PNNL-14591
Simulating the Flow Field
Upstream of the Dworshak
Dam Regulating Outlets
C.B. Cook
M.C. Richmond
March 2004
Prepared for Idaho Fish and Game, Ahsahka,
Idaho Dworshak Dam Assessment and Fisheries
Investigations Bonneville Power Administration
Project 1987-099-00
Prepared for the U.S. Department of Energy
under Contract DE-AC06-76RL01830
PNNL-14591
Simulating the Flow Field
Upstream of the Dworshak
Dam Regulating Outlets
C.B. Cook
M.C. Richmond
March 2004
Prepared for Idaho Fish and Game, Ahsahka, Idaho
Dworshak Dam Assessment and Fisheries Investigations
Bonneville Power Administration Project 1987-099-00
Pacific Northwest National Laboratory
Richland, Washington 99352
Cook and Richmond PNNL-14591
Summary
The Pacific Northwest National Laboratory numerically modeled flow conditions upstream of the
regulating outlets at Dworshak Dam, North Fork Clearwater River, Idaho. The Pacific Northwest
National Laboratory conducted this work for the Idaho Department of Fish and Game under
contract number 43857A in support of Idaho Department of Fish and Game Project 1987-099-00
with the Bonneville Power Administration.
Numerical simulations were performed using the computational fluid dynamics model Flow-3D,
a peer reviewed and validated three-dimensional Reynolds-averaged Navier-Stokes
hydrodynamic model. Results were studied to determine the impacts of water surface elevation
and discharge though the three regulating outlets on flow velocities in the reservoir forebay.
These simulations were in general support of a larger research program conducted by the Idaho
Department of Fish and Game that is evaluating the efficacy of strobe lights to deter fish from
entering the regulating outlets and powerhouse turbine intakes.
Simulation results indicate that large variations in forebay water velocities occur over the typical
range of regulating outlet operations and seasonal water surface fluctuations. As expected, water
velocities generally increase with larger outlet gate openings and higher water surface elevations.
Simulations span typical regulating outlet operations: forebay water surface elevations between
1460 ft and 1600 ft and regulating outlet gate valve openings between 1 ft and 10 ft open. In
addition, simulations examined flow conditions when only one or two of the three regulating
outlets were operating. The resulting matrix of 24 unique simulations have been distilled and
summarized in this report.
iii
Cook and Richmond PNNL-14591
Acknowledgements
We sincerely appreciate the cooperation, assistance, and dedication of the following persons:
Idaho Fish and Game
o Eric Stark (project leader)
o Melo Maiolie (program leader)
US Army Corps of Engineers, Walla Walla District
o Rick Emmert (structural drawings)
Pacific Northwest National Laboratory
o John Serkowski (graphics support)
o Julie Hughes (contract support)
v
Cook and Richmond PNNL-14591
Contents
1.0 Introduction ........................................................................................................................ 1
2.0 Numerical Model Components and Methods..................................................................... 3
2.1 Dworshak Dam Structural Components................................................................ 3
2.2 CFD Model Domain.............................................................................................. 7
2.3 Numerical Model................................................................................................... 9
2.3.1 CFD Model description............................................................................. 9
2.3.2 Boundary Conditions and Assumptions.................................................... 9
3.0 Computational Modeling Scenarios ................................................................................. 11
4.0 Results and Future Directions .......................................................................................... 13
5.0 References ........................................................................................................................ 15
Appendix A. Velocity Magnitude Simulation Results.............................................................. 17
vii
Cook and Richmond PNNL-14591
Figures
Figure 1 Location of Dworshak Dam in relation to other dams in the Snake and Columbia
River watershed. ......................................................................................................... 1
Figure 2 Schematic depicting the upstream face of Dworshak Dam.................................. 3
Figure 3 Plan View Drawing of Regulating Outlet 3 ......................................................... 5
Figure 4 Vertical Side View Drawing of a Regulating Outlet............................................ 6
Figure 5 CFD mesh domain extent ..................................................................................... 7
Figure 6 CFD mesh resolution............................................................................................ 8
Figure 7 Simulation results for Run 1............................................................................... 12
Figure 8 Simulation results for Run 2............................................................................... 12
Tables
Table 1 Matrix of simulations performed during the project............................................ 11
ix
Cook and Richmond PNNL-14591
1.0 Introduction
Dworshak Dam is a large hydroelectric dam located on the North Fork of the Clearwater River in
Western Idaho (Figure 1). Downstream of the dam, the Clearwater River enters the Snake River
near Clarkston,WA and Lewiston, ID. During late spring and summer water is released from
lower levels of the reservoir to help cool water temperatures in the Lower Snake River
downstream of the Clearwater and Snake River confluence. These cooler waters improve thermal
conditions for endangered salmon in the Lower Snake River, and often push water temperatures
below the state/federally mandated temperature criterion.
Figure 1 Location of Dworshak Dam in relation to other dams in the Snake and Columbia Rivers
watershed.
1
Cook and Richmond PNNL-14591
Three regulating outlet intakes (ROs) were placed in the dam during original construction. The
upstream intake openings are 22 ft tall by 16 ft wide at the face of the dam. These bell mouth
intakes taper once inside the dam to approximately 12.5 ft tall by 9 ft wide at the tainter gate
valve (approximately 30 ft inside the dam). Depending upon forebay elevation and tainter gate
opening, discharges through each RO can exceed 10,000 cfs.
Operation of the ROs depends upon downstream water needs and the quantity of flow entering
Dworshak Reservoir. Since hydroelectric power is not generated when flow passes through the
ROs, they are not the preferred method for passing water through the dam. The ROs are necessary
however during certain periods of the year when the forebay water surface is below the spillway
crest and outflows required from the dam surpass powerhouse capacity. When the ROs are
operated, fish passing through the ROs may be injured or killed. The Idaho Department of Fish
and Game is currently investigating the use of strobe lights, placed upstream of the RO entrances,
as a method to deter fish from entering the RO intakes.
Optimal placement of the strobe light apparatus requires understanding of the flow field upstream
of the RO intakes. To obtain this information, and to answer general questions regarding the scale
of velocity gradients near the RO entrances, a computational fluid dynamic (CFD) model was
constructed. Although originally it was proposed that field data be obtained to validate this
model, the scope of work was later changed. This scope change occurred primarily because the
ROs were not operated during 2003. As a result of this change, additional CFD simulations were
performed to satisfy the contract. These additional simulations span a wider range of water
surface elevations than what was originally proposed.
2
Cook and Richmond PNNL-14591
2.0 Numerical Model Components and Methods
2.1 Dworshak Dam Structural Components
The three regulating outlets (ROs) at Dworshak Dam are relatively deep and small structures on
the face of the dam. Figure 2 displays their location with respect to other structures along the
upstream face. In this figure, the water surface has been placed at elevation 1600 ft, although the
normal operational range is between elevations 1445 and 1600 ft (USACE, 1986). The ROs are
numbered sequentially from one to three, beginning with the left (looking downstream left/right
sign convention) outlet, following the engineering drawings supplied from the Walla Walla
District, US Army Corps of Engineers.
The upper elevation of the bell mouth intake is at elevation 1373 ft MSL (mean sea level). To the
right of RO3, the main structural components of the powerhouse penstock intakes were
constructed in the numerical model. Because the ROs are several hundred feet to the side of the
powerhouse intakes, the details of the selective withdrawal structures in front of the intakes were
not constructed in the CFD model, and hence are not displayed in the figure. The three operating
powerhouse intakes are visible in Figure 2, beginning approximately 230 ft to the right of RO3.
Figure 2 Schematic depicting the upstream face of Dworshak Dam
3
Cook and Richmond PNNL-14591
Specific details regarding the ROs are excerpted from USACE (1986):
Three outlets are available for evacuation of reservoir storage below the spillway
crest (elevation 1,545). The center outlet is located beneath the intermediate pier which
separates the two spillway bays. The other two outlets are also located beneath the
spillway, but these outlets flank the spillway gates. Discharge capacity of the three outlets
varies from 23,100 cfs at minimum pool elevation 1,445 feet to 39,750 cfs at full pool
elevation 1,600 feet. The outlets all seal at elevation 1,350 by hydraulically operated tainter
valves. The tainter valves are 9 feet wide by 12.5 feet high.
Discharge through these outlets is controlled through the use of three tainter gate valves located
approximately 30 ft inside the dam. Operation of these outlets is described in USACE (1986):
The tainter valves are an eccentric-trunnion type and are 9 feet wide by 12.5 feet
high. The principal elements of a valve are the skin-plate assembly, vertical girders,
struts, and trunnions. Modified low-alloy steel is used for the skin plate, and the balance
of the gate is fabricated from steel. The valve struts are bolted to the trunnion arms with
high-strength bolts for future removal of the valve, if necessary.
Valve operation consists of three phases: retraction, raising or lowering, and
sealing. Rotation of the eccentric-trunnion shafts by a pivoted hydraulic cylinder and
lever arm arrangement either advances or retracts the valve 3/4 inch. Each valve is raised
or lowered by another pivot-mouthed hydraulic cylinder. Individual hydraulic systems are
provided for each tainter valve and serve both the operating cylinder and the cylinder
which operates the eccentric.
Approximate valve operating times, including retraction and sealing operations,
are 2- minutes for raising completely and 3 minutes for lowering from the full raised
position. A complete set of operating controls, position sensing devices, and indicators
are provided for each valve machinery room. Remote controls and indicators are provided
in the powerhouse. A 100-kW emergency generator is located in the electrical room
(elevation 1,580) in monolith 25 for supplying power for operation of spillway and outlet
gates during an interruption in the regular power supply.
The engineering drawings shown in Figures 3 and 4 were obtained from Walla Walla District, US
Army Corps of Engineers. These drawings display the specific engineering details of the ROs.
Figure 3 is a plan view image specifically for RO3. ROs 2 and 1 are similar in construction,
although the centerline dimension changes from 5072 ft to 5000 ft and 4927 ft, respectively.
Figure 4 is a side view through a generic RO and details are approximately the same for all three
ROs.
4
Cook and Richmond PNNL-14591
Figure 3 Plan View Drawing of Regulating Outlet 3. Drawing scanned from USACE engineering drawing sheet 157, file no DWD-1-4-17/26. Sheet label
Dworshak Dam and Reservoir, Outlet Works, Bellmouth Intake, Horizontal and Vertical Control. Sheet drawn on May 2, 1967.
5
Cook and Richmond PNNL-14591
Figure 4 Vertical Side View Drawing of a Regulating Outlet. Drawing scanned from USACE engineering drawing sheet 157, file no DWD-1-4-17/26.
Sheet label Dworshak Dam and Reservoir, Outlet Works, Bellmouth Intake, Horizontal and Vertical Control. Sheet drawn on May 2, 1967.
6
Cook and Richmond PNNL-14591
2.2 CFD Model Domain
The extent of the three-dimensional numerical model is displayed in Figure 5. Although the ROs
are fairly small structures, the domain required by the CFD model was significant (approximately
1/3 of the dam width). This size of grid was required to ensure that model boundary conditions
did not influence solution results. The mesh domain extent was a hexahedral, 880 ft wide by 400
ft deep by 620 ft high. The vertical extent of the mesh extended from elevation 1000 ft to 1620 ft
MSL.
Figure 5 CFD mesh domain extent. Note: the mesh grid displayed in the figure is several times
coarser than actual (see Figure 6).
The time required to compute a simulation result is directly proportional to the number of
computational cells. Hence effort was put forth to construct a mesh that had sufficient resolution
to capture velocity gradients and turbulence characteristics in regions of interest, yet was coarse
in regions of either low velocity gradients or not of concern under the present project. The
resulting mesh grid is displayed in Figure 6.
Mesh resolution was finest at 1 ft grid cells near and inside the ROs. Upstream of the RO
entrances, the grid was coarsened to a uniform 3 ft grid. This region stretched from approximately
100 ft upstream of each RO centerline and 100 ft to the left of RO1 and to the right of RO3.
Outside of this finer resolution region, the mesh was uniformly sized with 10 ft grid cells.
7
Cook and Richmond PNNL-14591
Figure 6 CFD mesh resolution. The left graphic displays the mesh resolution between the downstream boundary (40 ft inside the dam) to 100 ft
upstream of the dam face. The right graphic displays the mesh resolution from 100 ft upstream of the dam face to the upstream CFD model boundary.
8
Cook and Richmond PNNL-14591
2.3 Numerical Model
2.3.1 CFD Model description
The commercial software package Flow-3D was selected as the numerical model for simulating
the ROs at Dworshak Dam. The model has a wide user base and has been previously tested and
validated under a wide range of applications.
Flow-3D uses the finite volume method to solve the Reynolds-averaged Navier-Stokes (RANS)
equations. The physical domain to be simulated must be decomposed into Cartesian grids
composed of variable-sized hexahedral cells. The domain can either be contained within a single
grid or several grids, often called multi-blocks. If several grids are used, each grid must either
be completely contained within a larger grid (nested) or be adjacent to another grid.
For each cell, average values for the flow parameters (pressure and velocity) are computed at
discrete times using a staggered grid technique (FSI, 2003). The staggered grid technique places
all dependent variables at the center of each cell except for velocities, which are located at cell
faces. This prevents the checkerboard solution that can result in incompressible flow
simulations when velocities and pressures become unlinked at adjacent computational nodes
(Patankar, 1980). Most terms in the equations are evaluated explicitly using the current time-level
values of the local variables. Although this explicit procedure is generally efficient and well
suited for free-surface wave propagation, it requires that the time-step size be limited to maintain
stability requirements. Time steps for most of the Dworshak Dam simulations were on the order
of 0.04 sec.
Flow-3D has several models for calculating the turbulent viscosity: Prandlt mixing length, one-
equation, two-equation - and Renormalization Group (RNG) -, and Large Eddy Simulation
(LES). These typically robust turbulence models have been well tested and documented in the
relevant technical literatures. Based upon prior experience with Flow-3D and the size of the
domain, the RNG - model was selected for all simulations.
The RNG model applies statistical methods for a derivation of the averaged equations for
turbulence quantities, such as turbulent kinetic energy and dissipation rate. The RNG model also
relies less on empirical constants versus the standard - model. Details on the turbulence model
can be found in Yakhot and S.A.Orszag (1986), Yakhot and L.M.Smith (1992) and Yakhot et al.
(1992).
2.3.2 Boundary Conditions and Assumptions
Flow-3D uses a technique called the Fractional Area/Volume Obstacle Representation (FAVOR)
technique to define solids within the model domain (FSI, 2003). Unlike the finite element or
another type of boundary fitted CFD model, the FAVOR technique requires the user to first
define a bounding hexahedral shape (see Figure 5 above). Then, inside of this hexahedron, the
dam structure and bathymetry are imported and placed to define solid objects. The model
determines at the onset of each simulation which cells within the hexahedron are fully blocked
and does not solve the equations of motion within these cells.
Boundary conditions must be specified along all faces of the bounding hexahedral domain shown
in Figure 5. Along the most upstream face a pressure boundary and a water surface elevation was
specified, allowing flow to smoothly enter the domain. The side boundaries were symmetry
planes of velocity magnitude, which forces water to only flow parallel to these faces. The bottom
boundary was the bathymetry and the top boundary was also a symmetry plane. Although Flow-
9
Cook and Richmond PNNL-14591
3D is capable of simulating free-surface flows, a rigid lid was placed at the top boundary to
suppress waves that may form along the water surface during model warm-up.
Several key assumptions implied by specifying these boundary conditions are: 1) there is zero
ambient flow in the lake and all flow entering the upstream pressure boundary starts from rest, 2)
the powerhouse turbine units are off and zero flow is passing through the powerhouse penstock
intakes, and 3) the lake was thermally homogenous. The first assumption was necessary because
observed velocity data from the forebay was unavailable at the time of this study. If observed
field data becomes available, alternative boundary conditions, such as specified velocity vectors
for both the upstream and side boundaries of the domain, may be more appropriate. The second
assumption was made for simplicity, and may be altered if desired for future simulations. Before
the powerhouse intakes could be simulated however, the selective withdrawal structures would
need to be incorporated into the CFD model grid. The third assumption ignores any influence of
stratification on water motions within the lake. This assumption was based upon 2002 water
profile data (Carroll and Barkow, 2003), which showed that at elevations below the thermocline
water temperatures in the lake were relatively uniform. Since the upper entrance of the ROs are at
elevation 1373 ft (approximately 100 ft to 200 ft beneath the water surface), these low level
outlets are generally below the thermocline and withdrawal characteristics are not expected to
change due to upper water column stratification. If future modeling of the selector gates is
performed this assumption could easily be changed within the numerical model.
10
Cook and Richmond PNNL-14591
3.0 Computational Modeling Scenarios
Twenty-four separate simulations were performed under the scope of this project. These
simulations fall into four water surface elevation groups: 1460 ft, 1500 ft, 1560 ft, and 1600 ft
MSL (see Table 1). For each water surface elevation, the tainter gate valves were opened either 1
ft or 10 ft. In addition, for each unique combination of water surface elevation and valve opening,
either one (center RO2 only), two (RO1 and RO 3 only), or all three ROs were operating. This
pattern of tainter valve operation follows the strict operational rules for Dworshak Dam, and is
required for smooth passage of flow down the spillway face.
Table 1 Matrix of simulations performed during the project.
Dworshak RO Simulations
Run # Elevation Gate Opening ROs Open Discharge/bay Total Q
2 1460 1 center 600 600
3 1460 1 ends 600 1,200
4 1460 1 all three 600 1,800
5 1460 10 center 5,650 5,650
6 1460 10 ends 5,650 11,300
7 1460 10 all three 5,650 16,950
8 1500 1 center 700 700
9 1500 1 ends 700 1,400
10 1500 1 all three 700 2,100
11 1500 10 center 6,700 6,700
12 1500 10 ends 6,700 13,400
13 1500 10 all three 6,700 20,100
14 1560 1 center 850 850
15 1560 1 ends 850 1,700
16 1560 1 all three 850 2,550
17 1560 10 center 8,100 8,100
18 1560 10 ends 8,100 16,200
19 1560 10 all three 8,100 24,300
20 1600 1 center 950 950
21 1600 1 ends 950 1,900
22 1600 1 all three 950 2,850
23 1600 10 center 9,100 9,100
24 1600 10 ends 9,100 18,200
1 1600 10 all three 9,100 27,300
This wide range of water surface elevations, number of ROs open, and tainter gate valve openings
produced a dynamic range of water velocities in the forebay. This range is illustrated graphically
in Figures 7 and 8 below.
11
Cook and Richmond PNNL-14591
Figure 7 Simulation results for Run 1. Brown is bathymetry, forebay water surface elevation is 1600
ft, all three ROs are open, tainter valves are open 10 ft, discharge per bay is 9,100 cfs, and the total
discharge is 27,300 cfs.
Figure 8 Simulation results for Run 2. Brown is bathymetry, forebay elevation is 1460 ft, only the
center ROs is open, tainter valve is open 1 ft, discharge for the center bay is 600 cfs, and the total
discharge is 600 cfs.
12
Cook and Richmond PNNL-14591
4.0 Results and Future Directions
Simulation results presented in Appendix A display velocity magnitude contour plots and X-Y
graphs of velocity magnitude versus distance from the dam face. The upper graphic on each page
displays results along a horizontal slice placed at elevation 1360 ft (approximate RO centerline).
Contours have been placed at 1 ft/s increments, between 1 and 10 ft/s (note: the 1 ft/s contour is
the most upstream contour shown in every figure).
The center graphic on each appendix page displays simulation results along the vertical centerline
in front of each RO. As for the horizontal contours, vertical slice contours of velocity magnitude
have been placed at 1 ft/s increments, between 1 and 10 ft/s (the farthest upstream is the 1 ft/s
contour). Black arrows have been placed in all contour plots at the entrance to each RO. These
arrows indicate the general flow direction if that particular RO was operating; they were not
scaled to represent velocity magnitude and have been placed at each entrance whether that
particular RO is operating or not for a specific simulation.
X-Y graphs of simulation results are presented on each page in Appendix A. Data displayed in
these graphs were constructed by sampling model results along a horizontal line extending out
from the dam face, starting at each RO intake center point.
As expected, water velocity magnitudes at the same distance away from the dam face increase
dramatically with larger tainter gate valve openings. Velocity contour lines have been shown
between 1 and 10 ft/s in the appendix figures. The location of small magnitude velocity contour
lines (i.e., contours less than 3 ft/s) could vary dramatically based upon ambient lake velocities,
powerhouse operations, and thermal stratification.
One recommendation to improve simulation results would be to confirm the underlying
assumptions used when constructing the numerical model. A method to confirm these
assumptions would be to collect of water velocity and temperature data in the forebay while the
ROs are operating. A second recommendation for model improvement would be to include the
selective withdrawal structure into the model and to simulate concurrent powerhouse operations
with RO operation. At present the impact of these assumptions are unknown, however
experience suggests that they could impact the location of the contour lines shown in the
appendix; especially smaller velocity magnitude contour lines. Likewise, the location of larger
velocity magnitude contour lines would be expected to move less due to these assumptions as
they are more directly influenced by RO operations.
13
Cook and Richmond PNNL-14591
5.0 References
Carrol, J and K. Barkow (2003) Preliminary Water Quality, Meteorological, Operations, and
Flow Data for the Snake River System. Carrol J. and K. Barkow, eds., U.S. Army Corps of
Engineers, March.
FSI (2003). Flow-3D Users Manual. Flow Science, Inc., Sante Fe, NM.
Patankar, S. (1980). Numerical Heat Transfer and Fluid Flow. Hemisphere, New York, New
York.
USACE (1986). Water Control Manual for Dworshak Dam and Reservoir, North Fork
Clearwater River, Idaho. U.S. Army Corps of Engineers, Walla Walla District, November.
Yakhot, V., and L.M.Smith (1992). The Renormalization Group, the e-Expansion and Derivation
of Turbulence Models. J. Scientific Computing, 7:3561.
Yakhot, V., S. Orszag, S. Thangam, T. Gatski, and C. Speziale, 1992. Development of turbulence
models for shear flows by a double expansion technique. Phys. Fluids, A4(7):15101520.
Yakhot, V., and S.A.Orszag, 1986. Renormalization Group Analysis of Turbulence. I. Basic
Theory. J. Scientific Computing, 1:151.
15
Cook and Richmond PNNL-14591
Appendix A. Velocity Magnitude Simulation Results
17
Cook and Richmond PNNL-14591
19
Cook and Richmond PNNL-14591
20
Cook and Richmond PNNL-14591
21
Cook and Richmond PNNL-14591
22
Cook and Richmond PNNL-14591
23
Cook and Richmond PNNL-14591
24
Cook and Richmond PNNL-14591
25
Cook and Richmond PNNL-14591
26
Cook and Richmond PNNL-14591
27
Cook and Richmond PNNL-14591
28
Cook and Richmond PNNL-14591
29
Cook and Richmond PNNL-14591
30
Cook and Richmond PNNL-14591
31
Cook and Richmond PNNL-14591
32
Cook and Richmond PNNL-14591
33
Cook and Richmond PNNL-14591
34
Cook and Richmond PNNL-14591
35
Cook and Richmond PNNL-14591
36
Cook and Richmond PNNL-14591
37
Cook and Richmond PNNL-14591
38
Cook and Richmond PNNL-14591
39
Cook and Richmond PNNL-14591
40
Cook and Richmond PNNL-14591
41
Cook and Richmond PNNL-14591
42
