Laserfiche WebLink
<br />transverse flows when the mean channel curvature, rc' is replaced by the <br />mean flow curvature, rf. Upon entering a bend, the mean flow curvature <br />increases with the growth in transverse circulation. In a bend, the <br />transverse flow becomes fully developed if the flow curvature approaches the <br />channel curvature. In the case of exiting from a bend to the downstream <br />tangent in which the channel curvature is zero, the flow curvature decreases <br />wi th the decay of transverse circulation. <br /> <br />The reason that the flow curvature lags behind the channel curvature <br />during circulation growth and decay is attributed to the internal turbulent <br />shear that the flow has to overccme in transforming from parallel flow into <br />the spiral pattern and vice versa. From the dynamic equation for the trans- <br />verse velocity, an equation governing the streamwise variation in transverse <br />surface velocity, V, was derived (Chang, 1984). In finite difference form, <br />the change in v over the distance & i.s given by <br /> <br />- r", <br />vi+l = lVi + <br /> <br />Fl(f) ~ exp(F2(f) ~s) ~sJ exp[-F2(f) ~s] <br />rc <br /> <br />(4) <br /> <br />where v is the transverse surface velocity along discharge centerline, U is <br />the average velocity of a cross section, i and i+l are s-coordinate indices, <br />Fl and F2 are functions of f (friction factor), and the overbar denotes <br />averaging over the incremental distance between i and i+l. Eg. 4 provides <br />the spatial variation in v, from which the mean flow curvature may be <br />obtained using the transverse velocity profile. From the transverse <br />velocity profile by Kikkawa, et al. (1976), the mean flow curvature, rf, is <br />related to the transverse surface velocity as <br /> <br />rf = Dc ~ (~~ _ ~ ~ ff) <br />I( v 3 I( 9J"2 <br /> <br />(5) <br /> <br />where Dc is the flow depth at discharge centerline (thalweg) and I( is the <br />Karman constant. At each time step, the mean flow curvature at each cross <br />section is obtained using Egs. 4 and 5. Accuracy of computation for the <br />finite difference equation (Eg. 4) is maintained if the step size ~ ~ 2Dc. <br />For this reason, the distance between two adjacent cross sections is divided <br />into snaller increments if necessary. Flow parameters for these increments <br />are interpolated from values known at adjacent cross sections. <br />12 <br />