Professional Documents
Culture Documents
FY&ZM JofHY (ASCE) 1943-7900
FY&ZM JofHY (ASCE) 1943-7900
FY&ZM JofHY (ASCE) 1943-7900
Abstract: This paper documents a prototype-scale investigation of cavitation damage to a newly resurfaced spillway being used opera-
tionally during phased construction work to replace all piers and radial gates. Initial concrete damage on the right abutment spillway steps
occurred in December 2015 as a result of over 300 h of continuous spilling from a new spillway gate. No cavitation damage had been
Downloaded from ascelibrary.org by Faizal Yusuf on 12/03/19. Copyright ASCE. For personal use only; all rights reserved.
observed on the original spillway constructed in the 1930s. Factors that contributed to the cavitation damage for this spillway are reviewed,
including the duration of continuous spill and the increased cavitation potential of both a smooth concrete surface compared to one that is
uniformly rough and sharp-edged steps compared to steps with rounded edges. Continued spilling through 2016 exacerbated concrete damage
to the spillway steps leading to an experimental interim measure of constructing a tetrahedron concrete wedge to reduce the magnitude of flow
separation and mitigate material loss in the location of greatest damage. Although damage to the steps continued to progress with ongoing
spilling, the concrete wedge was successful in reducing the extent of damage at that location. A numerical model was calibrated by comparing
simulated pressures to observed concrete damage allowing a long-term cavitation mitigation option to be designed with increased minimum
pressures of approximately 50% and greater at the spillway steps compared to the 2015 as-built geometry. DOI: 10.1061/(ASCE)HY.1943-
7900.0001671. © 2019 American Society of Civil Engineers.
channel, particularly when the outer gates are used. The spillway
chute terminates with a concrete apron at El. −10.13 m located
Ruskin Project downstream of old Bays 4 and 5, connecting the lowest benches
on either bank (Fig. 2).
Ruskin Dam is situated on the Stave River, about 65 km east The lower Stave River downstream of Ruskin Dam is an impor-
of Vancouver, Canada. The Stave River watershed is located at tant spawning area for anadromous fish, including chum, chinook,
the south end of the Coast Mountains at the edge of the Fraser coho and pink salmon, steelhead, and rainbow trout. To avoid
Lowland. The Ruskin project is the lower project in the Alouette- stranding fish and dewatering spawning habitat, sufficient dis-
Stave-Ruskin hydroelectric system and consists of Hayward Lake, charge must be released from the Ruskin facilities to maintain
Ruskin Dam, and Ruskin Powerhouse. Hayward Lake is a 6 km minimum tailwater elevations, i.e., El. 1.8 m from October 15 to
long reservoir impounded by Ruskin Dam with a surface area of November 30 and El. 1.7 m for the rest of the year. Tailwater el-
about 2.8 km2 at the maximum normal operating level of El. evations are a function of discharge from the Ruskin facilities and
42.91 m. Hayward Lake has a negligible contributing watershed, backwater from the Fraser River. During peak snowmelt periods
and its water supply comes almost entirely from the upstream Stave when the Fraser River is high (late May to mid-July), the minimum
Falls Dam spillway and generating station. The 105 MW Ruskin tailwater elevation requirements can generally be met without any
Powerhouse includes three generating units and is located below Ruskin discharge. For the rest of the year, the minimum tailwater
the left abutment of Ruskin Dam. Water discharged from the level is maintained by continuous Ruskin flow releases using a
Ruskin spillway and the powerhouse flows into the lower Stave combination of the powerhouse and spillway.
River. The confluence of the lower Stave River and the Fraser River In 2007, investigative work to upgrade and modernize the
is located about 3 km further downstream (Fig. 1). Ruskin spillway, abutments, and powerhouse was initiated.
Fig. 1. (Color) Location plan. (Image courtesy of BC Hydro, Burnaby, British Columbia.)
Fig. 2. (Color) Overview of Ruskin project: (a) seven-bay spillway before spillway upgrades. Old spillway bays are numbered from 1 to 7
(left to right looking at the photo). Old spillway radial gates and old piers are 10.1 and 2.1 m wide, respectively. Powerhouse access bridge
Downloaded from ascelibrary.org by Faizal Yusuf on 12/03/19. Copyright ASCE. For personal use only; all rights reserved.
and powerhouse shown in foreground; and (b) five-bay spillway and resurfaced concrete on spillway face and steps after upgrades. New spillway
bays are numbered from 1 to 5 (left to right looking at the photo). New spillway radial gates (SPOGs) and new piers are 13.5 and 3.5 m wide,
respectively. Photo date: April 2018. (Images courtesy of BC Hydro, Burnaby, British Columbia.)
Structural analysis indicated that the original spillway piers and The assessment included visual inspection, a delamination survey
gates were seismically deficient under the latest design criteria. by sounding using a chain-drag method combined with hammer
It was therefore decided to remove the original seven 10.1 m wide testing and impact echo testing, and drilling of several cores
radial gates and eight 2.1 m wide piers and replace them with five through the existing shotcrete and into the substrate concrete. The
new 13.5 m wide radial gates and six new 3.5 m wide piers. The cores were tested for density, compressive strength, volume of per-
powerhouse upgrade included replacement of the three generating meable voids, and absorption. The condition survey determined
units and associated ancillary equipment. In addition, the concrete that the shotcrete-faced spillway was still in reasonably good con-
on the spillway face and steps was partially resurfaced to extend the dition 35 years after the resurfacing work performed in 1973,
service life of the spillway surface for another 50 years. although there was some cracking, delaminations, and localized
A challenging aspect of this dam safety upgrade project was that areas of erosion and seepage through the shotcrete face that re-
the Ruskin facility had to be operational during construction to the quired repair. The stepped structure was in generally poor condition
extent required to maintain proper function and integrity of the with extensive delamination, erosion, and deterioration of the
Alouette-Stave-Ruskin hydroelectric system. This involved main- shotcrete overlay and near-surface deterioration in the substrate
taining flow passage during construction as the seven-bay spillway concrete. However, there were no signs of cavitation damage to
was converted to a five-bay spillway in phases with varying num- the right abutment concrete steps.
bers of old and new gates available during each phase concurrently Construction commenced in 2013 to upgrade the spillway gates
with the Ruskin Powerhouse upgrade that resulted in periodic and piers with the work progressing in phases. In Phase A of con-
unavailability of the generating units, which led to increased spill- struction, new spillway gates (SPOGs) 1 and 2 were installed, and
way operation. the spillway surface and right abutment spillway steps were resur-
Spillway capacity was significantly reduced during the multi- faced while old SPOGs 4 to 7 were available to pass required spill
year construction as a bulkhead was used to block flow through discharges. For the stepped structure on the right abutment, the
the spillway bays under construction. Consequently, a complex mesh reinforced shotcrete that was installed in 1973 was com-
strategy was developed to allow for passage of large floods without
pletely removed and replaced with cast-in-place concrete of
overtopping either Ruskin Dam or the upstream Stave Falls
200 mm nominal thickness and 15 mm diameter steel reinforcing
Dam and without incurring lost generation revenue at Stave Falls
bars and anchors, with a radius of 50 mm for step edges and 28-day
Generating Station (Finkenbine et al. 2017). In addition, continu-
minimum compressive strength of 40 MPa. The concrete minimum
ous spilling from available spillway gates was required to maintain
cover was 50 mm for the steel reinforcement. The resurfacing work
the required tailwater level while the powerhouse upgrade was in
for the right abutment steps was completed in 2015.
progress.
The new SPOGs 1 and 2 were operated for the first time on
December 2, 2015. A relatively steady discharge of 139 m3 =s was
Cavitation Damage Observed through May 2016 passed through each of SPOGs 1 and 2 between December 4–9,
A series of historical photos of the concrete spillway steps on the 2015, with the reservoir at El. 38.19 m. SPOG 2 was closed on
right abutment is shown in Fig. 3. The steps were originally con- December 9, while SPOG 1 releases continued until December
structed along with the concrete dam in 1930 with relatively sharp 15 but at a lower discharge of approximately 72 m3 =s. After
edges. A photo taken in 1967 does not show any cavitation damage SPOGs 1 and 2 were closed, concrete damage to the right abutment
to the spillway steps [Fig. 3(b)]. In 1973, a major program of com- spillway steps in the regions identified in Fig. 3(f) was observed.
pletely resurfacing the spillway face and spillway steps with Concrete damage was visible on the vertical face of Steps 11, 10, 9,
shotcrete was undertaken. The shotcrete was reinforced with 5 mm 8, and 7 just below the horizontal edge of the steps and close to the
diameter wire mesh, and the edges of the steps were rounded. spillway chute. The damage occurred in zones of flow separation
In 2008, a condition survey of the spillway surface and that are shown in Fig. 4.
steps was conducted to determine the location and nature of de- During Phase B of construction from December 2015 to January
fects and prepare designs for repair and rehabilitation to extend 2017, spillway upgrade work focussed on new SPOGs 3 and 4.
the service life of the spillway surface for another 50 years. Because the Ruskin powerhouse was being upgraded concurrently
(c) (d)
(e) (f)
Fig. 3. (Color) Right abutment spillway steps: (a) 1930s—original construction. Steps have sharp edges; (b) 1967—no cavitation damage to steps
below old spillway bays 1 and 2. Steps have sharp edges; (c) 1976—no cavitation damage to steps below old spillway bays 1 and 2. Steps have
rounded edges following shotcrete resurfacing in 1973; (d) 2008—no cavitation damage to steps below old spillway bays 1 and 2. Steps have rounded
edges; (e) December 2015—Steps have sharp edges following shotcrete resurfacing in 2015. New spillway gates (spillway bays 1 and 2) are shown;
and (f) December 15, 2015—Damage to concrete steps due to initial spill following shotcrete resurfacing in 2015. (Images courtesy of BC Hydro,
Burnaby, British Columbia.)
during this time and was periodically unavailable to pass flows Experimental Wedge to Mitigate Cavitation
coming into Hayward Lake to maintain required tailwater levels
for fisheries, new SPOGs 1 and 2 were operated for much longer In June 2016, repairs were performed for the damaged areas
durations than any other time in the dam’s history. Concrete dam- below SPOG 1 using dry-mix Target Superstick shotcrete with steel
age to the right abutment steps increased noticeably. By late May fiber reinforcement. Obviously, SPOG 1 was closed while the re-
2016, damage to the right abutment spillway steps had increased, as pairs were being performed, but SPOG 2 continued to operate.
shown in Fig. 5(a) with the largest zone of damage at Step 10, Fig. 5(b) shows the repairs, including a new 1.5 m high, 1.2 m wide,
which is below the central portion of SPOG 1. The approximate and 3 m long tetrahedron concrete wedge that was constructed
extent of concrete damage at Step 10 was nearly 1.5 m high, between Steps 9 and 10 as an interim experimental measure to
2.5 m long, and deep enough to expose the steel reinforcing bars. mitigate further cavitation damage. The tetrahedron wedge was
(a) (b)
(c) (d)
Fig. 5. (Color) Damage and repairs to right abutment steps between May 2016 and April 2017: (a) May 25, 2016—Damage to concrete steps;
(b) July 15, 2016—Repairs to concrete steps. Photo taken after ∼6 days of spilling following repairs in June 2016; (c) December 19, 2016—
Close-up view of concrete steps. Photo taken after ∼37 days of spilling following repairs in June 2016; and (d) April 3, 2017—Close-up view
of concrete steps. Photo taken after ∼124 days of spilling following repairs in June 2016. (Images courtesy of BC Hydro, Burnaby, British Columbia.)
and accuracy of simulated local pressures and velocities. Numerical numerical simulations were run for the 2015 as-constructed geom-
model results shown in Fig. 6(a) are based on a computational cell etry of the resurfaced spillway steps, and zones of simulated low
size of 125 mm for the spillway and right abutment concrete steps, pressure were reviewed in terms of cavitation potential. For nega-
while Fig. 6(b) shows results over a portion of the spillway steps for tive pressures less than the vapor pressure of water, cavitation
a finer cell size of 62.5 mm. would be expected.
The CFD modeled discharge of 288 m3 =s is within 3% of the Cavitation indices were calculated using Eq. (1) from numerical
rating curve discharge. General flow patterns of water cascading model results for low pressures approaching the vapor pressure of
over the spillway steps in the numerical model appear to be reason- water to gain insight into the potential for cavitation. The local min-
able, although spray and splashing are underestimated compared to imum pressure on the concrete surface was taken as the reference
prototype observations. Interestingly, the absence of noticeable pressure, and the modeled depth-averaged velocity near the loca-
spray and splashing in the numerical model simulation results al- tion of the reference pressure was used to calculate the cavitation
lows for more direct observation of flow patterns over the spillway indices. Empirical relationships between the cavitation index and
steps compared to what can be observed in the prototype during observed damage in spillways presented in Falvey (1990) are based
spill events. on mean flow velocities likely calculated from prismatic cross-
The physical processes of cavitation with surface material sections. Because the flow patterns over the Ruskin right abutment
erosion were not simulated with the CFD model. Instead, the spillway steps are quite complex with the interaction between flow
(a) (b)
Fig. 6. (Color) Numerical model simulated flow patterns for December 4–9, 2015 spill with a discharge of 139 m3 =s from each of spillway
bays 1 and 2: (a) CFD modelled area with a 125 mm computational cell size for the spillway and right abutment steps to show overall flow patterns;
and (b) refined CFD model mesh for steps below spillway bay 1 with a 62.5 mm computational cell size to investigate local hydraulics for assessment
of cavitation potential.
traveling down the spillway chute and water falling over the sides photo from late May 2016 [Fig. 5(a)]. This indicates that the incu-
of the steps, there is some uncertainty in the selection of appropriate bation period for cavitation was longer for Steps 13 and 12 com-
reference velocities. Because of the uncertainties involved, calcu- pared to the other steps below SPOG 1. There are no negative
lated cavitation indices presented in this paper are expressed to the pressures simulated for Steps 5 and 6 because they are directly
nearest 0.1. downstream of Pier 2 and would not be impacted by spillway re-
Modeled gauge pressures in pascals on the concrete surface of leases from either SPOG 1 or 2. No damage was observed on these
the right abutment spillway steps below the new SPOG 1 are shown steps [Fig. 3(f)].
in Fig. 7. These results, which represent the base case scenario for
assessment of concrete damage mitigation options for the right
abutment spillways steps, were produced with cubic computational
cells with sides of 62.5 mm. The pressure scale is intended to high-
light regions of minimum and maximum modeled pressure. The
lowest pressures occur on the vertical face of the spillway steps
near the interface with the spillway chute. Comparing Figs. 3(f)
and 7, we find very good agreement with the location of minimum
modeled pressures and observed concrete damage indicating excel-
lent numerical model calibration. The dark blue regions represent
pressures close to or below −100 kPa. Gauge pressures less than
about −100 kPa are below the vapor pressure of water and re-
present regions where cavitation would be expected.
The lowest modeled pressures occur on the vertical face of
Steps 10 and 7 with minimum values of approximately −177 and
−259 kPa, respectively. Because these pressures are well below the
vapor pressure of water, cavitation would occur as confirmed by the
observed concrete damage.
Step 11 has the highest minimum modeled pressure of about
−86 kPa for the steps below SPOG 1. The corresponding σ of
0.2 is at the threshold between minor and major damage. As shown
in Fig. 3(f), concrete damage did occur on the vertical face of Step
11 near the zone of negative pressures.
The numerical model simulates negative pressures at Steps 13
and 12, which would be expected to result in cavitation. Although Fig. 7. (Color) Modeled gauge pressures on right abutment steps based
on geometry just prior to the December 4–9, 2015 spill. Base case
the photo from December 15, 2015 [Fig. 3(f)] does not show any
scenario. Same spill discharge as shown in Fig. 6.
damage, there is significant damage to these steps, as shown in the
The spatial extent of pressures less than approximately −75 kPa Discussion
on the vertical face of the steps was also reviewed to give a general
indication of how large an area on the concrete surface could be According to Falvey (1983), the extent of spillway concrete dam-
affected by cavitation. For Step 10, the area of very low negative age due to cavitation depends on the duration of cavitating flow
pressures is approximately 70% smaller with Option R1 compared conditions, roughness and alignment of the boundary, and surface
to the base case. For Options R4 and R5, the areas of very low material strength. Testing of the cast in place concrete used for
negative pressures at Step 7 are about 85%–95% smaller relative resurfacing the right abutment spillway steps showed that it meets
or exceeds the specified 28-day compressive strength of 40 MPa
to the base case. The low-pressure region at Step 7 is 60% smaller
and field inspections showed that the resurfacing work was com-
for Option R5 compared to R4. A smaller region of very low neg-
pleted according to design specifications with no deficiencies that
ative pressures indicates that if cavitation does occur, it would be
would have contributed to the concrete damage. Therefore, the
quite localized.
remaining factors, the spill duration, and the effects of boundary
Tailrace flow patterns for the cavitation mitigation options were
roughness and alignment are discussed subsequently in terms of
also reviewed for Option R5 using the numerical model. The 6 m
the initial concrete damage that occurred to the Ruskin right abut-
long concrete prisms on the right abutment spillway steps would
ment spillway steps below SPOG 1 in December 2015.
reduce energy dissipation that could increase velocities in the tail-
race downstream of the spillway. Numerical model testing showed
that velocities near the bottom of the main channel close to the Duration
powerhouse access bridge were increased from around 1 m=s Prior to the recent upgrade work, the Ruskin spillway was used
for the base case scenario to about 2 m=s for Option R5. Overall, almost exclusively to pass floods, which generally consisted of less
tailrace flow patterns with Option R5 are similar as the base case than a few days of continuous spilling. Because discharge from a
scenario. Higher velocity flow near the bottom of the tailrace single generating unit is adequate to provide the required tailwater
channel indicates greater plunging of spill discharges, which can level, spilling to maintain downstream water levels was only re-
lead to increased total dissolved gas (TDG) values that are poten- quired in rare situations when all three units were forced out of
tially harmful to fish (Hughes et al. 2009). Therefore, some service. Therefore, the recent construction work to upgrade the
minor increase in TDG downstream of the spillway is possible for spillway and powerhouse has led to unprecedented extended
Option R5. operation of the Ruskin spillway gates.
Based on the results previously presented, Option R6 for the The original construction of the dam and spillway in 1930 had
right abutment spillway steps was developed by combining Options relatively sharp edges and a smooth surface for the right abutment
R1 and R5, as shown in Fig. 10. Three meter long tetrahedron concrete steps similar to conditions following the 2015 resurfacing.
wedges with 0 m offsets are applied to all steps below SPOG 1 Because there were no recorded observations of cavitation damage
except between Steps 6 and 7, where the polyhedron geometry prior to the resurfacing work in 1973, it is assumed that old SPOG 1
from Option R5 with an offset distance of 6 m is used to mitigate was rarely used in the first 43 years following initial construction.
the cavitation potential on the vertical face of Step 7. This location Spillway gate usage records from this period are not available for
had the largest spatial extent of pressures below −75 kPa for confirmation. It is likely that spills during that time were managed
Option R1 but showed a significant reduction in the area of low by preferentially using the central gates for more efficient energy
negative pressures for Option R5. Simulated pressure contours dissipation rather than the current operational strategy, which pla-
for Option R6 are shown in Fig. 10 and tabulated minimum ces greater emphasis on using the outer spill gates for improved
pressures and calculated cavitation indices are presented in total dissolved gas mitigation (Hughes et al. 2009). Anecdotal in-
Table 2. formation from a very large magnitude spill from November 1975
with a peak discharge of about 2,100 m3 =s noted that the old SPOG discharges below 80 m3 =s and full gate opening with releases
1 was not used during that event. around 139 m3 =s. The more severe damage below SPOG 1 that
Electronic records for Ruskin spillway operation are available was observed in late May 2016 prior to the interim measure involv-
from 1989 to present, while paper records for periods prior to ing wedge construction occurred after approximately 3,700 h of
1989 are very sparse with extended time periods noted as either operation (average discharge ¼ 109 m3 =s) with a maximum hourly
missing or destroyed by fire. Usage of the old SPOG 1 from 1989 discharge of 338 m3 =s.
to November 2015 and the new SPOG 1 from December 2015 and For comparison, Falvey (1990) presents data for several spill-
onwards are reviewed subsequently. ways showing observed cavitation damage after about 400 h of op-
The cumulative operation of the old SPOG 1 was approximately eration with flow velocities approaching or exceeding 40 m=s.
3,800 h from 1989 to November 2015, or less than 2% of the time, Flow velocities over the Ruskin right abutment spillway steps dur-
prior to the spillway upgrade work. Over 95% of the hours of op- ing the December 2015 spill were in the 15–20 m=s range based on
eration of the old SPOG 1 during this period was at a gate opening numerical model simulations. The 315 h of new SPOG 1 operation
of 2 m or less (∼3,580 h). Many of these spills were at very low in December 2015 that led to the initial cavitation damage is close
discharges during reservoir drawdown that were required during to the lower end of the range of observed cavitation damage noted
spillway crest block anchoring work or construction related to in Falvey (1990), although the flow velocities over the Ruskin right
the right abutment to address seismic deficiencies between 2006 abutment spillway steps are much lower than the data from Falvey
and 2013. For example, the old SPOG 1 was open 2 m for nearly (1990). The effects of boundary roughness and alignment are
800 h during June and July 2012 at a reservoir level below El. 36 m investigated in the following section to gain further insight.
resulting in an average discharge around 10 m3 =s. Such a low
discharge would not be expected to lead to cavitation damage. Boundary Roughness and Alignment
On the basis of available electronic records, it was noted that
Singular or isolated roughness elements can create sudden local
there were nine spill events for the old SPOG 1 longer than 3 days
changes in flow direction resulting in turbulence in the shear
with an average spill discharge greater than 70 m3 =s between 1989
zone and cavitation. However, for flow over a uniform roughness,
and November 2015. This is a somewhat arbitrary threshold, but it
there is a shear layer over the entire flow surface, which can
allows comparison of relatively large magnitude spill discharges
reduce the potential for cavitation damage. Falvey (1990) presents
that have greater potential to lead to cavitation. Table 3 summarizes
analysis showing 3.2 times reduction in cavitation damage poten-
these nine spill events for the old SPOG 1 as well as the spill in the
tial for the Glen Canyon Dam tunnel spillway with a very rough
first half of December 2015 for the new SPOG 1. surface (rugosity ¼ 1.5 mm) compared to a new, smooth surface
From December 2–15, 2015, the new SPOG 1 had an average (rugosity ¼ 0.015 mm). Nie (2001) summarizes experimental re-
discharge of 103 m3 =s for 315 h. The duration and volume of sults demonstrating that a uniformly rough surface upstream of
SPOG 1 spill releases in the first half of December 2015 were an irregularity reduces the pressure drop caused by the irregularity,
at least 30% and 50% greater, respectively, than any spill event which decreases cavitation damage potential.
from the old SPOG 1 in the previous 27 years corresponding to Misalignments of the boundary, such as offsets into or away
the period of reliable spillway gate records. from the flow, can create flow separation leading to a reduction
The new SPOG 1 was used for about 3,700 h between December in local pressures. For example, the misalignment of the spillway
2015 and May 2016, which is about 84% of the time. Therefore, the tunnel invert at Hoover Dam led to cavitation problems in the 1940s
new SPOG 1 was used for approximately the same duration in its (Pugh and Rhone 1988).
first 6 months of service compared to the previous 27 years of op- Prior to the 2015 resurfacing work, the Ruskin right abutment
eration for the old SPOG 1. There was also a significant increase in spillway steps had a rough surface with rounded edges [Fig. 3(d)].
the operation of larger gate openings, which generally led to higher The recent rehabilitation strategy involved replacing the shotcrete
spill discharges. Hours of operation for different gate opening installed in 1973 with cast-in-place concrete with a relatively
ranges were as follows between December 2015 and May 2016: smooth finish and sharper edges [Fig. 3(e)]. The specified radius
2,965 h for gate openings up to 2 m (81% of operation), 558 h for of the step edges constructed in 2015 is 50 mm, which creates much
2 m to 4 m gate opening (15% of operation), and 159 h for gate more flow separation compared to the previous step geometry.
openings greater than 6 m (4% of operation). Two tests were performed with the numerical model to inves-
The initial damage that was observed in mid-December 2015 tigate the effects of surface roughness and the step edge geometry
occurred after 315 h of spilling (average discharge ¼ 103 m3 =s), in terms of cavitation potential reduction for the base case scenario.
which included a mix of gate openings less than 2 m with Increasing the concrete surface roughness height in the numerical
reduction in cavitation potential due to these factors and provide early 2016 with 550 h of spilling greater than 180 m3 =s.
further insight into why cavitation damage was not observed prior As concrete damage progressed, an interim measure to mitigate
to the 2015 resurfacing work. the loss of material in the area of greatest damage was implemented
In this case, the rounded step edges have a much more signifi- with the construction of a tetrahedron concrete wedge to reduce the
cant role in reducing flow separation that leads to very low negative magnitude of flow separation. Although damage to the steps con-
pressures compared to the surface roughness. As discussed previ- tinued to progress with ongoing spilling, the concrete wedge was
ously, it would be very costly to implement a step edge radius of successful in reducing the extent of damage at that location.
300 mm while maintaining a minimum cover of 50 mm for the Numerical model results of negative pressure regions are con-
reinforcing steel bars used for the 2015 resurfacing. Also, it would sistent with observed initial concrete damage to the right abutment
be expensive to uniformly roughen the surface of the spillway steps spillway steps. Numerical analysis of the tetrahedron wedge in-
using either shotblasting, sandblasting, or water blasting in a con- stalled between Steps 9 and 10 as part of repairs in June 2016 shows
trolled manner. an increase in minimum pressures of nearly 50%. The cavitation
Bottom aerators can be used to protect against spillway cavita- index calculated based on simulation output indicates that damage
tion when high velocity flows in a chute lead to decreased pressures would still be expected, as was observed on the prototype after
and natural aeration does not satisfy the minimum air concentration several months of additional spilling following the repairs.
requirements to develop positive pressures [e.g., Rafi et al. (2012)]. Options using 3 m long concrete tetrahedron wedges with vari-
However, as discussed previously, the negative pressures that occur ous offset distances from the spillway chute were developed to mit-
in the flow separation zones at the Ruskin Dam right abutment igate concrete damage to the right abutment spillway steps with
spillway steps are due to the sharp transitions in boundary align- significantly higher simulated minimum pressures at all steps be-
ment rather than high velocity flows in a chute. Therefore, artificial low the new SPOG 1 compared to the base case scenario. In addi-
aeration would not be an effective means of cavitation mitigation in tion, the spatial extent of very low negative pressures that may lead
this case. to cavitation is substantially smaller. For example, the area of pres-
sures below −75 kPa at Step 10 is reduced by 70% for the recom-
mended mitigation option.
Summary and Conclusions With the recommended mitigation option, concrete damage due
to cavitation would only occur in the future as a result of an ex-
The Ruskin Dam spillway right abutment steps are intended to pro- tended duration of spilling or the cumulative effects of several large
tect against bank erosion and direct spill releases toward the center spill events. For reference, the 3 m long tetrahedron concrete wedge
of the downstream channel. Releases from spillway gate 1 flow installed between steps 9 and 10 in June 2016 was successful in
down the spillway chute and onto the right abutment spillway steps. preventing cavitation damage after at least 37 days of spilling at
A portion of the flow plunges over the sides of the steps with flow an average discharge of 93 m3 =s. This is a longer duration of spill-
separation occurring where the vertical face of the steps meet the ing than the cumulative operation of the Ruskin old SPOG 1 at
spillway chute (Fig. 4). This leads to local pressure drops and in- large spill events during the 20+ year period from the late 1980s
creased potential for cavitation. to 2010. Following the completion of the recent Ruskin Dam
Prior to recent phased construction work to resurface the spillway and powerhouse upgrade work, such extended periods
spillway concrete and replace all spillway piers and radial gates, of spilling are improbable.
the spillway had not shown signs of damage due to cavitation. Finally, the presented work shows the case in which observed
On the basis of the analysis described previously, the cavitation cavitation damage on a prototype spillway was used opportunisti-
potential was lower for the old spillway steps with a rougher sur- cally to calibrate a numerical model to help develop mitigation
face and rounded edges. It is postulated that relatively infrequent measures for future use while gaining valuable insight into the
spills from old SPOG 1 either did not result in cavitating conditions cavitation phenomenon.
or led to bubble formation associated with incipient cavitation that
did not exceed the incubation period.
The construction upgrades for the Ruskin Dam spillway and Data Availability Statement
powerhouse required an extended duration of spilling, which
was unprecedented in the dam’s 80+ year history based on available Some or all data, models, or code generated or used during the
records. The initial concrete damage to the right abutment spillway study are proprietary or confidential in nature and may only be pro-
steps that occurred in December 2015 is due to cavitation as a result vided with restrictions (e.g., anonymized data). Historical discharge
of over 300 h of continuous spilling from new SPOG 1 (peak spill data for the Ruskin spillway SPOG 1 discussed in this report may
Hughes, B., W. D. Clohan, and Z. Micovic. 2009. “Total dissolved gas as-
Notation sessment at the Ruskin dam.” Int. J. Hydropower Dams. 16 (4): 72–75.
Johnson, V. E., Jr. 1963. “Mechanics of cavitation.” J. Hydraulics Div.
The following symbols are used in this paper: 89 (3): 251–275.
Ax , Ay , and Az = fractional areas open to flow in subscript Kermani, E. F., G. A. Barani, and M. Ghaeini-Hessaroeyeh. 2013. “Inves-
direction; tigation of cavitation damage levels on spillways.” World Appl. Sci. J.
21 (1): 73–78. https://doi.org/10.5829/idosi.wasj.2013.21.1.2630.
G = gravitational force;
Lee, W., and J. A. Hoopes. 1996. “Prediction of cavitation damage for spill-
Po = reference pressure; ways.” J. Hydraul. Eng. 122 (9): 481–488. https://doi.org/10.1061
Pv = vapor pressure of water; /(ASCE)0733-9429(1996)122:9(481).
V f = fractional volume of fluid in each cell; Lesleighter, E. J. 1988. “Cavitation in hydraulic structures.” In Proc.,
V o = reference velocity; National Conf. on Hydraulic Engineering–Int. Symp. on Model-
f = viscous acceleration; Prototype Correlation of Hydraulic Structures. Reston, VA: ASCE..
Nie, M. 2001. “Cavitation prevention with roughened surface.” J. Hydraul.
p = average fluid pressure; Eng. 127 (10): 878–880. https://doi.org/10.1061/(ASCE)0733-9429
u, v, and w = velocities in x, y, and z directions; (2001)127:10(878).
ρ = fluid mass density; and Oukid, Y., V. Libaud, and C. Daux. 2015. “3D CFD modelling of spillways:
σ = cavitation index. Practical feedback on capabilities and challenges.” Int. J. Hydropower
Dams. 22 (6): 67–73.
Parsaie, A., S. Dehdar-Behbahani, and A. H. Haghiabi. 2016. “Numerical
References modelling of cavitation on spillway’s flip bucket.” Front. Struct. Civ.
Eng. 10 (4): 438–444. https://doi.org/10.1007/s11709-016-0337-y.
Amadour, A., M. Sanchez-Juny, and J. Dolz. 2009. “Developing flow Plesset, M. S., and R. E. Devine. 1966. “Effect of exposure time on cav-
region and pressure fluctuations on steeply sloping stepped spillways.” itation damage.” J. Basic Eng. 88 (4): 691–705. https://doi.org/10.1115
J. Hydraul. Eng. 135 (12): 1092–1100. https://doi.org/10.1061/(ASCE) /1.3645943.
HY.1943-7900.0000118. Pugh, C. A., and T. J. Rhone. 1988. “Cavitation in bureau of reclamation
Cassidy, J. J., and R. A. Elder. 1984. “Spillway of high dams.” Chap. 4. tunnel spillways.” In Proc., Int. Symp. on Hydraulics for High Dams,
In Developments in hydraulic engineering–Vol. 2. New York: Elsevier 645–652. Madrid, Spain: International Association of Hydro-
Applied Science Publishers. Environment Engineering and Research.
Falvey, H. T. 1983. “Prevention of cavitation on chutes and spillways.” Rafi, M., A. Ali, G. Qadir, and R. Ali. 2012. “Modeling the Mangla dam
In Proc., Conf. on Frontiers in Hydraulic Engineering, 432–437. spillway for cavitation and aerators optimization.” J. Water Resour.
Reston, VA: ASCE. Prot. 4 (12): 1051–1060. https://doi.org/10.4236/jwarp.2012.412121.
Falvey, H. T. 1990. Cavitation in chutes and spillways. Engineering mono- Rajasekhar, P., Y. V. G. Santhosh, and S. Soma Sekhar. 2014. “Physical and
graph 42. Denver, CO: US Bureau of Reclamation. numerical model studies on cavitation phenomenon—A study on
Finkenbine, J. F., Z. Micovic, and S. Vazinkhoo. 2017. “Construction flood Nagarjuna Sagar spillway.” Int. J. Recent Dev. Eng. Technol. 2 (1): 1–10.
management of extreme consequence Ruskin dam with reduced spill- Savage, B. M., and M. C. Johnson. 2001. “Flow over ogee spillway:
way capacity.” In Proc., CDA 2017 Annual Conf., Paper #6-03. Physical and numerical model case study.” J. Hydraul. Eng. 127 (8):
Toronto, ON, Canada: Canadian Dam Association. 640–649. https://doi.org/10.1061/(ASCE)0733-9429(2001)127:8(640).
Frizzell, K. W., and F. M. Renna. 2009. Cavitation potential of the Folsom Thiruvengadam, A., and H. S. Preiser. 1963. On testing materials for cav-
auxiliary stepped spillway. Denver, CO: US Bureau of Reclamation. itation damage resistance. Hydronautics Inc. Technical Report #233-3.
Frizzell, K. W., F. M. Renna, and J. Matos. 2013. “Cavitation potential of Fort Belvoir, VA: Defense Technical Information Center.
flow on stepped spillways.” J. Hydraul. Eng. 139 (6): 630–636. https:// Yakhot, V., and S. A. Orszag. 1986. “Renormalization group analysis of
doi.org/10.1061/(ASCE)HY.1943-7900.0000715. turbulence I: Basic theory.” J. Sci. Comput. 1 (1): 3–51. https://doi
Fujikawa, S., and T. Akamatsu. 1980. “On the mechanisms of cavitation .org/10.1007/BF01061452.
bubble collapse.” In Proc., International Association for Hydraulic Zhang, H., B. Han, X. G. Yu, and D. Y. Ju. 2013. “Numerical and exper-
Research 10th Symposium of Section for Hydraulic Machinery, Equip- imental studies of cavitation behavior in water-jet cavitation peening
ment and Cavitation, 91–102. Madrid, Spain: International Association processing.” Shock Vib. 20 (5): 895–905. https://doi.org/10.1155/2013
of Hydro-Environment Engineering and Research. /910613.