Document Type : Research Article
Hydraulic Structures Department-Shahid Chamran University of Ahvaz- Ahvaz -Iran
This paper describes the application of artificial neural networks to model the longitudinal dispersion coefficient, which is a key hydraulic parameter included in the advective-diffusion equation (ADE).
The dimensionless dispersion (Dl/HU*) has been modeled based upon four dimensionless hydraulic parameters, including: the relative roughness (ks/H), the aspect ratio (H/W), the relative shear velocity (U*/U) and the Reynolds number (UH/v). A total of 81 sets of measured data in small and large rivers from the current literature were used to establish and verify the ANN model. The final model has been compared favorably with two recent formulations published in the literature. Comparison of the results showed that the ANN model is able to predict the measured values more precise by (14%) than the other two considered models.