Water Science and Engineering 2019, 12(1): 62e72

H O S T E D BY Available online at www.sciencedirect.com

Water Science and Engineering

journal homepage: http://www.waterjournal.cn

Computational investigation of hydraulic performance variation with

geometry in gabion stepped spillways
Dominic E. Reeve*, Ali Adel Zuhaira, Harshinie Karunarathna
College of Engineering, Swansea University, Swansea SA2 8PP, UK
Received 18 May 2018; accepted 3 January 2019
Available online 8 April 2019

Abstract

Over recent years, there has been a clear increase in the frequency of reported flooding events around the world. Gabion structures offer one
means of flood mitigation in dam spillways. These types of structures provide an additional challenge to the computational modeller in that flow
through the porous gabions must be simulated. We have used a computational model to investigate the flow over gabion stepped spillways. The
model was first validated against published experimental results. Then, gabion stepped spillways with four different step geometries were tested
under the same conditions in order to facilitate inter-comparisons and to choose the best option in terms of energy dissipation. The results show
that normal gabion steps can dissipate more energy than overlap, inclined, and pooled steps. An intensive set of tests with varying slope, stone
size, and porosity were undertaken. The location of the inception point and the water depth at this point obtained from this study were compared
with those from existing formulae. Two new empirical equations have been derived, on the basis of a regression analysis, to provide improved
results for gabion stepped spillways.
© 2019 Hohai University. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).

Keywords: Computational modelling; Energy dissipation; Gabion stepped spillways; Inception point location; Skimming flow

1. Introduction foundations. Also, overtopping damage can occur, which can

cause significant problems for people in the surrounding areas
Spillways can be defined as structures, located over dams, (Novak et al., 2010).
whose function is to release any excess water during the Chanson (2002) has defined two types of flow over stepped
flooding seasons in order to reduce the probability of having spillways: non-aerated flow and aerated flow. Normally, non-
an overtopping failure (Novak et al., 2001). Many materials aerated flow takes place along the upper steps of spillways
can be used in the construction of spillways, including roller where there is no air entrainment, while aerated flow can be
compact concrete and gabions. Each material has advantages observed along the lower steps where air entrainment occurs
and disadvantages (Boes and Hager, 2003). Energy dissipation (e.g. Andre and Schleiss, 2004; Chanson, 1994, 1995). For the
can be considered one of the main design elements of stepped same geometry, when the discharge increases, the length of the
spillways because the high energy of the flow can cause many non-aerated zone also increases. Thus, the likelihood of
problems at the toe of the structure, such as the formation of generating cavitation will also increase. As noted by Husain
scour holes, which can lead to structural failure in spillway et al. (2013), cavitation can be extremely damaging and
even destroy spillways, leading to operational failure of the
steps and spillways. The aerated zone can be determined by
This work was supported by the Higher Committee for Education estimating the location of the inception point, which represents
Development (HCED) in Iraq.
* Corresponding author
the endpoint for the non-aerated zone and the start of the
E-mail address: [email protected] (Dominic E. Reeve). aerated zone. The inception point is the location where the
Peer review under responsibility of Hohai University. boundary layer intersects the free surface of the water

https://doi.org/10.1016/j.wse.2019.04.002
1674-2370/© 2019 Hohai University. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).
 Dominic E. Reeve et al. / Water Science and Engineering 2019, 12(1): 62e72 63

(Chanson, 1994, 1996). Flow over stepped spillways can be changed. It is important to note that the number of the steps in
classified into three hydraulic regimes: nappe flow, transition the previous studies affected the energy dissipation results
flow, and skimming flow (Zhang and Chanson, 2016b). more than the spillway slope. Therefore, more research is
Traditionally, spillways have been impermeable. However, required to investigate the impact of the number of steps.
the use of a permeable layer such as gabions offers a means to The sensitivity of the hydraulic performance to the char-
improve the hydraulic performance by enhancing energy acteristics of the material contained in gabions was investi-
dissipation. Gabions are containers that can be filled with gated experimentally by Chinnarasri et al. (2008). They used
gravel, cobbles, stones, and rocks, depending on the purpose of three stone types: (1) crushed stone of about 25e35 mm in
construction. Since 1879, gabions have been used in China and diameter; (2) rounded stone of about 25e35 mm in diameter;
Egypt. Gabions have also been used for different purposes such and (3) crushed stone of about 50e70 mm in diameter. The
as riverbed protection, bank stabilization, and retaining walls. results showed that the energy dissipation ratios over gabion
There are three different types of gabions: basket, mattress, and stepped weirs were greater than those over the corresponding
sack. All of these may be filled with gravel and/or cobble impermeable stepped weirs by nearly 7%, 10%, and 14% for
materials (Zhang and Chanson, 2014). The choice of gabion weir slopes of 30
, 45
, and 60
, respectively. Consequently,
type depends upon the application. For instance, basket gabions the outlet velocity was lower. Moreover, the results showed
are commonly used for stability purposes and to protect river that both the stone size and stone shape had a small effect on
beds and stream banks (Freeman and Fischenich, 2000). A brief the energy loss and flow velocity compared to the weir slope.
discussion of earlier studies conducted on gabion stepped The pressure on the step face of gabion stepped weirs was less
spillways is provided below. than that on the horizontal step due to the dampening influence
Gabions are one common construction element for spill- of filled stones. The average pressure difference was approx-
ways in the African Sahel (Peyras et al., 1992). Gabions have imately 29%. More recently, Wüthrich and Chanson (2014)
been used widely for water structures like small earth dams, carried out a laboratory study to investigate the hydraulic
retaining walls, intakes, and soil conservation work. Salmasi characteristics of flow, such as flow patterns, air-water flow
et al. (2012) stated that there are many benefits from using properties, and energy dissipation over normal and gabion
gabions, such as ease of construction, structural stability, stepped spillways with a 1:2 slope and a 0.1-m step height.
flexibility, and resistance to water load. The resistance to water This study was conducted with a wide range of flow rates in
load is likely to be related to flow through porous media. order to investigate nappe, transition, and skimming flows.
Porosity can help the water to drain faster and reduce the water They found that large velocities could be observed at the
load behind the structure (Zhang and Chanson, 2016a). downstream end of gabion stepped spillways, as well as low
Permeability is known to affect the flow properties of the free rates of energy dissipation over gabion steps in comparison to
surface in many cases (Manes et al., 2009). It has been sug- smooth impervious steps. It can be concluded that there are
gested that related flow mechanisms can play a vital role in many parameters with impacts on the performance of gabion
increasing or decreasing friction factors, thereby affecting the stepped spillways, such as the size of stones, water flow
shear penetration within the permeable bed, which can in turn conditions, and the air-flow entrainment. However, at present,
affect the boundary layer. Hence, increasing the flow resis- there is no clear picture of which of these parameters has the
tance may increase energy dissipation due to the momentum strongest influence on the efficiency of gabion stepped spill-
exchange between the surface and subsurface flows. ways. Thus, further investigations are needed to determine the
Stephenson (1979) performed a study on energy dissipation essential controlling factors.
over stepped gabions. Different configurations were examined, It is well established that stepped spillways (gabion or
such as stepped gabions with two to four steps and with four other) have an advantage over non-stepped spillways in terms
different slopes: 1:1, 1:2, 1:3, and 2:3. The energy dissipation of reducing cavitation damage due to air entrainment and by
was calculated using the differences in depths between the improving energy dissipation performance (Husain, 2013).
areas upstream and downstream. The results showed that the Although there is a well-developed understanding of the per-
relative energy dissipation ranged from 25% to 85%. They formance of normal stepped spillways, the same level of un-
also showed that the energy dissipation increased as the derstanding has not been developed for gabion stepped chutes
number of steps increased to three, but then decreased as the with more complex designs. Moreover, even though stepped
number of steps increased further. Concerns about the ability spillways have been extensively studied using computational
of gabion steps to resist damage under high flows were models, detailed numerical modelling studies on gabion
addressed by the experimental study of Peyras et al. (1992), stepped spillways have not been recorded. Also, gabion
who showed that gabion stepped weirs can withstand floods up porosity and gabion stone size, both of which affect the flow,
to 3 m3/(m∙s) without any damage. Kells (1994) studied the need to be investigated in detail in order to demonstrate their
energy dissipation over a gabion stepped weir as a function of impacts on other important parameters, such as the location of
the critical depth at and discharge over the crest. This exper- the inception point. Finally, finding an optimum gabion step-
imental study used two downstream slopes of 1:1 and 1:2, and ped spillway design can improve performance and offer an
the main finding was that 20% of the energy can be dissipated alternative for stepped spillway construction.
due to the through-flow. Moreover, no significant differences In this study, we used a computational model as a numer-
were noticed in the energy dissipation when the slopes ical flume to investigate the performance of gabion stepped
64 Dominic E. Reeve et al. / Water Science and Engineering 2019, 12(1): 62e72

spillways under a range of conditions that covered spillway solved the Reynolds-averaged Navier-Stokes (RANS) equa-
slope, step geometry, step height, gabion stone size, and tions with a turbulence sub-model. The code chosen was
porosity. In Section 2 details of the computational model are NEWFLUME, described in detail by Lin and Xu (2006),
described together with some validation results. In Section 3, which solves the RANS equations in two dimensions (2DV).
results of the computations along with discussion of their For completeness' sake, we provide a brief description of the
implications for hydraulic performance are presented, and the equations and solution technique here. The equations for flow
computational results obtained in this study are compared with outside the porous media are
those of some existing empirical equations for estimation of
the location of the inception point and water depth at that vui
¼0 ð1Þ
point. Finally, two new equations are proposed for gabion vxi
stepped spillways. !
vui vui 1 vP 1 v vui
þ uj ¼ þ gi þ m  ru0i u0j ð2Þ
2. Problem formulation and numerical model vt vxj r vxi r vxj vxj

2.1. Problem formulation where ui is the mean velocity in the xi direction (i ¼ 1 means
the horizontal direction, and i ¼ 2 means the vertical direc-
There are many issues that need to be addressed in the tion) (m/s), t is time (s), P is the mean pressure (kN/m2), r is
selection of a numerical model that can be used to simulate the the fluid density (kg/m3), gi is the gravitational acceleration in
flow over gabion stepped spillways (Fig. 1) such as the model's the xi direction (m/s2), m is the molecular viscosity (kg/(m∙s)),
capability to simulate the turbulent flow over the porous and u0i u0j represents the Reynolds stress (kN/m2).
media, the treatment of the interface between the fluid layer The mean viscous stress is written as
and porous medium, and the treatment of the free surface.
Since this study mainly aimed to investigate the location of the vui vuj
inception point, many cases have been tested in order to assess tij ¼ m þm ð3Þ
vxj vxi
the inception point location under different flow conditions,
step heights, and spillway slopes with wide ranges of gabion The Reynolds stress was calculated using a nonlinear eddy
stone sizes and porosity. The inception point determines the viscosity model. This model used mean velocity, turbulence
endpoint of the non-aerated zone; this point is immediately kinematic energy (k), and dissipation rate of turbulence (3 ).
followed by the aerated zone. Air entrainment is one of the The k-3 transport equation was used to calculate k and 3 (Lin
most important factors that can protect the hydraulic structures and Xu, 2006).
from cavitation damage due to the air-water flow. The incep- The mean flow in the porous media is governed by
tion point is defined as the point where the boundary layer
vu*i
intersects the free surface. The boundary layer thickness can ¼0 ð4Þ
be predicted by determining the point where the velocity is vxi
99% of its maximum value in the velocity distribution profile
at any particular location along the spillway (Schlichting, 1 þ CA vu*i u*j vu*i 1 vP n v2 u*i
þ 2 ¼ þ gi þ  gaP u*i 
1979; Husain et al., 2013). Thus, the water depth at that n vt n vxj r vxi n vxj vxi ð5Þ
point, measured perpendicular to the pseudo-bottom, can be qffiffiffiffiffiffiffiffiffi
obtained directly from the computed values of the free surface. gbP u*k u*k u*i

2.2. Numerical model where u*i is the mean velocity in the xi direction (m/s); n is the
porosity of the porous medium; n is the kinematic viscosity
As the flow in this study contained elements that were sub- (m2/s); g is the gravitational acceleration (m/s2); CA , aP , and
critical, supercritical, and highly turbulent, we used a code that bP are the coefficients of the porous medium; and the subscript
k refers to the summation of velocities in the two directions.
The coefficients in Eqs. (4) and (5) are given by Lin and Xu
(2006) as

1n
CA ¼ gP
n

ð1  nÞ2 v
aP ¼ a
n3 gD250
 
7:5 1  n 1
bP ¼ b 1 þ
KC n3 gD250
Fig. 1. Turbulent flow over gabion stepped spillway.
 Dominic E. Reeve et al. / Water Science and Engineering 2019, 12(1): 62e72 65

where D50 is the median diameter of gravel filling the gabions, is the discharge per unit width, and h is the water depth. The
qffiffiffiffiffiffiffiffiffiffi
critical section occurred where Fr ¼ 1. To ensure numerical
gP ¼ 0:34, a ¼ 200, b ¼ 1:1, and KC ¼ u*k u*k T=ðnD50 Þ,
stability, the initial time step was set to 0.001 s and the
with T being a calibration parameter. simulated period was 24 s. As in the laboratory experiments,
Free surface tracking was achieved through the application all of the boundaries were closed except the right boundary,
of the volume of fluid (VOF) method, which was originally which was open. It is important to mention that dam break
established by Hirt and Nichols (1981) and subsequently conditions were used in the numerical model rather than the
adjusted by Kothe et al. (1991). A finite difference method was steady state used in the experiments of Wüthrich and Chanson
employed to approximate all partial differential equations. The (2014). Dam break conditions were used in order to capture
calculation of the mean flow outside the porous media was different values of discharge at different times and also to
performed with the Navier-Stokes equation solver that was observe the development of the flow pattern, such as nappe
developed by Kothe et al. (1991). This method used a two-step flow, transient flow, and skimming flow, over the gabion boxes.
projection to solve the momentum equations. Initially the In the numerical model, dam break conditions were recreated
velocity was estimated without the pressure gradient term. In with the fluid at rest and an impermeable barrier placed at
the second step the velocity was corrected using the updated x ¼ 9.0 m, as shown in Fig. 2, in which velocity vectors are
pressure field, which was determined from the Poisson pres- plotted at regular grid points to give an indication of the fluid
sure equation. The governing equations for the flow through layer, free surfaces are contoured in blue continuous lines, the
the porous media had the same structure as the RANS equa- outline of the impermeable dam is shown as black lines, and
tions. However, Reynolds stresses were replaced by linear and gabion steps are shown as red lines. This barrier was removed
nonlinear friction terms. The two-step projection technique instantaneously at t ¼ 0 s and the fluid then moved under the
was also used for the flow calculation through the porous force of gravity.
media. On the interface between the porous media and outside The porosity of gravel was set equal to 0.375, representing
flow, the continuity of the pressure and velocity was satisfied. the average value used in the experimental study. The fluid
The central difference method was used to discretize all the was initially at rest. The volume of the upstream tank was
pressure and stress gradients in the projection method. The 9.0 m  1.4 m per unit width. The barrier at x ¼ 9.0 m was
upwind scheme was combined with the central difference removed instantaneously and the resulting flow was computed
method to discretize the advection terms for the k-3 transport (Fig. 2). The validation was conducted in terms of the location
equation. of the inception point. The velocity distribution over the outer
edge of the steps was used to determine the inception point. As
2.3. Sample validation illustrated in Fig. 3, the velocity profile over the steps in the
non-aerated zone is quasi-parabolic with no quasi-uniform
The capability of NEWFLUME was demonstrated by Lin segment in the upper part of the flow profile. The physical
and Xu (2006) through a series of case studies encompass- meanings of the variables in Fig. 3 can be found in Section 3.3.
ing a wide range of turbulent free surface flows, covering Results shown in Table 1 illustrate that the model was able
coastal and ocean engineering (e.g., breaking wave interaction to simulate the flow over the gabion steps and also to capture
with a seawall and wave forces on a submerged pipeline), the location of the inception point in agreement with experi-
hydraulic engineering (e.g., dam break flow and hydraulic mental results. Further details of the validation testing can be
jump), and environmental modelling (e.g., jets and pollutant found in Zuhaira et al. (2017).
transport).
As an additional validation exercise for the case of flow 3. Results and discussion
down a gabion stepped spillway, we compared the model re-
sults against the experimental work of Wüthrich and Chanson 3.1. Model set-up
(2014). Their test section consisted of a broad crested weir
with a length of 1.01 m and a height of 1.0 m, followed by ten In the numerical model, two arrangements were used to
identical impervious steps with a height of 0.1 m and a length accommodate the spillway geometry variations described in
of 0.2 m. Gabion steps were installed over the impervious Section 3.2 and to meet the requirement that the critical water
steps, which were made from marine plywood, with a 0.1-m depth be established over the weir crest. For the spillway with
height and a 0.3-m length. Gravel inside the gabions had a 0.06-m-high steps, the weir was placed at x ¼ 6.9 m, while for
D50 of 0.01 m. The porosity ranged from 0.35 to 0.40. spillways with step heights of 0.09 m and 0.12 m, the weir was
The same experimental conditions were established in the placed at x ¼ 7.4 m. In all cases the length of the weir crest
numerical model. The mesh sizes were set to 0.010 m and was set to 0.6 m.
0.005 m in the x- and y-directions, respectively. The initial The initial water depths were 1.58 m, 1.70 m, and 1.80 m
water depth was set to 1.4 m in order to achieve a required for spillways with step heights of 0.12 m, 0.06 m, and 0.09 m,
discharge. The discharge was calculated by determining the respectively, which ensured that the upstream tank contained
critical section over the broad crested weir. The critical section sufficient water to achieve the required discharge and also to
was determined by testing directly, from the model output, the
pffiffiffiffiffiffiffi mitigate the effects of transients. The total simulation time was
value of the Froude number, Fr, with Fr ¼ q= gh3 , where q set to 15.0 s for the 0.12-m step height, 16.5 s for the 0.09-m
66 Dominic E. Reeve et al. / Water Science and Engineering 2019, 12(1): 62e72

Fig. 2. Flow over gabion stepped spillway at different times.

step height, and 18.0 s for the 0.06-m step height. The time different values of porosity. A step height of 0.12 m with a 1:2
step was set at 0.0001 s to satisfy the stability criterion in all spillway slope was chosen initially in order to test the selected
cases. The mesh sizes in the x- and y-directions were 0.010 m gravel sizes and porosity. Comparisons were conducted for
and 0.005 m, respectively for 0.09-m and 0.12-m step heights. specific discharges at the skimming flow stage.
However, for the 0.06-m step height, they were reduced to Parameter values of typical gravel were used to represent
0.0075 m in the x-direction and 0.0030 m in the y-direction, the porous media component. According to the standards, the
respectively. The model domain had closed boundaries except minimum diameter of particles should be greater than 2 mm in
for the right-hand boundary, which was set as an open order to consider them gravel (Wentworth, 1922). Therefore,
boundary to let the water exit the flume. Overall, many cases the minimum value for the gravel size in this study was set to
of gabion stepped spillways were tested in the current study. 5 mm. The diameters of 10 mm, 15 mm, and 20 mm were
Each step height (0.12, 0.09, and 0.06 m) was examined tested as well. The porosity of gravel normally ranges between
individually with three spillway slopes (1:2, 1:2.5, and 1:3). 0.25 and 0.40. Therefore, values of 0.25, 0.30, 0.35, and 0.40
Then, for each of these nine spillway geometries, simulations were tested in this study.
were performed for four different gravel sizes and four To summarize, a set of numerical experiments were con-
ducted to compute the flow over gabion stepped spillways for
a dam break case in which the step height, spillway slope,
gravel diameter, gravel porosity, and gabion geometry varied
individually.

Table 1
Comparison between inception point locations of experimental and numerical
results.
Time (s) Discharge Inception point location
per unit Wüthrich and Numerical model
width (m2/s)
Chanson (2014)
5.57 0.11 Step 8 to step 9 At outer edge of step 8
5.98 0.10 Step 7 to step 8 At outer edge of step 7
6.43 0.08 Step 5 to step 6 At outer edge of step 5
7.01 0.06 Step 5 At middle of step 5
Fig. 3. Velocity distribution over different steps in non-aerated zone.
 Dominic E. Reeve et al. / Water Science and Engineering 2019, 12(1): 62e72 67

Fig. 4. Four different configurations of gabion steps considered in this study.

3.2. Different step geometries

The four different gabion step geometries, normal, overlap,

inclined, and pooled, are shown in Fig. 4 for the case in which
the step height was 0.12 m. All the configurations of the gabion
steps were installed over impervious steps. The same initial
conditions were applied in all configurations in order to deter-
mine the differences in terms of the time to establish skimming
flow, the location of the inception point, and energy dissipation.

3.3. Energy dissipation Fig. 5. Two-dimensional schematic view of stepped spillway showing
parameters required to estimate residual energy at outer edge of steps.
Energy dissipation is one of the most important parameters
from the hydraulic design point of view. To investigate the
energy dissipation for each of the gabion configurations, we The second point is located upstream of the weir and the
applied Bernoulli's equation, following Husain (2013). The head is calculated as
variables required to estimate the energy dissipation are
described in Fig. 5, where P is the pressure head at the top of V 20
H0 ¼ Y0 þ ð7Þ
the weir (m). The datum, or zero head line, is shown as a 2g
dashed horizontal line, and the total head is demonstrated by a
where Y0 is the flow depth above the horizontal surface of the
full horizontal line at the top of the figure. In brief, two points
step under consideration (m), and V0 is the approach flow
are required to determine the energy dissipation (DH ). At the
velocity (m/s).
first point (a point under consideration in the non-aerated
Hence, as shown in Fig. 5, the total energy dissipation with
zone), the head H is calculated as follows:
respect to the point under consideration is
aV 2
H ¼ y cosq þ ð6Þ DH ¼ H0  H ð8Þ
2g
The computational results of energy dissipation at
where y is the perpendicular depth of water over a pseudo- x ¼ 9.44 m were 0.099, 0.076, 0.069, and 0.046 m2/s2 for the
bottom (m), tan q is the spillway slope, a is the energy coef- spillways with normal, overlap, inclined, and pooled gabion
ficient (unitless), and V is the fluid velocity (m/s). steps, respectively. This demonstrates that for this situation
68 Dominic E. Reeve et al. / Water Science and Engineering 2019, 12(1): 62e72

normal gabion steps had the best performance in terms of non-aerated zone was observed for pooled gabion steps. Also,
energy dissipation, while pooled gabion steps exhibited the the results show that there was a small difference between
worst performance. To investigate this, further calculations of overlap gabion steps and inclined gabion steps in terms of the
the turbulent kinetic energy (TKE) of the flow were made. inception point location. Normal gabion steps exhibited the
TKE is the kinetic energy per unit mass of the turbulent best results in terms of energy dissipation and the location of
fluctuation in the flow. It is generally accepted that, for inception point.
spillway flow, the energy dissipation induced by turbulent
eddies is greater than the energy loss due to internal and 3.5. Sensitivities to slope, porosity, and gravel size
boundary frictions. Thus, there is a strong relation between the
energy dissipation and TKE, and overall energy dissipation is As the normal step configuration exhibited the best per-
prompted by creating turbulent flow (Husain, 2013). formance in terms of energy dissipation, discussion in this
The TKE was computed directly from the instantaneous section is restricted to this case. Three different step heights
velocities over the whole domain (see Fig. 6(a)). The plot and three different step widths for a given weir height were
shows that, from being initially at rest, the fluid gradually tested with the numerical code to investigate the impact of step
accelerates down the spillway, resulting in a rise in TKE as the configuration on the flow properties over gabion stepped
discharge changes from nappe to transition to skimming flow. spillways. Step widths were set to give spillway slopes of 1:2,
There are some fluctuations in TKE as skimming flow is 1:2.5, and 1:3. The porosity of the gabion steps was varied
established during the time between 2 s and 4 s after the dam from 0.25 to 0.40, while the gravel sizes were varied from
break, and then TKE gradually decreases as the head in the 0.005 m to 0.020 m. The selection of these values provided
upstream tank falls towards the level of the weir. wide-ranging physical characteristics likely to be encountered
The results reveal that normal gabion steps generated in real schemes in order to investigate the effects of the gabion
greater amounts of TKE, by inference greater overall energy porosity, gabion gravel size, spillway slope, and step height on
dissipation, than the other shapes. The finding is in agreement the flow characterization.
with the calculated results of energy dissipation for different Fig. 7 shows the effect of the step height on the location of
gabion geometries. For all the gabion geometries, there is a the inception point. The calculations of the developing
slight increase in TKE for 7.5 s < t < 8.5 s, arising from an boundary layer thickness were carried out at the end of each
increase in the discharge at this time due to the reflected step, in order to determine the location of the inception point.
transient in the upstream tank. Finally, as a further check, the In the plot, the dots denote the top of the boundary layer at the
turbulent energy dissipation rate, 3 , calculated directly from end of each step, and the inception point occurs where the
the k-3 equation, is shown in Fig. 6(b), and confirms that the boundary layer meets the free surface that is shown as the blue
turbulent kinetic energy and its dissipation rate have a similar line. It is clear that step heights affect the location of the
morphology over time. inception point and the length of the non-aerated zone.
The location of the inception point moves downslope as the
3.4. Inception point location discharge increases. In other words, the length of the non-
aerated zone increases with the flow rate. This is most likely
The computational results of the inception point location related to the discharge, as increasing discharges imply in-
were 9.75, 10.16, 10.20, and 10.32 m from the downstream creases in flow depth and/or velocity. Increasing the flow depth
edge of the weir for the spillways with normal, overlap, in- might increase the length of the boundary layer required to
clined, and pooled gabion steps, respectively. The results show intersect the free surface profile. For a given slope and
that the non-aerated zone for normal gabion steps was smaller discharge, it is observed that the location of the inception point
than those for other types of gabion steps, while the longest moves to the upstream side towards the weir crest when the

Fig. 6. Turbulent kinetic energy and its dissipation rate for different gabion geometries.
 Dominic E. Reeve et al. / Water Science and Engineering 2019, 12(1): 62e72 69

Fig. 7. Estimation of inception point location for spillway slope of 1:2 and unit discharge equal to 0.2 m2/s.

step height is increased (Fig. 7). This can be attributed to the However, for gabion stepped spillways, this kind of relation
fact that large step heights can have a stronger impact on the has not been reported before.
location of the inception point than small step heights as the Furthermore, the results show that both porosity and gravel
velocity of the overflowing water is decelerated when it flows size can have a significant impact on the location of the
over the outer step edge and hits the horizontal face. However, inception point. The computational results for high discharges
this effect might be expected to decrease with lesser reveal a general trend that, when the gravel size decreases, the
discharges. location of the inception point moves upslope towards the
In order to investigate the effect of the spillway slope on the weir. Thus, the length of the non-aerated zone will be shorter.
location of the inception point and the length of the non- A relationship between porosity, gravel size, and the length of
aerated zone, three different slopes (1:2, 1:2.5, and 1:3) were the non-aerated zone seems stronger for larger discharges. To
tested with three different step heights (0.06, 0.09, and quantify the relationship, we describe a nonlinear multiple
0.12 m). Three values of discharge, 0.25, 0.20, and 0.15 m2/s, regression analysis in the next section to determine the
were used for comparison purposes to explore how the strength and robustness of the putative relationships discussed
discharge influences the results. above.
The general trend of the numerical results reveals that the
location of the inception point may move upslope when the 3.6. Predictive formulae
spillway slope becomes steeper. Therefore, it can be
concluded that the length of the non-aerated zone has a Previous studies such as Chanson (1996), Carosi and
reciprocal dependence on the spillway slope (Fig. 8). In other Chanson (2008), Meireles and Matos (2009), Hunt and
words, a shorter length of the non-aerated zone will be ex- Kadavy (2011), and Husain et al. (2013) have developed
pected for a steep spillway slope in comparison with a flat nonlinear multiple regression equations to fit the results ob-
spillway slope. This can be explained by the fact that the tained in their work for normal stepped spillways to predict the
longitudinal velocity over steep slopes is higher than that over length of the non-aerated zone and the water depth at the
flat slopes. Therefore, the flow depth will be lower for steep inception point. Our computational results were compared
spillway slopes. This is likely to accelerate the development of with the results from those nonlinear multiple regression
the boundary layer and its expansion throughout the water equations. Some deficiencies were observed which provided
column to meet the free surface. All of the previous obser- the motivation to develop modified formulae appropriate for
vations agree with the experimental observation of Andre gabion stepped spillways. For this purpose, the values of the
(2004) who conducted a study on normal stepped spillways. length of the non-aerated zone, Li , and the corresponding

Fig. 8. Estimation of inception point location for 0.06-m step height and unit discharge equal to 0.15 m2/s.
70 Dominic E. Reeve et al. / Water Science and Engineering 2019, 12(1): 62e72

values of the roughness Froude number, F* , collected from the A further comparison was conducted against different
cases described above were used to obtain the following empirical equations for normal stepped spillways,
relationship: including (1) Li =ks ¼ 9:8ðsinqÞ0:08 F 0:71
* from Chanson (1994),
 0:1537 (2) Li =ks ¼ 1:05 þ 5:11F* from Carosi and Chanson (2008),
Li
¼ 2:2281ðsinqÞ
0:3172 1:2486 0:2831 D50
F* n ð9Þ (3) Li =ks ¼ 5:25F 0:95
* from Meireles and Matos (2009), (4) Li =
ks hs ks ¼ 6:1ðsinqÞ0:08 F 0:86
* from Hunt and Kadavy (2011), and (5)
Li =ks ¼ 5:19F 0:89
* from Hunt and Kadavy (2013). The objec-
where ks is the roughness, which is calculated as hs cos q, and tive of this comparison was to assess whether the empirical
hs is the step height. The roughness Froude number can be equations established for normal stepped spillways could be
calculated from used to find the location of the inception point over gabion
q stepped spillways.
F* ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð10Þ Comparisons of the predictions from these five formulae
g sinqk3s against the computational results of the present study are
In developing Eq. (9), we started with Chanson's (1994) shown in Fig. 10. It is evident that all the formulae over-
equation for normal stepped spillways and included extra estimate the location of the inception point in comparison with
terms related to gabion boxes. The coefficients of Eq. (9) were the computational results of the present study, with the for-
estimated using an iterative least squares estimation routine mula from Hunt and Kadavy (2013) showing the closest
available in MATLAB. For the fitting, results from the agreement.
computational model were split into two groups. The first It is important to mention that the formulae from these
group, which contained around 80% of the data, was used to earlier studies were derived from experimental results based
find the coefficients of the equation. The remaining data (20%) on physical models of normal stepped spillways. However, the
were used to test the equation. All the data were selected formula of the current study is based on the results of
randomly and were restricted to cases in which the step ge- computational modelling. The results in Fig. 10 show that
ometry was normal. The coefficient of determination, R2, was using gabion boxes increased the roughness of the spillway
used to measure how well the regression line approximated the surface and therefore accelerated the growth of the boundary
data. If the value of R2 approached 1.00, then the regression layer thickness. Hence, the length of the non-aerated zone was
line was considered to have provided a perfect representation reduced. Using the formulae for normal stepped spillways to
of the data. Fig. 9 shows that the value of R2 for Eq. (9) is predict the non-aerated length over gabion stepped spillways
0.9057, with a correlation factor of 0.952, which gives an will likely give an overestimation, leading to very conservative
indication that the formula describes the data well. designs.
The regression analysis indicates that the porosity term has The computational results were used to develop another
a negative power sign, which reveals a reciprocal relationship equation for estimation of the water depth at the inception
between porosity and the length of the non-aerated zone. Also, point for gabion stepped spillways. The expression is given as
a small value for the power, which is 0.2831, suggests that follows:
porosity has less impact on the length of the non-aerated zone  0:0091
Di 0:1175 0:7469 0:1188 D50
than either F* or the spillway slope. The positive exponent for ¼ 0:2569ðsinqÞ F* n ð11Þ
the gravel size shows a direct relationship between the gravel ks hs
size and the length of the non-aerated zone, and again the
where Di is the water depth at the inception point. Using the
value is quite small, which means that the dependence of Li on
same procedure as that for development of Eq. (9), the values
the gravel size is not very strong.
of R2 and the correlation factor for Eq. (11) were obtained as
0.9413 and 0.97, respectively.

Fig. 10. Comparison of computational results against published

Fig. 9. Data test against Eq. (9). formulae for length of non-aerated zone.
 Dominic E. Reeve et al. / Water Science and Engineering 2019, 12(1): 62e72 71

zone increased with decreasing step height and also with

decreasing spillway slope. A weaker dependence of the
location of the inception point on both porosity and gravel size
was found.
Two empirical equations were derived on the basis of the
regression analysis to estimate the location of the inception
point and the water depth. These two equations were
compared with other formulae developed on the basis of
physical experiments for the flow over normal stepped spill-
ways. The results showed that using gabion boxes would in-
crease the roughness of the spillway surface and therefore
accelerate the growth of the boundary layer thickness. It would
Fig. 11. Comparison of computational results against published
formulae for water depth at inception point. further reduce the length of the non-aerated zone. Overall,
using porous media over concrete steps, as simulated in this
study, would be expected to reduce the non-aerated length and
The results of Eq. (11) were compared with those of thus reduce the danger of cavitation damage over the steps.
empirical formulae, i.e., Di =ks ¼ ½0:4=ðsinqÞ0:04 F 0:64
* from
Chanson (1994) and Di =ks ¼ 0:28F 0:68 * from Meireles and
References
Matos (2009), as shown in Fig. 11. Both of the empirical
formulae were developed for normal stepped spillways, not for Andre, S., 2004. High Velocity Aerated Flows on Stepped-chutes with Macro-
gabion stepped spillways. As can be seen in Fig. 11, the for- roughness Elements, Communication 20. Laboratorie de Constructions
mula from Chanson (1994) consistently overestimates the 
Hydrauliques, Ecole Polytechnique Federale de Lausanne, Lausanne.
water depth, while the formula of Meireles and Matos (2009) Andre, S., Schleiss, A., 2004. High Velocity Aerated Flows on Stepped Chutes
with Macro-roughness Elements. Ecole Polytechnique Federale de Lau-
slightly underestimates the water depth, in comparison with
sanne, Lausanne. https://doi.org/10.5075/epfl-thesis-2993.
the computational results of the present study. Boes, R.M., Hager, W.H., 2003. Hydraulic design of stepped spillways. J.
It should be noted that both Eqs. (9) and (11) are valid for Hydraul. 129(9), 671e679. https://doi.org/10.1061/(ASCE)0733-
gabion stepped spillways. Both of the equations were obtained 9429(2003)129:9(671).
from the computational results for different step heights of Carosi, G., Chanson, H., 2008. Turbulence characteristics in skimming flows
0.06, 0.09, and 0.12 m (0.832  Yc =hs  3.080) with different on stepped spillways. Can. J. Civ. Eng. 35(9), 865e880. https://doi.org/10.
1139/L08-030.
spillway slopes of 1:2, 1:2.5, and 1:3. Moreover, these equa- Chanson, H., 1994. Hydraulics of skimming flows over stepped channels and
tions were applicable for porosity ranging from 0.25 to 0.40 spillways. J. Hydraul. Res. 32(3), 445e460. https://doi.org/10.1080/
and D50 in the range of 0.005e0.020 m. The maximum 00221689409498745.
discharge used was 0.25 m2/s. As ever with empirically Chanson, H., 1995. Air Bubble Entrainment in Free-Surface Turbulent Flows:
derived formulae, application outside the parameter ranges Experimental Investigations, Report CH46/95. University of Queensland,
Queensland.
used to develop the formulae is not recommended unless Chanson, H., 1996. Prediction of the transition nappe/skimming flow on a
further tests are performed. stepped channel. J. Hydraul. Res. 34(3), 421e429. https://doi.org/10.1080/
00221689609498490.
4. Conclusions Chanson, H., 2002. The Hydraulics of Stepped Chutes and Spillways. A. A.
Balkema Publishers, Lisse.
Chinnarasri, C., Donjadee, S., Israngkura, U., 2008. Hydraulic characteristics
The hydraulic performance of different gabion stepped of gabion-stepped weirs. J. Hydraul. Eng. 134, 1147e1152. https://doi.org/
spillways has been investigated with a state-of-the-art nu- 10.1061/(ASCE)0733-9429(2008)134:8(1147).
merical model based on solving the RANS equations. To test Freeman, G.E., Fischenich, J.C., 2000. Gabions for Streambank Erosion
the capability of the numerical model, a validation was con- Control, EMRRP Technical Notes Collection (ERDC TN-EMRRP SR-22).
ducted using the experimental data of Wüthrich and Chanson U.S. Army Engineering Research and Development Center, Vicksburg.
Hirt, C.W., Nichols, B.D., 1981. Volume of fluid (VOF) method for the dy-
(2014). Four different geometries of gabion steps were namics of free boundaries. J. Comput. Phys. 39(1), 201e225. https://doi.
investigated to compare the results in terms of energy dissi- org/10.1016/0021-9991(81)90145-5.
pation and the location of the inception point. These four Hunt, S.L., Kadavy, K.C., 2011. Inception point relationship for flat-slopped
geometries were normal gabion steps, overlap gabion steps, stepped spillways. J. Hydraul. Eng. 137(2), 262e266. https://doi.org/10.
inclined gabion steps, and pooled gabion steps. 1061/(ASCE)HY.1943-7900.0000297.
Hunt, S.L., Kadavy, K.C., 2013. Inception point for embankment dam stepped
The results showed different hydraulic characteristics for the spillway. J. Hydraul. Eng. 139(1), 60e64. https://doi.org/10.1061/(ASCE)
four types of gabion steps in terms of energy dissipation and the HY.1943-7900.0000644.
inception point location. The normal gabion step geometry Husain, S.M., 2013. Computational Investigation of Skimming Flow on
showed the best performance for the energy dissipation and the Stepped Spillways Using the Smoothed Particle Hydrodynamics Method.
location of the inception point. The length of the non-aerated Ph. D. Dissertation. Swansea University, Swansea.
Husain, S.M., Muhammed, J.R., Karunarathna, H.U., Reeve, D.E., 2013.
zone was explored for a range of step heights, spillway slopes, Investigation of pressure variation over stepped spillways using smooth
gabion porosity, and gabion gravel sizes using a computational particle hydrodynamics. Adv. Water Resour. 66, 52e69. https://doi.org/10.
model. The results showed that the length of the non-aerated 1016/j.advwatres.2013.11.013.
72 Dominic E. Reeve et al. / Water Science and Engineering 2019, 12(1): 62e72

Kells, J.A., 1994. Energy dissipation at a gabion weir with throughflow and Schlichting, H., 1979. Boundary Layer Theory. McGraw-Hil, New York.
overflow. In: Proceedings of Annual Conference of the Canadian Society Stephenson, D., 1979. Gabion energy dissipators. In: Transactions of the 13th
for Civil Engineering. Canadian Society for Civil Engineering, Winnipeg, International Congress on Large Dams. International Commission on
pp. 1e4. Large Dams, New Delhi, pp. 33e43.
Kothe, D.B., Mjolsness, R.C., Torrey, M.D., 1991. RIPPLE: A Computer Wentworth, C., 1922. A scale of grade and class terms for clastic sediments. J.
Program for Incompressible Flows with Free Surface, Rep. LA-12007-MS. Geol. 30(5), 377e392.
Los Alamos National Laboratory, Los Alamos. Wüthrich, D., Chanson, H., 2014. Hydraulics, air entrainment, and energy
Lin, P., Xu, W., 2006. NEWFLUME: A numerical water flume for two two- dissipation on a gabion stepped weir. J. Hydraul. Eng. 140(9), 04014046.
dimensional turbulent free surface flow. J. Hydraul. Res. 44(1), 60e64. https://doi.org/10.1061/(asce)hy.1943-7900.0000919.
https://doi.org/10.1080/00221686.2006.9521663. Zhang, G., Chanson, H., 2014. Two-phase flow on a gabion stepped spillway:
Manes, C., Pokrajac, D., McEwan, I., Nikora, V., 2009. Turbulence structure of Cavity and seepage air-water motion. In: Proceedings of the 19th Aus-
open channel flows over permeable and impermeable beds: A comparative tralasian Fluid Mechanics Conference. Australasian Fluid Mechanics So-
study. Phys. Fluids 21(12), 125109. https://doi.org/10.1063/1.3276292. ciety, Melbourne.
Meireles, I., Matos, J., 2009. Skimming flow in the non-aerated region of Zhang, G., Chanson, H., 2016a. Gabion stepped spillway: Interactions between
stepped spillways over embankment dams. J. Hydraul. Eng. 135(8), free-surface, cavity and seepage flows. J. Hydraul. Eng. 142(5). https://doi.
685e689. https://doi.org/10.1061/(ASCE)HY.1943-7900.0000047. org/10.1061/(ASCE)HY.1943-7900.0001120.
Novak, P., Mofatt, A.I.B., Nalluri, C., Narayanan, R., 2001. Hydraulic Zhang, G., Chanson, H., 2016b. Hydraulics of the developing flow region of
Structures, fourth ed. Taylor and Francis, London and New York. stepped spillways, Part I: Physical modelling and boundary layer devel-
Novak, P., Guinot, V., Jeffrey, A., Reeve, D.E., 2010. Hydraulic Modelling: An opment. J. Hydraul. Eng. 142(7). https://doi.org/10.1061/(ASCE)HY.1943-
Introduction. Spon Press, Abingdon. 7900.0001138.
Peyras, L., Royet, P., Degoutte, G., 1992. Flow and energy dissipation over Zuhaira, A.A., Karunarathna, H.U., Reeve, D.E., 2017. Numerical investiga-
stepped gabion weirs. J. Hydraul. Eng. 118(5), 707e717. https://doi.org/ tion of step dimensions impact over gabion stepped spillways. In: Pro-
10.1061/(ASCE)0733-9429(1992)118:5(707). ceedings of the 37th IAHR World Congress. International Association for
Salmasi, F., Chamani, M.R., Zadeh, D.F., 2012. Stepped gabion spillways with Hydro-Environment Engineering and Research, Kuala Lumpur.
low heights. IJST: Trans. Civ. Eng. 36(2), 253e264. https://doi.org/10.
22099/ijstc.2012.640.