# Turbulent rectangular compound open channel flow study using multi-zonal approach

- 138 Downloads

## Abstract

In this paper, an improved Shiono–Knight model (SKM) has been proposed to calculate the rectangular compound open channel flows by considering a multi-zonal (MZ) approach in modelling turbulence and secondary flows across lateral flow direction. This is an effort to represent natural flows with compound shape more closely. The proposed model improves the estimation of secondary flow by original SKM model to increase the accuracy of depth-averaged velocity profile solution formed within the transitional region between different sections (i.e. between main-channel and floodplain) of compound channel. This proposed MZ model works by sectioning intermediate zones between floodplain and main-channel for running computation in order to improve the modelling accuracy. The modelling results have been validated using the experimental data by national UK Flood Channel Facility. It has been proven to work reasonably well to model secondary flows within the investigated compound channel flow cases and hence produce better representation to their flow lateral velocity profile.

## Keywords

Shiono–Knight method Secondary flow Turbulence Natural flow Lateral velocity distribution Depth-averaged model## List of symbols

- b
Half of the in-bank channel width (m)

- B
Half of the whole channel surface width (m)

- D
_{r} Relative depth (−)

- f
Friction factor (−)

- g
Gravitational acceleration (m/s

^{2})- h
Height of in-bank channel (m)

- H
Flow depth (m)

- S
_{o} Bed slope (−)

- t
Time (s)

- U
_{*} Shear velocity (m/s)

- U
Velocity in x direction (m/s)

- U
_{d} Depth-averaged velocity (m/s)

- U
_{w} Width-averaged velocity (m/s)

- V
Velocity in y direction (m/s)

- W
Velocity in z direction (m/s)

- x
Streamwise coordinate (m)

- X
Water body forces (Pa)

- y
Lateral coordinate (m)

- z
Coordinate normal to bed (m)

- \( \rho \)
Density of water (kg/m

^{3})- \( \tau \)
Shear stress (Pa)

- \( \mu \)
Kinematic viscosity (m

^{2}/s)- \( \sigma \)
Normal stresses (Pa)

- \( \upvarepsilon \)
Eddy viscosity (kg/ms)

- \( \varGamma \)
Secondary flow parameter (kg/(ms

^{2}))- \( \lambda \)
Dimensionless eddy viscosity (−)

- \( {\bar{\varepsilon }}_{yx} \)
Depth-averaged eddy viscosity (m

^{2}/s)- \( \tau_{b} \)
Local boundary shear stress (Pa)

## 1 Introduction

In recent years, studies into compound channels have been undertaken due to its importance in representing the seasonal-impact rivers and floodplains. The naturally formed compound channel usually consisted of a deeper main-channel known as the in-bank and shallower floodplains known as the over-bank. This un-even channel could affect the flow velocity distribution, which could alter its sediment transport pattern [8]. In order to represent these flows, Shiono and Knight [11] introduced an analytical Shiono–Knight model (SKM) based on the simplified two-dimensional (2D) Reynolds Averaged Navier–Stokes (RANS) equations. The model ignored the effect of turbulence created from secondary currents. The SKM was improved when Shonio and Knight [12] included the effects of secondary flows within the SKM model, which they also found that the turbulence and secondary flow occurrences are complicated in compound channel flows due to the lateral exchange of flow momentum between the main-channel and floodplains.

Some compound channel studies further considered the channel-shape-generated turbulence as well as secondary flows to improve flow velocity profile representation (i.e. Ervine et al. [2] and Tang and Knight [13]). Studies by Yang et al. [17] showed that large-scale eddies are formed by instability in regions with high velocity variation, such as at the interface between main-channel and its floodplains of a compound channel flow. All these studies showed that the accurate calculation of secondary currents and hence turbulence are crucial for estimating the flow velocity.

To ease the consideration of secondary flows in SKM model, Shonio and Knight [12] assigned a secondary flow parameter, \( \varGamma \), into their model to calculate flow velocity, where this approach was commonly adopted in later studies [13, 14, 15, 18]. The parameter was chosen to represent each section in those studies to ensure the appropriate flow representation at main-channel and floodplain. Liao and Knight [3] and [4] further investigated \( \varGamma \) together with local friction factor (*f*) and dimensionless eddy viscosity (\( \lambda \)) for trapezoidal and rectangular compound channels respectively. They proved the importance of all these parameters in SKM model, in particular \( \varGamma \). From their studies, it has been proposed that velocity profile mis-representation could happen due to secondary flow assumption used by the SKM model. This mis-representation error usually occurs due to the lumping of single parameter within a section (i.e. within main-channel or floodplain) as a simplifying step at SKM modelling. Tang and Knight [13] considered the SKM’s boundary conditions associated with complex compound geometries and proved that by modifying SKM its velocity profile calculation can be improved for flows with different aspect ratios. Consequently, in Tang and Knight [14], it has been acknowledged that there is a lack of experimental and theoretical understanding to \( \varGamma \) which can lead to the inaccuracy of velocity computation.

In Tang and Knight [15], several analytical models have been reviewed and further tested, including by Ervine et al. [2], Castanedo et al. [1] and Van Prooijen et al. [16]. Ervine et al. [2] based their model by assuming the temporal-mean transverse velocity component in the flow momentum equation to be fraction of the streamwise depth-averaged velocity U_{d}, hence replacing \( \varGamma \) by KU _{d} ^{2} , where K is a weighted coefficient. In their model, K was used to account for complex 3D interfacing of the main-channel and floodplain, which includes the effect of horizontal shear layer, mass exchange in and out of each section and any expansion and contraction losses from non-prismatic channels. In Castanedo et al. [1], three different forms of the turbulent shear stress were investigated, however their model did not take into account the secondary flow effect. Van Prooijen et al. [16] proposed an analytical eddy viscosity model that included the effects of the horizontal coherent structures as well as the effects of 3D wall-induced turbulence. Tang and Knight [15] suggested that the large horizontal eddies induced by strong shearing in the sections’ mixing region play a significant role in the flow transverse mixing process. As a result, various eddy viscosity models that take into account the horizontal coherent structures and turbulence (i.e. Van Prooijen et al. [16]) have shown promising modelling outcomes.

From studies above, the turbulence and secondary flow are identified to be crucial for analytical model to represent velocity profile reasonably. The key forming of secondary flow usually takes place at the interval between sections (i.e. between main-channel and floodplain) of compound channel where its effect towards velocity is hard to be modelled accurately. Shonio and Knight [12] and Tang and Knight [13] have suggested that more complex sub-divided section members within a compound channel may be able to improve velocity modelling accuracy, but there is complexity to represent additional section’s turbulence and secondary flow. Due to this, a novel multi-zonal (MZ) model is proposed and investigated in this study for its suitability to improve common SKM model to represent velocity profile across rectangular compound channel. The proposed MZ model utilises extra section to improve the modelling of turbulence and secondary flow within the investigated compound channel flows and hence enhance the accuracy of velocity profile calculation. The proposed model has also been validated against the UK Flood Channel Facility (FCF) experimental data.

## 2 Model descriptions

### 2.1 Shiono–Knight (SKM) and turbulence modelling

Originated from the SKM model by Shiono and Knight [11, 12], Castanedo et al. [1] and Van Prooijen et al. [16] has proposed methods to improve the SKM’s turbulence modelling. However, their respective comparisons with experimental data demonstrated that the flow modelling improvement mainly restricted to flow regions away from compound sections’ interval. Yang et al. [18] considered a model attempting to represent flow velocity profile in rectangular compound channel. Their proposal was different to the SKM at designing secondary flows and eddy viscosity by considering momentum transfer.

Consequently, it appears that several studies for the compound channel flows were based around modifying the SKM model due to its modelling capability. Their proposed modifications to the SKM presented improved results from the original SKM model suggesting its versatility to adapt to different improved functions. Ergo, in this study, the SKM model will be adapted and improved.

_{d}, the depth-averaged momentum equation has to be solved for steady uniform turbulent flow in the streamwise direction. And by considering the dominant horizontal eddies across the channel, one can get that \( \partial U/\partial x = \partial \sigma_{xx} /\partial x = 0 \). Furthermore, it is assumed that the velocity variation is steady i.e. \( \partial U/\partial t = 0 \). As a result, the governing equations becomes (when combined with the continuity equation)

_{d}as

In Eqs. (8)–(10), \( \varGamma \) is secondary flow parameter; S_{o} is bed slope; g is gravitational acceleration; and A_{1}, A_{2}, k, \( \beta \) and \( \gamma \) are all coefficients used for depth-averaged velocity formulation.

### 2.2 Boundary conditions

*y*=

*b*can be written as

_{3}, A

_{4}, C

_{1}, C

_{2}, \( \gamma_{1} \) and \( \gamma_{2} \) are all coefficients for SKM boundary conditions.

## 3 New multi-zonal model and its validations

### 3.1 Modelling parameters

In analytical modelling, the parameters for secondary flow and eddy viscosity are crucial to ensure a precise representation to the compound channel flows. As proven in the calibration processes by Shiono and Knight [12], Tang and Knight [13, 14, 15], and Liao and Knight [3, 4], those parameters significantly influenced the performance of modelling calculation, in particular at near-wall and section transitional regions. Through the calibration using the UK-FCF experimental data, this study obtains a set of calibrated secondary flow and dimensionless eddy viscosity parameters. As for the friction factor, it has been obtained using the measured shear stress in the experiment, hence same coefficients used by the SKM model will be employed in this study [3, 14].

_{w}) for tests in Figs. 2, 3, 4, 5, 6 and 7 are described in Table 1, where the utilised B/H aspect ratio range from 1.98 to 15.99. For comparison purpose, the models by Tang and Knight [14], and Liao and Knight [3] are also plotted in same Figs. 2, 3, 4, 5, 6 and 7.

Experimental conditions for compound channel flow tests

Experiment no. | B/H (−) | S | b (m) | h (m) | B (m) | H (m) | U |
---|---|---|---|---|---|---|---|

FCF0804 | 15.99 | 0.001027 | 0.750 | 0.150 | 3.00 | 0.18760 | 0.469 |

FCF0807 | 11.99 | 0.001027 | 0.750 | 0.150 | 3.00 | 0.25020 | 0.783 |

ROS224 | 8.37 | 0.002024 | 0.400 | 0.050 | 0.60 | 0.07165 | 0.464 |

ROS245 | 6.93 | 0.002024 | 0.400 | 0.050 | 0.60 | 0.08660 | 0.692 |

DWK0404 | 2.67 | 0.000966 | 0.076 | 0.076 | 0.30 | 0.11420 | 0.379 |

DWK0406 | 1.98 | 0.000966 | 0.076 | 0.076 | 0.30 | 0.15380 | 0.497 |

Coefficients summary for compound channel flow tests

Tang and Knight [14] | Liao and Knight [3] | Present study | |||||
---|---|---|---|---|---|---|---|

Experiment no. | B/H | λ (1&2) | Γ (1&2) | λ (1&2) | Γ (1&2) | λ (1&2) | Γ (1&2) |

FCF0804 | 15.99 | 0.07, 0.07 | 0.25, − 0.10 | 0.01, 0.10 | 0.25, −0.15 | 0.10, 0.09 | 0.30, − 0.095 |

FCF0807 | 11.99 | 0.08, 0.07 | 0.60, − 0.25 | 0.01, 0.10 | 0.25, − 0.15 | 0.10, 0.09 | 0.65, − 0.25 |

ROS224 | 8.37 | 0.07, 0.07 | 0.25, − 0.10 | 0.01, 0.10 | 0.25, − 0.15 | 0.035, 0.055 | 0.24, − 0.10 |

ROS245 | 6.93 | 0.07, 0.07 | 0.35, − 0.25 | 0.01, 0.10 | 0.25, − 0.15 | 0.035, 0.055 | 0.40, − 0.30 |

DWK0404 | 2.67 | 0.07, 0.07 | 0.34, − 0.092 | 0.01, 0.10 | 0.25, − 0.15 | 0.10, 0.09 | 0.39, − 0.091 |

DWK0406 | 1.98 | 0.03, 0.07 | 0.55, − 0.187 | 0.01, 0.10 | 0.25, − 0.15 | 0.10, 0.09 | 0.60, − 0.18 |

From the model validation process, it has been found that the flow tests with high aspect ratio of B/H (i.e. FCF0804, FCF0807 and ROS224) can be represented quite accurately by SKM models (by Liao and Knight [3] and Tang and Knight [14]), and by the present calibrated model. For other flow tests with smaller B/H ratios, all models show incapability to represent the compound channel flows with high accuracy. The key inaccuracy happens at the flow interface between main-channel and floodplain prompting the use of different approach to model this region of flow is needed. The following section will describe a new multi-zonal (MZ) approach to improve compound channel study.

### 3.2 Proposed multi-zonal (MZ) approach

_{w}are outlined at Table 3. All tests in this section have aspect ratio from 1.98 to 7.10.

Experimental conditions for flows with small aspect ratios for multi-zonal (MZ) model tests

Experiment no. | B/H (−) | S | b (m) | h (m) | B (m) | H (m) | U |
---|---|---|---|---|---|---|---|

DWK0406 | 1.98 | 0.000966 | 0.076 | 0.076 | 0.30 | 0.1538 | 0.497 |

ROS255 | 6.36 | 0.002024 | 0.400 | 0.050 | 0.60 | 0.0950 | 0.783 |

ROS250 | 6.55 | 0.002024 | 0.400 | 0.050 | 0.60 | 0.0920 | 0.738 |

ROS245 | 6.93 | 0.002024 | 0.400 | 0.050 | 0.60 | 0.0866 | 0.692 |

ROS240 | 7.10 | 0.002024 | 0.400 | 0.050 | 0.60 | 0.0846 | 0.674 |

_{d}.

It has been found in this study that the interface zones 2 and 3 in Fig. 8 are accounted for about 10-25% of the distance for main-channel and floodplain respectively (agreed with findings by experimental studies on various channel flows in [6, 10]. In comparison, the shallower zone 3 is lengthier than zone 2 as the secondary flow has less depth of water to evolve vertically. Since both secondary flow \( \varphi_{1} \) and \( \varphi_{2} \) components act at the counter-rotational manner, they are estimated to have positive and negative impact to \( \varGamma_{1}^{'} \) and \( \varGamma_{2}^{'} \) respectively. In the exact middle of the meeting point between main-channel and floodplain, \( \varphi_{1} \) and \( \varphi_{2} \) are set at zero as both secondary flow components neutralised.

Root-mean square error (RMSE) analysis benchmarked by measured data at the main-channel and floodplain interface region

Test | Width-average of measured velocity (m/s) | Averaged RMSE for calculation Without MZ (m/s) | Averaged RMSE for calculation with MZ (m/s) | Comparative error fraction (non-MZ to MZ model) (−) |
---|---|---|---|---|

DWK0406 | 0.4968 | 0.036894 | 0.013479 | 2.737 |

ROS240 | 0.6737 | 0.097826 | 0.033446 | 2.925 |

ROS245 | 0.6922 | 0.066959 | 0.012258 | 5.462 |

ROS250 | 0.7381 | 0.125526 | 0.028812 | 4.357 |

ROS255 | 0.7834 | 0.108704 | 0.031258 | 3.478 |

The investigated narrow channel flow cases have been influenced by secondary currents which impact the flows towards left and right of the compound step. Compared to wide channel flows, the side-wall and compound step effects are crucial for narrow flow cases and it has been proven that the 2D depth-averaged model could not effectively represent them without combining with innovative approach (i.e. by Nezu and Nakagawa [5], Pu et al. [9], Pu [7]). In the tests of this study, the proposed MZ model has proven to reasonably improve the conventional SKM model deduced from 2D depth-averaged Navier–Stokes flow assumption. It has been shown that by additional sectioning of the flow model laterally, the compound channel 2D modelling accuracy could be improved. Theoretically, the MZ approach works by considering smaller width of the channel sections and hence able to reduce the inaccuracy of SKM calculation, by reducing error caused by representing whole section with a single simplified \( \varGamma \).

## 4 Conclusions

The Shiono–Knight model (SKM) has been investigated to estimate the flow velocity profile within rectangular compound channel which has been subjected to turbulence and secondary flows. Due to its 2D Navier–Stokes assumption, SKM has found to represent the narrow channel flow with less accuracy compared to wide channel flows. A multi-zonal (MZ) approach has been proposed to improve this modelling weakness. In the proposed MZ model, the investigated compound channel flows have been sectioned with extra zones in the interface region between main-channel and floodplain where the escalated secondary flow effect has been considered. To validate the proposed MZ model, it was used to calculate and compare with the experimental data by national UK-FCF facility. The comparison with measured data showed that the proposed MZ model represents the narrow flows with better accuracy than the SKM approach for the rectangular compound channel flow cases. This study has also proven that the modelling of narrow 3D-characterised secondary flow could be improved by considering extra zones’ turbulence and secondary current modelling in the analytical model.

## Notes

### Acknowledgement

The help of Ms. Faye Lynch (the author’s former undergraduate student) is greatly appreciated in the model testing and formulations.

## References

- 1.Castanedo S, Medina R, Mendez FJ (2005) Models for the turbulent diffusion terms of shallow water equations. J Hydraul Eng 131(3):217–223CrossRefGoogle Scholar
- 2.Ervine DA, Babaeyan-Koopaei K, Sellin RHJ (2000) Two-dimensional solution for straight and meandering overbank flows. J Hydraul Eng 126(9):653–659CrossRefGoogle Scholar
- 3.Liao H, Knight DW (2007) Analytic stage-discharge formulas for flow in straight prismatic channels. J Hydraul Eng 133(10):1111–1122CrossRefGoogle Scholar
- 4.Liao H, Knight DW (2007) Analytic stage-discharge formulae for flow in straight trapezoidal open channels. Adv Water Resour 30(1):2283–2295CrossRefGoogle Scholar
- 5.Nezu I, Nakagawa H (1993) Turbulent open-channel flows. IAHR Monograph: A.A. Balkema, AmsterdamGoogle Scholar
- 6.Nezu I, Tominaga A, Nakagawa H (1993) Field measurements of secondary currents in straight rivers. J Hydraul Eng 119:598–614CrossRefGoogle Scholar
- 7.Pu JH (2015) Turbulence modelling of shallow water flows using Kolmogorov approach. Comput Fluids 115:66–74CrossRefGoogle Scholar
- 8.Pu JH, Lim SY (2014) Efficient numerical computation and experimental study of temporally long equilibrium scour development around abutment. Environ Fluid Mech 14:69–86CrossRefGoogle Scholar
- 9.Pu JH, Shao S, Huang Y (2014) Numerical and experimental turbulence studies on shallow open channel flows. J Hydro Environ Res 8:9–19CrossRefGoogle Scholar
- 10.Pu JH, Tait S, Guo Y, Huang Y, Hanmaiahgari PR (2018) Dominant features in three-dimensional turbulence structure: comparison of non-uniform accelerating and decelerating flows. Environ Fluid Mech 18(2):395–416CrossRefGoogle Scholar
- 11.Shiono K, Knight DW (1988) Two-dimensional analytical solution for compound channel. In: Proceedings of third international symposium on refined flow modeling and turbulence measurements, Tokyo, Japan, pp 503–510Google Scholar
- 12.Shiono K, Knight DW (1991) Turbulent open-channel flows with variable depth across the channel. J Fluid Mech 222(1):617–646CrossRefGoogle Scholar
- 13.Tang X, Knight DW (2008) A general model of lateral depth-averaged velocity for open channel flows. Adv Water Resour 31(1):846–857CrossRefGoogle Scholar
- 14.Tang X, Knight DW (2008) Lateral depth-averaged velocity distributions and bed shear in rectangular compound channel. J Hydraul Eng 134(9):1337–1342CrossRefGoogle Scholar
- 15.Tang X, Knight DW (2009) Analytical models for velocity distributions in open channel flows. J Hydraul Res 47(4):418–428CrossRefGoogle Scholar
- 16.Van Prooijen BC, Battjes JA, Uijttewaal WSJ (2005) Momentum exchange in straight uniform compound channel flow. J Hydraul Eng 131(3):175–183CrossRefGoogle Scholar
- 17.Yang Z, Gao W, Huai W (2010) Secondary flow coefficient of overbank flow. Appl Math Mech 31(6):709–718CrossRefGoogle Scholar
- 18.Yang K, Nie R, Liu X, Cao S (2012) Modeling depth-averaged velocity and boundary shear stress in rectangular compound channels with secondary flows. J Hydraul Eng 139(1):76–83CrossRefGoogle Scholar

## Copyright information

**OpenAccess**This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.