Ecological Engineering 99 (2017) 324–332

Contents lists available at ScienceDirect

Ecological Engineering
journal homepage: www.elsevier.com/locate/ecoleng

A computational fluid dynamics modeling study of guide walls for

downstream fish passage
Kevin Brian Mulligan a,∗ , Brett Towler b , Alex Haro c , David P. Ahlfeld a
a
University of Massachusetts Amherst, 130 Natural Resources Rd., 18 Marston Hall, Amherst, MA 01003, USA
b
U.S. Fish & Wildlife Service, Northeast Region, 300 Westgate Center Dr., Hadley, MA 01035, USA
c
U.S. Geological Survey, Leetown Science Center, S.O. Conte Anadromous Fish Research Laboratory, 1 Migratory Way, Turners Falls, MA 01376, USA

a r t i c l e i n f o a b s t r a c t

Article history: A partial-depth, impermeable guidance structure (or guide wall) for downstream fish passage is typically
Received 18 September 2015 constructed as a series of panels attached to a floating boom and anchored across a water body (e.g. river
Received in revised form 6 October 2016 channel, reservoir, or power canal). The downstream terminus of the wall is generally located nearby to
Accepted 13 November 2016
a fish bypass structure. If guidance is successful, the fish will avoid entrainment in a dangerous intake
Available online 23 November 2016
structure (i.e. turbine intakes) while passing from the headpond to the tailwater of a hydroelectric facility
through a safer passage route (i.e. the bypass). The goal of this study is to determine the combination
Keywords:
of guide wall design parameters that will most likely increase the chance of surface-oriented fish being
Guide wall
Fish passage
successfully guided to the bypass. To evaluate the flow field immediately upstream of a guide wall, a
Downstream parameterized computational fluid dynamics model of an idealized power canal was constructed in ©
Computational fluid dynamics ANSYS Fluent v 14.5 (ANSYS Inc., 2012). The design parameters investigated were the angle and depth
of the guide wall and the average approach velocity in the power canal. Results call attention to the
importance of the downward to sweeping flow ratio and demonstrate how a change in guide wall depth
and angle can affect this important hydraulic cue to out-migrating fish. The key findings indicate that a
guide wall set at a small angle (15◦ is the minimum in this study) and deep enough such that sweeping flow
dominant conditions prevail within the expected vertical distribution of fish approaching the structure
will produce hydraulic conditions that are more likely to result in effective passage.
Published by Elsevier B.V.

1. Introduction Without the ability to freely move between and within each aquatic
ecosystem, the chance of a fish population’s long-term survival
Many fish species have evolved to use different types of envi- is greatly diminished (Limburg and Waldman, 2009; McDowall,
ronments over their life span in order to enhance the population’s 1987).
chance of survival. Each selected environment is well suited for a As a result of anthropogenic development on river systems, full
particular part of the life cycle for the fish (McDowall, 1997). For and partial barriers to fish movement commonly exist in water-
instance, anadromous clupeids (genus Alosa) are born in a fresh- sheds worldwide (Williams et al., 2012). These barriers typically
water river system where there are fewer predators, migrate as consist of small to large size dams, culverts, and other structures.
juveniles to the ocean where there is a more abundant food sup- Despite substantial efforts, issues related to passage of fish both
ply, then migrate as adults back to the fresh water river to spawn, up and downstream of dams are not yet fully resolved (Bunt et al.,
completing the life cycle (Weiss-Glanz et al., 1986). In addition, 2012; Enders et al., 2009). Even if a fishway structure is in place,
potamodromous fish perform migrations for the purposes of both poor design, predation, and degraded water quality can lead to
feeding and spawning, but only within freshwater river systems. fatigue, injury, fatality, or other hindrances to fish survival.
At a typical hydropower facility there are three primary routes
of downstream passage. The three routes, ordered by typical pro-
∗ Corresponding author. Present address: U.S. Geological Survey, Leetown Science
portion of average annual river flow, are 1) through the turbine
Center, S.O. Conte Anadromous Fish Research Laboratory, 1 Migratory Way, Turners
intakes, 2) over a spillway and 3) through a fish bypass (often con-
Falls, MA 01376, USA. structed as a sluice gate, weir, or pipe). The downstream bypass is
E-mail addresses: [email protected], [email protected] (K.B. Mulligan), typically constructed in close proximity to the turbine intakes to
brett [email protected] (B. Towler), [email protected] (A. Haro), reduce the number of fish passing through the turbines. The chal-
[email protected] (D.P. Ahlfeld).

http://dx.doi.org/10.1016/j.ecoleng.2016.11.025
0925-8574/Published by Elsevier B.V.
 K.B. Mulligan et al. / Ecological Engineering 99 (2017) 324–332 325

ing fish passed below a guide wall, possibly due to a strong vertical
Notation velocity component.
A guide wall is typically constructed of a series of floating
d Guide wall depth (ft.) partial-depth, impermeable panels. Depending upon the hydro-
d*(t*) Upper Guidance Zone depth (ft.) electric project configuration, the guide wall is anchored across a
DSR Downward to sweeping velocity ratio (−) river channel, reservoir, or power canal (Scott, 2012). Scott (2012)
DSRmin Minimum downward to sweeping velocity ratio at explains that the concept is based on knowledge that: 1) juvenile
each cross-section (−) anadromous fish tend to swim in the top portion of the water col-
H Water depth (ft.) umn (Whitney et al., 1997; Buckley and Kynard, 1985; Faber et al.,
 Approach distance (ft.) 2011), 2) some juvenile species have been shown to select a shal-
L Distance along the x-axis from the upstream to low rather than deep passage route when given the choice (Johnson
downstream ends of the guide wall (ft.) et al., 1997), and 3) anadromous juveniles tend to migrate down-
MMR Maximum to mean velocity ratio (−) stream in the river thalweg (Whitney et al., 1997). The concept
p Percent of the flow through the bypass relative to of a floating guide wall may have originated after dam opera-
the flow through the model inlet (%) tors observed fish accumulating along debris booms, similar to the
QB Total flow rate into bypass (ft3 /s) booms used for a floating guide wall.
QC Total flow rate under guide wall (ft3 /s) Novel to this study is the examination of the flow field upstream
QT Total flow rate through model inlet (ft3 /s) of a guide wall set at a wide range of depths and angles to flow and
t* Downward to sweeping velocity ratio threshold (−) subject to a wide range of average approach velocities, all within
V Average approach velocity (ft/s) an idealized power canal. New metrics, useful in the evaluation
Vx Mean velocity in the x-direction (ft/s) of guide walls, are presented. These metrics aim to explore the
Vy Mean velocity in the y-direction (ft/s) range of velocities and the strength of the downward flow sig-
Vz Mean velocity in the z-direction (ft/s) nal a fish may encounter while swimming along a guide wall. The
W Channel width (ft.) goal is to determine the combination of design parameters that will
 Angle of the guide wall relative to the side wall of most likely increase the chance of surface-oriented fish being suc-
the power canal (degrees) cessfully guided to the bypass. This analysis is performed through
sophisticated numerical modeling referred to as computational
fluid dynamics (CFD).

lenge is to either induce behaviorally or actively guide the fish into

2. Methodology
the bypass rather than the turbine intakes, which the bulk of the
flow in the power canal passes through (typically >90% when there
To evaluate the flow field immediately upstream of a guide wall,
is no spilling over the dam). Guidance technologies (e.g., louvers,
we used a parameterized CFD model of an idealized power canal
racks, screens, perforate plates, guide walls) are designed for this
(© ANSYS Fluent v 14.5, 2012). Fluent is a finite-volume code that
purpose.
iteratively solves the conservation of mass and momentum over a
Like other fish passage devices, guidance technologies rely on
set of discretized control volumes within the model domain until
the rheotactic response of fish (among other factors) to improve
convergence is reached. Section 2.1 describes the model domain (or
downstream passage efficiency and reduce migration delay (Schilt,
geometry of the model). Section 2.2 introduces the pertinent design
2007). Rheotaxis is defined as a fish’s behavioral orientation to the
parameters and details the range and interval over which each
water current (Montgomery et al., 1997). A fish’s movement with
is examined. Section 2.3 defines each of the boundary conditions
(or against) the water current is referred to as a negative (or pos-
applied to the model. These are the numerical conditions applied
itive) rheotaxis, respectively. In the case of a full-depth guidance
to the perimeter edges and faces of the model domain and must
structure (e.g. louvers and angled bar racks), the vertical velocity
be satisfied within the solution. Section 2.4 describes the mesh of
component upstream of the guidance structure is ignored and a
the CFD model. This pertains to the methods used to divide (or dis-
2-dimensional velocity vector is often used to inform the design.
cretize) the region within the model domain into a large number
These two velocity components are referred to as the sweeping
of small finite control volumes. Section 2.5 details the solvers (or
velocity (velocity component parallel to the guidance structure
numerical solution scheme) used to calculate the model results and
pointing in the direction of the bypass) and the normal veloc-
the convergence criteria applied to the solvers.
ity (velocity component perpendicular to the guidance structure
pointing directly at the face of the structure). A guidance struc-
ture installed at 45◦ or less to the upstream flow field will result in 2.1. Model domain
a sweeping velocity greater than or equal to the normal velocity,
thereby reducing the likelihood of impingement and entrainment. Fig. 2 displays the plan view of the power canal and a cross sec-
For this reason, guidance technologies are typically set at an angle tional view from the furthest downstream location at the bypass
of 45◦ or less to the flow field, thus creating a hydraulic cue designed entrance. The section downstream of the guide wall was not mod-
to elicit a negative rheotactic response from migrating fish. This cue eled to simplify the analysis. To accurately model head losses that
encourages their movement downstream towards the bypass. are incurred by the structure a more complex model than is pre-
In the case of a partial-depth guide wall (Fig. 1) that is aimed at sented here is required.
guiding surface-oriented fish, a strong downward vertical velocity For each scenario, the inlet location was fixed and the approach
component may be present upstream of the wall. The vertical veloc- distance  was held constant at 25 ft.(7.62 m). The longitudinal
ity component may compete with, or even overwhelm, hydraulic length of the guide wall, L, varies according to the angle of the guid-
cues created by the sweeping and normal velocities. Dominant ance structure, . The canal width, W, was 100 ft. (30.48 m) and
vertical velocities may encourage vertical fish movement and exac- the canal depth, H, was 40 ft. (12.192 m). The width of the bypass
erbate entrainment potential. NextEra Energy Maine Operating was 0.1 W or 10 ft. (3.048 m) The depth of the bypass opening was
Services, LLC, (2010), Kock et al. (2012), and Faber et al. (2011) 0.25H or 10 ft. (3.048 m). The total flow through the model inlet,
showed instances where a large proportion of downstream migrat- QT , the flow through the bypass outlet, QB , and the flow through
326 K.B. Mulligan et al. / Ecological Engineering 99 (2017) 324–332

Fig. 1. Partial-depth, floating, guide wall. The photo on the left (provided by Shane Scott) shows the panels with the floating boom. The photo on the right (taken from Google
Earth) shows an installed guidance device at the Bonneville Dam.

Fig. 3. The contour plot on the inlet of the CFD model geometry represents the veloc-
ity specified as a boundary condition in the case of V = 2 ft/s (0.607 m/s). Note the
Fig. 2. The schematic on the left shows the plan view of the idealized power canal. fully developed flow profile. Flow is in the positive x-direction. The model domain
The hatched area (upstream of the guide wall and bypass entrance) is the modeled is indicated by the black outline in this 3-D view.
region. The schematic on the right shows the cross-sectional view from A-A, the
furthest downstream location as seen on the plan view. The grey area is the guide
wall. The black area is the wall directly below the bypass entrance. Note the x-y-z
axis, the intersection of the x and y axis always occurs at the most upstream section set deeper than 20 ft. (6.096 m), the designs are less common and
of the guide wall, as shown above. On the x-axis, the bypass outlet is located at x = L are intended for use in deeper canals and forebays. The range of
and the model inlet is located at x = −.  is typical for surface guidance technologies and all guide walls
referred to in the literature are within this range. The range of V
Table 1 is also typical within a power canal, although 2 ft/s (0.607 m/s) is
Model Parameters.
more common. A value for V of 4 ft/s (1.219 m/s) is high for a typical
Parameter Range Interval power canal.
Depth of the Guide Wall (d), ft. 10–20 3.33
Angle of the Guide Wall (), deg 15–45 7.5
2.3. Boundary conditions
Average Inlet Velocity (V), ft/s 2–4 2

Three different types of boundary conditions were used in each

the main power canal outlet, QC , vary depending upon the aver- of the model scenarios. The first type of boundary condition was a
age approach velocity, V. The percent of the total flow through the velocity inlet. The inlet was defined using a velocity profile char-
bypass, p (equal to 100*QB /QT ), for all model runs was 5%. The size acteristic of a fully developed viscous flow with an average inlet
of the bypass opening and the percent of the total flow through velocity, V. The velocity profile for V = 2 ft/s (0.607 m/s) is shown
the bypass (p) are within the typical range for surface flow outlets in Fig. 3. To attain each developed flow profile, a rectangular chan-
(Johnson and Dauble, 2006) and p is also within the range of design nel CFD model was constructed, termed the Inlet Calculation CFD
criteria used by the US Fish and Wildlife Service in the Northeast Model (ICCM). The ICCM used a cross section at the inlet of the
(Odeh and Orvis, 1998). Idealized CFD Model and extruded it long enough such that fully
developed flow was achieved. In each ICCM run, the inlet was set
2.2. Model parameters to a uniform velocity equal to V and the outlet was specified as an
outflow carrying 100% of the flow. Identical solvers, described later,
The key parameters relevant to this work are the depth of the were used for both the ICCM runs and the Idealized CFD Model. The
guide wall, d, the angle of the guide wall, , and the average inlet velocity profile at the outlet of the ICCM was used as the velocity
velocity, V. There are a total of 40 scenarios. Table 1 displays the profile at the inlet of the Idealized CFD Model. In addition to the
ranges and intervals each parameter is evaluated on: velocity profile, the turbulence intensity (defined as the root-mean-
The range of d was chosen because it represents a set of typical square of the turbulent velocity fluctuations divided by the mean
values found within the literature. While guide walls have been velocity) was specified at 5%. © ANSYS Fluent v 14.5 (ANSYS Inc.,
 K.B. Mulligan et al. / Ecological Engineering 99 (2017) 324–332 327

2012) recommends the use of 5% in the event this value is unknown, 3. Results
as it was in this case.
The second type of boundary condition was a pressure outlet. To compare the 40 scenarios, several metrics were formulated
This outlet type is defined in two locations: 1) directly under the based on each scenario’s velocity output. Section 3.1 examines
guide wall and 2) through an entrance to a bypass. The two white trends found in the water velocity throughout each scenario and
areas in the cross-section A-A for Fig. 2 depicts each of the boundary shows in depth results for a single scenario. Section 3.2 introduces
locations. Each outlet was prescribed a hydrostatic pressure distri- a new metric referred to as the Maximum to Mean Velocity Ratio
bution and a target mass flow rate corresponding to the percentage (MMR), considered a possible indicator of fatigue and/or entrain-
of flow through the bypass, p. The streamlines were converging at ment. Section 3.3 presents the Downward to Sweeping Velocity
the pressure outlet specified below the guide wall; because of this Ratio (DSR), considered a possible indicator of guidance. Lastly, Sec-
a hydrostatic pressure distribution was not entirely accurate. How- tion 3.4 introduces the Upper Guidance Zone Depth (d*(t*)), a metric
ever, this likely has a minimal impact on the results as the pressure based off of a threshold DSR value, t*.
distribution should only be slightly different from hydrostatic.
The third type of boundary condition was a wall condition with 3.1. Velocity magnitude, components, and distribution
a specified shear and roughness height value. The water surface
was defined as a slip-condition with a specified shear stress of zero Fig. 4 displays the velocity magnitude and components (x-y-
and zero roughness because shear stress at the water-air interface z) on three vertical planes in the y-z axis for the scenario where
can be considered negligible. The channel walls and bottom were d = 10 ft. (3.048 m),  = 30◦ , and V = 2 ft/s (0.607 m/s). The three
defined as a no-slip condition, with a defined roughness height of planes are at x = 0.25 L, 0.5 L, and 0.75 L, where x was equal to 0 at the
1.64 × 10−2 . The face of the guide wall was also defined as a no-slip model inlet (the upstream boundary condition). The model bound-
condition, but the roughness height is 8.20 × 10−2 . An actual guide aries are shown in a sketched image around the contour plots. This
wall exterior is often composed of a rubber or stainless steel. figure shows several important points, all of which apply to each
of the 40 total scenarios. First, the maximum velocity magnitude
occurs immediately below the guide wall, while directly beside
2.4. Mesh the guide wall the water velocity magnitudes tend to be less than
the average inlet velocity, V. This drop in velocity correlates to an
In all scenarios for both the Idealized CFD Model and the ICCM, increase in the turbulence in the same region beside the guide wall.
the domains were divided into a number of finite volumes in the Second, the velocity component in the y-direction was shown to
form of tetrahedrons. Face and body sizing rules were applied in be negative in the upper portion of the water column and pos-
different regions of the domain. The smallest cells occur near the itive below the guide wall. This was expected as the guide wall
boundaries and guidance structure. The element face sizing on the was designed to create a strong sweeping velocity along the struc-
guidance wall ranged between 0.8 (0.024 m) and 1.6 ft (0.488 m). ture’s face toward the bypass. Third, the minimum velocity in the
The face sizing on the pressure outlets ranged between 1.0 (0.305 z-direction (a negative value) occurs directly at the bottom of the
m) and 1.6 ft. (0.488 m). Inflation layers were used to accurately guide wall. Fourth, the guide wall created a high velocity gradient
model the wall roughness effects on the flow field. The inflations along the z-axis at the face of the wall. Lastly, the velocity distri-
layers were applied at all boundaries of the model, including the bution beside and below the guide wall was very similar at each of
guide wall. The aspect ratio, orthogonal quality, and skewness were the locations.
the primary metrics used to evaluate mesh quality. Number of finite
volumes ranged from approximately 350,000–512,000. 3.2. Maximum to mean velocity ratio (MMR)

The MMR was calculated as the ratio of the maximum veloc-

2.5. Solver and convergence criteria ity magnitude on a specified plane to the average inlet velocity
magnitude (V). The specified plane was on the y-z axis at the lon-
All CFD runs performed in this analysis used the second order gitudinal midpoint of the guide wall (where x = 0.5L) and extends
upwind method to solve the conservation of momentum equa- from the water surface to the bottom of the guide wall. A value
tions for steady-state conditions. The runs were solved using the of the maximum velocity magnitude was determined for each of
SIMPLE scheme (Patankar and Spalding, 1972) as the pressure- the 40 scenarios based on the CFD output and then divided by the
velocity coupling method. The realizable k-⑀ turbulence closure average inlet velocity magnitude for the scenario. Fig. 5 shows the
model with standard wall functions was used to describe the tur- results in a contour plot for both V = 2 ft/s (0.607 m/s) and V = 4 ft/s
bulent kinetic energy and turbulent dissipation rate. Similar to (1.219 m/s) for all 40 scenarios.
momentum, the turbulence model was solved using the second Interestingly, the average approach velocity had minimal
order upwind method. However, in all scenarios each model was impact on the MMR. The values under all configurations range from
first solved using the first order upwind scheme. The results of the 1.14 to 1.62, with the lowest for a guide wall design of d = 10 ft.
first order upwind solving scheme were used as the initial solution (3.048 m) and  = 15◦ and the greatest for a design where d = 20 ft
to the second order upwind solver. This provided a means to reach (6.096 m) and  = 45◦ . Also, recalling from Fig. 4, the maximum
convergence quicker. Convergence criteria included the equation velocity magnitude occurs at the very bottom of the guide wall near
residuals for continuity, x-velocity, y-velocity, z-velocity, turbulent the face of the wall. This was consistent throughout all 40 scenarios.
kinetic energy, and turbulent dissipation rate. Additional monitors
included the integral of the velocity magnitude on the outlet below 3.3. Downward to sweeping velocity ratio (DSR)
the guide wall, integral of velocity magnitude on the outlet to the
bypass, total volume integral of the velocity magnitude in all fluid A problematic feature of some guide walls tested to date was
cells, the integral of the skin friction coefficient on the guidance that they can create a strong downward flow component which
face, and the total volume integral of turbulent kinetic energy in all can likely lead to a reduction in guidance efficiency. To evaluate
fluid cells. Additional details regarding the conservation of momen- this in the scenarios we tested, we formulated a metric that rep-
tum and turbulence solvers can be found in the © ANSYS Fluent v. resented the Downward to Sweeping Velocity Ratio (DSR), or the
14.5 code documentation manual (ANSYS Inc., 2012). ratio of the velocity in the z-direction to the magnitude of the x and
328 K.B. Mulligan et al. / Ecological Engineering 99 (2017) 324–332

Fig. 4. Contour plots of the velocity magnitude (far left), velocity in the x-direction (mid-left), velocity in the y-direction (mid-right), and velocity in the z-direction (far right)
for the scenario of d = 10 ft. (3.048 m),  = 30◦ , and V = 2 ft/s (0.607 m/s). The top row plots are for a plane located at x = 0.75L. The middle row plots are for a plane located at
x = 0.5L. The bottom row plots are for a plane located at x = 0.25L.

Fig. 5. Contour plots of the Maximum to Mean Velocity Ratio (MMR) for V = 2 ft/s (0.607 m/s) (left) and V = 4 ft/s (1.219 m/s)(right). The guide wall depth, d, is on the x-axis
and the guide wall angle, , is on the y-axis. The black circles indicate the data point locations corresponding to each combination of depth and angle run in the CFD analysis.
The contour lines are the result of a linear interpolation between data points.

y velocity components. To do this we assumed (based in part on the negative value indicates a downward flow, away from the water
rheotactic behavior of fish) that the larger the absolute value of the surface.
DSR, the more likely a fish will be to volitionally follow the down- Fig. 6 shows a typical distribution of the DSR taken at a plane at
ward current or be entrained below the guide wall. The DSR at each any x-location along the guide wall. There was a distinct DSR gra-
cell of the model was calculated using the following formula: dient that occurs along the face of the guide wall in the z-direction
where the values range from approximately 0 at the water surface
Vz to −0.825 at the bottom of the guide wall. This gradient exists for
DSR =  (1) each scenario, consisting of a DSR of approximately 0 at the water
Vx2 + Vy2
surface and a minimum value, DSRmin , occurring along the very
bottom of the guide wall, although the minimum value changes
Where Vz is the velocity in the z-direction, Vx is the velocity in the depending upon the depth and angle of the structure. The loca-
x-direction, and Vy is the velocity in the y-direction. The sweeping tion of DSRmin is the same location where the velocity magnitude
velocity (denominator of the DSR) at an elevation above the bottom reached its maximum value. Thus under this condition, a fish swim-
of the guide wall was always in the direction of the bypass whereas ming along the bottom of the guide wall might be more likely to be
the vertical velocity (numerator of the DSR) was always negative. entrained beneath it rather than safely guided to the bypass.
Fig. 6 displays a DSR contour plot on a vertical plane in the y-z By finding DSRmin for each scenario, we were able to state if the
axis at the longitudinal midpoint of the guide wall (x = 0.5 L) for the worst-case conditions along the guide wall are sweeping dominant
scenario of d = 10 ft. (3.048 m),  = 30◦ , and V = 2 ft/s (0.607 m/s). A
 K.B. Mulligan et al. / Ecological Engineering 99 (2017) 324–332 329

guide wall depth, d, changes dramatically from −0.8145 for a guide

wall design of  = 15◦ and d = 20 ft. (6.096 m) to −2.2715 for a guide
wall design of  = 45◦ and d = 20 ft. (6.096 m). This is also evident
when changing the guide wall depth as d*(t*) first equals d ranging
from −1.4965 to −2.2715 for guide wall designs where  = 45◦ . Note
that when d = d*(t*) there was a DSR greater than t* along the full
depth of the guide wall.
Also of note was that d*(t*) was nearly identical for each average
inlet velocity. This implies that when calculating the DSR a change
in velocity within the power canal was much less important than
the design parameters of the guide wall. However, the actual z-
component of the velocity changes in response to the prescribed
average inlet velocity, V.
Fig. 9 better illustrates the difference between d*(t*) and d for
all combinations of guide wall depths and angles with V equal to
2 ft/s (0.607 m/s) and t* equal to −1 (left), −0.67 (middle), and −0.33
(right). The transition depth alluded to in the previous Section 3.4
is represented in the left contour plot.
Most noticeable from Fig. 9 is that the difference between the
guide wall depth, d, and the Upper Guidance Zone Depth, d*,
increases as t* was reduced. This was expected as the threshold
Fig. 6. A contour plot of the DSR for the scenario of d = 10 ft. (3.048 m),  = 30◦ , and becomes more restrictive. This also shows the advantages of a lesser
V = 2 ft/s (0.607 m/s) taken at the longitudinal midpoint of the guide wall (x = 0.5L)
on a vertical plane in the y-z axis. The black rectangle in the top right indicates the
angle, particularly for the t* values closer to zero. For example, the
location of the guide wall. Recall the WSE = 40 ft. (12.192 m). difference in d*(t* = −0.33) for the scenario of  = 15◦ and d = 20 ft.
(6.096 m) and the scenario of  = 45◦ and d = 20 ft. (6.096 m) was
approximately 10 ft. (3.048 m). This difference was half of the guide
(DSRmin > −1.0) or downward dominant (DSRmin < −1.0). Therefore, wall depth for those scenarios. For these same two scenarios the
in the case that DSRmin was greater than −1.0, it was known that difference in d*(t* = −1) was approximately 6 ft. (1.829 m).
conditions from the water surface elevation (WSE) to the bottom of
the guide wall were sweeping dominant. However, if DSRmin indi-
cated that a specific scenario was downward dominant, then it was 4. Discussion & conclusion
known that there was a transition point somewhere between the
WSE and the bottom of the guide wall where the flow field shifts Considering the information gleaned from this study, a relatively
from sweeping dominant to downward dominant. This “transition small angle (the minimum was 15◦ ) appears more likely to produce
depth” (later referred to as d*(t* = −1)) was investigated in the fol- conditions favorable to efficient guidance. Both the metric related
lowing Section 3.4. to the maximum velocity (MMR) and the downward to sweeping
Fig. 7 displays two contour plots (for V = 2 ft/s (0.607 m/s) and velocity ratio (DSR) show that as the angle was increased 1) smaller
V = 4 ft/s (1.219 m/s)) which illustrate how DSRmin changes depend- juvenile fish should be more likely to be entrained below the guide
ing upon the depth and angle of the structure. The values range from wall and 2) larger adult fish should be more likely to volitionally
approximately −0.4 (d = 10 ft. (3.048 m),  = 15◦ ) to −2.3 (d = 20 ft. pass below the guide wall. Interestingly, lab-scale physical mod-
(6.096 m),  = 45◦ ). eling performed by the California Department of Water Resources
(CA DWR) found that guide wall panels oriented at 22◦ to the flow
3.4. Upper guidance zone depth (d*) and set at a depth of 5 feet (1.524 m) resulted in neutrally buoyant
beads guiding along and not passing under the guide wall (per-
Given a DSR threshold value (t*), the guide wall can be split sonal communication, Shane Scott, 3/14/14). Although this exact
from the water surface elevation (WSE = H = 40 ft. (12.192 m)) to the scenario was not tested in this analysis, it also shows the benefit of
guide wall depth, d, into two separate zones. For a given t*, the min- guide walls set at an angle near 15◦ .
imum depth (equivalent to the maximum elevation) at which the However, such a small angle may not always be required. In
DSR was equal to or less than t* is the Upper Guidance Zone Depth general, the authors suggest that the guide wall be set at an angle
(d*(t*)). For example, referring back to Fig. 6 and given a t* = −0.4, and depth such that d*(t* > −1) is greater than the maximum depth
d*(t* = −0.4)n≈ 7.5 ft. (2.286 m). The volume above the elevation at of the expected vertical distribution of all the target fish species at
depth d*(t*) possessed a DSR greater than t* and the volume below the site. The assumption of t* > −1 was applied to ensure sweeping-
possessed a DSR less than or equal to t*. The metric was based on dominant conditions and was designed to both take advantage of
the hypothesis that, due to a guide walls’ tendency to create strong the negative rheotactic fish response and to guide any passively
downward flows along its face, the guide wall can be split into an drifting juvenile fish. DSR threshold values closer to zero are likely
“Upper Guidance Zone” and a “Lower Guidance Zone”. The Upper to be more effective at reducing the number of fish that pass below
Guidance Zone was considered to be more likely to effectively guide the guide wall, although will require a longer (smaller angle) and/or
fish because of its reduced absolute value of the DSR. The Lower deeper wall to achieve.
Guidance Zone was considered to be less likely to effectively guide Without testing fish movement and behavior in response to
fish because of its greater absolute value of the DSR. Fig. 8 shows guide walls in real-world applications, it is difficult to predict
for V = 2 ft/s (0.607 m/s) and V = 4 ft/s (1.219 m/s) how the depen- how a fish will respond to the flow conditions. Although gener-
dent variable d*(t*) changes with the independent variable t*. The alized metrics partially based on the behavior known as rheotaxis
minimum d*(t*) is zero and the maximum is the depth of the guide were formulated, the results can in no way estimate actual fish
wall, d. behavior. Each of the metrics developed were based entirely on
The impact of changing the guide wall depth and angle on d*(t*) the velocity output data from the CFD analysis. Fish behavior was
is evident in Fig. 8. For instance, the value of t* where d*(t*) equals also impacted by hydraulic conditions such as acceleration and
330 K.B. Mulligan et al. / Ecological Engineering 99 (2017) 324–332

Fig. 7. Contour plots of DSR min for V = 2 ft/s (0.607 m/s) (left) and V = 4 ft/s (1.219 m/s) (right). The guide wall depth, d, is on the x-axis and the guide wall angle, , is on the
y-axis. The black circles indicate the data point locations, corresponding to each combination of depth and angle run in the CFD analysis. The black solid line is the contour
where DSRmin = −1.0. Scenarios above the line possess a sweeping dominant flow field along the entire depth of the guide wall whereas scenarios below the line possess a
lower section of the guide wall where a downward dominant flow field exists. The contour lines are the result of a linear interpolation between data points.

Fig. 8. Plots of d*(t*) versus the DSR Threshold, t*, for V = 2 ft/s (0.607 m/s) (left) and V = 4 ft/s (1.219 m/s)(right).

turbulence (Larinier, 1998), but fish also possess complex and the region around the downstream terminus of the guide wall).
unpredictable behaviors in response to environmental conditions However, the movement patterns also showed that the fish had
both inclusive and exclusive of hydraulics. Therefore, the authors a strong tendency to sound under the wall and on to the turbine
recognize that the inclusion of some of these variables in the eval- intakes where 33–52% of the fish by species passed downstream
uation of each scenario could make for a more sound approach to (the largest percentage of all the passage routes). Based on the
understanding how fish will behave near the guide wall. CFD analysis in this manuscript, the DSRmin for a guide wall at this
Field studies of guide wall installations that include detailed depth and angle is approximately −1.6 (see Fig. 7) and the transition
telemetry analysis are uncommon. One such study (referenced in depth, d*(t* = -1.0), is between 8 and 9 ft (see Fig. 9). It is likely that
the Introduction Section) was performed at the Cowlitz Falls Dam the guidance efficiency would have increased by either installing a
in 2011 (Kock et al., 2012) using radiotelemetry to track juvenile deeper guide wall or lowering the angle.
salmonids. The guide wall was constructed of steel panels attached CFD is based in physical laws and is capable of producing accu-
to a floating boom set at 10 ft. (3.048 m) deep and approximately rate and reliable results. Several other studies have been performed
45◦ to the approach flow. The study found that 40–63% of the fish using CFD as a means to better understand how a guide wall
by species arrived at the fish collection discovery area (defined as will impact the flow field in a forebay (Rakowski et al., 2006,
 K.B. Mulligan et al. / Ecological Engineering 99 (2017) 324–332 331

Fig. 9. Contour plots of the Upper Guidance Zone Depth, d*(t*) for t* = −1.0 (left), t* = −0.67 (middle), and t* = −0.33 (right). The guide wall depth, d, is on the x-axis and the
guide wall angle, , is on the y-axis. The average inlet velocity, V, is equal to 2 ft/s (0.607 m/s). The black circles indicate the data point locations, corresponding to each
combination of depth and angle run in the CFD analysis. The contour lines are the result of a linear interpolation between data points.

2010; Lundstrom et al., 2010). Lundstrom et al. (2010) examined likely to produce hydraulics favorable for efficient guidance. Future
ten guide wall configurations (different lengths, curvatures, and work is necessary, particularly to investigate other guide walls con-
depths) upstream of a spillway and turbine intakes at a hydro- figurations and perform more rigorous full-scale, field tests with
electric facility. An important metric used in this analysis was the the various fish species of interest.
acceleration along the guide wall and the acceleration downward
upstream of the guide wall. The authors argued that a high accel- 5. Disclaimer
eration downward immediately upstream of the guide wall would
improve guidance efficiencies juvenile fish tend to avoid regions The information, data or work presented herein was funded in
of high acceleration (Haro et al., 1998; Kemp et al., 2005; Johnson part by an agency of the United States Government. The Depart-
et al., 2000; Taft, 2000). The authors were satisfied with the perfor- ment of Energy, the Hydro Research Foundation, the U.S. Fish and
mance of the guide wall because the acceleration along the device Wildlife Service, and the University of Massachusetts Amherst, nor
was much smaller than that going downward, meaning the fish any of their employees, makes and warranty, express or implied, or
would choose the route along the device. While this may be true in assumes any legal liability or responsibility for the accuracy, com-
certain cases, we argue caution because a downward acceleration pleteness, or usefulness of any information, apparatus, product, or
that is too high may entrain the weak swimming juvenile fish and process disclosed, or represents that its use would not infringe pri-
force them under the wall towards the turbines. vately owned rights. Any use of trade, product, or firm names is
Furthermore, the authors acknowledge several limitations to for descriptive purposes only and does not imply endorsement by
this study. First, the selected model domain of a rectangular power the U.S. Government. The views and opinions of authors expressed
canal was not truly representative of a real hydropower project, herein do not necessarily state or reflect those of the Department of
which likely has much more complex hydraulics. When possible in Energy, the Hydro Research Foundation, the U.S. Fish and Wildlife
practice, the authors recommend applying the derived metrics to Service, and the University of Massachusetts Amherst.
a site-specific CFD model in order to determine proper depths and
angle. Second, the use of a single phase model results in a loss of
Acknowledgments
model resolution near the water surface boundary layer, although
this is not expected to make a substantial difference in the results
The information, data, or work presented herein was funded
and is a common simplification when wave action is not integral
in part by the Office of Energy Efficiency and Renewable Energy
in the analysis. Third, physical aspects of the structure have been
(EERE), U.S. Department of Energy, under Award Number DE-
ignored. The forces applied to a guide wall may create a vertical
EE0002668 and the Hydro Research Foundation. In addition, this
tilt such that the guidance wall is not perpendicular to the water
work was partly funded by the Perrell family who generously
surface and/or a curvature may develop when looking from plan
offered support in the first author’s final semester at the University
view. Ideally, strengthening of the structure and anchoring it to the
of Massachusetts.
bottom could minimize the deflection. More research is needed to
investigate the hydraulics of tilted/deflected guide walls.
References
In conclusion, guide walls have been utilized to improve down-
stream passage survival for anadromous fishes including salmonids ANSYS, Inc, 2012. ANSYS FLUENT Theory Guide. Southpointe, Canonsburg, PA.
and alosines for more than 20 years. Less frequently implemented Buckley, J., Kynard, B., 1985. Vertical Distribution of Juvenile American Shad and
than other surface guidance technologies (e.g. louvers, bar racks, Blueback Herring During the Seaward Migration in the Connecticut River.
Massachusetts Cooperative Fishery Research Unit, Department of Forestry and
screens, among others), they are gaining popularity, particularly in
Wildlife Management, Amherst, MA.
the northwestern United States. This body of research focuses on Bunt, C.M., Castro-Santos, T., Haro, A., 2012. Performance of fish passage structures
the basic design parameters and begins to answer the question of at upstream barriers to migration. River Res. Appl. 28, 457–478, http://dx.doi.
org/10.1002/rra.1565.
which configuration might enhance fish guidance. A CFD approach
Enders, E.C., Gessel, M.H., Williams, J.G., 2009. Development of successful fish
was used to answer this fundamental question. The key findings passage structures for downstream migrants requires knowledge of their
indicated that a guide wall set at a small angle and deep enough behavioural response to accelerating flow. Can. J. Fish. Aquat. Sci. 66,
such that sweeping-dominant conditions (or d*(t* > −1)) covers the 2109–2117, http://dx.doi.org/10.1139/F09-141.
Faber, D.M., Ploskey, G.R., Weiland, M.A., Deng, D., Hughes, J.S., Kim, J., Fu, T.,
expected vertical distribution of the approaching fish was more Fischer, E.S., Monter, T.J., Skalski, J.R., 2011. Evaluation of Behavioral Guidance
332 K.B. Mulligan et al. / Ecological Engineering 99 (2017) 324–332

Structure on Juvenile Salmonid Passage and Survival at Bonneville Dam in Migration Season. NextEra Energy Maine Operating Services, LLC, Hallowell,
2009. Pacific Northwest National Laboratory, Richland, WA. ME.
Haro, A., Odeh, M., Noreika, J., Castro-Santos, J., 1998. Effect of water acceleration Odeh, M., Orvis, C., 1998. Downstream fish passage design considerations and
on downstream migratory behavior and passage of Atlantic salmon juvenile developments at hydroelectric projects in the Northeast USA. In: Jungwirth,
salmonids and juvenile American shad at surface bypasses. Trans. Am. Fish. M., Schmutz, S., Weiss, S. (Eds.), Fish Migration and Fish Bypasses. Fishing New
Soc. 127, 118–127. Books, Oxford, UK, pp. 267–280.
Johnson, G.E., Dauble, D.D., 2006. Surface flow outlets to protect juvenile salmonids Patankar, S.V., Spalding, D.B., 1972. A calculation procedure for heat: mass and
passing through hydropower dams. Rev. Fish. Sci. 14, 213–244. momentum transfer in three-dimensional parabolic flows. Int. J. Heat Mass
Johnson, G. E., Giorgi, A. E., Erho, M. W., 1997. Critical assessment of surface flow Transfer 15, 1787–1806.
bypass development in the lower Columbia and Snake rivers. Completion Report Rakowski, C. L., Richmond, M. C., Serkowski, J. A., Johnson, G. E., 2006. Forebay
for the U.S. Army Corps of Engineers, Portland and Walla Districts. computational fluid dynamics modeling for the Dalles Dam to support behavior
Johnson, G.E., Adams, N.S., Johnson, R.L., Rondorf, D.W., Dauble, D.D., Barila, T.Y., guidance system siting studies. Final Report. Prepared for the U.S. Army Corps of
2000. Evaluation of the prototype surface bypass for salmonid juvenile Engineers Portland District, Portland, Oregon Under a Related Services
salmonids in spring 1996 and 1997 at Lower Granite Dam on the Snake River, Agreement with the U.S. Department of Energy Contract DE-AC06-76RL01830.
Washington. Trans. Am. Fish. Soc. 129, 381–397. Rakowski, C. L., Richmond, M. C., Serkowski, J. A., 2010. Bonneville Powerhouse 2 3D
Kemp, P.S., Gessel, M.H., Williams, J.G., 2005. Fine-scale behavior responses of CFD for the Behavioral Guidance System. Prepared for the U.S. Army Corps of
pacific salmonid smolts as they encounter divergence and accleration of flow. Engineers Portland District, Portland, Oregon Under a Contract
Trans. Am. Fish. Soc. 134 (2), 390–398. DE-AC05-76RL01830 with the U.S. Department of Energy.
Kock, T.J., Liedtke, T.L., Ekstrom, B.K., Tomka, R.G., Rondorf, D.W., 2012. Behavior Schilt, C.R., 2007. Developing fish passage and protection at hydropower dams.
and Passage of Juvenile Salmonids During the Evaluation of a Behavioral Appl. Anim. Behav. Sci. 104, 295–325, http://dx.doi.org/10.1016/j.applanim.
Guidance Structure at Cowlitz Falls Dam, Washington, 2011. U.S. Geological 2006.09.004.
Survey Open-File Report1030–2012. Scott, S., 2012. A postive barrier fish guidance system designed to improve safe
Larinier, M., 1998. Upstream and downstream fish passage experience in France. downstream passage of anadromous fish. In: 9th ISE 2012, Vienna.
In: Jungwirth, M., Schmutz, S., Weiss, S. (Eds.), Fish Migration and Fish Taft, E.P., 2000. Fish protection technologies: a status report. Environ. Sci. Policy 3,
Bypasses. Blackwell Science Ltd Publisher, pp. 127–145. 5349–5359.
Limburg, K.E., Waldman, J.R., 2009. Dramatic declines in North Atlantic Weiss-Glanz, L. S., Stanley, J. G., Moring J. R., 1986. Species profiles: life histories
diadromous fishes. Bioscience 59 (11), 955–965, http://dx.doi.org/10.1525/bio. and environmental requirements of coastal fishes and invertebrates (North
2009.59.11.7. Atlantic): American shad. U.S. Fish and Wildlife Service Biological Report
Lundstrom, T.S., Gunnar, J., Hellstrom, I., Lindmark, E.M., 2010. Flow design of 82(11.59). U.S. Army Corps of Engineers, TR EL-82-4. 16 pp.
guiding device for downstream fish migration. River Res. Appl. 26, 166–182. Whitney, R., Calvin, L., Erho, M., Coutant, C., 1997. Downstream Passage for Salmon
McDowall, R.M., 1987. Evolution and importance of diadromy. Am. Fish. Soc. Symp. at Hydroelectric Projects in the Columbia River Basin: Development,
1, 1–13. Installation, and Evaluation. Northwest Power Planning Council, Portland, OR.
McDowall, R.M., 1997. The evolution of diadromy in fishes (revisited) and its place Williams, J.G., Armstrong, G., Katopodis, C., Larinier, M., Travade, F., 2012. Thinking
in phylogenetic analysis. Rev. Fish Biol. Fish. 7, 443–462. like a fish: a key ingredient for development of effective fish passage facilities
Montgomery, J.C., Baker, C.F., Carton, A.G., 1997. The lateral line can mediate at river obstructions. River Res. Applic. 28, 407–417, http://dx.doi.org/10.1002/
rheotaxis in fish. Nature 389, 960–963. rra.1551.
NextEra Energy Maine Operating Services, LLC, 2010. NextEra Energy Diadromous
Fish Passage Report for the Lower Kennebec River Watershed During the 2009