Prediction of the flow resistance in non-prismatic compound channels

Abstract

Achieving an accurate estimation of the flow resistance in open channel flows is crucial for resolving several critical engineering difficulties. In instances when there is excessive flow on both banks of a river, it results in the breach of the primary channel, leading to the discharge of water into the adjacent floodplain. The alteration of floodplain geometry occurs as a consequence of agricultural and developmental practises, leading to the emergence of compound channels that exhibit converging, diverging, or skewed characteristics throughout the course of the flow. The efficacy of conventional equations in accurately forecasting flow resistance is limited due to their heavy reliance on empirical approaches. As a result of this phenomenon, there persists a significant need for methodologies that possess both novelty and precision. The objective of this work is to use the support vector machine (SVM) technique for the estimation of the Manning's roughness coefficient in a compound channel with converging floodplains. Statistical indicators are used to validate the constructed models in the experimental investigation, enabling the assessment of their performance and efficacy. The findings indicate a significant correlation between the Manning's roughness coefficient predicted by SVM and both experimental data and prior research outcomes. HIGHLIGHTS Flow resistance is crucial across various engineering contexts, such as planning irrigation systems, managing drainage networks, and constructing flood control structures.; This study delved into assessing flow resistance in non-prismatic compound channels by employing support vector machines.; The results indicate a strong alignment between the predicted Manning's roughness coefficient and the actual observed values.;

Keywords

• geometric and flow parameters
• manning's roughness coefficient
• non-prismatic compound channel
• statistical analysis
• support vector machine