Degree Name

Doctor of Philosophy


Department of Mechanical Engineering


Natural convection in enclosures arising from localised heating and cooling of side walls and/or the floor is relevant to many engineering situations. Passive solar heating of buildings is one situation in which such natural convection processes play an important role. This thesis includes a description of experimental, analytical and computational studies carried out on water-filled scale models of small rooms. Two types of rooms were investigated: i) one with a conventional heater mounted below a cold window on the same wall (other sides being adiabatic), which is referred to as the "colliding boundary layer model" throughout this thesis; ii) the second room considered had a part of the floor heated and a cold window located on one of the vertical walls, henceforth referred to as the "heated floor model".

This work was undertaken to clarify the nature of the complex flows arising in enclosures with localised heating and cooling and in particular to model experimentally and theoretically the thermal stratification arising from colliding natural convection boundary layers. This was achieved in the following three stages: i) experimental investigation, development of an analytical flow element model and iii) simulation of the problem using a computational fluid dynamics (CFD) package.

The major achievements of the present work include the following

i) Measurement of three-dimensional temperature and fluid flow fields in waterfilled enclosures was successfully achieved using necessary thermocouple technology and particle tracking velocimetry (PTV), respectively.

ii) Stratification in the colliding boundary layer model was found to differ from the results predicted by the earlier numerical models of the same situation.

iii) The experiments provided useful quantitative data on the complicated phenomenon of "fingering".

iv) A simple flow element analytical model was developed using boundary layer equations, which modelled the stratification in the enclosure reasonably as comprising two mixed layers.

v) problems were simulated using a commercial CFD software package PHOENICS. The temperature predictions using PHOENICS matched the experimental results reasonably well, but a substantial quantitative difference was observed in the experimental and CFD velocity fields.

The first chapter of this thesis covers a general the subject and points the objectives of the work. The second chapter overviews earlier research work carried out by others in related areas, including natural convection with localised heating and cooling of surfaces. This is followed by a discussion of various aspects of passive solar heating and a review of literature on velocity and temperature measurements in natural convection.

Design and the fabrication of the scale models is discussed in Chapter 3. Three dimensional temperature measurements were performed using very fine thermocouples and two velocity measurement methods were used i) a pH indicator flow visualisation technique using phenolphthalein and ii) particle tracking velocimetry (PTV). PTV was found to be the more useful technique both quantitatively and qualitatively.

Chapter 4 reports on the experimental results followed by discussion of these results. Experiments were carried out for a constant window temperature of about 12°C and the heater temperatures were varied between 40 and 60°C which was equivalent to Rayleigh numbers up to order IO1'. Both enclosures were found to be stably stratified which was contradictory to some recent computational studies by other researchers. The effect of Rayleigh number on velocity field and thermal stratification is also discussed.

A flow element analytical model developed for the colliding boundary layer situation is discussed in Chapter 5. Established correlations were used to represent the various flow elements in the enclosure (boundary layer flows, laminar plumes or turbulent plumes). The model predicted a two-layer stratification which gave good agreement with the experimental results.

A commercial CFD software package (PHOENICS) was used to simulate the enclosure flows reported in Chapter 6. These computational results were compared with the present experimental results and computational results of other researchers. For complicated boundary conditions such as localised heating or cooling of a single surface, the CFD results were found to be at odds with the present experimental results.

In the concluding chapters application of experimental and flow element model data to real world situations is discussed. The present experimental and analytical results provide useful information regarding the air movement and the expected thermal stratification in the rooms such as those considered here. Conclusions drawn from the present work and suggestions for further work are reported in the last chapter of this thesis.



Unless otherwise indicated, the views expressed in this thesis are those of the author and do not necessarily represent the views of the University of Wollongong.