Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)

Legacy Department

Civil Engineering


Kaye, Nigel B

Committee Member

Aziz , Nadim M

Committee Member

Khan , Abdul A

Committee Member

Pang , Wei C


This research focuses on an experimental and theoretical investigation of windborne debris emanating from loose gravel on built-up roofs. During severe storms, windborne debris can cause considerable physical harm and property damage. One of the major sources of flying debris in large commercial areas is loose gravel on built-up roofs. Such loose gravel can be responsible loss of life and significant property damage. Despite the high risk of windborne debris, their flight mechanics are poorly understood. To better understand windborne debris flight, a series of experiments were conducted in the Clemson University Boundary Layer Wind Tunnel. These experiments were designed to quantify the conditions under which gravel became airborne, the rate at which it was removed, and the resulting flight distance of the debris.
In order to conduct experiments in the Boundary Layer Wind Tunnel it is important to understand how to model the atmospheric boundary layer (ABL). A new curve fitting method is presented for calculating the ABL logarithmic velocity profile parameters i.e. shear velocity, surface roughness and zero plane displacement. The new method uses only the time averaged velocity profile and requires no iteration. Comparison with existing methods shows that the new approach has equal or better accuracy than existing curve fitting and geometric approaches with fewer calculation steps.
Debris flight is a highly stochastic process with uncertainty and variability in the debris particle the turbulent wind field. However, current models are almost entirely deterministic. A series of Monte Carlo simulations based on existing debris flight equations were run to quantify the impact if input uncertainty on flight outcome (flight distance and impact kinetic energy). Results indicate that failure to account for parameter variability will result in under predicting the mean flight distance and kinetic energy, and ignoring outcome variability / uncertainty. A full quantification of the relationship between input variability and outcome variability is presented for roof gravel blow-off.
A series of new experimental methods have been developed to measure the conditions under which blow-off occurs, the rate of gravel removal, and the downwind flight distance for two-dimensional buildings. The critical condition for blow off is parameterized in terms of the particle densimetric Froude number, particle Reynolds number and building geometry. A series of non-dimensional plots of the critical Froude number versus Reynolds number for different parapet heights are presented. The results indicate that the current approach for scaling result from laboratory to full scale is flawed and that full scale experiments are required to fully understand this process.
The rate of removal, or mass flux, varies over time. The removal process exhibits an initial high mass flux regime followed by a period of reduced blow-off rate. Dimensionless plots of both regimes mass flux versus Particle Froude number for different parapet heights are presented. The results show that increasing the parapet height usually decreases the mass loss rate, though this is not the case for very small parapets. Further, the transition time from the initial to secondary blow-off regimes is independent of the building geometry. Finally, the initial mass flux is approximately four times that of the secondary loss rate, and that this ratio is independent of both the building geometry and the Froude number.
Experimental results indicate that the wake behind the building dominates the downwind transport of debris. The flight distance is a function of the building height, particle Froude number (written in terms of a Tachikawa number), and the parapet geometry. A full characterization of the down-wind debris field requires a detailed analysis of the wake behind the building that is beyond the experimental capability of the current facility. Further, scaling of the results to full scale is again problematic, and therefore full scale testing is recommended.