Multi-variable regression models for prediction of discharge and approach velocity coefficients in flow measurement flumes with compound cross-section

Abstract

In this paper, two multi-variable regression models have been developed to predict the discharge and approach velocity coefficients from relevant independent variables. The regression models are developed based on relevant experimental data obtained from testing nine different flow measurement long-throated flumes with symmetrical rectangular compound cross-sections. The long-throated flume was used in the compound cross-section to experimentally estimate the discharge and approach velocity coefficients using mainly head measurements and cross-section dimensions as required by the stage-discharge equations. The independent variables used in predicting the two coefficients represent dimensionless parameters defined using the gauged head at the head measurement section, floodplain depth, length of throat in the direction of flow, and cross-section geometry at the control section. Several statistically-based analyses were performed to verify the reliability of the developed multi-variable regression models. All deployed analyses have indicated that the two regression models are associated with high predictive strength. Therefore, the main contribution of this paper is the development of regression-based models to predict the discharge and approach velocity coefficients that can be used in conjunction with stage-discharge equations to estimate the flow in a symmetric rectangular channel with compound cross-section.