Degree Name

Doctor of Philosophy


Faculty of Engineering


Plumes generated from hot surfaces may contain fumes and other contaminants that constitute major health and environmental hazards. Current design techniques of contaminated plume control for area sources of heat have limited applicability, and provide little information on plume characteristics such as vertical velocity distribution, density distribution, etc. in the zone near the source.

This study is an experimental and computational (CFD) investigation into the fundamental processes of plume generation and dispersion from a hot area source, such as a hot metal bath, blast furnaces etc. The main aims are to provide quantitative data and theoretical models that will enable engineers/designers to greatly improve the efficiency of exhaust systems and reduce exposure of workers and the community to harmful contaminants.

The main aim of this project was to conduct a fundamental investigation into the complex processes of plume generation from a hot surface of finite size and its subsequent dispersion. Particular objectives included: 1. Use of experimental techniques to investigate plume characteristics such as velocity and density/temperature fields. 2. Computational fluid dynamics (CFD) analysis of velocity and temperature fields. 3. Validation of the CFD simulations, using analytical solutions and specific experiments of increasing complexity.

The experimental work was conducted in two distinct programs. In the first part, saline plumes generated from a uniform area source descending into quiescent water were studied. This represented a ‘cold flow’ analogy of thermal plumes ascending in quiescent air. A 1.4m cubed environmental tank was designed, built and commissioned in which experiments were carried out. Experiments are described in which the velocity and salt concentration profiles in the plume were measured. Two-dimensional velocity profiles were determined using a PTV (Particle Tracking Velocimetry) technique. The plume shape was recorded on video tape and also analysed using the DigImage PTV analysis software. A special conductivity probe with digital position control was used to measure the concentration of salt solution (or plume density distribution) both inside and outside the plume at different levels within a 1.4m cubed environmental tank. In the CFD simulations, two geometries were used to determine plume characteristics and were compared with experimental results.

In the second program, thermal plumes generated from a uniform temperature area source were investigated inside a glass enclosure. Both vertical and horizontal velocity profiles were measured using the LDV (Laser Doppler Velocimetry) technique. Fine thermocouples were used to measure the temperature distribution of in the plume inside the enclosure. CFD analysis of confined and unconfined thermal plumes was carried out.

Saline plume experiments showed that the plume flow from an area source can be modelled as two regions, one from the area source to the plume neck and the second region beyond the neck. The saline plume radius was measured using the Shadowgraph and the Particle Tracking Velocimetry (PTV) technique. The plume radius determined by the Shadowgraph technique and image analysis was greater than the plume radius determined by the vertical velocity analysis by approximately 50%.

It was found that the Gaussian equation is suitable as a model of the vertical velocity and density distributions where the plume reaches the self-similarity region but in the region near an area source, a proposed modification to the Gaussian equation showed a better fit than the standard Gaussian equation. The centreline vertical velocity as a function of height by experiment showed two distinct regions and this supports the work of Colomer, et al. (1999). In the first region, the present experimental results showed the same pattern as the model of Colomer, et al. but in the second region the model of Colomer, et al. is in doubt because the value of their centreline velocity increases with distance away from the area source, which is in contrast to the present experimental results and contradicts to a decreasing velocity that one would expect on dimensional grounds.

In the experimental study of thermal plumes in an enclosure, the spread of the plume vertical velocity radius was found to be wider than the plume temperature radius. The Gaussian equation also provided a good fit to the vertical velocity and temperature data in the far-field region.

The modified Gaussian equation was used to fit the data in the near-field region. In the numerical study, it was found that the choice of the value of constants in the turbulence model had the effect on the results of simulation. For example, when the value of Cε3 and the value of turbulent Prandtl number (Prt) were set to 0.6 and 0.65, respectively the numerical results showed a good match with experimental results of both saline plumes and thermal plumes.

Plumes from the ‘hot’ processes such as metallurgical operations (e.g. foundries, furnace tapping, charging), are often generated from area sources and such plumes frequently do not reach a self-similar state due to the space over the ‘hot’ processes is not high enough. Therefore, the main aim of the present study was to determine plume characteristics, such as temperature or density distribution, velocity profile and plume width in the near-field region where the plume profiles are not self-similar.

The results of this study provide new information on the fundamental behaviour of plumes from area sources and will assist in the improvement of the design procedures for industrial ventilation systems and other building ventilation system applications.

02Whole.pdf (4181 kB)