The origin and evolution of spatially stationary streamwise vortical s
tructures in plane mixing layers with laminar initial boundary layers
were recently examined quantitatively (Bell & Mehta 1992). When both i
nitial boundary layers were made turbulent, such spatially-stationary
streamwise structures were not measured which is indicative of the hig
h sensitivity of these structures to initial conditions. In the presen
t study, the effects of four different types of spanwise perturbations
at the origin of the mixing layer were investigated. The wavelengths
of the imposed perturbations were chosen to be comparable to the initi
al Kelvin-Helmholtz wavelength. For the first two perturbations, the b
oundary layers were otherwise left undisturbed. A serration on the spl
itter plate trailing edge was found to have a relatively small effect
on the formation and development of the streamwise structures. The int
roduction of cylindrical pegs in the high-speed side boundary layer no
t only generated a regular array of vortex pairs, but also affected th
e mixing-layer growth rate and turbulence properties in the far-field
region. For the other two perturbations, the initial boundary layers w
ere tripped on the splitter plate. An array of vortex generators mount
ed in the high-speed boundary layer and a corrugated surface attached
to the splitter plate trailing edge had essentially the same effects.
Both imposed a regular array of relatively strong streamwise vortices
in counter-rotating pairs upon the mixing layer. This resulted in larg
e spanwise distortions of the mixing layer mean properties and Reynold
s stresses. While the vorticity injection increased the growth rate in
the near-field region as expected, the far-field growth rate was redu
ced by a factor of about two, together with the peak Reynolds stress l
evels. This result is attributed to the effect of the relatively stron
g streamwise vorticity in making the spanwise structures more three-di
mensional and hence reducing entrainment during the pairing process. T
he imposed streamwise vorticity did not follow the pattern of increasi
ng spanwise spacing seen in the 'naturally occurring' streamwise vorti
city. The mean streamwise vorticity decayed with increasing streamwise
distance in all cases, albeit at different rates.