Some effects of history on turbulent flow.

Few physical phenomena can be adequately described without reference to their histories. This maxim applies not only to the field of experimental engineering but throughout academic disciplines to geography, biology, human relations and almost every sphere of physical activity. With reference to the field of fluid dynamics the interpretation of this statement means that few real flow situations can be described solely in terms of local values and relationships and that the previous history of the flow must be taken into account. An example of this is the work of Cockrell , Diamond and Jones ( 1 ) *, who showed that for a given diffuser , the performance varied even though the inlet boundary layer thickness parameters were maintained constant , and inferred that the manner in which the boundary layers had been produced has an appreciable effect on diffuser performance. This thesis has three major objectives. The first is to use our knowledge of flow history to provide an explanation of the 'overshoot' phenomenon experienced by the growth of the boundary layer thickness parameters in duct flow. This manifests itself by the manner in which a thickness parameter grows until it reaches a stationary point, whereupon it subsequently decreases. The second objective is to attempt to expand our knowledge of flow history effects by simulating a mean velocity profile at the duct entry and observing the subsequent development in both this and the turbulence parameters of the flow. This second objective has practical applications in the production of atmospheric boundary layers and other shear flows for the purpose of model Figures in brackets refer to reference listing. testing. The third objective is to use the experimental results obtained from duct flow in the auxiliary equation of an integral method of boundary layer calculation. This then should provide a realistic allowance for the flow history effects. Since the lack of any allowance of flow history effects is very often a major failing of the use of integral techniques it is possible in this instance to attempt to assess the overall usefulness of such a method. In addition to these major objectives, appendices are presented on the experimental behaviour of the local relationships, often used in calculation methods, between the turbulent shear stresses and the mean velocity profile. The distributions of these mixing lengths and eddy viscosities are compared with some of the formulations available in the literature. A description of the experimental facilities is given which includes a section on the use of hot wire anemometers for the measurement of turbulence. Some of the likely sources of error encountered here, together with their possible magnitudes, are also presented.