Study of transverse mixing in non uniform open channel flow

Estimating the transverse mixing in rivers is essential for a better estimation of pollutant dispersion, and the determination of concentration levels in order to anticipate sanitary risks. It is usually estimated using a simple model that supposes uniform flow conditions in the main flow direction (constant flow velocity and depth along the direction of the flow), and introduces a 'transverse mixing' coefficient that quantifies transverse mixing efficiency. River studies usually assume uniformity, and report values of the coefficient much higher than in laboratory studies for which the uniformity condition is verified, without explanation. The work presented here aims at verifying if the flow longitudinal non-uniformity (accelerated or decelerated), often observed in rivers and not studied in laboratory, can increase the transverse mixing. In situ tracing experiments are undertaken in the Durance river, a piedmont river that exhibits a strong longitudinally non uniform flow. A first field experiment reveals the increase of the transverse mixing coefficient as the non-uniformity parameter increases. A new equation including the flow non-uniformity parameter fits our experimental data with high accuracy, and explains the high values of the transverse mixing coefficient in rivers. A second field experiment focuses on the difficulties of properly estimating the characteristic quantities of the mean flow and its impacts on the measured value of the transverse mixing coefficient. Finally, in situ tracing experiments in a straight laboratory channel with controlled conditions confirms the increase of the transverse mixing efficiency in an accelerated flow compared to a uniform flow. Further experimentations are required in order to understand the physical mechanism responsible for this increase in a non uniform flow, and to determine a more general empirical law for predicting the value of the transverse mixing coefficient in a river. (author)