Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

Ocean Engineering 234 (2021) 109326

Contents lists available at ScienceDirect

Ocean Engineering
journal homepage: www.elsevier.com/locate/oceaneng

Bridge pier geometry effects on local scour potential: A comparative study


Aly Mousaad Aly a, *, Erin Dougherty a, b
a
Windstorm, Impact, Science, and Engineering (WISE) Research Lab, Department of Civil and Environmental Engineering, Louisiana State University, Baton Rouge, LA,
70803, USA
b
United States Coast Guard, USA

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

Keywords: Scour, or the localized loss of soil around the foundation, is a dominant factor contributing to failure. Scour at
Scour bridge piers, columns of elevated buildings, and wind turbines with monopile foundations can cause damage
Bridge pier after flooding and natural hazards. However, due to the lack of sophisticated modeling, scour is not fully un­
Countermeasure
derstood. This study utilizes computational fluid dynamic (CFD) simulations to understand the performance of
Flow altering
Monopile wind turbine
different flow countermeasures in reducing the scour potential. We look at altering the flow around the pier and
Computational fluid dynamics thereby alleviating the horseshoe vortices and downflow issues. CFD models are developed and validated with
Coastal infrastructure available experimental data. The recommended models, with turbulence closures, are employed to predict the
Coastal protection and rehabilitation bed shear stress for different pier configurations: streamlined shape, tapered sheath, delta vane, and guide wall
with slanting vanes. Reductions in the maximum bed shear stress are 30% for both the angled plate and the delta
vane, 20% for the tapered sheath, and 15% for the guide wall with slanting vanes. By reducing the bed shear
stress, these countermeasures demonstrate their capability to mitigate scour. The proposed solutions have a
potential to minimize the accelerated deterioration and protect bridges, elevated buildings, wind turbines, and
coastal and offshore infrastructure against scour-induced failures.

1. Introduction These flow physics are often difficult to measure experimentally. How­
ever, scour modeling and prognosis are indispensable for the safe and
Scour is one of the most significant factors contributing to failure in economical design of bridges, wind turbines, elevated buildings, and
bridges (Niedoroda and Dalton, 1982). The analysis of 500 bridges in the other inland and offshore infrastructures. In addition, scour modeling is
United States (between 1989 and 2000) shows that 53% of failures were crucial for the design and analysis of potential mitigation devices.
attributed to floods and scour (Wardhana and Hadipriono, 2003). Scour Research on local scour has been enjoying a renewed interest over
may also affect the stability of offshore monopile wind turbines. the past 50 years. Empirical formulas, deriven from field and laboratory
Monopile offshore wind turbines often suffer from scour. Several measurements, may prognosticate the maximum scour depth (Chortis
research studies on the effects of scour on wind turbines in terms of et al., 2020; Najafzadeh et al., 2015). Because of the complex interaction
analysis, measurements, prognosis, dynamic modeling of the structure, among several parameters in the sediment transport process, these for­
and mitigation countermeasures are available in the literature (Lin et al., mulas often serve as a rough estimate of scour depth and may not be
2019; Prendergast et al., 2015; Trojnar, 2020; B. Yang et al., 2020; Yuan used for accurate prognosis. With advances in computational fluid dy­
et al., 2017). namics (CFD) modeling, scour can now be adequately simulated in three
As floodwater passes around an obstruction, the flow accelerates, dimensions. Most CFD models rely on solving the Reynolds Averaged
losing and transporting the soil. Structural supports, such as bridge piers Navier-Stokes (RANS) equations, although Large Eddy Simulations are
and wind turbines, depend on the foundation to transfer the loads to the emerging.
ground. When the sediment supporting a foundation is lost, the structure To assess the scour prevention potential of various flow-altering
may become unstable and collapse (Fig. 1-a). One countermeasure may countermeasures, CFD provides many benefits over laboratory experi­
decrease scour by minimizing the downflow at the face of a pier, ments. CFD requires less physical space and can provide instantaneous
whereas another method may weaken the horseshoe vortices (Fig. 1-b). flow characteristics anywhere within the computational domain.

* Corresponding author.
E-mail address: aly@lsu.edu (A.M. Aly).

https://doi.org/10.1016/j.oceaneng.2021.109326
Received 23 October 2020; Received in revised form 7 June 2021; Accepted 8 June 2021
Available online 18 June 2021
0029-8018/© 2021 Elsevier Ltd. All rights reserved.
A.M. Aly and E. Dougherty Ocean Engineering 234 (2021) 109326

However, numerical errors are always present, and validating the CFD tapered sheath, delta vane, and guide wall with slanting vanes. Re­
model with experimental data will ensure meaningful and accurate re­ ductions in the maximum bed shear stress are in the order of 30% with
sults. CFD simulations can assist in the understanding of how the the angled plate and the delta vane, 20% with the tapered sheath, and
countermeasures may reduce scour. CFD analysis of piers, with different 15% with the guide wall with slanting plates. By reducing the bed shear
geometries, may indicate which geometry has the best performance in stress, these countermeasures demonstrate their potential to mitigate
reducing scour. Mitigation devices can alter the flow velocities, bed scour. The presented solutions can reduce the accelerated deterioration
shear stress, and hence the ability to transport sediment. and protect bridges and other coastal and offshore infrastructure against
Due to the potentially disastrous consequences of local scour at collapse due to scour. The paper layout is as follows. Section 2 provides
bridge piers and other elevated structures, researchers have been an overview of scour and its classification. Section 3 presents recent
attempting to reduce scour effects via different methods (Pandey et al., efforts on scour processes by scaled laboratory models. Computational
2020a; Yang et al., 2019; Yao et al., 2020). These methods are known as modeling, with its potential to resolve the scale issues, is presented in
flow-altering (active) or bed-armoring (passive) countermeasures Section 4. Scour countermeasures are reviewed in Section 5. Section 6
(Wang et al., 2017). Both methods act to alter the primary flow char­ presents CFD modeling to prognosticate local scour. The CFD validation
acteristics, including the strong downflow, when the flow encounters the study is presented in Section 7. In Section 8, numerical investigations of
face of the pier and the horseshoe vortex, entailing vortices that shed off different flow countermeasures are presented. Reynolds number effects
the sides of a pier (Melville and Coleman, 2000). The primary method on the performance of the delta vane countermeasure are addressed in
often used to reduce local scour is the armoring of the bed around a pier, Section 9. Section 10 summarizes the main findings of the paper.
such as with riprap. Recently, however, researchers started to assess the
combination of flow-altering and bed-armoring techniques to synergis­ 2. An overview of scour
tically reduce scour effects (Gaudio et al., 2012). It is worthy to mention
that, the first author of the current paper realized the role of certain Scour can be classified as general, contraction, and local (Wang et al.,
aerodynamic features in reducing wind pressures on buildings (fluid 2017). General scour occurs when the water flow removes sediment
mechanics problem) (Aly et al., 2017; Aly and Bresowar, 2016). This without structure impediment. Contraction scour arises under the nar­
increased our interest to understand whether other features can work in rowing of the width of a river or a channel. Local scour occurs at an
water to reduce the bed shear stress (fluid mechanics problem), and obstacle obstructing the flow, such as a pile or a pier, and generally
hence decrease the scour potential. The literature review shows that a causes ten times the depth due to general scour (Johnson et al., 2013;
streamlined pier shape may decrease the scour depth by as much as Parker et al., 1998). Additionally, a river bed may experience aggrada­
50–57% under various flow velocities (Al-Shukur and Obeid, 2016). tion and degradation, i.e., long-term bed elevation changes due to the
Under clear water conditions, the delta-wing plate experimentally deposition or removal of sediment from natural or man-made causes
reduced the maximum scour depth by 67% (Gupta, 2002). This design (Arneson et al., 2012). Aggradation and degradation differ from general
has shown a maximum scour reduction of up to 32% in unsteady flow; scour. The general scour may be cyclic or caused by short-term events
however, this reduction is highly dependent on the flow depth (Tafar­ such as flooding. The contributions of all of these forms should be
ojnoruz et al., 2010). In general, such devices lack their performance accounted for when evaluating the total scour at piers and abutments.
during floods (live-bed conditions), and they may adversely increase During the local scour process, water flow encounters a structure and
scour. A guide wall with slanting vanes may decrease the maximum accelerates, creating a horseshoe vortex, which loses sediment in the
depth by 90% (Daido and Yano, 1995). Both methods become ineffec­ flow direction (Wang et al., 2017). The horseshoe vortex in Fig. 1-(b)
tive once the sediment is cleared out, below the desired installation provides a visual representation of the flow structure during the local
depth. By armoring the bed below these flow countermeasure devices, scour process. The directed flow pushes and transports sediment at the
scour can be further reduced, compared to the case with bed armoring base. The lee wake vortices generated downstream erode the sediment
alone. behind the pier (Breusers and Raudkivi, 1991). As the depth of the
Due to the lack of sophisticated modeling, scour is not fully under­ resulting scour hole increases, the strength of the horseshoe vortex de­
stood. This study aims to employ the CFD capabilities to understand the creases, which subsequently decreases the transport rate of sediment
performance of different flow countermeasures in reducing scour. The (Arneson et al., 2012).
scope of the study is clear-water scour, with a non-cohesive sand bed, Scour is greatly affected by the type of sediment. Loose and granular
and specific ranges of D50. The complex set of flow equations is reduced soils are more easily eroded by flowing water, while more cohesive or
by employing the two-equation k-ε and k–ω turbulence closures. Both cemented soils are more scour resistant (Arneson et al., 2012). However,
turbulence closures, in validated CFD models, are used to prognosticate the scour hole can be deep in cohesive sediment as in sand beds if the
the bed shear stress for different pier configurations: streamlined shape, time is long enough to reach the maximum depth. While sand may

Fig. 1. Scour: (a) bridge collapse due to scour/erosion (Banks, 2010), and (b) scour vortices around a cylindrical pier (adapted from Roulund et al., 2005).

2
A.M. Aly and E. Dougherty Ocean Engineering 234 (2021) 109326

develop a maximum scour depth within hours, limestone may not materials; (2) models with a hypothesis that the depth of flow is more
experience a maximum depth for years (Wang et al., 2017). Another important than the velocity for sediment transport; and (3) models with
factor related to sediment transport, which affects local and contraction the velocity expressed as a function of Reynolds or Froude number. The
scour, is being clear-water or live-bed scour. Clear-water scour occurs limitations of these models include accuracy only over the region of the
when the water flow does not transport bed sediment or that the amount empirical data, the primary use of uniform, non-cohesive soils, and the
of sediment transported is less than the capacity of the flow. This type of lack of field measurements to correlate equations with laboratory
scour occurs in coarse bed materials, armored streams, flatbeds with low testing.
flow, heavily vegetated beds, and in the case of riprap around bridge To develop better predictive models, numerous studies were con­
piers where rocks are much larger than the transported sediment ducted (Breusers and Raudkivi, 1991; Melville and Coleman, 2000;
(Arneson et al., 2012). Live-bed scour transports bed sediment from Sumer and Fredsøe, 2001b). The normalized scour depth, or the scour
upstream to scour hole. Live-bed scour is cyclic, meaning that the depth divided by the pile diameter (ds/D), for uniform diameter sedi­
developed scour hole (for example, during a flooding stage) fills back ment, is closely correlated to the flow intensity, normalized velocity,
during flood falls. For instance, in a river with coarse bed materials, the water depth normalized by pile diameter, and pile diameter normalized
riverbed may be subjected to a clear-water scour, before and after floods, by median sediment grain diameter (D/D50). Two general equations to
and live-bed scour during flooding or when the discharge rates are the determine local scour were developed (Arneson et al., 2012; Melville
highest. Clear-water scour causes approximately a 10% increase in and Coleman, 2000).
depth, compared to live-bed scour, and may occur over a longer time. ⎫
ds = KyB KI Kd Ks Kθ KG Kt
The critical velocity of the median particle size D50 refers to the speed ⎪

at which the motion of a particle initiates and is used to determine ys
( )0.65
a (1)
0.43 ⎪
= 2.0 K1 K2 K3 Fr1 ⎭
whether clear-water or live-bed scour will occur. If the mean upstream y1 y1
flow velocity, Ū, is less than or equal to the critical velocity of D50, then
each K in Eqn. (1) is a parameter that influences the scour depth (ds
the local scour can be classified as clear-water (Melville and Coleman,
or ys), KyB accounts for the depth of a pier, KI accounts for the flow in­
2000). The critical velocity depends on the flow depth and the size of the
tensity, Kd depends on the sediment size, Ks is for the pier shape, Kθ is for
bed material. If the ratio of the shear velocity of the flow to the fall
the pier alignment, KG is for channel geometry and Kt accounts for time.
velocity of D50 exceeds two, live-bed scour may occur. In clear-water
The scour depth ys, over the flow depth y1, is a function of several pa­
scour, sand beds with dunes tend to increase the equilibrium scour
rameters – K1, a factor of pier shape, K2, a factor of the angle of attack of
depth by 30% (compared to live-bed scour), but, in large rivers, the
the flow, and K3, a factor of bed condition, a is the pier width, and Fr1 is
dunes tend to plane out. The clear-water condition further affects local
the upstream Froude number. Two empirical expressions have been
scour in that scouring continues until the shear stress caused by the
found to better fit the time history data of local scour, in clear water
horseshoe vortex equals the critical shear stress of the sediment at the
(Sheppard et al., 2004).
bottom of the scour hole. Under the live-bed condition, however,
[ ] [ ]⎫
scouring continues until the inflow sediment equals its outflow (Arneson 1 1 ⎬
ds (t) = a 1 − +c 1−
et al., 2012). (1 + abt) (1 + cdt) (2)
Local scour can be further classified as occurring due to wave, cur­ ⎭
ds (t) = a[1 − exp(− bt)] + c[1 − exp(− dt)]
rent, and a combination of the two. In wave-induced scour, the
Keulegan-Carpenter number is the main influential parameter (Sumer in Eqn. (2), the constants a, b, c, and d produce the best least-squares
and Fredsøe, 2001a). The Keulegan-Carpenter number, Kc, is defined as fit of the experimental data. These equations yield more accurate equi­
the amplitude of the flow velocity oscillation times the period of oscil­ librium scour depth. The duration of the test should be at least the same
lation divided by a characteristic length scale such as the hydraulic as that of the experiment conducted in (Sheppard et al., 2004), to pro­
diameter of the pier (similar to the reduced velocity in wind tunnel duce accurate estimates of the equilibrium depth.
testing (Aly, 2014)). A bridge pier may be subjected to forces in multiple The time required to reach the equilibrium scour depth is a vital
directions due to wave and tidal actions. Although wave and tidal ac­ parameter in predictive models. Scour depth was expressed as a function
tions can influence local scour at the piles and piers of elevated struc­ of time, using cylindrical and square bridge pier models in a laboratory,
tures, in this paper, we will focus on steady flow local scour, under clear-water conditions and uniform bed materials (Yanmaz and
representing inland waterway bridge systems experiencing flow pri­ Altinbilek, 1991). A semi-empirical time-dependent analysis of local
marily in one direction. scour was conducted using sediment continuity equations. The vortex
field was calculated in front of the cylinder and at the trailing
3. Laboratory studies for scour modeling wake-vortex system (Graf and Istiarto, 2002). By using Acoustic Doppler
Velocimetry (ADV), the horseshoe vortex was studied, within a devel­
Experiments carried out on local scour are categorized in three areas: oping scour hole, for cylindrical and square piers (Dey and Raikar,
hydraulic, structural, and geotechnical (Wang et al., 2017). Hydraulic 2007). It was observed that, for the two piers, the flow structure and
studies usually focus on the flow mechanisms of local scour; however, turbulence intensities in the horseshoe vortex, in a developing scour
local scour depends on the structure; therefore, hydraulic studies should hole, are reasonably similar.
account for the structural system. Geotechnical investigations usually Experiments on bridge pier scour were carried out under steady and
address the variables related to sediment. Flume tests are often needed unsteady clear-water conditions, with uniform and non-uniform sedi­
to measure the scour depth and rate under steady current flows (Wang ments. A scheme was proposed for computing the scour-depth evolution
et al., 2017; Zhao et al., 2010). Field measurements are usually (Chang et al., 2004). The temporal evolution of jet-driven scour depth,
employed to develop predictive models. in a pothole laying on a cohesionless granular bed, was investigated by
using diverse approaches. Ordinary differential equations were devel­
3.1. Laboratory-driven predictive models oped in (Bombardelli et al., 2018), and the theory was validated using
datasets. The temporal variation of scour depth around a vertical wall
Sediment, bed patterns, and scour depth are measured to derive re­ spur dike was assessed and the influencing parameters were identified
lationships between several variables (Ni et al., 2020; Yilin Yang et al., (Pandey et al., 2020b). A modified Sheppard-Melville method and a new
2020). These variables include the flow depth, effective pier width, corrected time-scale equation were recommended to predict clear-water
Froude number, shear stress, and critical shear stress (Jones, 1984). The scour depth and scour evolution for complex bridge piers (Yifan Yang
models can be grouped into three categories: (1) models for fine bed et al., 2020).

3
A.M. Aly and E. Dougherty Ocean Engineering 234 (2021) 109326

New data on the temporal evolution of downstream clear-water 1997). A useful expression to calculate the critical mean channel ve­
scour depth at a submerged weir were presented (Wang et al., 2020). locity, Vc, was established to understand the sediment motion (Melville
A dimensionless equilibrium timescale was defined and its dependence and Coleman, 2000).
on these parameters was studied. The temporal evolution of scour depth ( )
Vc y
in cohesive sediments was extensively studied in the literature (Bom­ = 5.75*ln 5.53 (5)
u*c D50
bardelli et al., 2018; Chaudhuri et al., 2018; Hamidifar and Omid, 2017;
Lodhi et al, 2017, 2018; Muzzammil et al., 2015). Local scour in In Eq. (5), y is the flow depth. In a later study, the Shields-Rouse
clay-sand mixed cohesive sediment beds was also covered in several equation was adapted for the calculation of the critical shear stress,
studies (Debnath and Chaudhuri, 2010, 2011, 2012; Khalid et al., 2018; for a dimensionless sediment diameter (Guo, 2002). The critical shear
Najafzadeh and Barani, 2014). Local scour was measured around a stress calculated from this equation matched the experimental data, for
group of piles, for different spacings and skew angles. It was noted that diverse types of sediment, with different diameters.
predictive models do not agree with measurements (Salim and Jones, Like the scaling difficulties observed in the structural aspects of local
1996). A method was then developed to predict scour for pile arrange­ scour, as well as aerodynamic modeling (Aly, 2016; Aly and Bitsuamlak,
ments. Local scour was measured around three adjacent piles, with 2013), addressing the scaling issues in sediment modeling is chal­
different arrangements, to find the arrangement that creates the smallest lenging. The effects of the sediment size on the scour depth were
scour depth (Wang et al., 2017). A relationship between the aspect ratio investigated under three uniform sediment sizes at three different model
of submerged piles and the timescale was developed (Yao et al., 2018). scales (Lee and Sturm, 2009). A distortion was realized between labo­
Additional experimental studies were conducted to understand the ef­ ratory and field measurements, due to improper scaling of the horseshoe
fect of pier shape on the scour depth (Al-Shukur and Obeid, 2016). vortex. Scaling scour in laboratory settings, to accurately represent field
observations, still poses a challenge for researchers (Wang et al., 2017).
3.2. Scale issues However, computational fluid dynamics offer an attractive means for
predicting flow velocities and scour at full scale.
Researchers have yet to agree upon a scaling method that takes into
account all relevant parameters (Zhao et al., 2010). Equilibrium scour 4. Scour prognosis via CFD modeling
depth was measured under large-scale turbulence (Ettema et al., 2006).
The normalized scour depth was increased when the pier diameter was One of the main factors limiting the accurate prediction of scour by
decreased. A scour depth adjustment factor was calculated to scale local physical modeling is the scaling from laboratory settings to field settings
scour. However, difficulties still arise in scaling the flow characteristics, and vice-versa (scale issues). The Froude number and Reynolds number
as well as the sediment size. both can affect scour, but usually are not accounted for simultaneously.
Of critical importance to local scour research, is the dependence on One of the two numbers often sacrifices the other. However, one of the
the sediment type and the mechanisms behind sediment transport. The advantages of numerical simulations of local scour is to partially resolve
answer to fundamental questions, such as when the erosion initiates, the scale issues. Other advantages include the ability to model complex
where particles are transported to, and how the particles interact with conditions that experiments cannot achieve, observe the mechanical
each other, may vary depending on the type and structure of the soil, for behavior of the process in detail, and reduce the physical space required
instance, particle size, adhesion, composition, and the degree of satu­ to simulate the process (Wang et al., 2017). Most numerical simulations
ration (Wang et al., 2017). The sediment size is the most influential use computational fluid dynamics (CFD) to model the scour process (Liu
parameter since it is a primary factor in determining the scour shape and et al., 2019; Yazdanfar et al., 2021; Yu and Zhu, 2020), (Zhao et al.,
the fall velocity (Melville and Coleman, 2000). The median particle size, 2021). The limitations of this method include the long computational
or D50, can adequately describe alluvial sediment in uniform materials. time, as per the need for a refined mesh, to accurately represent the
The dimensionless sediment size, d* , is another vital parameter (Julien, sediment grains. The difficulty in approximating the interaction pa­
2010). rameters, such as particle porosity and drag coefficients, and the many
[ ]1/3 assumptions made to predict the scour depth, are discussed in the
d* = D50
(G − 1)g
(3) literature (Wang et al., 2017).
Most CFD studies combine hydrodynamic and morphologic models
v 2

to simulate the flow field and sediment transport. Hydrodynamic models


where G is the specific gravity of the sediment, υ is the kinematic vis­ are classified into three categories: steady Reynolds-Averaged Navier-
cosity of the fluid, and g is the gravitational acceleration. The initiation Stokes (RANS) equations, unsteady RANS (URANS), or Large Eddy
of the sediment motion relies on the critical value of the Shields Simulation (LES). Navier-Stokes (N–S) equations describe the motion of
parameter, τ*c, which depends on the dimensionless sediment size, d*. a fluid, which is usually associated with turbulence (Aghaee and
Eq. (4) expresses the Shields parameter τ*, as the ratio of the bed shear Hakimzadeh, 2010). Direct numerical simulation (DNS) of N–S equa­
force to the submerged particle weight. tions can resolve the spatial and temporal range of turbulence. Although
τ0 u* 2 hSf DNS would be the most accurate method for scour modeling, it is limited
τ* = = ≅ (4) to low Reynolds numbers due to its inhibitive computational cost.
(γs − γ)ds (G − 1)gds (G − 1)ds
Compared to DNS, a simplified model would combine the continuity
In Eq. (4), τ0 = ρ u2* is the bed shear stress, u* is the shear velocity, γs is equation and the RANS equations (Zhou, 2017).
the specific weight of sediment particles, γ is the specific weight of
∂ρU ( )
water, h is the flow depth, and Sf is the friction slope (Julien, 2018). The ∇⋅U=0⋅ + ∇ ⋅ ρUUT = − ∇p + ∇ ⋅ (2μm S + ρR) + ρf (6)
critical shear velocity, u*c, or the velocity of the incipient motion, is ∂t
√̅̅̅̅̅̅̅̅̅
defined as u*c = τc /ρ. For sand with a mean sediment size of 0.5 mm, here, ρ is the fluid density, U is the fluid velocity vector, S is the mean
for example, the critical velocity is approximately 0.016 m/s and the strain-rate tensor, μm is the fluid dynamic molecular viscosity, R is the
critical shear stress is about 0.27 Pa. Estimates in the literature show that Reynolds stress tensor which may be calculated for various turbulence
a shear stress of 0.1 Pa can cause incipient motion in silts, but not in models, p denotes pressure, and f is the external force. The left side of Eq.
sand. Also, a shear stress of 1 Pa can cause incipient motion in sand, but (6) includes the local time derivative of velocity and convective accel­
not in gravel (Julien, 2010). eration. The right side of the equation includes the effects of viscosity,
Different forms of the above equation are available in literature. An pressure gradient, and external forces. Under the Boussinesq eddy
earlier formula for the critical shear velocity was proposed (Melville,

4
A.M. Aly and E. Dougherty Ocean Engineering 234 (2021) 109326

viscosity assumption, the eddy viscosity is modeled by momentum except for the eddy viscosity, μt, which no longer has Cμ as a constant in
transfer due to turbulent eddies. This is based on the analogy between Eq. (8). For the realizable k-ε model, the term Cμ is calculated as follows.
molecular-motion-based laminar momentum and scalar transport, and
1
eddy-motion-based turbulent momentum and scalar transport (Kundu Cμ = * (12)
A0 + As kUε
et al., 2012). The molecular viscosity describes how the momentum is
transferred by molecular motion. In the RANS equations, R represents
where A0 and As are constants and U* is the square root of the sum of the
the mean viscous stress and is defined as follows.
square of the mean rate of the state tensor and the mean rate of the
R = − u’ u’ T = 2 μρt S − 23 kδij ; rotation tensor. The model is called ‘realizable’ because it satisfies the
1( ) (7) physics in turbulent flows, as well as certain mathematical constraints
S= ∇U + (∇U)T on the Reynolds stresses. Some advantages include, more accurate pre­
2
dictions of the spread rate of both planar and round jets, as well as the
where μt is the eddy viscosity, k is turbulence kinetic energy, u′ is the capability of modeling flows involving rotation, flows with boundary
fluctuating component of the velocity and δij is the Kronecker delta. layers under strong adverse pressure gradients, flows with separation,
Reynolds stresses develop from the nonlinear advection term of the and flows with recirculation (Spalart, 2000).
conservation of momentum equation and represent the average stresses The k-ω model improves upon the k-ε model in that the damping
induced by the flow fluctuations (Kundu et al., 2012). Reynolds stresses functions are no longer employed and the Dirichlet boundary conditions
may be considered as the rate of mean momentum transfer due to flow allow for better numerical stability (Menter, 1993; Wilcox, 1993). The
fluctuations. Reynolds stresses tend to be much larger than viscous k-ω turbulence closure significantly improves the modeling of flows with
stresses, except close to a solid surface where the fluctuations vanish, adverse pressure gradient, although it is sensitive to the value chosen for
and the mean flow gradients are large. the free-stream variable. Eq. (13) solves for the turbulence kinetic en­
To solve the RANS equations, the two-equation eddy-viscosity ergy, k, and Eq. (14) solves for the dissipation of the turbulence energy,
models are utilized: one equation for the turbulence kinetic energy, and ω. The parameters k, ω, and ε are defined in Eq. (15).
the other equation is related to the turbulent length scale (Menter,
∂ρk
1992). The primary models in RANS simulations are the k-ε and k-ω + ∇ ⋅ (ρUk) = ∇ ⋅ [(μm + σ k μt )∇k] + R∇U − β* ρkω (13)
∂t
turbulence closures. The k-ε model was developed in 1973 and has been
improved over time (Jones and Launder, 1973). In the standard k-ε ∂ρω γ
equation, the eddy viscosity, μt, is expressed as follows. + ∇ ⋅ (ρUω) = ∇ ⋅ [(μm + σω μt )∇ω] + R∇U − ρβω2
∂t vT
ρσω2
k2 + 2(1 − F1 ) ∇k⋅(∇ω)T (14)
μt = C μ ρ (8) ω
ε
′ ′
where k is the turbulence kinetic energy, ε is the dissipation rate, and Cμ 1 ′ ′ ε ∂u ∂u
k = ui ui , ⋅ ω = * , ⋅ ε = v i i (15)
is a constant. The rate of production of turbulence, Pk, is given by: 2 kβ ∂xk ∂xk

Pk = μt (∇ × U)⋅(∇ × U)T (9) the new parameters μm , σ k , β* , σω β, σ , ω2 , and γ are constants defined
in the literature, except for vT which is the turbulence kinematic vis­
The two equations of the k-ε turbulence closure are: cosity and F1 which is a blending function.
[( ) ] The Mentor Baseline (BSL) model blended the k-ε and the k-ω models
∂ρk μ
+ ∇ ⋅ (ρUk) = ∇ ⋅ μm + t ∇k + Pk − ρε (10) and helped reduce the ambiguity of the free-stream term. Like the k-ε
∂t σk
model, the k-ω model under predicts the flow separation. The Shear-
∂ρε
[( )
μt
]
ε ε2 Stress Transport (SST) model improves upon the standard k-ω model
+ ∇ ⋅ (ρUε) = ∇ ⋅ μm + ∇ε + C1 Pk − C2 ρ (11) and assumes that the turbulence shear-stress is proportional to the tur­
∂t σε k k
bulence kinetic energy in the logarithmic and wake regions of the
in these equations, Cμ, C1, C2, σk, and σε are constants, primarily boundary layer, and the eddy viscosity is limited by taking the maximum
given as: Cμ = 0.09, C1 = 1.44, C2 = 1.92, σk = 1.0, and σε = 1.3 (Launder term in the denominator of the eddy viscosity equation (Menter, 1993).
and Spalding, 1974). The numerical simulations underpredicted the scour depth by 15% up­
Although being widely used, the k-ε model has several drawbacks stream of the pier and up to 30% downstream. Similarly, a 10–20%
including a lack of sensitivity to adverse pressure gradients. The model underprediction of the scour depth was found for steady flow around
over-predicts the shear stress, which is often a challenge in the modeling submerged cylinders (Zhao et al., 2010). The underprediction may arise
of flow separation (Menter, 1993). Another disadvantage is that the from the steady-state flow modeling which did not account for the un­
equations in the model may be stiff, and convergence may be difficult steady effects of the fluctuating velocity components of the horseshoe
due to the damping terms. Several researchers have shown the effec­ and the lee-wake vortices. However, with the consideration of
tiveness of the k-ε turbulence closure under certain conditions. flow-altering countermeasures, more reliable designs can be achieved.
The scour hole created by an inundated bridge deck was experi­
mentally measured under steady flow at a high Reynolds number 5. Flow altering countermeasures
(Tulimilli et al., 2010). The hydrodynamic model was coupled with a
sediment transport model that simulates the scour of the initially flat One of the earliest comprehensive studies on scour countermeasures
bed, in iterative increments. Sediment transport is initiated when the was published by the National Cooperative Highway Research Program
shear stress exceeds its critical value. The flow mechanism was studied (NCHRP) in 1998 (Parker et al., 1998). The study provided recom­
for three piles (with different arrangements) under steady flow, using mendations on the use of scour countermeasures, other than riprap.
RANS simulations with the k-ε turbulence closure, and sediment scour Although the preferred method was bed armoring with riprap or cable
model. The k-ε turbulence model was used to reduce the computational tied blocks over a geotextile, other countermeasures include sacrificial
cost. The simulated results fairly agreed with the experimental data piles, upstream sheet piles, collars and horizontal plates, vanes or plates,
(Zhang et al., 2017). modified pier shape or texture, slots in piers or pier groups, and suction
An improvement upon the standard k-ε model is the realizable k-ε applied to the pier. Since then other flow altering techniques have been
model. The realizable k-ε equations follow the standard k-ε equations, developed, including internal connecting tubes, surface guide panels,

5
A.M. Aly and E. Dougherty Ocean Engineering 234 (2021) 109326

threading, and sleeve and downstream bed sill (Tafarojnoruz et al., vertical plates can separate the flow around a pier, and hence reduce the
2010). Scour countermeasures may be categorized as pier geometry intensity of the horseshoe vortex (Parker et al., 1998). Vertical plates
modification or openings, pier attachments, bed attachments, among with an aspect ratio of 3:1 may reduce scour by 70–80%, and they are
other techniques. effective in low and high flow rates (Tafarojnoruz et al., 2010).
One of the more successful vane designs includes the delta vane,
5.1. Pier geometry and pier openings which disrupts the flow at the base of the pier and produces vortices that
rotate against the horseshoe vortex (Gupta, 2002). This design has
Research regarding the effects of the geometry of the pier (or the shown scour reduction of up to 32% in unsteady flow, and 67% in
shape) itself shows that reductions in the scour depth can be realized. clear-water laboratory settings. However, this design is highly depen­
Studies on the pier shape have been conducted with a general under­ dent on the flow depth (Tafarojnoruz et al., 2010). The use of a vertical
standing that the scour depth can be reduced as the pier becomes more guide wall and vertically inclined plates at 45◦ showed up to 90%
streamlined (Melville and Coleman, 2000). The scour depth is reduced reduction in clear-water scour, with similar limitations as the delta vane.
(or increased) by the following factors for the following pier shapes: 1.1 Another pier attachment was developed, which consists of a guide
for square, 1.0 for circular (round nose, or group of circular piers), and wall with vanes (Daido and Yano, 1995). The guide wall is intended to
0.9 for piers with a sharp nose (Arneson et al., 2012). In an experimental decrease the intensity of the horseshoe vortex by disturbing the
study, a larger variety of pier shapes were considered (Al-Shukur and approaching flow. Slanting vanes are used to further decrease the in­
Obeid, 2016). Compared to a circular pier, the streamlined pier reduced tensity of the horseshoe vortex. The results showed that at a 90%
the scour depth by 50–57% for three different clear-water flows. The reduction in the scour depth was realized, compared to the unprotected
next best scour-reducing geometries were the hexagonal shape circular pier. This design was tested under a few flow velocities. In a
(28–40%), the sharp nose (23–29%), and the elliptical shape (7–20%) later study, this configuration reduced scour by only 50%, compared to
(Al-Shukur and Obeid, 2016). the unprotected circular pier (Parker et al., 1998). The effect of sub­
Experimental testing and CFD modeling were conducted for circular, merged vanes on the flow pattern around a pier was numerically studied
diamond, and square piers and the findings agreed with previous liter­ by testing combinations of 2, 4, and 6 vanes angled at 20◦ and 30◦ (Azizi
ature that scour increases for the diamond shape, and even more for the et al., 2016). Three plates angled at 30◦ significantly reduced the ve­
square shape, when compared to the circular pier (Khosronejad et al., locity near the bed. This decrease in velocity is associated with a
2012). The optimization of the pier geometry relies on the flow angle, decrease in the ability of the flow to transport sediment.
which may not capture the realistic field conditions (Tafarojnoruz et al., Most experiment discussed thus far focused on scour in steady flow.
2010). The effectiveness of splitter plates and threaded piles under waves and
Scour was investigated for complex pier shapes that arise when current was investigated (Dey et al., 2006). The main scour mechanism
scouring exposes pile caps and the piles themselves (Baghbadorani et al., under waves is vortex shedding. Attaching a splitter and threading the
2018). Fifty-two tests were conducted on four different complex pier pile or wrapping cables in a helical pattern around the pile, can disrupt
models, in clear-water conditions, to help improve the accuracy of the vortex shedding (this approach is known in wind engineering applica­
published HEC-18 equations (Richardson and Davis, 2001). The exper­ tions as well, for example, the older antenna of the car had a spiral shape
iments agreed with the equations which predict reduced scour depth as to disrupt vortex shedding and hence reduce the vortex shedding ef­
the pile cap is exposed to sediment that interrupts and minimizes the fects). With a cable to pile diameter ratio of 0.75, a maximum reduction
downflow effects. A collar located at the base of a pier, or below the level of scour under waves of 51% was realized. Under steady current, a
of the bed, works under the same mechanism. The performance of col­ maximum scour depth reduction of 46% was realized for a cable to pile
lars and exposed pile caps rely on a minimum diameter to reduce scour, diameter ratio of 0.1. However, the splitter plate reduced scour by 62%.
otherwise, the scour depth may increase (Parker et al., 1998). Although A combination of flow-altering countermeasures may prove to be
a pile cap is not meant to be a countermeasure, it is beneficial for the even more effective. The following combinations were experimentally
safety of a bridge as the last expedient, in case other countermeasures, investigated in clear water: submerged vanes and a bed sill, slot and
inspection, or maintenance plans fail. sacrificial piles, collar and sacrificial piles, slot and collar, and bed sill
The performance of a pier with a vertical slot was experimentally and collar (Gaudio et al., 2012). Most combinations did not reduce scour
investigated, to understand its ability to reduce scour by diverting the more than the individual countermeasures. However, the slot and collar
flow away from the bed, or by reducing the downflow effects (Chiew, combination reduced scour by 82%, but its efficiency was reduced when
1992). A reduction of 20% was realized in the maximum scour depth, for the collar rim was exposed (for example, during degradation or when the
a slot length of twice the diameter of the pier and a width of at least slot is blocked by debris). When a sill was positioned upstream of a pier
one-quarter of the diameter. A reduction of 45% in the maximum scour with a collar, scour was reduced by 63–76%, but degradation is a po­
depth was achieved, after experimentally testing different sizes of tential concern.
openings in different shapes of piers (EL-Ghorab, 2013). A significant In summary, there are many countermeasures, including active
scour depth reduction of 89% was realized by utilizing a sacrificial countermeasures that attempt to alter the flow pattern around the pier,
perforated pile upstream of a perforated main pile (each pile has a hole with a potential to reduce scour. Each method, such as altering pier
that is 43% of the pile diameter, orientated at 45◦ ) (Elnikhely, 2017). geometry, sacrificial piles, pier attachments, or openings, has potential
Other studies had shown that sacrificial piles alone may reduce the benefits, but may have some concerns. There is no single solution for the
maximum scour depth by 44%, in the field, and up to 60% under lab­ local scour problem, so different countermeasures and solutions may be
oratory conditions (Parker et al., 1998). Nevertheless, pier openings considered in the design process, especially for unique projects. In the
cause practical problems including potential blockage by debris and a following section, we will create and test computational models that will
reduction in the pier’s strength. be used to study different scour countermeasures.

5.2. Pier attachments 6. Case study of CFD application to predict local scour

Experiments with plates and vanes to reduce scour depth began as Here we model and predict the bed shear stress and velocity around a
early as the 1960s (Gaudio et al., 2012). Before that time, vertical plates pier. We evaluate different turbulence closures and validate the model
installed in the channel, were primarily used to prevent lateral channel with available experimental data. Later, the model with the proper
migration, although a reduction in scour had been noticed. The effec­ turbulence closures will be employed to investigate the performance of
tiveness of vertical plates and vanes was compared, showing that proposed scour countermeasures.

6
A.M. Aly and E. Dougherty Ocean Engineering 234 (2021) 109326

6.1. Computational domain in the vertical direction is smaller towards the bed so that the height of
the first cell from the bed surface is 1.0 × 10− 4 m. The cell height also
A computational domain was created using the same dimensions as becomes smaller towards the pier surface. The height of the first cell on
those in (Roulund et al., 2005). A schematic representation of the the pier wall is 6.3 × 10− 5 m. This size was determined from the y+ value
computational domain is shown in Fig. 2. The dimensions of the domain to resolve the viscous sublayer (White, 2002). The k-ω SST model was
are given in terms of the diameter of the pier, D. A symmetry boundary the initial turbulence closure, which requires a y+ value of less than 1
condition was considered at the side boundaries. This condition assumes (ANSYS Inc, 2021).
a zero flux of all quantities across a symmetry plane. The shear stress is
y Uτ
also set to zero, so a surface with symmetry boundary conditions may be y+ = ρ ;
μ
a slip surface. The flow along the sidewalls of the flume in the experi­ √̅̅̅̅̅
τw (16)
ment was a no-slip surface, but because of the distance of the side walls Uτ = ;
ρ
to the cylinder, the effect of the side walls was not anticipated to impact 1
the results significantly. The blockage ratio is 1D/20D or 5%. The
2
τw = Cf ρU∞ ;
2
experiment was performed with a width of 5.6D which has a blockage Cf = 0.058Re−l 0.2

ratio of approximately 18%. The blockage effect did not pose any sig­
nificant problems because the flow velocity was increased by less than Cf is the skin friction, and all other quantities including the Reynolds
3%, close to the sidewalls (Roulund et al., 2005). number, viscosity, density, and flow velocity are known (Table 1). For
Symmetry condition was applied at the top surface of the domain. the standard and realizable k-ε models, enhanced wall treatments were
This will place a “lid” on the fluid surface since the free surface is not employed. These treatments bring improvements over the standard wall
modeled. Since a lower Froude number will decrease the amount of functions that are not recommended for use under boundary layer sep­
elevation of the free surface of an object in a flow, such as a wake at the aration (Khaled et al., 2021). The enhanced wall treatment is a y+
bow of a boat, lower Froude numbers may be modeled with a lid con­ insensitive approach and will act as a wall function if the first grid point
dition. Two experiments with Froude numbers of 0.2 and 0.5 were is in the log layer. To predict the physics in the near-wall region, this
performed for a depth of flow of 26 cm and a 31 cm pier diameter method formulates the law-of-the-wall as a single wall law for the entire
(Roulund et al., 2005). A height difference of 22%, compared with the wall region, by combining the linear (laminar) and logarithmic (turbu­
water depth, was found for a Froude number equal to 0.5. The height lent) laws-of-the-wall (ANSYS Inc, 2021). The wall shear stress, like the
difference was only 3.8% for a Froude number of 0.2. The velocity normal velocity gradient at the wall, is dependent on the formulation of
profiles in both cases were also compared and the Froude number of 0.2 the near-wall region using the enhanced wall treatment.
matched closely with the free surface velocity distributions. The rigid lid The generated mesh is shown in Fig. 3. Notice how the mesh ele­
approximation is therefore applicable when the Froude number is of the ments decrease towards the cylinder and bottom surface. The number of
order of 0.2. elements was then optimized following a grid independence study, and
The boundary layer thickness to pier diameter (δ/d) was kept con­ the results were compared with available experimental data. The mesh
stant. The value of δ/d is about 1. The CFD analysis was carried out for a characteristics for each trial are listed in Table 2. The ‘element count’
δ/d value that is similar to the reference experimental data (Roulund refers to the number of elements in each direction, over height (z-di­
et al., 2005). It is worth mentioning that the separation of the bed rection), width (y-direction), and length (x-direction). The O-grid
boundary layer (forming the horseshoe vortex) will be delayed if the around the cylinder, for extra grid refinement, was not altered from test
boundary layer thickness to pier diameter ratio, δ/d, is small (δ/d < 0.5). to test since the y+ value depends on ensuring the first cell height re­
This may lead to a more uniform velocity distribution in the upstream mains the same.
boundary layer flow, which can lead to a smaller-size horseshoe vortex.
For small values of δ/d, the boundary layer may not separate (no 6.3. Discretization
horseshoe vortex).
The boundary condition for the inlet was a velocity inlet. A velocity The flow governing equations were solved using a pressure-based
profile based on the average velocity was applied to the inlet, as well as a finite volume discretization process. A control-volume-based tech­
constant absolute velocity equal to the main flow velocity. The turbu­ nique was employed, which converts a general scalar transport equation
lence was set to 5% with a length scale of 0.54 (Roulund et al., 2005). to an algebraic equation that can be solved numerically (ANSYS Inc,
The bottom surface and pier surface are wall conditions with no slip. 2021). This control volume technique consists of integrating the trans­
Wall functions dependent on the turbulence closure were employed as port equation to yield a discrete equation that expresses the conserva­
will be discussed in the next subsection. The outlet was set as a pressure tion law. A second-order upwind scheme was applied. The momentum
outlet. Zero gradient and Neumann conditions were applied. equations were also discretized. The SIMPLE algorithm is used to couple
velocity and pressure for mass conservation, to obtain the pressure field.
6.2. Mesh generation A predictor-corrector method is then used to predict and correct the
pressure and flux fields.
A structured O-grid mesh was created (Fig. 3). The mesh cell height The models were run for a varying number of iteration steps until the
scaled residuals were decreased and the solution was converged. The
drag coefficient of the pier leveled out to constant and the mean value of
the lift coefficient was zero (for steady-state RANS simulations) (Blevins,
1984).

7. CFD validation

A velocity profile was applied at the inlet with the mean flow velocity
as the main scaling parameter and run in an empty domain with the k-ω
SST turbulence closure. The resulting velocity profile at the location of
the cylinder, in the middle of the computational domain, was then
compared to the undisturbed experimental velocity profile. The velocity
Fig. 2. Computational domain in dimensions of the pier diameter D. profiles, for different values of D, at a constant inlet velocity, were also

7
A.M. Aly and E. Dougherty Ocean Engineering 234 (2021) 109326

Fig. 3. Structured mesh: (a) isometric view, (b) section view at the axis of symmetry for y = 10D, and (c) top view. Note that the mesh element size decreases toward
the pier and the bottom surface (walls).

one of the first iterations with the unstructured mesh, we used a


Table 1
maximum global element size of 0.2 m. We added three prism layers to
Parameters of the rigid bed experiment (Roulund et al., 2005).
the bottom surface and the pier surface, with a maximum element size of
water depth, h (cm) 54 0.1 m, a maximum near wall height of 0.01 m, and a growth rate of 1.2.
boundary layer thickness, δ (cm) 54 The unstructured mesh was computed with 350k tetrahedral-shaped
mean flow velocity, ū (cm/s) 32.6 elements. When this unstructured mesh was used to simulate the flow
pier diameter, d (cm) 53.6 with the k-ε realizable turbulence closure, the solution converged. The
Rel = ūdρ/μ 1.7 × 105
friction velocity, uτ (cm/s) 1.3
drag coefficient leveled at a constant value, but the lift coefficient was
Froude number 0.14 irregular before nearing zero. To improve upon the coarseness of the
mesh, without significantly increasing the number of elements, we
placed a mesh density around the cylinder (pier).
The unstructured mesh requires more elements, compared to the
Table 2
Grid details, with the total number of elements (structured mesh). structured mesh, to produce consistent results. To increase the number
of elements without significantly increasing the computation time, the
total elements height count width count length count
size of the domain was optimized. The original size was chosen as 20D x
113,324 20 60 60 20D x 2D, to safely assume the effects of the walls did not impact the
277,264 20 100 100
results of flow around the cylinder and close to the bottom surface. The
436,796 32 100 100
476,504 35 100 100 domain size was decreased to a width of 11D, a height of 1.5D, and the
width was further decreased to 9D, to check if the results would change.
Each simulation was run using the k-ε realizable turbulence closure,
compared. The model was simulated using the RANS equations, with the since this model performed well for the structured mesh case. The results
k-ω SST turbulence closure, for each mesh in Table 3. The experimental of the mesh optimization are shown in Fig. 4, for the x-component of
data include the x-component of the velocity, at a distance 2D upstream velocity.
and downstream of the cylinder, in the plane of symmetry, or at y = 10D In Fig. 4, it is shown that the change in the domain size has a mini­
(Roulund et al., 2005). mum effect the ability to predict the x-component of the upstream ve­
We investigated the ability of structured and unstructured grids to locity flow. Because the physical experiment was conducted in a flume
replicate the flow characteristics from the experiment. A structured of 5.6D (width), there were likely effects from the flume walls that are
mesh may be more spatially efficient, but the fluid domain should be now incorporated into the numerical model by decreasing the domain
subdivided to control the mesh intensity, for accurate CFD simulations. width. This effect is further exaggerated in the downstream region of the
An unstructured mesh may use differently shaped elements, such as pier where the 9D width yields result close to those of the experiment.
tetrahedral shapes, to generate the volume mesh, with inputs such as the
maximum global size and surface size. With fewer inputs, the unstruc­
tured mesh is easier to apply to complex geometries. However, creating
an unstructured mesh is an iterative process, and the grid is usually
refined by trial and error.
We used the same pier and domain size to create the unstructured
mesh, as shown in Fig. 2. The maximum global element size, and the
maximum element size on any surface, can be set. ANSYS suggests
adding prismatic layers at solid surfaces to capture the shear stress and
other essential flow physics in the boundary layer (ANSYS Inc, 2021). In

Table 3
Parameters of two different unstructured meshes (size in m).
total no. mesh inner outer pier bottom pier
of density surface surface surface prism prism
elements size size size size height height

457k None 0.05 0.1 0.01 0.005 0.001


449k 0.1 0.05 0.1 0.05 0.005 0.001
Fig. 4. Domain size comparison of the x-velocity component at z = 9.3%D.

8
A.M. Aly and E. Dougherty Ocean Engineering 234 (2021) 109326

The simulation results for the reduced domain width, other than 9D,
were relatively close to each other in the downstream zone. In general,
only the 9D width domain seems to affect the simulation of the x-
component of velocity for the regions considered. Considering the
benefits of predicting the x-velocity component for a reduced domain
size of 11D, which also predicted the z-velocity components downstream
of the pier well when combined with a 1.5D height, this combination
seems to be the best optimum domain size, to be used for the investi­
gation of different flow countermeasures in the coming section.
Once an optimum domain size was determined, the mesh parameters
were optimized, to resolve the viscous sublayer. If the unstructured
mesh could adequately resolve the viscous sublayer, the standard k-ω,
and k-ω SST may prove to predict the flow velocities better than the k-ε
models since they are better at modeling adverse pressure gradients. The
Fig. 6. Unstructured mesh model comparison of the x-velocity component at z
wall and cylinder’s first cell height should be small, so a mesh was made
= 9.3 %D.
with smaller elements on the surface around the cylinder to reduce the
total number of elements and computational time. Another mesh was
made with a mesh density with a maximum size of 0.1 m at the same approximate the experimental data better with the unstructured mesh
location as the inner bottom surface. The parameters of the two different for a greater y+ value. This may be due to the improvement of the
meshes are listed in Table 3. Each mesh has five prism layers. Each mesh tetrahedral shape over the rectangular prisms of the structured mesh or
was used with the k-ω SST model since both k-ω models performed achieving a y + value close to 1.0 may not be that critical.
relatively similar compared to experimental data and the k-ω SST is The parameters of each mesh are listed in Table 4, and they are
regarded as an improvement upon the k-ω model. The density mesh was labeled mesh 1, 2, and 3. Each mesh has a mesh density to refine ele­
then used with the standard and realizable k-ε turbulence models, with ments around the pier. The unstructured mesh was optimized for the
enhanced wall treatment. The x and z velocity components from each realizable k-ε model so that the mesh parameters that most closely
simulation were compared with experimental results. match the experimental data could be applied to other pier shapes (to
In Figs. 5–6, all models yield the x-velocity component upstream of investigate the performance of different flow countermeasures in a
the pier closely although the results vary more when z = 1.9%D. This is comparative study).
likely due to the mesh near the bottom surface not adequately capturing
the flow at this height. For the downstream region of the pier, the k-ω 8. Numerical investigations of different countermeasures to
realizable and the standard models predict the flow better than the case reduce local scour
of using the structured mesh. This could be a feature of the tetrahedral
element shape which may provide better vertical coverage of the We investigated four different countermeasures to reduce local scour
boundary layer, since the elements do not form distinct layers in the (Fig. 8). The proposed countermeasures are: (1) tapered streamlined
vertical direction, compared to the structured mesh. The standard k-ε sheath, (2) delta vane, (3) guide wall with slanting vanes, and (4) angled
and realizable k-ε closures more closely match the experimental data plate footing.
downstream of the pier for the x-velocity components. The realizable k-ε
tends to predict the downstream velocity slightly better than the stan­ 8.1. Tapered streamline sheath
dard k-ε as shown in Figs. 5–6.
Another finding of the mesh simulations using the parameters in The tapered streamlined sheath was modeled in AutoCAD (Fig. 8-a).
Table 3 include the difference in the wall y+ values, between structured The pier has the same dimensions as those in (Roulund et al., 2005). In
and unstructured grids. Fig. 7-(a) shows the wall y+ values are between terms of the diameter, the width of the sheath is 1D and the total length
0.1 and 1.0 for the structured mesh. The unstructured mesh would is equal to 3D. The sheath extends 0.5D vertically where it meets the
require many more elements than the structured mesh to achieve a pier. The countermeasure was then meshed using tetrahedral-shaped
similar wall y+ value. The unstructured y+ values are shown in Fig. 7- elements with the same parameters as those of the unstructured mesh.
(b), where they vary between 1.0 and 17. The two k-ω models tended to Three prism layers with a growth rate of 1.2 are applied at the wall. The

Fig. 5. Unstructured mesh model comparison of the x-velocity component at z = 1.9%D.

9
A.M. Aly and E. Dougherty Ocean Engineering 234 (2021) 109326

pier. This confirms the ability of the streamlined sheath to reduce scour
at the face of the pier, compared to the cylindrical pier.
Fig. 11 shows that the velocity at z = 1.9% is closely correlated with
the bed shear stress in Fig. 10. The streamlined sheath reduced the
magnitude of the velocity by approximately 10% (from 0.5 to 0.45 m/s)
which reduces the ability of the flow to initiate sediment movement. The
streamlined shape allows the velocity field in this plane to be more
constant compared to the cylindrical pier. Since this plane is approxi­
mately one cm above the bed, the velocity will decrease logarithmically
to zero. At this height, sand with a critical velocity ranging between
0.01 m/s and 0.1 m/s can be transported, except for the dark blue re­
gions (Fig. 11). The velocity is 0.35 m/s around the aft half of the
tapered sheath where it is sharply decreased to 0.002 m/s for the aft half
of the cylindrical pier. Accordingly, more sediment can be transported in
this region, for the streamlined pier.
Based on the reduction of bed shear stress by 20% and the decrease in
the velocity magnitude by 10%, the streamlined tapered sheath would
be a good flow-altering scour countermeasure, especially upstream of
the pier. Furthermore, the sheath at the face of the pier acts as a bed
armoring device as the downflow deflects from it instead of the sedi­
ment. For this mean flow velocity, one would expect sand to be trans­
ported along the sides of the pier just forward of center for the
cylindrical pier, and only finer sands along the sides for the streamlined
tapered pier. The streamlined tapered pier does not have a decrease in
flow velocity directly downstream of the pier to the extent of the cy­
lindrical pier. Immediately after the aft of the cylindrical pier, silt could
be transported based on the value of the critical shear stress, whereas
sand with a critical shear stress of 0.3 Pa could be transported aft of the
Fig. 7. Wall y+ values for the cylindrical pier: (a) structured mesh, and (b) streamlined tapered sheath. At this flow rate, however, the tapered
unstructured mesh. streamlined sheath can effectively reduce the transport of coarser sand
and finer sediments in front of the pier.
total elements are 1.36 million. The generated mesh is shown in Fig. 9.
The streamlined tapered pier was simulated using the model that best
represented the experimental data, the RANS equations with the real­ 8.2. Delta vane
izable k-ε turbulence closure. The boundary conditions remained the
same from the previous structured and unstructured simulations so that The delta vane was modeled in AutoCAD (Fig. 8-b). The dimensions
the resulting flow structures may be compared. for the delta vane are 2D from the apex to the opposite side and 1.5D for
Contours of bed shear stress and velocity were obtained after the the length of the side opposite to the apex. The apex is attached to the
simulations were completed. Each generated figure was compared bed upstream of the pier and the opposite side is raised and attached to
against the pier that most closely approximated the experimental data. the pier so that the vane forms a 15◦ angle with the bed. A spinal rib is
The bed shear stress is a critical parameter in determining an incipient also attached under the vane for support. The countermeasure was then
motion of sediment. Fig. 10 shows that the cylindrical pier experiences meshed using tetrahedral-shaped elements with the same parameters as
shear stress of about 1 Pa, whereas the streamlined tapered sheath has those of the unstructured mesh. Three prism layers with a growth rate of
maximum shear stress of 0.8 Pa (20% reduction). There is a larger region 1.2 are applied at the wall. The total elements are 1.5 million. The
of higher shear stress for the cylindrical pier, compared to the stream­ generated mesh is shown in Fig. 12. The delta vane attached pier was
lined pier. Sand could be transported in the red regions for the cylin­ simulated using the RANS equations with the realizable k-ε turbulence
drical pier, whereas only finer sands could be expected to be transported closure. The boundary conditions remained the same from the previous
along the sides of the streamlined tapered sheath. Both piers could cause structured and unstructured simulations so that the resulting flow
sediment with a critical shear stress of 0.5 Pa to be transported about the structures may be compared.
same amount based on the green portion of the contour. Transport Contours of velocity, pressure coefficient, bed shear stress, and tur­
occurring most along the sides of the piers is consistent with the previous bulence intensity were obtained after the simulation was complete. Each
figures. The earlier figures indicated transport might also occur down­ generated figure was compared against the cylindrical pier. Fig. 13
stream of the streamlined tapered pier. From this figure, sediment with a shows promising results in the delta vane’s ability to minimize shear
critical shear stress of 0.3 (coarse sand) could be transported just aft of stress on the bed. The highest shear stress for the delta vane attached
the tapered pier but only silts could be transported aft of the cylindrical plate is 0.7 Pa at the tip of the apex. This is approximately 30% less than
the maximum shear stress for the cylindrical pier. Furthermore, this is a

Table 4
Parameters of the unstructured mesh, optimized for the realizable k-ε model (size in m).
mesh total no. of mesh bottom surface bottom prism bottom prism pier surface pier pier prism
number elements density size height layers size prism layers
height

1 317k 0.1 0.1 0.005 5 0.05 0.001 10


2 372k 0.1 0.1 0.005 5 0.01 0.005 5
3 391k 0.1 0.1 0.005 5 0.01 0.001 5
4 1.2m 0.1 0.05 0.005 3 0.01 0.001 3

10
A.M. Aly and E. Dougherty Ocean Engineering 234 (2021) 109326

Fig. 8. Different shapes are designed and attached to the cylindrical pier: (a) streamlined tapered sheath, (b) delta vane, (c) guide wall with slanting plates, and (d)
angled plate footings.

small region at 0.7 Pa, which indicates sediment with critical shear stress It should be noted that altering the orientation of the delta vane to
less than about 0.5 Pa (sand) will be transported. Both piers have about the bed or the pier will affect the resulting flow characteristics and bed
the same area with 0.5 Pa shear stress and so would transport a similar shear stress. If implemented on an actual bridge pier, the design orien­
amount of sediment for this shear stress. The delta vane experiences a tation would need to be maintained. This could be accomplished by
larger zone bed shear stress of 0.3 Pa, compared to the cylindrical pier, armoring the bed beneath the vane to ensure the apex of the vane re­
which indicates that more sediment could be transported. However, mains in contact with the bottom surface. The combination of a flow-
unlike the tapered streamline pier, the delta vane decreases the bed altering method with a bed-armoring method could extend the oper­
shear stress in the wake zone to about zero Pa, similar to the cylindrical ating life or inspection intervals of the bed armoring technique by
pier. This means that both the cylindrical pier and the delta vane reducing the velocities and shear stress the bed armoring device is
countermeasure may not transport significant sediment in the wake exposed to. Further studies would need to assess whether the cost of
zone. The delta vane countermeasure may direct the approaching flow construction of a delta vane on a pier could outweigh the potential
upward, away from the bed. savings in maintenance costs for the bed armoring method.
Fig. 14 shows the effectiveness of the delta vane to reduce the ve­ The delta vane seems to be successful at directing the flow up and
locity. The velocity magnitude is the largest at the apex of the vane, away from the bed. There is a 30% reduction in maximum bed shear
where the flow is directed upward, away from the bed. The maximum stress for the delta vane compared to the cylindrical vane, which means
magnitude of the velocity is about 0.45 m/s which is 10% lower than the larger sediment may be transported such as coarser sand in this case. The
corresponding maximum velocity experienced with the cylinder (0.5 m/ maximum velocity just above the bed is reduced 10% for the delta vane,
s). The velocity along the side of the pier is greatly reduced compared to however, the delta vane has a larger region of 0.35 m/s which could
the delta vane, 0.33 m/s versus 0.5 m/s for the cylindrical pier. Based on mean transporting more medium-sized sand at the bed. Both will
Fig. 14, the delta vane cause sediment transport with similar critical transport approximately the same amount of finer sediment such as silt
velocities as the cylindrical pier in the wake (0–0.16 m/s) but will lead based on the bed shear stress contours. The delta vane is an improve­
to more sediment transport with a critical velocity equal to 0.35 m/s ment over the streamlined taper sheath pier because it provides bed
than the cylindrical pier. These numbers will decrease exponentially armoring in the case of a strong downflow like the streamlined pier, but
towards the bed. Larger sediment requiring a higher critical velocity it also reduces the velocity and bed shear stress in the wake like the
such as coarse sane may be transported in the case of the cylindrical pier cylindrical pier.
but not for the delta vane pier.

11
A.M. Aly and E. Dougherty Ocean Engineering 234 (2021) 109326

Fig. 9. Mesh of the streamlined tapered sheath: (a) top view of the domain, (b) zoomed top view showing the mesh around the countermeasure, and (c) a 3D view.

Fig. 10. Comparison of bed shear stress for the cylindrical pier (a) and the streamlined tapered sheath (b).

8.3. Guide wall with slanting vanes the unstructured mesh results. The global mesh was set at 0.4 m, the
surface mesh for the pier was 0.02 m, and 0.1 m for the bottom surface.
The guide wall with slanting vanes was modeled in AutoCAD (Fig. 8- Prism layers were not added to the pier since they produced a poorer
c). The length of the guide wall is 0.2D and the slanting vanes are placed quality mesh. Three prism layers of growth rate 1.2 were applied to the
every 0.2D. The slanting vanes taper to the sides of the pier so that the bottom surface with a maximum height of 0.005. The total elements
effective width of the pier is not increased. Additionally, the guide wall were approximately 460k. The guide wall with slanting vanes was
and slanting vanes are all 3 mm thick (0.5%D). The pier was meshed simulated using the RANS equations with the realizable k-ε turbulence
using tetrahedral-shaped elements with the same parameters found in closure. The boundary conditions remained the same from the previous

12
A.M. Aly and E. Dougherty Ocean Engineering 234 (2021) 109326

Fig. 11. Comparison of the velocity magnitude for the cylindrical pier (a) and the streamlined tapered sheath (b), at z = 1.9%D.

Fig. 12. Mesh of the delta vane countermeasure: (a) top view of the domain, (b) zoomed top view showing the mesh around the countermeasure, and (c) a 3D view.

structured and unstructured simulations so that the resulting flow The amount of sediment with a critical shear velocity equal to 0.5 would
structures may be compared. be transported less for the guide wall with slanting vanes as well as this
Contours of bed shear stress and velocity were obtained after the region is a little smaller in the guide wall pier contour. The guide wall
simulation was complete. Each generated figure was compared against with slanting vanes seems to reduce the magnitude of shear stress along
the cylindrical pier. In Fig. 15, the maximum bed shear stress is reduced the sides of the pier. This may be from the guide wall separating the flow
from about 1.0 Pa to 0.85 Pa or a 15% reduction. The area that has a bed similar to the streamlined tapered sheath pier. Both piers would trans­
shear stress equal to 0.85 is small, compared to the area equal to this port similar amounts of sediment, such as silt in the wake.
value or higher for the cylindrical pier. This means the guide wall with The guide wall with slanting vanes reduces the maximum velocity at
slanting vanes pier could cause little transport of sediment with a critical z = 1.9% from 0.5 m/s to 0.45 m/s or about 10% (Fig. 16). This indicates
shear stress of 0.85, whereas the cylindrical pier could cause transport of the cylindrical pier could allow transport of sediment with a critical
sediment equal to and up to 1.0 Pa (or coarse sand versus coarse sand). shear velocity 10% higher than the sediment allowed by the guide wall

13
A.M. Aly and E. Dougherty Ocean Engineering 234 (2021) 109326

Fig. 13. Comparison of bed shear stress for the cylindrical pier (a) and the delta vane (b).

Fig. 14. Velocity magnitude for the cylindrical pier (a) and the delta vane (b), at z = 1.9%D.

Fig. 15. Comparison of the bed shear stress for the cylindrical pier (a) and the guide wall with slanting vanes (b).

14
A.M. Aly and E. Dougherty Ocean Engineering 234 (2021) 109326

Fig. 16. Comparison of the velocity magnitude for the cylindrical pier (a) and the guide wall with slanting vanes (b), at z = 1.9%D.

with slanting vanes pier. The wake looks similar for both, indicating for the guide wall with slanting vanes. The wake region remained similar
sediment would likely be transported similarly between the two piers in between the guide wall and slanting vanes pier and the cylindrical pier.
this region. It is worth mentioning that one experimental study on the guide wall
The guide wall with slanting vanes reduced the maximum bed shear with slanting vanes was considered. The mentioned efficiencies may not
stress by 15% and the maximum velocity at z = 1.9% by 10% indicating be reliable, as this device loses its efficiency during high flow intensities,
that this pier reduces the maximum size particle that could be trans­ since the horseshoe vortex develops a few centimeters upstream of the
ported. In this case, coarse sand to coarse sand. Even though the pier. To show how effective is this countermeasure during high flow
reduction in magnitude was not as high as in some of the other pier intensities by numerical simulations, Large Eddy Simulations (LES)
designs, there is a reduction in the area of shear stress equal to 0.5 Pa. should be considered (Tafarojnoruz and Lauria, 2020).
There will be less sediment transported at or below this critical velocity

Fig. 17. Mesh of the angled plate footings: (a) top view of the domain, (b) zoomed top view showing the mesh around the countermeasure, and (c) a 3D view.

15
A.M. Aly and E. Dougherty Ocean Engineering 234 (2021) 109326

8.4. Angled plate footings maximum velocity above the bed for the pier with plate footings, but this
pier extends the region of lower velocity both upstream by about 0.5D
The angled plate countermeasure was modeled in AutoCAD. There and downstream compared to the cylindrical pier. The angled plates
are four plates with two angled at 30◦ , facing the inlet, and the other two may act to capture sediment as an active scour control measure, so this
are angled at 30◦ , facing the outlet of the domain. Each plate extends might be a useful test to perform with sediment in a laboratory study.
0.5D from the outside surface of the pier. The height is also 0.5D. The
countermeasure was then meshed using tetrahedral-shaped elements 9. Delta vane countermeasure at higher Reynolds number
with the same parameters as those of the unstructured mesh. Three
prism layers with a growth rate of 1.2 are applied at the wall. The total In Section 7, the delta vane pier attachment demonstrated a signifi­
elements are approximately 450k. The generated mesh is shown in cant decrease in bed shear stress, indicating a strong potential to reduce
Fig. 17. The angled plate footing pier was simulated using the RANS scour. All investigated piers were analyzed under a Reynolds number of
equations with the realizable k-ε turbulence closure. The boundary 1.7 × 105. This was based on the Reynolds number used to validate the
conditions remained the same from the previous structured and un­ CFD model with experimental results. Even though the delta vane was
structured simulations so that the resulting flow structures may be effective at this Reynolds number, it is crucial to ascertain that the shear
compared. stress will be reduced at a higher Reynolds number (full-scale).
Contours of bed shear stress and velocity were obtained after the To determine a realistic, yet high, Reynolds number to reanalyze the
simulation was complete. Each generated figure was compared against delta vane pier attachment, the estimated depth-averaged flow velocity
the cylindrical pier. In Fig. 18, the magnitude of shear stress is reduced was determined using measured data from the United States Geological
by about 30% for the pier with plate footings since there is a small region Survey (USGS) (USGS, 2019). Flood data from the Mississippi River in
of shear stress equal to approximately 0.7 Pa along the sides of the pier Baton Rouge are used to calculate a high Reynolds number that can be
with plate footings versus 1.0 Pa along the upstream sides of the cylin­ experienced in this location. Historical maximum flow discharges for the
drical pier. Since this region is so small, one would expect sediment of Mississippi River in Baton Rouge were considered to be about 41,626
critical shear stress of 0.5 Pa or lowered to be transported. This region of m3/s (USGS, 2019). The depth-averaged velocity for the highest
0.5 Pa is a little larger than the corresponding region for the cylindrical discharge was calculated by dividing the discharge by the estimated
pier, indicating more sediment with this critical shear velocity (sand) cross-sectional area located at Mississippi River mile marker 232. The
will be transported for the pier with plate footings. Both piers perform area is approximately 13,470 m2, and the average velocity is about 3
well in the wake with 0–0.1 Pa, which is conducive for silt transport. m/s (Little and Biedenharn, 2014).
The velocity magnitude above the bed corresponds well with the If the diameter of a bridge pier or support structure is 1.6 m, then the
shear stress contour in Fig. 18. The maximum velocity at the front sides corresponding Reynolds number for a velocity of 3 m/s is 5.1 × 106. The
of the cylindrical pier is reduced from 0.5 m/s to 0.35 m/s (30% modeled pier is 0.536 m in diameter, so a velocity of 9.5 m/s was applied
reduction). There is a larger region of slower velocity at the front of the at the inlet to obtain the same Reynolds number. Because of this high
pier with plate footings, which showed the velocity magnitudes of the Reynolds number, the mesh was refined so that the delta vane config­
piers at the plane of symmetry. The lower velocity in the wake extends uration has 2.6 million elements, and the cylindrical pier has 2.1 million
farther downstream than the cylindrical pier, indicating this pier will elements. Fig. 20 shows the bed shear stress for the cylindrical pier
minimize sediment transport downstream more than the cylindrical versus the delta vane for a Reynolds number of 5.1 × 106. The calculated
pier. The angle of the plate footings may act to capture sediment maximum bed shear stress for the cylinder is approximately 450 Pa and
colliding with the base of this pier. This region at the upstream base of approximately 350 Pa for the delta vane, which indicates a 22%
the pier also has a low velocity so as seen in Fig. 19 extending about 0.5D reduction. This is less than the 30% reduction seen at the lower Reynolds
farther upstream than the cylindrical pier, so sediment may build up in number. It may be that at higher Reynolds numbers the delta vane re­
this region. This may act counter or prevent future scour, but the sedi­ duces the scour potential less, but it still causes a significant reduction in
ment transport is still likely along the sides of this pier. maximum bed shear stress compared to the cylindrical pier. Fig. 21
The angled plate footings pier reduced the maximum bed shear stress shows the velocity magnitude for the plane at 1 cm above the bed. The
by 30%, but most of the shear stress for this pier was about 0.5 Pa simulated maximum velocity for the cylinder was approximately 15 m/
indicating that sediment with a critical shear stress of this value or s. The maximum velocity for the delta vane is approximately 14 m/s.
lower, such as sand, will be transported. There is a 30% reduction in This indicates a 7% reduction, which is less than the 10% reduction seen

Fig. 18. Comparison of bed shear stress for the cylindrical pier (a) and the angled plate footings (b).

16
A.M. Aly and E. Dougherty Ocean Engineering 234 (2021) 109326

Fig. 19. Velocity magnitude for the cylindrical pier (a) and the angled plate footings (b), at z = 1.9%D.

Fig. 20. Bed shear stress for the cylinder (a) and the delta vane (b) at Re = 5.1 × 106.

Fig. 21. Velocity magnitude above the bed for Re = 5.1 × 106 and z = 1.9%D: (a) cylinder, and (b) delta vane.

at the lower Reynolds number. Like the bed shear stress, the delta vane above the bed by a significant amount which correlates to a reduction in
seems to be less effective at reducing scour potential at a high Reynolds scour potential. This higher Reynolds number was estimated using the
number, but there is still a noticeable reduction. maximum parameters for a bridge pier in the Mississippi River near
The delta vane configuration reduced maximum bed shear stress by Baton Rouge, Louisiana. It is likely that in most cases, for a given bridge
22% compared to the cylindrical pier and the velocity above the bed by pier, the Reynolds number would be much smaller than this value. The
7% for a Reynolds number equal to 5.1 × 106. Although both values are reduction of scour potential for the delta vane could be likely larger.
less than the reductions at the lower Reynolds number, the delta vane It is worth to mention that the study of different sediment is beyond
configuration still decreases the maximum bed shear stress and velocity the scope of the current paper. The CFD investigation is conceptual, to

17
A.M. Aly and E. Dougherty Ocean Engineering 234 (2021) 109326

compare different scour countermeasures under the same flow condi­ CRediT authorship contribution statement
tions. However, by reducing the bed shear stress, the proposed coun­
termeasures demonstrate their potential to mitigate scour. These Aly Mousaad Aly: Conceptualization, Methodology, Software,
solutions have a strong potential to reduce the accelerated deterioration Validation, Investigation, Writing. Erin Dougherty: Conceptualization,
and can protect bridges and other coastal and offshore infrastructure Methodology, Software, Validation, Investigation, Writing.
against collapse due to scour.
Declaration of competing interest
10. Conclusions
The authors declare that they have no known competing financial
Scour around bridge piers and monopile foundations, such as interests or personal relationships that could have appeared to influence
offshore wind turbines, is a significant issue that contributes to the the work reported in this paper.
failure and instability of structures. We investigated different turbulence
closures, including the standard k-ε, k-ε realizable, standard k-ω, and k-ω Acknowledgments
SST, to understand their capability to predict the bed shear stress . We
then assessed the performance of different flow countermeasures in a The first author received funds from the Louisiana Board of Regents,
comparative study. The k-ω equations can simulate adverse pressure RCS program (contract no. LEQSF (2021-22)-RD-A-30). The second
gradients; however, the k-ε equations consistently performed better in author would like to thank the United States Coast Guard for funding her
approximating the experimental data. In particular, the k-ε realizable education at Louisiana State University. The findings and conclusions
model predicted the components of velocity well. The use of unstruc­ are those of the authors and do not necessarily represent the official
tured mesh allowed the consideration of more complex geometry. Four position of the sponsors.
scour countermeasures are investigated at a Reynolds number of 1.7 ×
105: streamlined tapered sheath, delta vane, guide wall with slanting
References
plates, and angled plate footings. The findings are as follows.
Aghaee, Y., Hakimzadeh, H., 2010. Three dimensional numerical modeling of flow
➢ The optimized unstructured mesh may have y+ values varying be­ around bridge piers using LES and RANS. River Flow 2010: Proceedings of the
International Conference on Fluvial Hydraulics. Bundesanstalt für Wasserbau,
tween 1 and 17. The k− ω models approximate the experimental data
Braunschweig, Germany, pp. 211–218. https://hdl.handle.net/20.500.11
more, with the unstructured mesh, compared to the structured mesh, 970/99648.
for greater y+ values. This improved performance may be due to the Al-Shukur, A.H.K., Obeid, Z.H., 2016. Experimental study of bridge pier shape to
tetrahedral shapes, compared to the rectangular prisms, or that minimize local scour. Int. J. Civ. Eng. Technol. 7, 162–171. https://iaeme.com/Mas
terAdmin/Journal_uploads/IJCIET/VOLUME_9_ISSUE_7/IJCIET_09_07_120.pdf.
achieving a y + value close to 1 is not that critical for the flow physics Aly, A.M., 2014. Atmospheric boundary-layer simulation for the built environment: Past,
considered in the current study. present and future. Build. Environ. 75, 206–221. https://doi.org/10.1016/j.
➢ The delta vane and the angled plate footings reduced the maximum buildenv.2014.02.004.
Aly, A.M., 2016. On the evaluation of wind loads on solar panels: The scale issue. Sol.
bed shear stress by 30%. The two countermeasures reduced the Energy 135, 423–434. https://doi.org/10.1016/j.solener.2016.06.018.
maximum velocity around the pier by about 10% and 30%, respec­ Aly, A.M., Bitsuamlak, G., 2013. Aerodynamics of ground-mounted solar panels: Test
tively. In the wake of the pier, both countermeasures yield a bed model scale effects. J. Wind Eng. Ind. Aerod. 123, 250–260. https://doi.org/
10.1016/j.jweia.2013.07.007.
shear stress close to zero. The delta vane represents an upstream Aly, A.M., Bresowar, J., 2016. Aerodynamic mitigation of wind-induced uplift forces on
physical barrier, which may armor the bed in the case of a strong low-rise buildings: A comparative study. J. Build Eng. 5, 67–76. https://doi.org/
downflow. 10.1016/j.jobe.2016.01.007.
Aly, A.M., Chokwitthaya, C., Poche, R., 2017. Retrofitting building roofs with
➢ The streamlined tapered sheath reduced the maximum bed shear
aerodynamic features and solar panels to reduce hurricane damage and enhance eco-
stress and the maximum velocity by 20% and 10%, respectively. The friendly energy production. Sustain. Cities Soc. 35, 581–593. https://doi.org/
sheath armors the bed and deflects the downflows. The reductions in 10.1016/j.scs.2017.09.002.
ANSYS Inc, 2021. Introduction to Ansys ICEM CFD. ANSYS Inc. https://engineering.
the shear stress and the flow velocity at the front of the pier suggest
purdue.edu/~scalo/menu/teaching/me608/tutorial.pdf. (Accessed 13 June 2021).
that the streamlined tapered sheath may reduce scour at this loca­ ANSYS Inc, 2021. Ansys CFD (Fluent and CFX) Turbulence Modeling(Self-paced Learning
tion. However, the streamlined tapered sheath does not yield a Available). ANSYS Inc. https://www.ansys.com/training-center/course-catalog/fl
decrease in the flow velocity, directly downstream. Sediment such as uids/ansys-cfd-fluent-and-cfx-turbulence-modeling, 2021. (Accessed 13 June 2021).
Arneson, L.A., Zevenbergen, L.W., Lagasse, P.F., Clopper, P.E., 2012. Evaluating Scour at
sand, with a critical shear stress of 0.3 Pa, can be transported aft of Bridges, 5th. National Highway Institute, Arlington, Virginia, USA, FHWA-HIF-12-
the streamlined tapered sheath. 003. http://ponce.sdsu.edu/hec-18-scour.pdf.
➢ The guide wall with slanting vanes was the least effective counter­ Azizi, S.H., Farsadizadeh, D., Arvanaghi, H., Abbaspour, A., 2016. Numerical simulation
of flow pattern around the bridge pier with submerged vanes. J. Hydraul. Struct. 2,
measure. This countermeasure reduced the maximum bed shear 46–61. https://doi.org/10.22055/jhs.2016.12856.
stress and the maximum velocity above the bed by 15% and 10%, Baghbadorani, D.A., Ataie-Ashtiani, B., Beheshti, A., Hadjzaman, M., Jamali, M., 2018.
respectively. Prediction of current-induced local scour around complex piers: review, revisit, and
integration. Coast. Eng. 133, 43–58. https://doi.org/10.1016/j.
➢ The performance of the delta vane countermeasure was further coastaleng.2017.12.006.
investigated at a Reynolds number of 5.1 × 106. At such a high Banks, G., 2010. Safer levees and bridges thanks to new erosion and scour detector
Reynolds number, the delta vane reduced the maximum bed shear [WWW Document], 9.20.11. https://newatlas.com/safer-levees-bridges-thanks-ero
sion-scour-detector/17020/.
stress and maximum velocity by 22% and 7%, respectively. Overall,
Blevins, R.D., 1984. Applied Fluid Dynamics Handbook. Van Nostrand Reinhold
the findings reveal that the delta vane countermeasure can reduce Company, New York.
the scour potential at low and high Reynolds numbers. Bombardelli, F.A., Palermo, M., Pagliara, S., 2018. Temporal evolution of jet induced
scour depth in cohesionless granular beds and the phenomenological theory of
➢ To conclude, hydrodynamic models are crucial for the assessment
turbulence. Phys. Fluids 30, 85109. https://doi.org/10.1063/1.5041800.
and performance comparison of scour countermeasures. The delta Breusers, N.T., Raudkivi, A.J., 1991. Scouring. Hydraulic Structures Design Manual, 2.
vane and the angled plate footings performed well in reducing the Taylor & Francis, New York.
bed shear stress, followed by the streamlined tapered sheath. The Chang, W.Y., Lai, J.S., Yen, C.L., 2004. Evolution of scour depth at circular bridge piers.
J. Hydraul. Eng. 130, 905–913. https://doi.org/10.1061/(ASCE)0733-9429(2004)
angled plate footing may capture sediment, and hence actively act as 130:9(905).
a scour countermeasure. Chaudhuri, S., Singh, S.K., Debnath, K., Manik, M.K., 2018. Pier scour within long
contraction in cohesive sediment bed. Environ. Fluid Mech. 18, 417–441. https://
doi.org/10.1007/s10652-017-9560-x.
Chiew, Y.-M., 1992. Scour protection at bridge piers. J. Hydraul. Eng. 118, 1260–1269.
https://doi.org/10.1061/(ASCE)0733-9429(1992)118:9(1260).

18
A.M. Aly and E. Dougherty Ocean Engineering 234 (2021) 109326

Chortis, G., Askarinejad, A., Prendergast, L.J., Li, Q., Gavin, K., 2020. Influence of scour Lodhi, A.S., Jain, R.K., Karna, N., 2017. Temporal variation of scour at wake of the spur
depth and type on p–y curves for monopiles in sand under monotonic lateral loading dikes in cohesive sediment mixtures. In: 22nd International Conference on
in a geotechnical centrifuge. Ocean. Eng., 106838 https://doi.org/10.1016/j. Hydraulics, Water Resources and Coastal Engineering. LD College of Engineering,
oceaneng.2019.106838. Ahmedabad, India, pp. 625–632. https://www.researchgate.net/publication/3226
Daido, A., Yano, S., 1995. Local scour around bridge piers andits protection with guide 83427_TEMPORAL_VARIATION_OF_SCOUR_AT_WAKE_OF_THE_SPUR_DIKES_IN_C
wall and slanting plate and piers surface. In: Proc. 6th Intl. Symp. River OHESIVE_SEDIMENT_MIXTURES.
Sedimentation, pp. 1181–1187. Lodhi, A.S., Jain, R.K., Sharma, P.K., Karna, N., 2018. Influence of cohesion on scour at
Debnath, K., Chaudhuri, S., 2012. Local scour around non-circular piers in clay–sand wake of partially submerged spur dikes in cohesive sediment mixtures. ISH J.
mixed cohesive sediment beds. Eng. Geol. 151, 1–14. https://doi.org/10.1016/j. Hydraul. Eng. 1–12. https://doi.org/10.1080/09715010.2018.1525325.
enggeo.2012.09.013. Melville, B.W., 1997. Pier and Abutment Scour: Integrated Approach. J. Hydraul. Eng.
Debnath, K., Chaudhuri, S., 2011. Effect of suspended sediment concentration on local 123, 125–136. https://doi.org/10.1061/(ASCE)0733-9429(1997)123:2(125).
scour around cylinder for clay-sand mixed sediment beds. Eng. Geol. 117, 236–245. Melville, B.W., Coleman, S.E., 2000. Bridge Scour. Water Resources Publications, LLC,
https://doi.org/10.1016/j.enggeo.2010.11.003. Highlands Ranch, CO.
Debnath, K., Chaudhuri, S., 2010. Bridge pier scour in clay-sand mixed sediments at Menter, F.R., 1993. Zonal Two Equation k-w Turbulence Models For Aerodynamic Flows.
near-threshold velocity for sand. J. Hydraul. Eng. 136, 597–609. https://doi.org/ In: 23rd Fluid Dynamics, Plasmadynamics, and Lasers Conference. American
10.1061/(ASCE)HY.1943-7900.0000221. Institute of Aeronautics and Astronautics:, Orlando, FL, U.S.A., pp. 1–21. https://doi.
Dey, S., Raikar, R.V., 2007. Characteristics of horseshoe vortex in developing scour holes org/10.2514/6.1993-2906
at piers. J. Hydraul. Eng. 133, 399–413. https://doi.org/10.1061/(ASCE)0733-9429 Menter, F.R., 1992. Improved Two-Equation k-w Turbulence Models for Aerodynamic
(2007)133:4(399). Flows. In: NASA Technical Memorandum 103975. NASA, Moffett Field, CA. https
Dey, S., Sumer, B.M., Fredsøe, J., 2006. Control of scour at vertical circular piles under ://ntrs.nasa.gov/api/citations/19930013620/downloads/19930013620.pdf?attach
waves and current. J. Hydraul. Eng. 132, 270–279. https://doi.org/10.1061/(ASCE) ment=true.
0733-9429(2006)132:3(270). Muzzammil, M., Alama, J., Danish, M., 2015. Scour prediction at bridge piers in cohesive
EL-Ghorab, E.A.S., 2013. Reduction of scour around bridge piers using a modified bed using gene expression programming. Aquat. Procedia 4, 789–796. https://doi.
method for vortex reduction. Alexandria Eng. J. 52, 467–478. https://doi.org/ org/10.1016/j.aqpro.2015.02.098.
10.1016/j.aej.2013.04.001. Najafzadeh, M., Barani, G.A., 2014. Experimental Study of Local Scour around a Vertical
Elnikhely, E.A., 2017. Minimizing scour around bridge pile using holes. Ain Shams Eng. Pier in Cohesive Soils. Sci. Iran. 21, 241–250. http://scientiairanica.sharif.edu/art
J. 8, 499–506. https://doi.org/10.1016/j.asej.2016.06.016. icle_1628.html.
Ettema, R., Kirkil, G., Muste, M., 2006. Similitude of large-scale turbulence in Najafzadeh, M., Barani, G.A., Hessami-Kermani, M.R., 2015. Evaluation of GMDH
experiments on local scour at cylinders. J. Hydraul. Eng. 132, 33–40. https://doi. networks for prediction of local scour depth at bridge abutments in coarse sediments
org/10.1061/(ASCE)0733-9429(2006)132:1(33). with thinly armored beds. Ocean. Eng. 104, 387–396. https://doi.org/10.1016/j.
Gaudio, R., Tafarojnoruz, A., Calomino, F., 2012. Combined flow-altering oceaneng.2015.05.016.
countermeasures against bridge pier scour. J. Hydraul. Res. 50, 35–43. https://doi. Ni, X., Xue, L., An, C., 2020. Experimental investigation of scour around circular
org/10.1080/00221686.2011.649548. arrangement pile groups. Ocean. Eng., 108096 https://doi.org/10.1016/j.
Graf, W.H., Istiarto, I., 2002. Flow pattern in the scour hole around a cylinder. oceaneng.2020.108096.
J. Hydraul. Res. 40, 13–20. https://doi.org/10.1080/00221680209499869. Niedoroda, A.W., Dalton, C., 1982. A review of the fluid mechanics of ocean scour.
Guo, J., 2002. Hunter Rouse and Shields Diagram. In: Advances in Hydraulics and Water Ocean. Eng. 9, 159–170. https://doi.org/10.1016/0029-8018(82)90011-7.
Engineering: Proceedings of the 13th IAHR–APD Congress, Singapore, 6 – 8 August Pandey, M., Azamathulla, H.M., Chaudhuri, S., Pu, J.H., Pourshahbaz, H., 2020a.
2002. World Scientific, pp. 1096–1098. https://doi.org/10.1142/978981277 Reduction of time-dependent scour around piers using collars. Ocean. Eng. 213,
6969_0200. 107692 https://doi.org/10.1016/j.oceaneng.2020.107692.
Gupta, A.K., 2002. Hydrodynamic modification of the horseshoe vortex at a vertical pier Pandey, M., Valyrakis, M., Qi, M., Sharma, A., Lodhi, A.S., 2020b. Experimental
junction with ground. Phys. Fluids 30, 1213. https://doi.org/10.1063/1.866270. assessment and prediction of temporal scour depth around a spur dike. Int. J.
Hamidifar, H., Omid, M.H., 2017. Local scour of cohesive beds downstream of a rigid Sediment Res. 36, 17–28. https://doi.org/10.1016/j.ijsrc.2020.03.015.
apron. Can. J. Civ. Eng. 44, 935–944. https://doi.org/10.1139/cjce-2016-0398. Parker, G., Toro-Escobar, C., Voigt, R., 1998. Countermeasures to Protect Bridge Piers
Johnson, E.B., Testik, F.Y., Ravichandran, N., Schooler, J., 2013. Levee scour from from Scour. In: NCHRP 24-7, 1. St. Anthony Falls Laboratory, Minneapolis. https://c
overtopping storm waves and scour counter measures. Ocean. Eng. 57, 72–82. onservancy.umn.edu/handle/11299/108221.
https://doi.org/10.1016/j.oceaneng.2012.09.006. Prendergast, L.J., Gavin, K., Doherty, P., 2015. An investigation into the effect of scour
Jones, J.S., 1984. Comparison of prediction equations for bridge pier and abutment on the natural frequency of an offshore wind turbine. Ocean. Eng. 101, 1–11.
scour. In: Bridge Engineering Conference, 2nd, Washington, DC, USA. https://doi.org/10.1016/j.oceaneng.2015.04.017.
Transportation Research Board, pp. 202–209. http://onlinepubs.trb.org/Onlinepu Richardson, E.V., Davis, S.R., 2001. Evaluating Scour at Bridges, 4th. Federal Highway
bs/trr/1984/950/950v2-022.pdf. Administration. Hydraulic Engineering Circular No. 18, FHWA NHI 01-001.
Jones, W.P., Launder, B.E., 1973. The calculation of low-Reynolds-number phenomena Roulund, A., Sumer, B.M., Fredsoe, J., Mitchelsen, J., 2005. Numerical and experimental
with a two-equation model of turbulence. Int. J. Heat Mass Tran. 16, 119–1130. investigation of flow and scour around a circular pile. J. Fluid Mech. 534, 351–401.
https://doi.org/10.1016/0017-9310(73)90125-7. https://doi.org/10.1017/S0022112005004507.
Julien, P.Y., 2018. River Mechanics, second ed. Cambridge University Press, Cambridge. Salim, M., Jones, J.S., 1996. Scour around Exposed Pile Foundations. In: North American
https://doi.org/10.1017/9781316107072. Water and Environment Congress & Destructive Water. American Society of Civil
Julien, P.Y., 2010. Erosion and Sedimentation, second ed. Cambridge University Press, Engineers, New York, NY, pp. 2202–2211. https://cedb.asce.org/CEDBsearch/reco
Cambridge. https://doi.org/10.1017/CBO9780511806049. rd.jsp?dockey=0099802.
Khaled, M.D., Aly, A.M., Elshaeir, A., 2021. Computational efficiency of CFD modeling Sheppard, D.M., Odeh, M., Glasser, T., 2004. Large scale clear-water local pier scour
for building engineering: an empty domain study. J. Build Eng., 102792 https://doi. experiments. J. Hydraul. Eng. 130, 957–963. https://doi.org/10.1061/(ASCE)0733-
org/10.1016/j.jobe.2021.102792. 9429(2004)130:10(957).
Khalid, M., Muzzammil, M., Alam, J., 2018. Reliability analysis of local scour at bridge Spalart, P.R., 2000. Strategies for turbulence modelling and simulations. Int. J. Heat
pier in clay-sand mixed sediments. Aquademia 2, 1. https://doi.org/10.20897/awet/ Fluid Flow 21, 252–263. https://doi.org/10.1016/S0142-727X(00)00007-2.
86715. Sumer, B.M., Fredsøe, J., 2001a. Wave scour around a large vertical circular cylinder.
Khosronejad, A., Kang, S., Sotiropoulos, F., 2012. Experimental and computational J. Waterw. Port, Coast. Ocean Eng. 127, 125–134. https://doi.org/10.1061/(ASCE)
investigation of local scour around bridge piers. Adv. Water Resour. 37, 73–85. 0733-950X(2001)127:3(125).
https://doi.org/10.1016/j.advwatres.2011.09.013. Sumer, B.M., Fredsøe, J., 2001b. Scour around pile in combined waves and current.
Kundu, P.K., Cohen, I.M., Dowling, D.R., 2012. Fluid Mechanics, fifth ed. Fluid J. Hydraul. Eng. 127, 403–411. https://doi.org/10.1061/(ASCE)0733-9429(2001)
Mechanics. Academic Press. https://doi.org/10.1016/B978-0-12-382100-3.10008- 127:5(403).
3. Tafarojnoruz, A., Gaudio, R., Dey, S., 2010. Flow-altering countermeasures against scour
Launder, B.E., Spalding, D.B., 1974. The numerical computation of turbulent flows. at bridge piers: a review. J. Hydraul. Res. 48, 441–452. https://doi.org/10.1080/
Comput. Methods Appl. Mech. Eng. 3, 269–289. https://doi.org/10.1016/0045- 00221686.2010.491645.
7825(74)90029-2. Tafarojnoruz, A., Lauria, A., 2020. Large eddy simulation of the turbulent flow field
Lee, S.O., Sturm, T.W., 2009. Effect of sediment size scaling on physical modeling of around a submerged pile within a scour hole under current condition. Coast Eng. J.
bridge pier scour. J. Hydraul. Eng. 135, 793–802. https://doi.org/10.1061/(ASCE) 62, 489–503. https://doi.org/10.1080/21664250.2020.1807453.
HY.1943-7900.0000091. Trojnar, K., 2020. Simplified design of new hybrid monopile foundations for offshore
Lin, X., Zhang, Jisheng, Wang, R., Zhang, Jing, Liu, W., Zhang, Y., 2019. Scour around a wind turbines. Ocean. Eng. 219, 108046 https://doi.org/10.1016/j.
mono-pile foundation of a horizontal axis tidal stream turbine under steady current. oceaneng.2020.108046.
Ocean. Eng. 192, 106571 https://doi.org/10.1016/j.oceaneng.2019.106571. Tulimilli, B.R., Majumdar, P., Kostic, M., Lottes, S.A., 2010. Development of CFD
Little, C.D., Biedenharn, D.S., 2014. Mississippi river hydrodynamic and delta simulation for 3-D flooding flow and scouring around a bridge structure,
management study (MRHDM)–Geomorphic assessment [WWW document]. https pp. 129–135, 3rd WSEAS International Conference on URBAN PLANNING AND
://www.mvn.usace.army.mil/Missions/Environmental/Louisiana-Coastal-Area TRANSPORTATION. http://mdx2.plm.automation.siemens.com/sites/default/files/
/Mississippi-River-Hydrodynamic-and-Delta-Managemen/. conference_proceeding/pdf/WSEAS-UPT2010-19_Bhaskar_Scouring.pdf.
Liu, M., Jin, X., Wang, L., Yang, F., Tang, J., 2019. Numerical investigation of local scour USGS, 2019. USGS 07374000 Mississippi river at Baton Rouge, LA [WWW document].
around a vibrating pipeline under steady currents. Ocean. Eng. 221, s https://doi. https://waterdata.usgs.gov/usa/nwis/uv?site_no=07374000.
org/10.1016/j.oceaneng.2020.108546.

19
A.M. Aly and E. Dougherty Ocean Engineering 234 (2021) 109326

Wang, C., Yu, X., Liang, F., 2017. A review of bridge scour: mechanism, estimation, Yao, W., An, H., Draper, S., Cheng, L., Harris, J., 2018. Experimental investigation of
monitoring and countermeasures. Nat. Hazards 87, 1881–1906. https://doi.org/ local scour around submerged piles in steady current. Coast. Eng. 142, 27–41.
10.1007/s11069-017-2842-2. https://doi.org/10.1016/j.coastaleng.2018.08.015.
Wang, L., Melville, B.W., Whittaker, C.N., Guan, D., 2020. Temporal evolution of clear- Yao, W., Draper, S., An, H., Cheng, L., Harris, J.M., Whitehouse, R.J.S., 2020. Effect of a
water scour depth at submerged weirs. J. Hydraul. Eng. 146, 6020001 https://doi. skirted mudmat foundation on local scour around a submerged structure. Ocean.
org/10.1061/(ASCE)HY.1943-7900.0001712. Eng. 218, 108127 https://doi.org/10.1016/j.oceaneng.2020.108127.
Wardhana, K., Hadipriono, F.C., 2003. Analysis of recent bridge failures in the United Yazdanfar, Z., Lester, D., Robert, D., Setunge, S., 2021. A novel CFD-DEM upscaling
States. J. Perform. Constr. Facil. 17, 144–150. https://doi.org/10.1061/(ASCE) method for prediction of scour under live-bed conditions. Ocean. Eng. 220, 108442
0887-3828(2003)17:3(144). https://doi.org/10.1016/j.oceaneng.2020.108442.
White, F.M., 2002. Fluid Mechanics, fifth ed. McGraw HillBoston, Boston. Yu, P., Zhu, L., 2020. Numerical simulation of local scour around bridge piers using novel
Wilcox, D.C., 1993. Comparison of two-equation turbulence models for boundary layers inlet turbulent boundary conditions. Ocean. Eng. 218, 108166 https://doi.org/
with pressure gradient. AIAA J. 31, 1414–1421. https://doi.org/10.2514/3.11790. 10.1016/j.oceaneng.2020.108166.
Yang, B., Wei, K., Yang, W., Li, T., Qin, B., 2020. A feasibility study of reducing scour Yuan, C., Melville, B.W., Adams, K.N., 2017. Scour at wind turbine tripod foundation
around monopile foundation using a tidal current turbine. Ocean. Eng., 108396 under steady flow. Ocean. Eng. 141, 277–282. https://doi.org/10.1016/j.
https://doi.org/10.1016/j.oceaneng.2020.108396. oceaneng.2017.06.038.
Yang, S., Shi, B., Guo, Y., Yang, L., 2019. Investigation on scour protection of submarine Zhang, Q., Zhou, X.L., Wang, J.H., 2017. Numerical investigation of local scour around
piggyback pipeline. Ocean. Eng. 182, 442–450. https://doi.org/10.1016/j. three adjacent piles with different arrangements under current. Ocean. Eng. 142,
oceaneng.2019.04.090. 625–638. https://doi.org/10.1016/j.oceaneng.2017.07.045.
Yang, Yifan, Melville, B.W., Macky, G.H., Shamseldin, A.Y., 2020. Temporal evolution of Zhao, E., Dong, Y., Tang, Y., Sun, J., 2021. Numerical investigation of hydrodynamic
clear-water local scour at aligned and skewed complex bridge piers. J. Hydraul. Eng. characteristics and local scour mechanism around submarine pipelines under joint
146, 4020026 https://doi.org/10.1061/(ASCE)HY.1943-7900.0001732. effect of solitary waves and currents. Ocean. Eng. 222, 108553 https://doi.org/
Yang, Yilin, Qi, M., Wang, X., Li, J., 2020. Experimental study of scour around pile 10.1016/j.oceaneng.2020.108553.
groups in steady flows. Ocean. Eng. 195, 106651 https://doi.org/10.1016/j. Zhao, M., Cheng, L., Zang, Z., 2010. Experimental and numerical investigation of local
oceaneng.2019.106651. scour around a submerged vertical circular cylinder in steady currents. Coast. Eng.
Yanmaz, A.M., Altinbilek, H.D., 1991. Study of time-depenbent local scour around bridge 57, 709–721. https://doi.org/10.1016/j.coastaleng.2010.03.002.
piers. J. Hydraul. Eng. 117, 1247–1268. https://doi.org/10.1061/(ASCE)0733-9429 Zhou, L., 2017. Numerical Modelling of Scour in Steady Flows. Université de Lyon, Lyon,
(1991)117:10(1247). France. https://tel.archives-ouvertes.fr/tel-01598600.

20

You might also like