Flow Measurement and Instrumentation 80 (2021) 102004

Contents lists available at ScienceDirect

Flow Measurement and Instrumentation

journal homepage: www.elsevier.com/locate/flowmeasinst

Experimental and numerical study of flow over a broad-crested weir under

different hydraulic head ratios
Hanifeh Imanian a, b, *, Abdolmajid Mohammadian b, Pouneh Hoshyar c
a
Department of Civil Engineering, Alzahra University, Tehran, Iran
b
Department of Civil Engineering, University of Ottawa, Ottawa, Canada
c
Department of Civil and Environmental Engineering, Amirkabir University of Technology (Tehran Polytechnic), Tehran, Iran

A R T I C L E I N F O A B S T R A C T

Keywords: Herein, the free surface flow over a broad-crested weir under different hydraulic heads was investigated. A series
Experimental study of experiments were conducted in a 5-m-long-flume to measure the 3D velocity profile around a rectangular weir
Numerical modeling using Acoustic Doppler Velocimetry (ADV). The efforts were undertaken to simulate the hydrodynamic field over
Broad-crested weir
the weir using open-source toolbox OpenFOAM with five turbulence models including standard k-ε, RNG k-ε,
Turbulence models
OpenFOAM
realizable k-ε, k-ω SST and LRR. The numerical results were compared to the experimental data obtained from
laboratory measurements. It demonstrated that the k-ω SST and LRR models had the best performance in cases
dealing with vortices. Then, separation zone patterns at three locations: behind, over and in front of the weir
under different head ratios were examined. It was found that changing the head ratio does not have a notable
effect on the relative dimension of separation flow over the weir, whereas the separation zone located behind and
in front of the weir grew linearly. After raising the head ratio, the downstream separation zone was split into two
significant vortices rotating oppositely. Discharge coefficients are calculated for a wide range of head ratios. It
was concluded that Cd equaled 0.85 for head ratios up to 0.3 and then grew to 1.0 by increasing the head ratio.

1. Introduction because of their durability, are commonly employed for regulation of

water level and flow measurements in small to medium-sized channels
As a type of hydraulic structure, a weir is generally employed for and rivers. The upstream vertical face of the weir would cause a major
flow measurement in open channels, flow diversion to a reservoir, and loss of head and sediment deposition [4]. This may decrease the accu­
flow regulation in hydropower plants [1]. According to the relative racy of discharge measurements and raise operating costs [5]. Offsetting
length of the weir crest, weirs are commonly classified into four types the structure of the weir and forming a structure of weir-culvert is
[2]: proposed as a solution for sediment deposition problems in the weir’s
⎧ upstream [6,7]. Decreasing the undesirable flow conditions is also
⎪ 0 < H0 /Lw ≤ 0.1 long − crested ​ weirs
⎪
⎪
⎪ possible by using a suitable curvature of streamlined weirs [8]. The
⎪
⎪
⎪
⎪ upstream and downstream of the circular-crested weirs have the lowest
⎨ 0.1 < H0 /Lw ≤ 0.4 broad − crested ​ weirs
(1) and highest bed shear stresses, respectively. According to this, the up­
⎪
⎪ stream zone of the weir with a relative eccentricity of unity in com­
⎪ 0.4 < H0 /Lw ≤ 1.5 short − crested ​ weirs
⎪
⎪
⎪
⎪
⎪ parison to lower relative eccentricity the has superior potential of
⎩
1.5 < H0 /Lw sharp − crested ​ weirs sedimentation. The advantage of the streamlined weir dominates the
circular-crested weir, even though, the discharge coefficient is decreased
where H0 is the total upstream head and Lw is the length of the weir crest by rising the height of the weir [8]. However, the broad-crested weir can
in the streamwise direction (see Fig. 2). deal with these issues by providing discharge efficiency and better hy­
Some advantages of broad-crested weirs such as insensitivity to draulic characteristics because of its structural design flexibility [5].
tailwater submergence, low cost, and structural stability, make them Several researchers have investigated experimentally and numeri­
preferable to other types of weirs [2,3]. Moreover, broad-crested weirs, cally the free surface flow over broad-crested weirs in their studies

* Corresponding author. Deh-Vanak Ave., Tehran, 1993893973, Iran.

E-mail address: [email protected] (H. Imanian).

https://doi.org/10.1016/j.flowmeasinst.2021.102004
Received 10 January 2021; Received in revised form 10 May 2021; Accepted 28 June 2021
Available online 2 July 2021
0955-5986/© 2021 Elsevier Ltd. All rights reserved.
H. Imanian et al. Flow Measurement and Instrumentation 80 (2021) 102004

[9–17]. Afshar et al. [9] discussed the predictive ability of CFD models studied broad-crested weirs with openings and their influence on water
over a broad-crested weir using the obtained values of basic experi­ free surface profiles, hydraulic jump length, distributions of velocity,
ments. The numerical results of 3D simulation of water level profile and and discharge coefficient [32,38–41]. Other researchers investigated the
streamlines with different turbulence models were in good agreement effect of weir surface roughness on the discharge coefficient [42,43].
with experimental data in which the RNG model had the best perfor­ They found that the discharge coefficient decreases by increasing the
mance [9]. Haun et al. [18] simulated the flow over a trapezoidal roughness of the weir’s crest; however, the upstream and downstream
broad-crested weir using Flow-3D and SSIIM 2 numerical codes with the roughness of the embankment weirs did not have a significant effect on
experimental study [4]. Considering types of grid and algorithms of the discharge coefficient [42]. Lv et al. [44] applied a newly developed
water free-surface were the major differences in their numerical study. free surface flow model to predict the challenging overflow problem for
The total computational time of Flow-3D is much less than SSIIM 2 broad-crested weirs.
because Flow-3D’s scheme is discretized more accurately [18]. Har­ The key purpose of this study is to determine to what extent the
greaves et al. [19] investigated the flow characteristics over a vertical chosen turbulence models can improve flow field prediction over the
faced weir and evaluated the accuracy of their CFD model with an broad-crested weir. Literature review showed most of the numerical
experimental study [2]. Considering that turbulence models have an investigations focused on linear eddy viscosity models and there is a
effective role in resolving detailed separated areas, standard k-ε model is room for improvement in simulation to apply more sophisticated Rey­
not accurate enough in estimating these regions. The variation in the nolds stress transport turbulence model like LRR. In this regard, five
downstream discharge, which is related to the highly turbulent area at turbulence closures of standard k–ε, RNG k–ε (Renormalized Group),
just the weir’s downstream, diminished by increasing the distance of the realizable k–ε, k–ω SST (Shear Stress Transport) and LRR (Launder-
defined downstream boundary from the turbulent region [19]. The Reece-Rodi) with extensive usage and competitive advantages were
longitudinal and transverse free-surface profiles over the simple geom­ used to assess the performance of the RANS models in flow simulation
etry of a rectangular broad-crested weir were investigated experimen­ over the broad-crested weir. Meanwhile, experimental tests were con­
tally and numerically by Sarker et al. [20]. Helmi [21] considered ducted in a straight 5-m-long flume to investigate the flow field around a
different turbulence models (one-equation, k-ε, RNG k-ε, and k-ω) for broad-crested weir. The flow velocity components were measured in 3
simulating the full free surface profile of flow over a broad-crested weir directions for 8 water depths at 8 cross sections instantaneously using an
floodway from upstream to downstream. 1D analysis was able to predict Acoustic Doppler Velocimeter (ADV). In the next step, the flow domain
the water level with acceptable accuracy, which is significant at the around the broad-crested weir was simulated numerically using Open­
floodway’s location. However, there was no considerable difference in FOAM [45]. The recorded velocity profiles were used to validate the
the accuracy of estimation for the different considered turbulence numerical simulation results and determine the optimal turbulence
models, because of the low turbulence intensity at the weir’s upstream models with the best measurement agreement. Then, additional com­
part [21]. Guven et al. [17,22] investigated the hydraulic characteristics putations were carried out to evaluate whether changing the hydraulic
of simultaneous flow over the broad crested weir and through square head ratio significantly affects the flow field and discharge coefficient
culverts experimentally. Broad-crested weirs made of metal and presented in the literature [46].
tempered hardboard were used in Doeringsfeld and Baker’s [23] labo­ Within the aforementioned outline, the other parts of the paper are
ratory study. The effects of rounded upstream corners of a broad-crested structured as follows. The experimental setup and laboratory procedure
weir were investigated in Smith’s [24] experimental tests. Discharge are described in the next section. Afterwards, the configuration of the
coefficient results predicted by Rao and Muralidhar [25], using a simulation including governing equations, turbulent closures and nu­
square-edged long-crested weir and a broad-crested weir, were consis­ merical modeling features are provided. In the fourth section, the
tently higher than the results available in the literature. model’s validation is explained. The simulation results and discussions
A separation zone of the flow over a rectangular broad-crested weir are described in the next part. Lastly, some concluding remarks com­
was numerically investigated by Paik et al. [26] employing RNG k-ε for plete the study in the last section.
turbulence simulation in the model. The results of the model were in
good agreement with physical models. Madadi et al. [5] experimentally 2. Material and method
investigated the effect of an upstream slope of a trapezoidal
broad-crested weir on discharge coefficient, development of separation 2.1. Experimental setup
zone, and water surface profiles. Decreasing the upstream slope led to
decrease in pressure drop and water surface and also preventing the The experiments were conducted on a horizontal, glass walled and
separation zone development [5]. Goodarzi et al. [27] conducted smooth bed flume of 512 cm total length and 480 cm of working length.
experimental tests to investigate the effect of an upstream slope on the The rectangular flume cross section area is 61 cm wide and 32 cm deep.
discharge coefficient, the surface pattern of flow, separation zone, and The weir block was a sharp corner plexiglas box with 45 cm crest length,
velocity profile. A correction factor for predicting the discharge coeffi­ 12 cm height and the same width of the flume. Both upstream and
cient and a mathematical relationship for the surface profile of flow downstream sides of the weir block were vertical. It was placed at a
were proposed by Goodarzi et al. [27]. Kamath et al. [28] numerically distance of 2 m away from the entrance of the flume. Fig. 1 shows the
studied the turbulence damping of complex free surface flow over a weir laboratory flume and the weir physical model.
and near bridge piers using RANS (Reynolds Averaged Navier-Stokes) In the closed-circuit system, a pump-controlled hydraulic system
simulations. In that model, an additional boundary condition for dissi­ delivered the flow rate, enabling an accurate discharge adjustment. The
pation of turbulence was defined to restrict the free surface. This pro­ upstream and downstream ends of the flume were connected to the
vided a better estimation of the flow features, especially when the flow stilling basins. Screens were provided at the upstream basin to produce a
conditions were changing quickly. Flow measurements over the weir smooth and uniform inlet flow. Water surface levels on the upstream
were significantly affected by the upstream face slope [29]. The influ­ side of the weir were measured on the centerline of the flume with a
ence of a broad-crested weir upstream and downstream face slope on the mechanical point gauge with an accuracy of 1 mm. The Reynolds
discharge coefficient was considered by many studies [3,4,30–33]. The number was sufficiently high in all experiments, so the flow regime
findings showed that the slope of the weir face and discharge coefficient could be considered fully turbulent.
have a close relationship [34–37]. When the slope of the upstream face The flow velocity profile was measured using an Acoustic Doppler
decreases from 90◦ to 10◦ , the discharge coefficient values grow and the Velocimeter (ADV). The fixed-stem downward-looking Vectrino ADV
separation zone is dissipated [27]. Also, the critical depth occurs nearer measures instantaneous velocity components in 3 dimensions at a single-
to the upstream edge of the broad-crested weir [29]. Some researchers point. In this experiment, ADV was installed on a mount which could

2
H. Imanian et al. Flow Measurement and Instrumentation 80 (2021) 102004

Fig. 1. (a) View of the laboratory flume (b) the weir block used in the experiment (c) schematic of the physical model.

move vertically and horizontally along the flume. Mean velocity was 2.3. Turbulence models
obtained by post-processing of instantaneous ADV velocity data. The
velocity profile was measured at the centerline of six sections at dis­ Determining the accurate flow characteristics highly depends on
tances of 4, 8, 14, 20, 40 and 110 cm upstream of the weir. In each selecting the correct turbulence model in the simulation [1]. Therefore,
section, velocity was measured at 8 depths from the water free surface to different turbulence models are investigated in the simulation:
near the flume bed. two-equation RANS models such as standard k–ε, RNG k–ε, realizable
After some examinations, a frequency of50 Hz was selected for data k–ε, k–ω SST and also LRR, which is a second-order model and not
collecting. Also, it was found that a 3-min measurement can be well classified in linear eddy-viscosity closures, unlike the prior models. In
represented the flow. The minimum signal-to-noise ratio was 30 and the the two-equation eddy-viscosity turbulence models, two independent
correlation was at least 95% for all the cases. variables are derived from solving two distinct transport equations
which are related to the time scales and turbulence length directly.
2.2. Governing equations The standard k–ε model is widely employed to compute the turbulent
kinematic viscosity in turbulence models. The equations of standard
The incompressible continuity and conservation of momentum transport are as follows [50]:
equations are employed, which are given as [47,48]: [( ) ]
∂k ∂kui ∂ ν ∂k
+ = ν+ T + Gk − ε (5)
∇.U = 0 (2) ∂t ∂xi ∂xi σk ∂xi
[( ) ]
∂ρU ∂ε ∂εui ∂ νT ∂ε ε ε2
+ ∇.(ρUU) = − ∇p + ρg + ∇.τ (3) + = ν+ + C1ε Gk − C2ε (6)
∂t ∂t ∂xi ∂xi σ ε ∂xi k k

where U is the velocity vector of fluid flow; t is the time; p is the k2

pressure; ρ is the density; g is the gravitational acceleration vector and τ νT = Cμ (7)
ε
is the shear-rate tensor.
The RANS (Reynolds averaged Navier–Stokes) equations, time- where k is the turbulent kinetic energy, ε is the turbulent kinetic energy
averaged form of momentum equation (Eq. (3)), were used for solving dissipation rate, ν is the kinematic viscosity, νT is the turbulence kine­
the fluid flow motion in the model, and can be written as [49]: matic viscosity and σk , σε , C1ε , C2ε , Cμ are constant.
The RNG k– ε turbulence model is the k– ε family, which uses the
∂ui ∂ui 1 ∂p ∂2 ui ∂u′i u′j
+ uj = − +ν − (4) modified transport equation of the turbulent kinetic energy dissipation
∂t ∂xj ρ ∂xi ∂xj ∂xj ∂xj rate (ε) to calculate the eddy viscosity as follows [48]:
[( ) ] ( )
where xi and xj are the Cartesian coordinate axes, ui and uj are the ∂ε ∂εxi ∂ ν ∂ε ε Cμ η3 (1 − η/ηo ) ε2
+ = ν+ T + C1ε Gk − C2ε + (8)
averaged flow velocity components, ν is the kinematic viscosity, and ui ∂t ∂xi ∂xi σε ∂xi 1 + β η3
′
k k
uj is the flow velocity fluctuation about the time averaged value.
′
The realizable k– ε turbulence model modifies the dissipation rate
formulation by using a few restrictions on normal turbulent stresses to

3
H. Imanian et al. Flow Measurement and Instrumentation 80 (2021) 102004

prevent them from being negative in the solution [51]. The developed models and several numerical schemes. InterFoam multiphase solver in
realizable k– ε model is given as follows [49]: OF240 was used for the simulation. Velocity-pressure coupling of RANS
[( ) ] equations were solved using the PIMPLE algorithm.
∂k ∂kui ∂ ν ∂k
+ = ν+ T + Gk − ε + Gb − YM + Sk (9) Numerical simulations were employed to investigate the flow char­
∂t ∂xi ∂xi σ k ∂xi
acteristics based on the physical modeling conducted by the authors as
[( ) ] explained in the previous section. The simulated horizontal rectangular
∂ε ∂εui ∂ νT ∂ε ε ε2
+ = ν+ + C1ε C3ε Gb + C1 Sε − C2 √̅̅̅̅̅ + Sε flume was 3.335 m long and 0.325 m tall. The height and the length of
∂t ∂xi ∂xi σε ∂xi k k + νε
the broad-crested weir were 0.12 m and 0.45 m, respectively. The ge­
(10) ometry of the simulation domain can be seen in Fig. 2. The distance from
the inlet boundary to the weir was much more than the height of the
where Sε is the mean strain rate, Gk is the generation of turbulence
weir to minimize the influence of the inlet boundary. Also, the tail water
kinetic energy due to the mean velocity gradients, Gb is the generation of
was kept low to prevent submerged flow condition.
turbulence kinetic energy due to buoyancy, YM is the contribution of the
The defined BCs (boundary conditions) applied in the numerical
fluctuating dilatation in compressible turbulence to the overall dissi­
simulation are depicted in Fig. 3. Two distinct inlet BCs were defined for
pation rate, Sk and Sε are source terms, and σ k , σ ε C1ε , C3ε , C1 , C2 are
the air and water inflow. The streamwise inflow water velocity was
constants.
equal to 0.20 m in compliance with the experimental test. A fixed value
The k− ω SST turbulence model employs a k– ω turbulence model to
BC was assigned for other parameters at the inlet water BC. No slip
compute the inner boundary layer flow and a k- ε turbulence model in
condition for the velocity and standard wall function for other param­
the free-shear flow [52]. This model also exhibits good behavior with
eters were determined at the bottom of the flume. At the outlet patch, a
respect to separating flow and adverse pressure gradients [53].
zero gradient BC was determined. For the atmosphere boundary, the
[ ]
∂k ∂k ∂ ∂k inlet-outlet boundary condition was allotted. Note that the weir has a
+ ui = (ν + σ k νT ) + Pk − β * k ω (11)
∂t ∂xi ∂xi ∂xi free wide crest and walls that do not significantly affect the flume’s
[ ] centerline results.
∂ω ∂ω ∂ ∂ω 1 ∂k ∂ω The developed model was performed with three grid sizes of 5, 2.5
+ ui = (ν + σω νT ) + αS2 − βω2 + 2(1 − F1 )σω2
∂t ∂xi ∂xi ∂xi ω ∂xj ∂xj and 2 mm for sensitivity analysis. The calculated streamwise velocity
(12) profile just behind the weir is shown in Fig. 4. It can be seen that the
horizontal velocity results were changed by decreasing the mesh size
νT =
a1 k
(13) from 5 to 2.5 mm. This confirms that model results still were sensitive to
max(a1 ω, SF2 ) the grid size. Using finer grids from 2.5 to 2 mm, the velocity profile did
The LRR turbulence model is a type of the standard Reynolds Stress not show any variations and made the model results independent from
Model (RSM) in which six extra transport equations are solved for each the grid size. Therefore, the mesh size of 2.5 mm is the rational choice
time step instead of using the eddy viscosity approach. Convergence for for grid size considering computational cost reduction.
the RSM in comparison to the k- ε model would be slower because of its The computational domain with structured mesh is presented in
complicated source terms [54]. Fig. 5. Applying a uniform mesh to the whole domain is computationally
very expensive, so discretization was performed with a high resolution

[ ] ( ) ( )
∂ ( ′ ′) ∂ ( ) ∂ [ ′ ′ ′ )] ∂ ( ′ ′)
′ ′
( ∂ ′ ′ ∂uj ′ ′ ∂ui ∂u′i ∂uj ∂u′ ∂uj
ρui uj + ρuk u′i u′j = − ρui uj uk + p′ δkj u′i + δki u′j + μ uu − ρ ui uj + uj uk + p′ + − 2μ i
∂t ∂xk ∂xk ∂xk ∂xk i j ∂xk ∂xk ∂xj ∂xi ∂xk ∂xk
( )
(14)
′ ′
− 2ρΩk uj u′m εikm + ui u′m εjkm

around the weir. The total number of cells is almost 29 000. This value
increases to 47 000 for the LRR turbulence model because of its sensi­
Further information about turbulence models and equations can be tivity to grid uniformity.
found in Refs. [55,56]. OpenFOAM employed the finite volume method. The defined time
step was 0.001s, but the adjustable time step was assigned, which de­
pends on the Courant number. In this way, the software was capable of
2.4. Numerical model features calculating the value of time step in each computational cycle, and if it is
essential, according to solution stability, the time step value can be
In this study, 2DV OpenFOAM (Open Field Operation And Manipu­ changed. The Courant number was less than 0.2. In the computations,
lation) version 2.4.0 (OF240) was used for the simulation of flow over a the residual error, as a convergence criterion was within 10− 5 for all
broad-crested weir. This open source CFD software includes packages computed variables including Ux , Uy , k, ε, ω. Different numerical
using C++ and has the capability of implementing different turbulence

Fig. 2. The geometry of the flume and the weir in the numerical simulation.

4
H. Imanian et al. Flow Measurement and Instrumentation 80 (2021) 102004

Fig. 3. Boundary conditions in the numerical simulation.

Fig. 4. Grid size sensitivity analysis for (a) horizontal velocity results with different grid size (b) closer view.

Fig. 5. Computational mesh (a) entire view (b) closer view (c) near the weir crest.

Fig. 6. Water and air phases and simulated free surface (a) entire view (b) near the weir.

schemes such as Euler, backward, and Crank Nicolson for time schemes, Gauss linear upwind grad for Divergence scheme of velocity, Gauss
linear and linear Upwind with/without limited Grad for Gradient linear for Laplacian schemes led to better performance in accuracy and
schemes, and linear, linear Upwind, vanLeer, Cubic, QUICK for Diver­ numerical stability. After 30 s of simulation time, the flow field results
gence schemes were investigated to choose the most appropriate simu­ became steady. For the sake of conservativism, the simulations were
lation features. The temporal term was discretized using the Crank performed up to 35 s. The Volume of Fluid (VOF) method was employed
Nicolson scheme. It is observed that performing Gauss linear with to capture the interface of water and air. The simulated water free sur­
limited Grad of cell Limited Gauss linear 1 for Gradient schemes, and face and fraction of water and air phases are shown in Fig. 6. The initial

5
H. Imanian et al. Flow Measurement and Instrumentation 80 (2021) 102004

Fig. 9 (e) and (f), the normalized velocity profiles tended to be uniform
throughout most of the depth at sections far from the weir (x = − 0.40 m
and − 1.10 m), even though they illustrated a steep change in velocity
values close to the flume’s bottom because of the no-slip boundary
condition in which the velocity should be equal to zero. The effect of the
weir’s existence on the flow field was more visible at sections near the
weir (x = − 0.04 m and − 0.08 m) as shown in Fig. 9 (a) and (b).
To indicate the quality of numerical simulation, a scatter plot of
numerical velocity results against experimental data was produced in
Fig. 10. In Fig. 10, the solid line represents a complete fit of numerical
results to experimental data, and the band between two black dashed-
lines shows the ±20% error at 80% confidence level. An exact look at
Fig. 10 shows that LRR, k-ω SST, and standard k-ϵ models provided
better agreement with the laboratory observations. In contrast, the
points corresponding to the RNG k-ϵ and realizable k-ϵ results deviated
Fig. 7. Free surface profiles calculated by different turbulence models.
farther from the matching line.
To quantify the difference between the laboratory and simulated
water elevation of the simulation was set to 15 cm for the full flume level velocity values, some indicators are used, which are given by:
behind the weir. √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
1 ∑ n
( )2
RMSE = uexp − ucalc (15)
3. Results N i=1

3.1. Free surface validation RMSE

NRMSE = ( ) ( ) (16)
max uexp − min uexp
The main objective of this paper is to investigate the flow behind a
broad-crested weir. To achieve this, the developed numerical model was RMSE2
NMSE = ( ( ) ( )) ) (17)
investigated using five turbulence models of standard k–ε, RNG k–ε, max uexp − min uexp .(max(ucalc ) − min(ucalc )
realizable k–ε, k–ω SST and LRR. The water free surface profiles calcu­
lated by different turbulence models are presented in Fig. 7. n ⃒ ⃒
1 ∑ ⃒uexp − ucalc ⃒
It can be seen in Fig. 7 that all turbulence models computed the free MAPE = ⃒
⃒
⃒ (18)
N i=1 uxexp ⃒
surface similarly. The free surface profile experiences a slowly
descending part at the beginning of the weir, and a sharply descending 1 ∑n ⃒ ⃒
part after the weir which is like a hydraulic drop. Regarding Fig. 7, it is MAE = ⃒uexp − ucalc ⃒ (19)
N i=1
clear that the different turbulence models have nearly the same value of
free surface elevation of water which is equal to 20 cm upstream of the where uexp is the experimental measurement and ucalc is the calculated
flume. This confirms that it is required to employ a more accurate flow result.
characteristic like the vertical velocity profile rather than the water free The results of the error analysis at six sections are presented in
surface elevation to assess the accuracy of the turbulence models’ Table 1. The error values in Table 1 suggested that the standard k-ϵ, k-ω
performance. SST, and LRR turbulence models have better performance, especially
when far from the weir.
3.2. Velocity profile validation
3.3. Critical depth validation
The laboratory test data collected by the authors were employed to
validate the developed model results. Six sections upstream of the weir For the next step of validation, the flow critical depths computed by
are considered for verification purposes (see Fig. 8). The first five sec­ numerical simulation using various turbulence models were compared
tions are closer to the weir at 4, 8, 14, 20 and 40 cm in the upstream with the analytical one, which is given as follows:
direction. The last section is far away from the weir at a distance of 110 √̅̅̅̅̅ √̅̅̅̅̅̅̅̅̅̅̅̅̅
cm. 2 2
3 q 3 (U.h)
yc = = (20)
Vertical profiles of streamwise velocity at the indicated sections g g
modeled with different turbulence closures were compared with
experimental data measured by the authors in Fig. 9. The velocity values where yc is the critical depth, q is the flow rate per unit flume width, g is
are made dimensionless by dividing them by the discharge velocity, and gravitational acceleration, U is uniform inflow velocity and h is the
the vertical axis is divided by the water depth. water depth.
According to Fig. 9, the standard k-ϵ, RNG k-ϵ, realizable k-ϵ, k-ω In the present work, the inflow velocity and water depth were 0.19
SST, and LRR turbulence models show a similar trend and they are in a m/s and 0.20 m, respectively. Thus, the analytical critical depth will be
good agreement with the available experimental data. Considering 0.053 m.

Fig. 8. The locations of six specified sections at the flume for validation.

6
H. Imanian et al. Flow Measurement and Instrumentation 80 (2021) 102004

Fig. 9. Comparison of normalized velocity profiles using different turbulence models and experimental data at (a) x = − 0.04m (b) x = − 0.08m (c) x = − 0.14m (d)
x = − 0.20m (e) x = − 0.40m (f) x = − 1.10m.

The flow critical depths computed by numerical simulation using

various turbulence models were compared with the analytical and
relative errors and are presented in Table 2, which are given by:
yanal − ycalc
RE = (21)
yanal
Error analysis in Table 2 shows that standard k-ϵ turbulence model
simulates the critical depth of the flow field accurately, while the LRR
model presents an overestimated prediction. Applying other turbulence
models results in very similar error values.

4. Discussion

Based on the results presented in section 3, the standard k-ϵ, k-ω SST
and LRR turbulence models had better performance in flow predictions.
For additional study of the predictive abilities of the numerical simu­
lation of flow field around the broad-crested weir, turbulence charac­
teristics and the pressure domain simulated by these selected turbulence
models were studied in detail.

Fig. 10. Comparison of the normalized velocity simulated by different turbu­

4.1. Flow field and streamlines
lence models and the laboratory measurements.

The streamlines and velocity magnitude contours of selected turbu­

lence models are presented in Fig. 11. Fig. 11 shows that the velocity
predicted by the LRR turbulence model is greater than the velocity

7
H. Imanian et al. Flow Measurement and Instrumentation 80 (2021) 102004

Table 1
Error analysis of the computed streamwise velocity at six specified sections using different turbulence models.
Location (m) Turbulence model RMSE NRMSE NMSE MAPE MAE R2

x = -0.04 standard k-ϵ 0.151 0.107 0.011 0.123 0.119 0.906

RNG k-ϵ 0.126 0.089 0.008 0.093 0.093 0.926
realizable k-ϵ 0.119 0.084 0.006 0.113 0.093 0.956
k-ω SST 0.119 0.084 0.007 0.082 0.088 0.938
LRR 0.132 0.094 0.008 0.102 0.098 0.918
x = -0.08 standard k-ϵ 0.122 0.157 0.017 0.113 0.087 0.981
RNG k-ϵ 0.132 0.170 0.020 0.130 0.097 0.960
realizable k-ϵ 0.195 0.251 0.038 0.196 0.145 0.918
k-ω SST 0.186 0.240 0.035 0.185 0.138 0.937
LRR 0.132 0.170 0.020 0.128 0.096 0.965
x = -0.14 standard k-ϵ 0.020 0.039 0.001 0.019 0.016 0.987
RNG k-ϵ 0.028 0.054 0.003 0.027 0.024 0.973
realizable k-ϵ 0.052 0.102 0.009 0.048 0.041 0.919
k-ω SST 0.023 0.046 0.002 0.022 0.020 0.981
LRR 0.021 0.040 0.002 0.018 0.018 0.987
x = -0.20 standard k-ϵ 0.068 0.162 0.018 0.065 0.060 0.935
RNG k-ϵ 0.066 0.159 0.017 0.063 0.057 0.953
realizable k-ϵ 0.108 0.257 0.036 0.095 0.081 0.974
k-ω SST 0.086 0.206 0.025 0.078 0.069 0.947
LRR 0.060 0.144 0.015 0.058 0.055 0.930
x = -0.40 standard k-ϵ 0.026 0.087 0.006 0.019 0.016 0.935
RNG k-ϵ 0.029 0.099 0.008 0.020 0.017 0.953
realizable k-ϵ 0.039 0.134 0.013 0.027 0.024 0.974
k-ω SST 0.022 0.074 0.005 0.015 0.013 0.947
LRR 0.019 0.064 0.004 0.015 0.014 0.930
x = -1.10 standard k-ϵ 0.046 0.204 0.031 0.041 0.039 0.880
RNG k-ϵ 0.046 0.207 0.031 0.041 0.038 0.888
realizable k-ϵ 0.045 0.199 0.030 0.040 0.038 0.889
k-ω SST 0.044 0.196 0.029 0.040 0.038 0.881
LRR 0.039 0.175 0.024 0.036 0.034 0.892

weir by all turbulence models.

Table 2
The third rotation zone was formed at the downstream corner of the
Error values of the computed critical depth using various turbulence models.
weir as visualized in Fig. 12. Although all turbulence models predicted
Turbulence model Computed yc (m) RE (%) this zone with the same dimensions, the shape of eddies are different. As
standard k-ϵ 0.054 1.9 discussed in the literature [19], the standard k-ϵ model was not capable
RNG k-ϵ 0.055 3.8 of resolving the separated regions well enough. However, the k-ω SST
realizable k-ϵ 0.049 7.5 and LRR models, as more sophisticated turbulence models than the
k-ω SST 0.055 3.8 standard k-ϵ model, are able to predict these regions near the walls more
LRR 0.062 16.8
accurately. According to Fig. 12, the standard k-ϵ model just showed a
simple eddy and had poor performance in capturing the separation zone
computed by the standard k-ϵ and k-ω SST models. precisely. As shown in Fig. 12, k-ω SST and especially LRR models were
When the flow passes over the weir with a square-edged entrance, capable of predicting some small rotation zones sticking to the weir’s
three separation zones around the weir were generated [57], as captured vertical wall. This shows that LRR is a beneficial method to indicate this
in Fig. 11. The eddy structures formed into separation zones where the phenomenon. Consequently, the k-ω SST and LRR turbulence models
flow is separated and then passes over the zones [58]. The flow sepa­ predict the separation zone more accurately than the standard k-ϵ
ration zone over the weir had a striking effect on the capacity of the weir model.
[58] that will be discussed in the following sections. This zone can be
seen in Fig. 11 at the early part of the weir. Moreover, the developed
model could predict another rotation zone at the upstream corner of the

Fig. 11. The predicted streamlines and velocity contours using (a) standard k-ϵ (b) k-ω SST (c) LRR turbulence model.

8
H. Imanian et al. Flow Measurement and Instrumentation 80 (2021) 102004

Fig. 12. Predicted downstream vortices computed by (a) standard k-ϵ (b) k-ω SST (c) LRR turbulence model.

Fig. 13. The predicted eddy viscosity contours using (a) standard k-ϵ (b) k-ω SST (c) LRR turbulence model.

Fig. 14. The predicted pressure field using (a) standard k-ϵ (b) k-ω SST (c) LRR turbulence model.

4.2. Turbulence characteristics 4.3. Pressure field

The contours of eddy viscosity around the broad-crested weir for the The pressure contours predicted by selected turbulence models are
selected models are shown in Fig. 13. Since velocity was low for all presented in Fig. 14.
turbulence models before the weir, it can be seen that flow was not When the negative flow pressure occurs, the surface of the weir be­
considerably turbulent; whereas the velocity increased at the water’s comes vulnerable to cavitation. In the aforementioned turbulence clo­
surface of the reservoir and eddy viscosity grew slowly. Transiting over sures, LRR is the optimal tool for capturing this issue. As displayed in
the weir, velocity increased and hence, a turbulent area developed after Fig. 14, a small negative pressure zone was seen forming downstream of
the weir. According to Fig. 13, different turbulence models do not pre­ the weir in the simulation using the LRR turbulence model.
dict a similar area of high eddy viscosity. The eddy viscosity value
predicted by the standard k-ϵ model is much more than the values
4.4. Flow field under different head ratios
predicted by the k-ω SST and LRR models.
As discussed in previous sections, the capacity of the weir was

9
H. Imanian et al. Flow Measurement and Instrumentation 80 (2021) 102004

Fig. 15. Separation zones at upstream behind the broad-crested weir for different head ratios (a) H/h = 0.27 (b) H/h = 0.37 (c) H/h = 0.41 (d) H/ h = 0.45 (e) H/
h = 0.52 (f) H/h = 0.6.

Fig. 16. Separation zones over the broad-crested weir for different head ratios (a) H/h = 0.27 (b) H/h = 0.37 (c) H/h = 0.41 (d) H/h = 0.45 (e) H/ h = 0.52 (f) H/
h = 0.6.

affected by the separation zones [59]. Moreover, the existence of In Fig. 17, a changing pattern for rotating zones was observed. For a
rotating flows around the weir may cause pulsation of surface water by small head ratio, an eddy has stuck to the top of the weir vertical face as
affecting the water flowing over the separation zones [60]. Due to the shown in Fig. 17 (a). Increasing the head, a rotating flow formed at the
mentioned drawbacks, it is essential to investigate the shape and front corner of the weir, which can be seen in Fig. 17 (b)–(e). This eddy
behavior of these zones. grew rapidly and became the main rotating zone in front of the weir. In
As mentioned earlier, the k-ω SST model showed good behavior in addition, the initial eddy stuck to the weir wall started shrinking. For a
adverse pressure gradients and separating flow. In addition, this tur­ high head ratio in Fig. 17 (f), the weir wall eddy was eliminated and two
bulence model proved its satisfactory performance in simulation of flow major eddies were recognizable on the bed downstream. The rotation
over a broad-crested weir in the validation section. Therefore, the above direction of these eddies is illustrated in Fig. 18 using velocity arrows. It
discussions on separation zones and vortices apply to the results can be seen that the larger eddy turned clockwise, while the swirl di­
computed by the k-ω SST turbulence model as well. rection of the other one was counter-clockwise.
In this section, the effects of head ratio (H /h) on separation zone To improve understanding of head ratio effects on the rotating zones
patterns at three cross sections of the upstream, over, and downstream of size, relative dimension changes of eddies are considered in Fig. 19. LW
the broad-crested weir are investigated (marked in Fig. 11). H and h are is the weir length, P is the weir height (see Fig. 2) and LX andLZ are the
the hydraulic head on the weir and the water depth, respectively, which length and height of eddy, respectively.
vary from 0.25 to 0.6 in the present study. The shape and dimension of Based on Fig. 19 (a), the length ratios of rotating flows upstream and
rotating flows around the weir formed under the different head ratio are downstream have a direct relationship with the head ratio. The up­
presented in Fig. 15 to Fig. 17. stream rotating flow length ratio changed much less than the down­
As shown in Figs. 15–17, when increasing the head ratio, the di­ stream one. Considering Fig. 19 (b), the height ratios of the upstream
mensions of eddies grew in all cases. The generated rotating flow at the and downstream rotating flows have a direct and inverse relationship
upstream had two cores that are recognizable in Fig. 15. Fig. 16 shows with the head ratio, respectively.
that parallel streamlines approach the weir and flow over the eddy According to Fig. 19 (c), the variation of head ratio did not have an
formed on the crest. essential effect on the length and height ratios of the recirculation zone

10
H. Imanian et al. Flow Measurement and Instrumentation 80 (2021) 102004

Fig. 17. Separation zones at downstream in front of the broad-crested weir for different head ratios (a) H/h = 0.27 (b) H/h = 0.37 (c) H/h = 0.41 (d) H/ h = 0.45 (e)
H/h = 0.52 (f) H/h = 0.6.

following correlation for discharge coefficient [46]:

Cd = 0.873(H/h)2 − 0.3(H / h) + 0.878 (24)

The discharge coefficients computed by the developed numerical
model for different head ratios were compared with the values available
in the literature and the aforementioned correlation and its changes
versus head ratio are depicted in Fig. 20.
A precise look at Fig. 20 indicates that the discharge coefficient
variation is divided into two main parts. At the first part, Cd has an
approximately fixed value for a head ratio less than 0.3. The next part
shows that the discharge coefficient relates to the pressure on the weir
directly, in a way that it increases up to 1.0 for a head ratio of 0.6. The
obtained results of the present numerical model followed perfectly the
same trend. Furthermore, the RMSE statistical parameter equals 1.5%,
which confirms the accurate prediction of the developed model.

Fig. 18. Velocity arrows and rotation direction of eddies downstream of the 5. Conclusions
weir for.
In this research, the flow field over a broad-crested weir was inves­
over the weir. Therefore, the dimensions of the rotating flow over the tigated. An experimental study was conducted in a 5 m flume to measure
weir were approximately constant by increasing the head ratio. Di­ the flow characteristics over the weir using Acoustic Doppler Velocim­
mensions varied 12%–23% for the models in the range of broad-crested etry (ADV). The experimental data was used for validation of numerical
weirs. simulations. The efforts were undertaken to study the hydrodynamic
domain over the weir using open-source package, OpenFOAM. The
4.5. Discharge coefficient computed results from a 2DV numerical simulation with five turbulence
models (standard k–ε, RNG k–ε, realizable k–ε, k–ω SST and LRR) were
The discharge coefficient indicates the weir’s efficiency and exclu­ compared to the experimental measurements obtained from the physical
sively depends on the actual head. This coefficient for the broad-crested model. Flow characteristics and separation zones for the selected tur­
weir under the free flow condition is given as follows [29]: bulence models under different head ratios were investigated.
√̅̅̅̅̅̅̅̅̅̅̅̅̅ The main outcomes of this research are summarized as follows:
q = Cd 2gH0 3 (22)
• The comparisons of the computed velocity profile around the broad-
/ crested weir with laboratory measurements demonstrated that the
2
H0 = H + q 2gh 2
(23)
standard k-ϵ, k-ω SST and LRR model had better performance. The k-
where q is discharge per unit width; Cd is discharge coefficient; g is ω SST and LRR closures demonstrated their advantages in cases
gravitational acceleration; H0 is total overflow head and h is water depth dealing with streamline curvature and vortices, which are the stan­
(see Fig. 2). dard k-ϵ turbulence model’s limitations.
The discharge coefficient for the broad-crested weir was measured in • The separation zones formed at three locations: behind, over and in
various laboratorial efforts. Azimi and Rajaratnam (2009) collected the front of the weir under different head ratios were investigated. It was
experimental results available in the literature and proposed the noticed that changes in head ratio did not have a notable effect on

11
H. Imanian et al. Flow Measurement and Instrumentation 80 (2021) 102004

Fig. 19. Relative dimensions of rotating zone versus head ratio (a) eddy length (b) eddy height (c) eddy dimension over the weir.

Author statement

Hanifeh Imanian: Formal analysis, Investigation, Resources, Writing

– review & editing, Project administration, Abdolmajid Mohammadian:
Conceptualization, Methodology, Supervision, Funding acquisition,
Pouneh Hoshyar: Software, Validation, Writing – original draft,
Visualization.

Declaration of competing interest

The authors declare that they have no known competing financial

interests or personal relationships that could have appeared to influence
the work reported in this paper.

References
Fig. 20. Variation of discharge coefficient versus head ratio.
[1] B.F. Fulasa, Investigation on Different Shapes of Broad-Crested Weirs by Means of
CFD, NTNU, 2019.
the relative dimension of separation flow over the weir. However, [2] W.H. Hager, M. Schwalt, Broad-crested weir, J. Irrigat. Drain. Eng. 120 (1) (1994)
13–26.
the effect of variation of head ratio was more obvious on the sepa­
[3] L. Jiang, M. Diao, H. Sun, Y. Ren, Numerical modeling of flow over a rectangular
ration zone configuration located behind and in front of the weir. broad-crested weir with a sloped upstream face, Water 10 (11) (2018) 1663.
With increasing hydraulic head, the separation zone grew linearly. [4] J.E. Sargison, A. Percy, Hydraulics of broad-crested weirs with varying side slopes,
J. Irrigat. Drain. Eng. 135 (1) (2009) 115–118.
• By raising the head ratio, the eddy stuck to the weir’s downstream
[5] M.R. Madadi, A.H. Dalir, D. Farsadizadeh, Investigation of flow characteristics
face shrunk and the separation zone in front of the weir split up into above trapezoidal broad-crested weirs, Flow Meas. Instrum. 38 (2014) 139–148.
two significant vortices rotating in opposite directions. [6] S. Salehi, A.H. Azimi, Discharge characteristics of weir-orifice and weir-gate
• Discharge coefficients, as a practical parameter for weirs, were structures, J. Irrigat. Drain. Eng. 145 (11) (2019), 04019025.
[7] S. Salehi, A.H. Azimi, H. Bonakdari, Hydraulics of sharp-crested weir culverts with
computed for a wide range of head ratios according to the results of downstream ramps in free-flow, partially, and fully submerged-flow conditions,
the developed numerical model. It can be concluded that the Irrigat. Sci. 39 (2) (2021) 191–207.
discharge coefficient was approximately constant, equal to 0.85 for [8] S. Bagheri, A. Kabiri-Samani, Simulation of free surface flow over the streamlined
weirs, Flow Meas. Instrum. 71 (2020) 101680.
head ratios up to 0.3 and then grew to 1.0 by increasing the head [9] H. Afshar, S.H. Hoseini, Experimental and 3-D numerical simulation of flow over a
ratio. rectangular broad-crested weir, Int. J. Eng. Adv. Technol. 2 (6) (2013) 214–219.
[10] S.A. Al-Hashimi, H.M. Madhloom, R.M. Khalaf, T.N. Nahi, N.A. Al-Ansari, Flow
over broad crested weirs: comparison of 2D and 3D models, J. Civil Eng. Archit. 11
The results of the current study encourage researchers to investigate (8) (2017) 769–779.
the impact of the slope of faces and greater head ratios in the range of [11] A.H. Azimi, N. Rajaratnam, D.Z. Zhu, A note on sharp-crested weirs and weirs of
sharp-, short-crested weirs on the vortices using CFD. In addition, the finite crest length, Can. J. Civ. Eng. 39 (11) (2012) 1234–1237.
[12] A.H. Azimi, N. Rajaratnam, D.Z. Zhu, Submerged flows over rectangular weirs of
implementation of other advanced turbulence models like LES and DES finite crest length, J. Irrigat. Drain. Eng. 140 (5) (2014), 06014001.
can be considered in future work.

12
H. Imanian et al. Flow Measurement and Instrumentation 80 (2021) 102004

[13] S.S. Hakim, A.H. Azimi, Hydraulics of submerged triangular weirs and weirs of [37] H. Tong, K. Ai, X. Ding, Discharge capacity of broken-line practical weir, J. Yangtze
finite-crest length with upstream and downstream ramps, J. Irrigat. Drain. Eng. River Sci. Res. Inst. 19 (2) (2002).
143 (8) (2017), 06017008. [38] G. Abozeid, A.M. El-Belasy, S.M. Shehata, Simulation of flow over weirs with
[14] S.A. Jalil, S.A. Sarhan, Performance of flow over a weir with sloped upstream face, bottom pipes (case study: Bahar hasan Wasef weir), J. Eng. Sci. 42 (14) (2014)
J. Pure Appl. Sci. 29 (3) (2017) 43–54. 891–904.
[15] S.I. Khassaf, A.N. Attiyah, H.A. Al-Yousify, Experimental investigation of [39] R. Daneshfaraz, O. Minaei, J. Abraham, S. Dadashi, A. Ghaderi, 3-D Numerical
compound side weir with modeling using computational fluid dynamic, Int. J. simulation of water flow over a broad-crested weir with openings, J. Hydraul. Eng.
Energy Environ. 7 (2) (2016) 169. (2019) 1–9.
[16] R. Maghsoodi, M.S. Roozgar, H. Sarkardeh, H.M. Azamathulla, 3D-simulation of [40] A.M. El-Belasy, Developing Formulae for combined weir and orifice (case study:
flow over submerged weirs, Int. J. Model. Simulat. 32 (4) (2012) 237–243. EL-Fayoum weirs), Alexandria Eng. J. 52 (4) (2013) 763–768.
[17] N.O. Tănase, D. Broboană, C. Bălan, Free surface flow over the broad—crested [41] N.Y. Saad, E.M. Fattouh, Hydraulic characteristics of flow over weirs with circular
weir, in: 2015 9th International Symposium on Advanced Topics in Electrical openings, Ain Shams Eng. J. 8 (4) (2017) 515–522.
Engineering (ATEE), IEEE, 2015, pp. 548–551. [42] S. Felder, N. Islam, Hydraulic performance of an embankment weir with rough
[18] S. Haun, N.R.B. Olsen, R. Feurich, Numerical modeling of flow over trapezoidal crest, J. Hydraul. Eng. 143 (3) (2017), 04016086.
broad-crested weir, Eng. Appl. Comput. Fluid Mat. 5 (3) (2011) 397–405. [43] J. Pařílková, J. Říha, Z. Zachoval, The influence of roughness on the discharge
[19] D. Hargreaves, H. Morvan, N. Wright, Validation of the volume of fluid method for coefficient of a broad-crested weir, J. Hydrol. Hydromechanics 60 (2) (2012)
free surface calculation: the broad-crested weir, Eng. Appl. Comput. Fluid Mat. 1 101–114.
(2) (2007) 136–146. [44] X. Lv, Q. Zou, D. Reeve, Numerical simulation of overflow at vertical weirs using a
[20] M. Sarker, D. Rhodes, Calculation of free-surface profile over a rectangular broad- hybrid level set/VOF method, Adv. Water Resour. 34 (10) (2011) 1320–1334.
crested weir, Flow Meas. Instrum. 15 (4) (2004) 215–219. [45] O.U. Guide, Available online, OpenFOAM Foundation, London, UK, 2011, https
[21] A.M. Helmi, Assessment of CFD turbulence models for free surface flow simulation ://www.openfoam.com/documentation/user-guide/. (Accessed 14 November
and 1-D modelling for water level calculations over a broad-crested weir oodway, 2018).
WaterSA 45 (3) (2019) 420–433. [46] A.H. Azimi, N. Rajaratnam, Discharge characteristics of weirs of finite crest length,
[22] A. Guven, M. Hassan, S. Sabir, Experimental investigation on discharge coefficient J. Hydraul. Eng. 135 (12) (2009) 1081–1085.
for a combined broad crested weir-box culvert structure, J. Hydrol. 500 (2013) [47] T. Holzmann, Mathematics, Numerics, Derivations and OpenFOAM®, Holzmann
97–103. CFD, Loeben, Germany, 2016.
[23] H. Doeringsfeld, Pressure-monentum theory applied to the broad-crested weir, [48] H. Imanian, A. Mohammadian, Numerical simulation of flow over ogee crested
Trans. ASCE 106 (1941) 934–946. spillways under high hydraulic head ratio, Eng. Appl. Comput. Fluid Mat. 13 (1)
[24] C. Smith, Open channel water measurement with the broadcrested weir, Bull. Int. (2019) 983–1000.
Commn Irrig. Drain (1958) 46–51. [49] E. Furbo, Evaluation of RANS Turbulence Models for Flow Problems with
[25] N.G. Rao, D. Muralidhar, Discharge characteristics of weirs of finite-crest width, La Signigicant Impact of Boundary Layers, 2010.
Houille Blanche (5) (1963) 537–545. [50] B.E. Launder, D.B. Spalding, Mathematical Models of Turbulence, Academic press,
[26] J. Paik, N.J. Lee, Numerical modeling of free surface flow over a broad-crested 1972.
rectangular weir, J. Korea Water Resour. Assoc. 48 (4) (2015) 281–290. [51] T.-H. Shih, W.W. Liou, A. Shabbir, Z. Yang, J. Zhu, A New K-Epsilon Eddy Viscosity
[27] E. Goodarzi, J. Farhoudi, N. Shokri, Flow characteristics of rectangular broad- Model for High Reynolds Number Turbulent Flows: Model Development and
crested weirs with sloped upstream face, J. Hydrol. Hydromechanics 60 (2) (2012) Validation, 1994.
87–100. [52] M. Richmond, A. Antoniadis, L. Wang, A. Kolios, S. Al-Sanad, J. Parol, Evaluation
[28] A. Kamath, G. Fleit, H. Bihs, Investigation of free surface turbulence damping in of an offshore wind farm computational fluid dynamics model against operational
RANS simulations for complex free surface flows, Water 11 (3) (2019) 456. site data, Ocean Eng. 193 (2019) 106579.
[29] T. Xu, Y.-C. Jin, Numerical study of the flow over broad-crested weirs by a mesh- [53] C.-H. Lee, Rough boundary treatment method for the shear-stress transport k-ω
free method, J. Irrigat. Drain. Eng. 143 (9) (2017), 04017034. model, Eng. Appl. Comput. Fluid Mat. 12 (1) (2018) 261–269.
[30] A.O. Aksoy, M. Doğan, Experimental investigation OF the approach angle effects [54] M. Tajnesaie, E. Jafari Nodoushan, R. Barati, M. Azhdary Moghadam, Performance
ON the discharge efficiency for broad crested weirs, Anadolu Üniversitesi Bilim Ve comparison of four turbulence models for modeling of secondary flow cells in
Teknoloji Dergisi A-Uygulamalı Bilimler ve Mühendislik 17 (2) (2016) 279–286. simple trapezoidal channels, J. Hydraul. Eng. 26 (2) (2020) 187–197.
[31] S.A. Al-Hashimi, H.M. Madhloom, T.N. Nahi, Experimental and numerical [55] S.B. Pope, Turbulent Flows, IOP Publishing, 2001.
simulation of flow over broad crested weir and stepped weir using different [56] H.K. Versteeg, W. Malalasekera, An Introduction to Computational Fluid
turbulence models, J. Eng. Sustain. Develop. 21 (2) (2017) 28–45. Dynamics: the Finite Volume Method, Pearson education, 2007.
[32] A.H. Azimi, N. Rajaratnam, D.Z. Zhu, Discharge characteristics of weirs of finite [57] W.H. Hager, Dischage Measurement Structures, EPFL-LCH, 1986.
crest length with upstream and downstream ramps, J. Irrigat. Drain. Eng. 139 (1) [58] Z. Zachoval, I. Mistrová, L. Roušar, J. Šulc, P. Zubík, Zone of flow separation at the
(2013) 75–83. upstream edge of a rectangular broad-crested weir/Oblast odtržení proudu na
[33] H.M. Fritz, W.H. Hager, Hydraulics of embankment weirs, J. Hydraul. Eng. 124 (9) návodní hraně pravoúhlého přelivu se širokou korunou, J. Hydrol.
(1998) 963–971. Hydromechanics 60 (4) (2012) 288–298.
[34] K. Badr, D. Mowla, Development of rectangular broad-crested weirs for flow [59] U.S. Tim, Characteristics of Some Hydraulic Structures Used for Flow Control and
characteristics and discharge measurement, J. Civil Eng. 19 (1) (2015) 136–141. Measurement in Open Channels, Concordia University, 1986.
[35] Y. Chen, Z. Fu, Q. Chen, Z. Cui, Discharge coefficient of rectangular short-crested [60] I. ČSN, 3846 (25 9332), Liquid Flow Measurement in Open Channels by Weirs and
weir with varying slope coefficients, Water 10 (2) (2018) 204. Flumes, Rectangular Broadcrested Weirs, 1994.
[36] J. Farhoudi, N. Shokri, Flow from Broad Crested Rectangular Weirs with Sloped
Downstream Face, 32nd IAHR Congress, Venice, Italy, 2007.

13