EFFECT OF BED AND SIDE ROUGHNESS ON DISPERSION AND DIFFUSION IN OPEN CHANNELS

Abstract

The present research program was designed to study the effect of large-scale bed and side roughness on both longitudinal dispersion and transverse diffusion in a straight rectangular channel. The tracer material used consisted of salt solution in water rendered neutrally buoyant with the help of denatured spirit. For dispersion runs the tracer was injected instantaneously as a line source across the full width of the flume, and for diffusion runs it was injected continuously through a point source at the centreline of the flume. Concentra tion vs time records for dispersion and diffusion runs were obtained by using an appropriate conductivity probe, associated electronic circuit and an 'Encardiorite' strip chart recorder. Longitudinal dispersion coefficients^) were first computed by change of moments method. A computer program, employing Gauss-Hermite Quadrature technique, for Routing Procedure was developed for obtaining the appropriate values of dispersion coefficients. For calculating transverse diffusion coefficients, the method of moments was employed. The first phase of the study consisted of dispersion and diffusion experiments in a flume having smooth bed as well as smooth sides. The main aim of this phase was to test the validity of the methodology adopted - the results ill obtained compared well with those of Fischer (1966b), and McQuivey and Keefer (1971), and 0koye(l970), and Lau and Krishnappan (1977a), etc., thereby confirming the correct ness of instrumentation and calculation procedures. The second and third phase of the experimental program consisted of dispersion and diffusion studies respectively, on rough bed and rough sides channel. It was shown that the datum shift (i.e., bed shift) in case of rough bed runs was necessary to establish the standard resistance law; and, in the case of side spurs, the side shift (i.e., effective width of flow) - akin to bed shift in rough beds - was similarly necessary to obtain a standard resistance law. In the case of rough beds it was observed that the non-dimensional dispersion coefficient (It/RUO is a function of Chezy's roughness coefficient (C/Yg) and the micro structural details of roughness elements, like, relative spacing, and relative height, etc. Similar results were obtained for side spur cases - showing the analogous behaviour of bed+side roughness conditions. Finally, a generalised roughness coefficient (P ) was proposed for the unification of dispersion data, and the data of other investigators (flume data as well as field data) showed a satisfactory agreement with the proposed iv model - the field data, expectedly for channel irregulari ties and implied average values for flow data, showing some scatter. Lastly, it was observed that the nondimensional diffusion coefficient (&jV U) - for all smooth,rough bed and rough sides channels is primarily a function of C/fg and aspect ratio (W/Y) for lower values of these parameters (i.e., when the channel is rougher). It was, however, observed that for C/fg greater than about 18.00 the value of D /W U asymptotically, for all values of W/Y, approaches a constant value of about 3.0 x 10 to 4.0 x 10""4. Amodel (i.e^D^/WU =0.0295(W/Y)""0*3475 (G/fg)"* * ^Vas proposed to unify the diffusion data of the present study as well as the data of other investigators. Further, it was seen that the distribution of time-averaged concentration (c) vs transverse distance, at an observation stationj and the nature of deviations of concentration from the time-averaged concentration (c), at a given observation point, are Gaussian.