigated through shaking table tests and numerical modeling for large amplitude excitation. The results confirm earlier work undertaken for small amplitude excitation and considerably contribute to the body of knowledge describing tank behavior under large amplitude excitation. The random choice method of solving the fully nonlinear shallow-water-wave equations captures the underlying physical phenomenon adequately, including wave breaking, for most of the frequency range of interest. The tank behaves as a hardening spring system due to the liquid sloshing motion, and this trend is enhanced as excitation amplitude increases.

Influence of non-Gaussian local pressures on cladding glass

Abstract The importance of the non-Gaussian nature of local pressure fluctuations is assessed through a numerical procedure to evaluate glass cladding damage. The dimensionless damage function developed by Reed and Fuller [1] is used in conjunction with a pressure distribution developed by Matsui et al. [2] to assess this influence. It is shown that a significant difference in damage estimation occurs for a turbulence intensity of 20% on rectangular cladding panels. These limited results suggest that the damage incurred during storms where large pressure fluctuations occur may be considerably greater than that estimated using a Gaussian assumption. This conclusion confirms the earlier work of Holmes [3] who used a different approach.

Fast vortex methods for predicting wind-induced pressures on buildings

The random vortex method is a Lagrangian particle-based numerical simulation scheme especially appropriate for fluid flows characterized by high Reynolds numbers and complex geometries. When coupled with a fast solver for computing vortex interactions, it becomes ideally suited for wind engineering simulations. Fast vortex methods have several advantages over grid-based methods because they do not suffer from numerical diffusion, are simpler to implement particularly as the geometry becomes more complex, and are likely to exploit the architecture of distributed-memory computers more effectively. In this paper, initial results are presented for a fast vortex two-dimensional simulation of the atmospheric boundary layer wind flow around a bluff body under open terrain conditions. The results compare favorably with full-scale measurements for the well known instrumented Texas Tech Building.

Expert systems in wind engineering

Abstract Knowledge-based expert systems (KBES) have generated a great deal of excitement in the engineering community. Many civil engineering applications have been identified. Recent developments in the wind engineering field include WINDLOADER and WISER. Wind-induced damage assessment is also under investigation. Because many of the design and analysis problems facing wind engineers are ill structured, it is anticipated that expert systems will play an important role in the wind engineering field in the next decade. The objective of this paper is to explore the use of expert systems in wind engineering.

Structural safety assessment

Abstract A prototype knowledge-based system for structural safety assessment with emphasis on wind effects is under development. The system consists of three main modules: a rule-based module for guidance and wind engineering background information of a qualitative nature; a module for causal network modeling of the structure under consideration; and another for the numerical evaluation of the causal network. Heuristics for wind loading and damage records are being incorporated into the rule-based module for including qualitative aspects of the loading formulation and structural safety assessment. The causal network has been introduced previously as an alternative approach to estimating the probability of failure of a multicomponent assemblage such as a structure from individual component probability of failure values. The knowledge-based system is being developed on the Symbolics 3640 in a lisp-based environment.