Effects of changing ambient temperature on finite element model updating of the Dowling Hall Footbridge

Abstract

In this paper, effects of changing ambient temperatures on finite element (FE) model updating of the Dowling Hall Footbridge are investigated. The Dowling Hall Footbridge is located on the Tufts University campus in Medford, Massachusetts. The footbridge is equipped with a continuous monitoring system that records vibration and temperature of the bridge once an hour or when triggered by large vibrations. Natural frequencies, mode shapes, and modal damping ratios of the structure are extracted from measured ambient vibration data using an automated data-driven stochastic subspace identification algorithm. The identified natural frequencies and mode shapes are then used for calibration/updating of an initial FE model of the bridge. However, the identified natural frequencies show significant variability with changing ambient temperature. This variability propagates through the FE model updating process and therefore yields uncertainty in the FE model updating results. A static polynomial model is estimated to represent the relationship between identified natural frequencies and measured temperatures. This model is then used to “remove” the temperature effects from the identified natural frequencies. Two sets of FE models are updated in this study based on 17 weeks of hourly-identified modal parameters, before and after removing the temperature effects. The proposed approach is successful in minimizing the effects of changing ambient temperature on FE model updating of the Dowling Hall Footbridge. Accounting for the temperature effects in the FE model updating process reduces the variability of temperature-sensitive updating parameters and therefore decreases the probability of missed identification of damage.

Introduction

Major structural failures in recent years have focused public attention on the need for improved infrastructure monitoring and maintenance [1]. In January 2009, the American Society of Civil Engineers (ASCE) issued its latest Report Card for America’s Infrastructure [2], the fourth since 1998. This report asserts that our current infrastructure is poorly maintained, is unable to meet current and future demands, and is in some cases, unsafe. Deteriorating conditions and inflation have added hundreds of billions to the total cost of repairs, needed upgrades and replacements. In this report, bridges receive a grade of C. More than 26% of the nation’s bridges are either structurally deficient or functionally obsolete. An estimated $17 billion annual investment is needed to substantially improve current bridge conditions. As part of the solution, ASCE proposes that owners of the infrastructure should be required to perform ongoing evaluations and maintenance to keep them functioning at a safe and satisfactory level. To manage the nation’s infrastructure system, it is essential to understand the true state of structural health and rate of degradation of each significant bridge structure. This often cannot be determined from visual inspections alone. Vibration-based structural health monitoring (SHM) provides information that is complementary to visual inspections.

The basis for vibration-based SHM is that the dynamic parameters of a structure are functions of its physical properties (mass, damping, and stiffness). Therefore, changes in these physical properties due to “structural damage” will be reflected by changes in dynamic parameters such as natural frequencies, damping ratios and mode shapes. Numerous methods for vibration-based damage assessment of structures have been proposed in the literature. Extensive reviews on vibration-based damage identification have been provided in [3], [4], [5]. Sensitivity-based FE model updating is among these methods [6], [7]. In this method, the physical parameters of a FE model of the structure are updated to match the measured modal properties of the structure as damage evolves, and the updated modeling parameters are used to detect, locate, and quantify damage. In some recent studies, FE model updating methods have been successfully applied for damage identification of real-world, large-scale structures [8], [9], [10], [11]. However, the accuracy and spatial resolution of the damage identification results depend significantly on the accuracy and completeness of the identified modal parameters [12]. The estimation variability/uncertainty of the modal parameters can be influenced by several factors. One of the most important factors (and one of the few that can be measured) is changing environmental conditions, such as ambient air temperature [13], [14], [15], [16], [17]. Therefore, separation methods are needed to remove the effects of changing ambient temperatures from system identification (e.g., natural frequencies) and damage identification (e.g., model calibration factors) results. Even though researchers have underlined the importance of environmental effects in structural identification, little work has been done to quantify these effects on damage identification results.

The focus of this study is (1) to quantify the variation of FE model updating results for the Dowling Hall Footbridge induced by the measured ambient temperatures, and (2) to reduce this variation through removing the temperature effects from identified natural frequencies. The paper is organized in the following order. In Section 2, the Dowling Hall Footbridge and its continuous monitoring system are introduced. A brief review of the automated system identification process and modeling of the identified natural frequencies versus measured temperatures are provided in Section 3. In Section 4, the initial and reference FE models of the footbridge as well as the sensitivity based FE model updating process used in this study are reviewed. Two series of FE model updating are performed using the hourly-identified natural frequencies before and after removing the temperature effects. Variation in the FE model updating results before and after removing the temperature effects and a discussion of the observations are presented in Section 5. Finally, some concluding remarks are offered in Section 6.