A framework for estimating the dynamic response of structures subject to thunderstorm downbursts

Song, Jing ORCID: 0000-0003-3269-1865 (2022). A framework for estimating the dynamic response of structures subject to thunderstorm downbursts. University of Birmingham. Ph.D.

Song2022PhD_Redacted.pdf
Text - Redacted Version
Restricted to Repository staff only until 13 July 2024.
Available under License All rights reserved.
Download (96MB) | Request a copy

Abstract

Thunderstorm (TS) downburst is a type of extreme wind event that are statistically non-stationary with rapid variation in time. Therefore, they cannot be regarded as the same as the stationary boundary layer wind for the purposes of engineering design. In this research, a framework to evaluate the influence of TS downburst wind load on structures has been established. Firstly, a theoretical model, named the vortex model, is introduced and shown to be capable of simulating the process of six full-scale TS downburst events. These six sets of full-scale TS downburst data were recorded by the Texas Tech University Wind Engineering Research Field Laboratory (WERFL) and the Joint Airport Weather Study (JAWS). The simulations of the vertical distribution of horizontal wind speed are validated with the data from WERFL and JAWS and demonstrate good consistency. Secondly, a combination of the quasi-steady (QS) theory and concepts from the partial turbulence simulation (PTS) method has been applied to estimate the wind load on generic building shapes arising from downburst types of events. Using this approach, the wind-induced pressures arising from an isolated downburst-like flow on the building surface can be estimated by the dynamic pressures of a specific downburst event and a database of pressure coefficients for wind flows with a range of azimuths, elevation angles and turbulence intensities; the latter was measured during the experimental part of this study, using a standard, atmospheric boundary layer wind tunnel (WT). In these experiments, the building models were positioned on a rotational support in uniform flow which was produced by a passive turbulence generator located at different upstream positions to obtain different turbulence intensities. As preliminary validation of the technique, velocity data from the University of Birmingham Transient Wind Simulator (TWS) has been combined with these pressure coefficients to predict the wind pressure time series, which is compared to those measured directly in the TWS. The results demonstrate that the predicted pressure time series have a similar tendency compared with the TWS measurements; and the root-mean-square (RMS) errors of pressure time series between these are less than 50 Pa which is less than one-third of the maximum pressure of TWS. Thirdly, the concept of the design spectrum is adopted subject to thunderstorm downburst. The present investigation builds on previous research to develop wind design spectra (WDS), addressing the potential occurrence of TS downbursts and their effect on structures. This algorithm revisits random vibration theory for assessing the dynamic response of the oscillators that filters input acceleration induced by wind, to then conform design spectra of structures exposed to TS downbursts. A parametric analysis was completed to establish the variability of outflow velocity, structural dimensions, mass, and natural frequency, and their influence on the design spectra, in addition to variations induced by different damping ratios and terrain categories. The outcome of the referred spectral technique is then used to estimate the dynamic performance of a benchmark building. The results are further validated by comparing them with their equivalent calculation based on numerical integration methods, fed with simulated wind time series. The simulated wind field is also applied to the benchmark building whose modelling enhances with finite element software. The superiority of the characteristics of this technique enables the effective and rapid application for the engineering structural design with variable mechanical and aerodynamic properties. According to the built algorithm for the thunderstorm downburst wind design spectra (TWDS), the generalisation of the spectra is carried out through non-linear regression modelling, which can be regarded as an alternative to estimate the dynamic response of structures subjected to transient non-stationary winds. Finally, this research presents a computational method for investigating the ductility demands imposed by transient non-stationary wind loads induced by TS downburst to steel building structures that have been previously damaged during natural hazard events (earthquakes, winds). The new numerical framework enables an inter-phase between the finite element (FE) analysis software ABAQUS and MATLAB reverse optimisation subroutines. A nonlinear static pushover analysis is initially performed to impose the ductility demands due to the primary hazard event. The relationship between the base shear force and lateral roof drift is obtained from the linear elastic range up to a target plastic deformation. The method uses the pushover analysis to bring the structure to any pre-defined ductility level, including the descending branch of the load-deformation curve, thus allowing for a better control on the initially imposed damage. Then, the transient wind loads can be subsequently applied as an externally applied dynamic loading and the new displacement ductility demand can be directly determined. Accordingly, the global plastic deformation mechanism that leads to collapse can be identified. The proposed computational algorithm is applied to a tall steel building structure subjected in its damaged stage to eleven downburst synthetic records. The results demonstrate the non-negligible effects of such wind events on the ductility demands.

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):

Supervisor(s)	Email	ORCID
Martinez-Vazquez, Pedro	UNSPECIFIED	UNSPECIFIED
Skalomenos, Konstantinos	UNSPECIFIED	orcid.org/0000-0002-0734-3992
Jesson, Michael	UNSPECIFIED	orcid.org/0000-0002-6798-1371
Sterling, Mark	UNSPECIFIED	orcid.org/0000-0003-2119-592X