You are here

Prediction of Wind Induced Damage Using Prior Knowledge and Monitored Data

Title: Prediction of Wind Induced Damage Using Prior Knowledge and Monitored Data.
239 views
197 downloads
Name(s): Alduse, Bejoy Paruthyvalappil, author
Jung, Sungmoon, professor directing dissertation
Liang, Zhiyong Richard, university representative
Vanli, Omer Arda, 1976-, committee member
Mtenga, Primus V., committee member
Rambo-Roddenberry, Michelle Deanna, committee member
Florida State University, degree granting institution
College of Engineering, degree granting college
Department of Civil and Environmental Engineering, degree granting department
Type of Resource: text
Genre: Text
Issuance: monographic
Date Issued: 2014
Publisher: Florida State University
Place of Publication: Tallahassee, Florida
Physical Form: computer
online resource
Extent: 1 online resource (132 pages)
Language(s): English
Abstract/Description: Wind induced damage is observed in different types of civil engineering structures. There are several methods in use to predict damage. Researchers and stakeholders rely on these methods to quantify damage, which helps to schedule maintenance and to estimate financial loss. These damage prediction methods utilize the knowledge on properties of the wind or the wind load resistance of the material that constitutes the structure. However, recently, researchers have pointed out several shortcomings in these approaches. One such shortcoming is the inability of these methods to address the uncertainty in the data. A typical method for damage prediction rely on the accuracy of statistic of the wind load or the material property used in the analysis. If uncertainty exists in the data, then the statistic obtained from the data will give overconfident inferences. As a result the final predicted damage, will be biased and will not reflect the uncertainties involved in the actual data. In this research, an approach is proposed to enhance the damage prediction model. In order to address the uncertainties in damage prediction, the approach integrates monitored data and existing knowledge, which gives probabilities of damages rather than a single number. The advance in sensors and wireless technologies has enabled much easier access to high-quality monitored data. The monitored data can be used to enhance the accuracy of damage prediction. While using monitored data, the proposed approach also seeks to fully utilize existing damage prediction models. These models provide a strong framework based on theories of mechanics and knowledge gained from many years of research. In order to integrate existing damage models and additional monitored data, a Bayesian approach is adopted. The Bayesian approach provides a sound framework for integrating the existing model and the additional data. In the Bayesian approach the existing model is termed as the prior. The prior is systematically updated using additional monitored data, in order to provide the posterior. In this research two case studies are considered. These are complete sealant failure of three tab asphalt shingles under wind load and fatigue damage of slender structures due to turbulence and wind structure interaction. In case of asphalt shingles, wind vulnerability is determined using a sensor based strength monitoring and integrating the existing data. The sealant in the shingle, helps to resist the wind load acting on the shingle. After installation of asphalt shingle, the sealant deteriorates over time and loses bond with the shingle. Consequently the wind uplift capacity is reduced and larger area of the shingle is exposed to higher wind load. A complete failure of sealant due to the wind load acting on it is defined as the failure of the shingle. A sensor mechanism is proposed to monitor the deterioration of the sealant and wind vulnerability of the asphalt shingle. Existing knowledge and monitored data is integrated to estimate the uplift capacity and the wind load acting on the shingle. The vulnerability of the shingle at each wind speed is expressed in terms of the sensor reading. MC simulation is carried out to determine the failure contour on the roof and fragility curves of roof at different ages. It is observed that, the fragility curve for a 2% area of roof failure at 100 mph for a 10 year old roof from this study compares well with the results of fragility of roof cover from Cope, 2014. In case of long span bridges, the wind data from existing and monitored data are integrated to determine the possible statistic of wind data and damage is predicted using this data. Accuracy of fatigue damage prediction depends on the accuracy of the wind speed and direction statistic. Conventional approaches rely on initial wind statistics only, which result in a single fatigue damage value. The proposed approach systematically updates the prior and wind statistic using the monitored data of wind for one year. This is used to determine the possible values of wind speed and direction statistic at the location. Fatigue analysis provides the probability distribution of different fatigue damage values. A long span bridge and long span beam were studied using the conventional and proposed approach. For the long span bridge, the fatigue damage from conventional approach is 0.002 and the mean fatigue damage from proposed analysis is 0.002. For the long span beam it is 0.392 and 0.397 respectively. The results from the proposed approach will give the designers and retrofitters a comprehensive view of the possible values of damage at any location on the bridge, thus helping in planning a maintenance task.
Identifier: FSU_migr_etd-9226 (IID)
Submitted Note: A Dissertation submitted to the Department of Civil and Environmental Engineering in partial fulfillment of the requirements for the degree of Doctor of Philosophy.
Degree Awarded: Fall Semester, 2014.
Date of Defense: October 28, 2014.
Keywords: Asphalt shingles, Bayesian approach, Damage prediction, Long span bridges, Monte Carlo simulation, Wind engineering
Bibliography Note: Includes bibliographical references.
Advisory Committee: Sungmoon Jung, Professor Directing Dissertation; Arda Vanli, Committee Member; Primus V. Mtenga, Committee Member; Michelle Rambo-Roddenberry, Committee Member.
Subject(s): Civil engineering
Persistent Link to This Record: http://purl.flvc.org/fsu/fd/FSU_migr_etd-9226
Owner Institution: FSU

Choose the citation style.
Alduse, B. P. (2014). Prediction of Wind Induced Damage Using Prior Knowledge and Monitored Data. Retrieved from http://purl.flvc.org/fsu/fd/FSU_migr_etd-9226