Abstract:
Tropical cyclones are intense storms that have in recent times incurred up to billions of dollars worth of damage in a number of developed regions of the world. The smaller island economies are also regularly devastated from frequent occurrences of such storms. Post-cyclone
investigations reveal that a significant proportion of damage is commonly suffered to low-rise buildings. While the issue of quality in construction dominates the prospects for mitigation of damage to low-rise buildings, a number of technical issues however confront the scientific community. In the present study, the issues dealing with the tropical cyclone wind characteristics and some aspects of building wind loads, particularly where "gaps" exist in current knowledge,
are investigated. Since wind loads are determined primarily by the wind, an investigation of the structure of the tropical cyclone wind is firstly made from a review of past measurements. This shows
conclusively that convective activity in such storms makes the wind unstable, causing turbulence intensities to be enhanced and velocity profiles to be flatter than those obtained in strong winds under neutral stability conditions. While the Australian wind loading code, AS1170.2-1989, makes special provisions for wind characteristics in cyclone prone regions, it is shown that there is however some scope for improvement. In particular, increases in turbulence intensities are
recommended and some suggestions are made with regards mean and gust wind profiles, and the form of the turbulence energy spectrum. Some unexplained phenomena regarding surface pressure fluctuations, such as the shear layer instabilities on the roof that cause extreme suctions, is revealed by further development of the quasi-steady theory to be caused by Reynolds stresses. Drawing on measured values of the gradient of surface pressure with wind azimuth and wind elevation, the theory suggests that Reynolds stresses may also suppress fluctuations in surface pressure quite significantly. This
immediately explains the apparent loss of mid-frequency response in roof pressures observed in full-scale measurements. In addition to the effects of distortion and blockage of turbulent eddies, it is shown that
the high frequency attenuation in fluctuating surface pressures is effected by the deflection of eddies with the mean flow around bluff-bodies. Furthermore, area pressure fluctuations that drive fluctuations in internal pressure through dominant openings, are shown to be additionally effected by the filtering effects of area-averaging that leads to further attenuation. Empirical equations are proposed for windward wall pressure admittance functions that account for the high-frequency attenuation.
The study on internal pressure begins with the application of computational and
experimental modelling techniques in the investigation of the transient response behaviour
following a sudden opening. This enables the determination of the previously ill-defined
parameters in the governing equation for internal pressure. It is shown that computational
modelling provides an accurate prediction of the Helmholtz frequency, and the damping
characteristics are estimated with fair accuracy. Computational flow visualisation is used to show
that the unsteady flow through the opening forms a vena-contracta which may persist through
most sections of the oscillating flow cycle. The Helmholtz resonance phenomenon in buildings with a dominant opening makes the
quasi-steady assumption invalid for internal pressures, since fluctuations are determined by the
frequency response. Model-scale experiments in the present study show that the quasi-steady
internal pressure coefficients (ie. coefficient of peak internal pressure to peak gust pressure) are
usually larger than the mean coefficients as a result of Helmholtz resonance. A gust-factor
method is thus developed for the calculation of peak internal loads. At oblique wind angles for
wall openings, experiments are used to demonstrate the importance of the previously unfamiliar
"Helmholtz resonance under oblique flow" phenomenon that results in a very severe response, so
that peak loads may approach those obtained under normal flow. Furthermore, predictions based
upon analytical models developed separately for the inclusion of the effects of internal
partitioning and building flexibility on internal pressure and the net load developed across the
envelope, are shown to agree well with experimental results.
While a study on the characteristics of internal pressure is important in itself, the manner
in which it combines with the envelope external pressure to produce the net load is of greater
significance in the design process. The characteristics of the net load developed across the roof
with wall openings, and across the wall with roof openings is therefore determined through
model-scale experiments. Correlation coefficients between internal and external pressures
measured to range from 0.4 to 0.8, can be applied with the gust-factor equations in the
calculation of the peak load across the envelope. An interesting outcome of this part of the
investigation is that the worst-case situation in terms of the net roof or wall load always occurs at
oblique wind angles and not under normal flow as has been assumed in past studies.
The effects of enhanced turbulence levels and internal pressure on the fatigue
characteristics of wind loads are studied using a design cyclone approach. A set of loading
sequences are proposed for wall and roof pressures, and a method is proposed for calculating the
accumulation of fatigue over several tropical cyclones. The present results indicate that the
loading sequences in the Australian wind loading code might not be conservative; a result that is
in agreement with other recent studies in Australia.
Finally, a set of conclusions and recommendations are drawn from the study both in terms
of fundamental interest, as well as for improvements in the relevant wind loading codes. These
include increased turbulence intensities in tropical cyclones, a gust-factor method for the
calculation of peak internal pressure and net envelope loads in the presence of dominant
openings, the effect of a number of variables on these loads, and fatigue loading cycles for net
envelope loads. The need for further investigation is highlighted, particularly on some aspects of
tropical cyclone winds, remaining issues on internal pressure resulting mainly from some
questions that have arisen, and fatigue characteristics of overall loads acting on the envelope.