Experimental and numerical studies of complex open channel flows

Abstract

The flow phenomena in three channel types, i.e., open channel, confluence channel, and bend channel are investigated in this project. Furthermore, flow characteristics in channels with gradual width expansion and contraction are also studied in both experimental and numerical ways. The flows, that rapidly change either directions or widths, induce strong secondary circulation, surface slope variation, and boundary separation. The flow structures in these channels are rather different from those in straight channels. In experimental study, a three dimensional, supersonic velocity detector (Acoustic Doppler Velocimeter, ADV) is used to obtain components of instantaneous velocity. A series of flume tests have been carried out to study turbulence, velocity distribution and energy dissipation of flows in the selected channels. The results not only provide detailed flow properties but sound base for verification and validation of numerical models. The main difficulties in modeling complex open channel flows are dealing with the free surface and eddy viscosity. In this study, both two-dimensional, depth-averaged models and three-dimensional models are adopted to simulate the flows in the three typical channels. For two dimensional simulations, a self-written program coupled with Large Eddy Simulation (LES) and implemented by several turbulence models is developed to investigate different scales of flow properties. For three dimensional simulations, commercial software Fluent coupled with self-programmed, free surface tracking is used. During the simulation, several turbulence models are adopted to estimate their impact on the simulation accuracy. Investigations of flow in confluence channels show that the strength of secondary circulation is flow-condition-dependent and is affected by the ratio flow rate of the branch channel to the main channel. The circulation is weak when the low flow rate ratio is low, confluence flow exhibits two-dimensional characteristics and can be approximately simulated by two-dimensional models. Under the case of high flow rate ratio, the circulation becomes strong, where the effects of secondary circulation cannot be neglected in the simulation. Reasonable results can be obtained only by using three-dimensional models together with proper free surface tracking. The key elements for simulation of confluence flow are also presented. The exploration of flow in bend channel demonstrates that the strength of secondary circulation and the transverse surface slope largely depend on the water level near the tail gate under the same flow rate. When the level is high, flow in bend channel moves slowly, circulation is weak, and the transverse surface is nearly flat. This situation can be modeled by the two-dimensional model without correction of secondary circulation; good agreement is found with the measurement. On the contrary, when the water level is low, flow in bend channel is faster, with strong secondary circulation and steep transverse surface slope. Hence, in numerical simulations, high accuracy can only be achieved by using three-dimensional or two-dimensional models with proper correction of the secondary circulation. Studies on flows in channels with changing widths indicate that the flow performs like hydraulic jump near the expansion and hydraulic fall in the vicinity of contraction. The transition from hydraulic fall to hydraulic jump generates rapid changes in water surface and velocity and induces strong secondary circulation. Investigations of flow in confluence channels show that the strength of secondary circulation is flow-condition-dependent and is affected by the ratio flow rate of the branch channel to the main channel. The circulation is weak when the low flow rate ratio is low, confluence flow exhibits two-dimensional characteristics and can be approximately simulated by two-dimensional models. Under the case of high flow rate ratio, the circulation becomes strong, where the effects of secondary circulation cannot be neglected in the simulation. Reasonable results can be obtained only by using three-dimensional models together with proper free surface tracking. The key elements for simulation of confluence flow are also presented. The two-dimensional simulations of water level present good agreement with the experimental data, but the simulated velocity profiles at expansion parts agree poorly with the experimental data. The accuracy of two-dimensional simulation is also flow-condition-dependent. The flow is also studied by three-dimensional simulations and the self-programmed codes perform well in tracking the water surface. The first major contribution of this study is that the characteristics of flows in three typical open channels are investigated, which provide references for further studies and solving similar problems in engineering applications. The second contribution is to develop and verify the self-written codes to tackle the depth-averaged shallow water equations and the complex open channel flows. Furthermore, codes for free surface tracking in three-dimensional simulations are also developed, which enhance accuracy and save computational time and resources in simulations compared to those utilizing the surface capturing methods.

Date of Award	3 Oct 2012
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	W. Z. LU (Supervisor)