AN EXPERIMENTAL INVESTIGATION OF THE FLOW OF DILUTE POLYMER-SOLUTIONSTHROUGH CORRUGATED CHANNELS

Measurements of the velocity using Laser Doppler Velocimetry and norma l stress are made for the flow of dilute polymer solutions through a c hannel with corrugated top and bottom plates. (Since we are dealing wi th non-Newtonian fluids, there can be significant contributions to the normal stress from non-linear terms in the constitutive expression, e ven when the flow is slow. The measurements being made are the normal stresses and not the ''pressure''.) The surfaces of the plates are sin usoidal. A Reynolds number based on half the average plate spacing as the length scale and the characteristic velocity as the velocity scale was used and the range of Reynolds numbers studied was 50 < Re < 1000 . The centerline velocities indicate that the experiments were perform ed in the inertial regime, as confirmed by the asymmetry of the center line velocities along the channel length. The velocity profiles at the trough near the wall, for a channel with wavelength of 2.54 cm, indic ate the presence of secondary flow. Sinusoidal plates with nearly iden tical aspect ratios (ali,) allowed for dramatic changes in the way in which the friction factor varied with Reynolds number, in that, in one case the friction factor associated with the fluid without polymer wa s higher than the friction factor associated with the fluid with polym er, while in others it was just the opposite. This would call into que stion the use of aspect ratio as an appropriate parameter for studying such problems. Changes in plate wavelength either increased or decrea sed the friction factor depending on the Reynolds number. Increasing p late amplitude increased the friction factor of the fluid for the rang e of values for the Reynolds number that was considered. The amplitude associated with the dimensionless normal stress increased with decrea sing wavelength, for particular Reynolds numbers, irrespective of the fluid studied. Increasing the polymer concentration in the fluid decre ased the difference in the amplitude of the dimensionless normal stres s, the Reynolds number being fixed. Increasing the plate amplitude inc reased the amplitude of the normal stress. while an increase in plate wavelength decreased the amplitude of the normal stress.