Characterization and Modelling of the mixing in the Athabasca River downstream of a pulp mill
Odigboh, Ifeanyi S.
Methods for predicting the transverse mixing coefficient, Ez in rivers based solely upon estimates or measurements of the channel geometry, channel slope and flow parameters are not completely reliable. Therefore, it is generally necessary to perform field tracer tests in order to determine Ez for a particular river reach at a given stage and flow. Characterizations of the transverse mixing in the Athabasca River downstream of a pulp mill located near Boyle, Alberta are described herein. The characterization of the mixing is based upon analysis of four tracer tests conducted on the river (three continuous input tests and one slug test). The tracer tests were conducted on different dates and cover a range of flow conditions (October 1994, 270 m3/s; February 1995, 84 m3/s, and August 1997, 960 m3/s and 876 m3/s). The February test was conducted under ice-covered conditions. Beak Consultants Ltd. conducted the October 1994 and February 1995 tests. The tracer input for each field test consisted of injection of Rhodamine WT fluorescent dye at the mill outfall location. In the continuous input tests, the tracer was injected into the mill effluent stream at a constant rate and entered the river via the effluent diffuser structure. The diffuser is 52 m long, oriented perpendicular to the flow, and located close to the left bank of the river. In the slug input, test a known mass of tracer was instantaneously dumped directly into the river at approximately the mid-point along the length of the diffuser. For each test the dye plume was sampled across transects oriented perpendicular to the river flow at a number of downstream locations stretching over a 32-km reach of the river. Hydrographic surveys were conducted at each sampled transect and at several other transects to determine channel geometry and flow parameters. The hydrographic survey information and the tracer input conditions were required for numerical modelling of the mixing in the river reach. The implementation of the two-dimensional river mixing modelling procedure used in the present study was written by Putz (1996), based upon the descriptions given by Beltaos and Arora (1988). The numerical model utilizes a streamtube approach and a numerical procedure employing an advection optimized grid to limit numerical errors. The model is capable of simulating continuous input and unsteady input conditions. The modelling package includes two preprocessing programs for generating the calculation grid based upon channel characteristics, a two-dimensional, transient mass input mixing program, and a post-processing program for output of data at selected locations. The transverse mixing coefficient was determined for each field test using the model fit of predicted tracer concentrations compared to the concentration of the samples collected in the field. The mixing simulations provide a very good representation of the measured concentration distributions. The four field verification studies demonstrate that the Advection Optimized Grid method can be applied to two-dimensional, steady and unsteady source mixing problems in natural streams satisfactorily. Nondimensionalized transverse mixing coefficients determined from the modelling procedure are compared to the range of values reported in other locations. These comparisons show that the values of ÃŸ which characterize the transverse mixing in the reach of the Athabasca River studied fall well within the range of reported ÃŸ values from other studies. Reach-averaged dimensionless mixing coefficient, ÃŸ, is fairly consistent (in the range of 0.34 to 0.48) for the range of flow conditions represented by the four tests. The overall weighted average of ÃŸ for the different ranges of flow was found to be 0.41. The results demonstrate that the dimensionless mixing coefficient measured at one flow condition may be used with the appropriate flow parameters to estimate mixing for other flow conditions for the same reach. The results of the modelling procedure are also used to assess the variation in mixing with discharge and due to the presence of an ice cover.