A method for the numerical solution of the two-dimensional Euler equations on unstructured grids has been developed. The cell-centred symmetric finite-volume spatial discretisation is applied in a general formulation that allows the use of arbitrary polygonal computational cells. The integration in time, to a steady-state solution, is performed using an explicit, multi-stage procedure, with standard convergence acceleration techniques such as local time stepping, enthalpy damping and implicit residual smoothing. Accuracy of solution, in terms of minimising spurious entropy production, is achieved through careful treatment of the artificial dissipative terms near boundaries. Standard test cases for both subcritical and supercritical flows, including single- and multi-element aerofoils have been used to validate the method.

A numerical method, based on the panel technique, has been developed for calculating the time-averaged separated flow about a wedge. The mathematical model, with constant vorticity in the wake, gives an infinite extension of the wake and constant base pressure while incorporation of the dispersion of vorticity in the wake into the mathematical model leads to its finiteness and non-uniform under-pressure. Experimental tests have been conducted in a low speed wind tunnel on models haying different wedge angles. The theoretical and experimental results agree closely for different configurations and specific values of dispersion factors. However, the agreement for base pressure is poor.

This paper reports on the measurement of aerodynamic forces that act on a square cylinder which was downstream to an identical cylinder. When the two cylinders were in tandem formation, a critical spacing equal to about four times the side length D of the square cylinder, was found to exist. For L/D < (L/D)cri, where L is the streamwise centre to centre spacing between the two cylinders, the boundary layers which separate from the upstream cylinder reattach onto the downstream cylinder. Only the latter sheds vortices and it is subjected to a thrust rather than a drag force. For LID > (L/D)cri, both cylinders shed vortices and both are subjected to drag forces. When the two cylinders were in staggered formation, due to the presence of various lift generating mechanisms, the lift force that acts on the downstream cylinder can be either positive which points away from the wake centre line of the upstream cylinder or negative which points towards the wake centre line of the upstream cylinder. It was found that in the range 4 ⩽ LID ⩽ 12, an unstable regime exists in that the lift force generated pushes the downstream cylinder away from the upstream cylinder’s wake centre line at small dimensionless transverse spacing (Y/D) where Y is the transverse centre to centre spacing between the two cylinders and reverses its direction only at Y/D > 2. At L/D = 29, the largest streamwise spacing that can be investigated in the present study, the influence that the upstream cylinder has on the aerodynamic forces acting on the downstream cylinder was found to be still fairly appreciable.

For conical bodies, at moderate angles of attack, the flow separates from the lee side, forming two vortices. Although the vortex lift contribution is highly desirable, as the angle of attack increases, the vortex system becomes asymmetric, and eventually the vortices breakdown. Thus, some control of the separation process is necessary if the vortex lift is to be exploited at higher angles of attack.

The theoretical model which is used in this analysis has three parts. First, the ‘single line-vortex’ model is used within the framework of ‘slender-body theory’ to compute the outer inviscid field for specified separation lines. Second, the 3-D boundary layer is represented by a momentum equation for the cross-flow, analogous to that for a plane boundary layer and a von Karman/Pohlhausen approximation is applied to solve this equation. The cross-flow separation for both laminar and turbulent layers is determined by matching the pressure at the upper and lower separation points. This iterative procedure yields a unique solution for the separation lines and consequently for the positions of the vortices and the vortex lift on the body. Third, control of separation is achieved by blowing tangentially from slots located symmetrically along cone generators.