ML23109A084

NRC-2022-000160 - Resp 2 - Final, Agency Records Subject to the Request Are Enclosed, Part 7 of 7

ML23109A084
Person / Time
Issue date:	04/13/2023
From:	NRC/OCIO
To:
Shared Package
ML23109A075	List: ML23109A076 ML23109A077 ML23109A078 ML23109A079 ML23109A080 ML23109A082 ML23109A083 ML23109A084
References
NRC-2022-000160
Download: ML23109A084 (1)

Text

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l l l lis file CUI 1tai1I! GRl~GEII.

Review of Hypothetical Flooding at the Oconee Nuclear Station as a Result of a Potential Breach of Jocassee and Keowee Dams Working Draft December, 2010 Neil Coleman (ACRS staff) and Rex Wescott (NMSS staff)

View of hydroelectric powerhouse (right center) as seen from western crest of Jocassee Dam.

1 SEP~SITIV E SECURITY-RELATED IP~FORMATIOP~

CRITICAL ENERGY/ELECTRICAL IP~FRASTRUCTURE IP~FORMATIOP~

Offieial LJ3e Ol'lly Geeu~ity Relateel ll'lfo~l'l"lfttiol'l Disclaimer At the request of NRC's Office of Nuclear Reactor Regulation, the authors have independently reviewed the reports related to Jocassee Dam, Keowee Dam, and the Oconee Nuclear Station. This is a limited-scope analysis based largely on security-related documents that are not publically available. We applied approximation methods to get a feel for the behavior of the hydrologic system and the general effects of regional rainfall and flooding events. We have not used the HEC-RAS model or the 2-D hydrologic model developed by Duke Power's consultants because they were not available to us. We also were unable to obtain a runable version of the HEC-1 model used by Duke's consultants to evaluate the spillway inflow design flood. Our general conclusions are based on observations from an onsite visit to Jocassee Dam, Keowee Dam, and the Oconee Nuclear Station, and review of reports made available to us, as well as various technical references we obtained to better understand the regional geology and hydrology. Our conclusions about the low likelihood of runoff-induced overtopping and catastrophic dam failure are based on conservative approximations , the reservoir configuration (saddle dams), and undocketed information obtained during the site visit.

2

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l Executive Summary Issue to be evaluated: "How susceptible is the Oconee Nuclear Plant to surface water flooding, in the event of a hypothetical dam failure upstream at Jocassee Reservoir, leading to possible failure of the Keowee Dam at Oconee? Is this kind of analysis risk informed, and how does it relate to "reasonable ass urance" evaluations by NRC for site protection?"

• Based on our site visit to Jocassee and review of materials provided by Duke, the dam appears to be well-constructed, stable, and well-maintained.

• Jocassee Dam is licensed by the Federal Energy Regulatory Commission (FERC) and regularly monitored for safety conditions, including seepage and subsidence.

• The dam has operated safely for more than 40 years.

• Duke has performed hypothetical dam breach assessments to support FERC's emergency planning. T he latest assessment for NRC is conservative and can be used to analyze flood inundation at the Oconee site under " sunny day conditions.

• The risk of a seismically-induced "sunny-day failure for Jocassee Dam, which was low to start with, is substantially reduced. This is new info and a key observation.

• The lake level for analyzing a hypothetical sunny-day dam breach should be ::;; (bX7XF). (bX3):16 u.s.c § 8240-l(d) r )(7)(F). (bX3):16 U.S.C. § 8240-l(d) 1

• The 1991 and 1993 probable maximum flood (PMF) analyses by Duke were evaluated by FERC and found to be acceptable. The analyses evaluated the potential for overtopping. The FERC review included consideration of an antecedent storm and antecedent soil moisture.

• The PMF is not a risk-informed concept. The PMF is the theoretical maximum flood that can happen at a given place and represents an extreme upper bound. The PMF represents such a rare event that no realistic probability is assigned to it. For discussion purposes it could be considered less than 10-6/year and may be less than 10-s/year.

• For much of its operating history, Jocassee lake has reportedly remained below its normal full pool level. This fact should be conside red in hypothetical flooding and PRA studies.

• Independent analysis by NRR based on the FERC approved runoff model provides reasonable assurance that a PMF with antecedent storm can be safely contained and released in Jocassee Reservoir without overtopping of the dam.

0 r 7)(F), (b)O) '6 USC§ 82M(O) 3

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l r )()XF), C,)(7),16 U.S.C. § 82<0-l(d)

• Overtopping is a highly unlikely concern under standard hydrologic design scenarios for Jocassee Dam, but a " sunny day random failure is a higher probability safety issue that is being evaluated by Duke.

• The NRC staff's analys.is of generic failure rate for Jocassee Dam is based on six dam failures that are not representative of conditions at Jocassee. We are not aware of any failures of zoned rockfill darns that are truly comparable to the physical and operational conditions at Jocassee. Our independent analyses of the generic failure rate yield two numbers - one a 95% upper bound and the other an indication of reduced failure rate over time.

• Combined with the staff's generic failure rate (1 E-4 to SE-4 per dam year; also see previous bullet) for Jocassee Dam, an estimated joint probability (includes a PMF probability of ~1 E-5 to 1E-6/year) would be <1 E-8/year; therefore the highly speculative scenario of a PMF combined with random piping failure has no safety significance.

,e (b)(7)(F), (b)(3):16 U S.C. § 8240-l(d)

Based on our visit to Jocassee Dam and review of materials provided by Duke Energy, the dam appears to be a well-constructed, stable, and well-maintained engineering structure. It is regularly monitored for any significant change in seepage rates and embankment or foundation stability.

Past work to minimize and control seepage appears to have been successful.

Duke Energy has performed several dam break analyses for Jocassee Dam to support inundation CbX7)(F), CbX3):16 U.S.C. § 8240-l(d) e lake level for such an event should be :,; CbX7XF), CbX3) 16 us c § s240-1(d)

°ll!!l!ll!!!!ll!!!!IIIIJ!ll!l!!!l!lll!I~,...;,..:.;,:;.,;..:;:..:..;.:;.,,:.::..:-=..:....::..::..:.,.:

CbX7)(F). CbXJ):16 use§ s24o-l(d) he lake will always remain at or belo Cb) ). ithout human Intervent1on un ess in ows tot e lake are large, which would represent cipitation events and not "sunny day" conditions .

Duke Ener and its contractors are now evaluating a larger dam breach than in their earlier analysis.

CbX7)(F), (bX3):16 U.S C. § 8240-l(d)

. Most of the dam

~="1'1'"11"1T,";:l'l'l'l'<Pi'rn'""""ru'Fi'ivl';:-::;,.::;i~~"rom n:-::a;;:n;-;e:;-;m

~pn:1;::

nc;::--:a~r::;r::~=~r:::

aIT.

i le~d'.i°'.;d:T.a:;-;m

~s, most of which were 4

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l earthen dams, not higher-quality rockfill dams like Jocassee. Approximately 75% of the failed dams in the available databases are relatively small, having heights less than 15 m, and therefore not representative of large reservoirs like Jocassee Lake. Available predictive methods will therefore tend to overestimate breach widths and underestimate breach formation times when applied to Jocassee Dam.

The pool level of Lake Keowee will produce flooding tailwater effects downstream of Jocassee Dam that would limit the depth to which a hypothetical breach could form . Lake Keowee is a very large lake that seemingly could safely hold a large influx of water from an upstream dambreak. However, Keowee Lake is analogous to an hourglass, with a narrow waterway that restricts the flow rate between two large lake segments. This means that a large influx of water will preferentially raise the water level in the northern half of the lake, possibly leading to overtopping of Keowee Dam near the Oconee Nuclear Station In considering the likelihood of a "sunny-day" failure, reviewers should note that the seismic hazard for western South Carolina, where Jocassee Dam is located, has been reduced. The older 2002 mapping data suggested that for a probability of recurrence of 2% in 50 years, the peak ground acceleration at the dam site would be 0.197 g. The latest update from the USGS (Petersen et al.,

2008, Fig. 28) lowers this estimate to <0.1 g, which significantly reduces the risk of a seismic event inducing a piping failure in the dam face. Thus, the risk of a seismk:ally-induced "sunny-day failure for Jocassee Dam, which was low to start with, is substantially reduced. This is a key observation.

A second scenario should be considered for all large dams - this is the overtopping failure. The greatest catastrophe that can befall an earthen or rockfill dam is prolon ed overto in of the crest leadin to intense erosion and formation of a ra idly growing breach.

(bX7)(F), (bX3);16 U.S.C. § 8240- l(d) s Ima es o e ro a e xImum oo in ux o ocassee a e ave risen

.,1i"i'Trm-r~-.t:; years. In 1972 the PMF hydrograph for inflow to Jocassee Lake had a peak of

~245,000 ft3/s. The PMF peak that is now reported is more than twice that rate. It is important to recognize that the PMF is not a risk-informed concept based on flood frequency analysis. The PMF is the theoretical maximum flood that can happen at any place, and represents an extreme upper bound.

CbX7)(F), (bX3):16 U.S.C. § 8240-l(d) r X7)(F). (b)(3):16 U.S C § 8240-l(d) 5

Offieial LJ3e Ol'lly Geeu~ity Relateel ll'lfo~l'l"lfttiol'l CbX7)(F), (bX3):16 U.S.C. § 8240-l(d)

(bX7XF), (bX3):16 U.S.C. § 8240-l(d) 6

Offieial LJ3e Ol'lly Geeu~ity Relateel ll'lfo~l'l"lfttiol'l Table of Contents Introduction Features of Jocassee Dam Data on Jocassee Dam, Reservoir, and Watershed Recent Water Levels in Jocassee and Keov.ee Lakes Regional Geology Geotechnical Data (Dam Monitoring)

Independent Dam Breach Assessment for Jocassee Dam Uncertainty in Dam Breach Assessments Data on Large Historical Dam Breaches National Performance of Dams Program Insights About Probability of Dam Breach NRC Evaluation of Failure Rate for Jocassee Dam Independent Assessment of Random Failure Probability Scenario of Piping Failure during a PMF Probable Maximum Precipitation and Probable Maximum Flood Evaluation of Curve Number The significance of using a higher curve number Conclusion regarding Spillway Design Flood (PMF)

Upstream Reservoirs Seepage at Jocassee Dam Potential Role of Saddle Dikes to Eliminate the Possibility of Overtopping Failure of Jocassee Dam Notes about Keov.ee Lake and Dam Comments on Bureau of Reclamation Technical Letter Report (Feinberg, 2009)

Review Conclusions References Acknowedgment Appendix I - Dam Facts for South Carolina Appendix II - Potential for Dam Failure Appendix Ill Inflow Design Flood Probability Appendix IV Spillway Discharge Rating Curve 7

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l Introduction Issue to be evaluated: "How susceptible is the Oconee Nuclear Plant to surface water flooding, in the event of a hypothetical dam failure upstream at Jocassee Reservoir, leading to possible failure of the Keowee Dam at Oconee? Is this kind of analysis risk informed, and how does it relate to "reasonable assurance" evaluations by NRC for site protection?"

ANS VANS-2.8-1992 states that "Nuclear reactor safety from flooding needs to be ensured not only in floods from extreme precipitation but in floods from other causes as well. Surges from upstream dam failures from nonhydrologic causes constitute potential threats." ANSVANS-2.8-1992 includes, for the events that shall be considered to determine the controlling flood elevations (single or in combination), the failures of upstream dams from "hydrologic, seismic, or other causes."

FERC has required Duke to evaluate inundation areas that would result from dam failures. These are intended to be "worst-case" analyses to help guide and support emergency preparedness for evacuation plans, etc. These kinds of worst-case analyses are not derived from probabilistic risk assessments and are therefore not risk-informed or risk-based.

Duke Energy has performed several dam break analyses for Jocassee Dam to support inundation studies and emergency planning efforts by FERC. Their main anal sis is of ah othetical "sunn da "failure in which the dam would fail from The I ke level for such an eve--n -~ st!'"o

~u"'TT~ e-".la)(7)(F), (bX3) 16 u.s.c § 8240-l(d)

(bX7)(F). (b)(3):i .----------

The lake will always remain at or below 1--_ _, ithout human

~,=nr::e~rv~e~nT!1=-

on~ u=- n~e=-

ss,,....,.,,

m=ow = s-r.o,,....,_e lake are large, which would represent arge precipitation events and not "sunny day" conditions.

In terms of actual risk, it should be noted that, in addition to providing recreation and electrical power (with minimal CO2 release), Jocassee Dam provides flood protection for downstream dams and for th Jocassee Dam provides an engineering mechanism to regulate and safely re ease water

.__o=:w-:-:~nstream to Keowee Lake.

Historic Example The greatest natural catastrophe that can befall an earthen or rockfil l dam is prolonged overtopping of the dam crest, leading to intense erosion and formation of a rapidly growing breach. The main cause of such overtopping isl (bX7)(F), (b)(3)16 U.S C. § 8240- l(d)

\

A classic example of inadequate discharge capacity is the tragedy of the Johnstown Flood of 1889 when the South Fork dam failed. This was the largest earthen dam in the world at that time. It included two spillways excavated in bedrock. When the dam was no longer needed to supply water to canals it fell into disrepair. The stone culvert in the center of the dam failed in 1862, washing out a section of embankment. This section was later rebuilt by the South Fork Hunting &

Fishing Club to form a recreational lake. Unfortunately, three radical changes were made to the dam. First, the five 24-in. discharge pipes at the base of the dam were removed. A critical design feature for draining the reservoir was now lost. Second, the portion of the dam that washed away in 1862 was replaced, but there is no evidence that it was rebuilt with the low permeability clay core as in the original design. And third, the dam crest was lowered ~2 feet, probably to quickly and 8

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l cheaply get material for dam repairs. This final (and fatal) change lowered the dam crest below the elevation of the emergency spillway at the southwestern end of the dam . Now only the main spillway remained to discharge water, and it was partly obstructed by fish screens that collected debris during the flood and further reduced the discharge capacity. Water and debris surged into the reservoir during the heavy rainfall event of May 30-31, 1889, overwhelmed the partly obstructed spillway, overtopped a long segment of the dam crest, and eroded the downstream face until the dam center suddenly collapsed. A large breach rapidly for med. More than 2200 people died in the ensuing flood wave.

Features of Jocassee Dam Jocassee Dam and its engineered features are very different from the South Fork dam example.

Jocassee is a large, zoned, rockfill dam, with multiple discharge points, unlike the earthen embankment at South Fork. This means it is far less susceptible to catastrophic piping failures in which fine materials would be carried out of the dam core by transport in streaming groundwater, leading eventually to large-scale embankment collapse and breaching.

Although Duke Energy has analyzed a hypothetical "sunny day" failure for Jocassee Dam, the potential for overtopping of the Jocassee Dam is also important to address. Key to this question is whether there is adequate capacity to hold and safely discharge the influx from a hypothetical probable maximum precipitation and runoff into the reservoir. There are two engineered features for discharge from Jocassee Dam . 1 (b)(7XF), (b)(3)16 U.S.C. § 8240-l(d)

Participants in NRC staff visit to Jocassee Dam on June 15, 2009.

9

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l Data on Jocassee Dam, Reservoir, and Watershed (b)(7)(F). (bX3):16 U.S.C § 8240-l (d)

SRlfCEII (bX7)(F). CbX3):16 U.S.C. § 8240-l(d)

SRI/C~I I Gate structure for Jocassee spillway.

NRG management and staff conducted a site visit at the Oconee Nuclear Station (ONS) and Jocassee Dam on June 15, 2009. Observations from the site visit are documented in a trip report

[

(NRG , 2009). / (bX7)(F). (b)(3):16 U S.C. § 8240- l (d)

\_

10 S.iM~ITIVa ~iCURITY RliilwATaO l~JfiOR~4ATIO~J CRITICAL El~ERC,'r'iELEC"fRIC)BcL ll~FR,e.cS"fRUC"fURE ll~FORMMIOI~

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l The reoort bv FEI /2004) discusses ootential failure modes /PFM) of Jocassee Dam. I (bX7)(F), (b)(3):16 U.S.C. § 8240-l(d)

I SRh'CEII Spillway structure for Jocassee Reservoir (NRC photo).

11 sa, SITfVE SECURITY RBbc\TeD ~WOR~ 1ATlON CR::lTICAL E?mRGY,'£LECTR:ICJ'\£ ll'lFR::hSTR:UCTUR£ mror~M ATION

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l (bX3):16 USC § 8240-l(d), (bX7XF)

SRll'CEI I (bX7)(F), (b)(3):16 U.S.C. § 8240-l(d)

Recent Water Levels in Jocassee and Keowee Lakes As shown in the diagrams below, as of July 2009, the level of Jocassee Lake was ~5 ft below the normal full pond elevation of 1110 ft. The level in Lake Keowee was ~4 ft below its normal full pond elevation of 800 ft.

Lake Jocassee 3-Month Lake Lev el History ISelect Date Range ..:J Lake Message Updated: 611012009

--==;~~!'-"ll-~..oiill~~

100 1i;

... I - .J/' [ j 21 ~ !!ll-..,_~_,,-

JI S 1/4 ti ,,_"i!!l...-~~~":.!.-:'!":.':_':_':_:_~~llll-t s 1 0 0

~

90 90 - 90 80 85 75 80 ..)-_ _ _ _ _ __ _ _ _ _ _ _ _...,._ _ _ _ _ _ _ _ _ _...,._ ~ 75 80 70 - 70 6ll 60 60 55 ~ 55 50 50 45 T* rrt"1""n1-rrr-rTT1 rr--t1"Tn7 rtrn1 7 ~ ,~ , *T.--.r-n-rrttTrn-, 1-n-m-rrr1fn r't"!-M17"1TT1"T1-i---rnj1-n 45 04101109 0,116/09 0SA>1i09 05116109 05/31'°9 06115109 06/30109 Jocassee Lake water levels (relative to normal full pool) through mid-July, 2009.

Nonna/ Full Pond Elevation= 100.0 ft= 1110.1 ft (AMSL) 12 OEtdGITl'u'E SECURITY RELATED l ~JFORMATIO~J CR:FflC,e,;L EP ER8'1'1ELEOTRIOAL l~JFR,AiST RUGTURE l~lFORMATIG~I

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l Lake Jocassee 3-Month Lake Level History

~i ISelect Date Range .::.J 100 Lake l,lessage Updated 6/1012009

-i-""'l'll:-_..._:!"_-- ~:_"'!!!_!"_~_,,_-_'!::- '2 _.._~ '"°'!'~-= ~"'"")!! . .~ !K!5!'-"'""".!

~- ~l!""-!"~ lll""'-!f,--@

fl R!!!!!-"f!Lli

. .-!!l

-~

!!!!""""lff[l!lllllllt"'!!'P,~100

~

90 - 90 85 -85 75 __,_______

80 70

'U!"'_ _ ~-i------------.---'l-75 -80 t-70 65 - 65 60 55 50 -50 45 1 I I I I I I I I r I T I I 1 1 I I I I I I 117 1 I 11 I I I I I T l t"I I I I I I I 1 1 1 l I 11 I I n I T r T r n I n rl I I 1 1 17 T"f 1 T I 1 1 l I I l I I I r-I1I I f' n f'I I I l r r i-1 rIr l 11 I I~ 45 06,0U:)9 06116109 07101109 07116109 07/31,09 08115,09 08/30.,9 09/14,09 3-month water level history at Lake Jocassee during June to September, 2009.

Lake Keowee 3-Month Lake Level History ISelect Date Range .:J Lake Me ssage Updated* 6/1012009

~ ~

~ ~

~ ~

~ ~

86 ie-""!111!"""1. . . -<< 86 84 k I. k ,_a Q i! li. £ ill,.- . - - ~ ::

82 80 80 78 n~, ~, 1 ~, 1 ~, ,, ,- ,rrrr 11 -,--,--y,---.rrrrrrn7 m , ---rrr, ,. r1~ r1T rn ,..---,--r , 171 nTT"rrTTn r1 r r ,1 1 rTT1--rrn-rTrflrr1 , ,--rr, 78 0 4101109 0 4116109 05101109 05116109 0513 1109 06/15109 06130109 Lake Keowee water levels (relative to normal full pool).

Nonna/ Full Pond Elevation = 100.0 ft= 800.0 ft (AMSL)

[Webs ite for lake levels: http://www.duke-energy.com/lakes/levels.asp]

Regional Geology Jocassee Dam is on the border of Pickens and Oconee Counties, approximately 1O miles south of the North Carolina border. The dam site is in Piedmont terrain, close to the boundary between the Walhalla Thrust Sheet and the Chauga belt. The Chauga belt cons ists of metavolcanic and metaplutonic rocks. The Walhalla Thrust Sheet reportedly consists mostly of gneisses, which are metamorphic rocks. If these are indeed the primary materials from which the Jocassee Dam was 13

Offieial LJ3e Ol'lly Geeu~ity Relateel ll'lfo~l'l"lfttiol'l built, then the dam consists off strongly indurated rocks that are not susceptible to physical or chemical degradation over time by submersion beneath the water table. The rip rap on the dam and the outcrop beside the power station consist of biotite gneiss, a strong and stable metamorphic rock.

Generalized Geologic lVIap of South Carolina 2005 Rf\istd by

\\*illougbby, Howard~a od Xystrom, 2005 DESCRIPTION OF MAP UNITS Ofi&innl co1npUafion by COASTAL PLAIN lhybin and Xysrrom., 1997 QUATERNARY D HolOoene D Pleistocene TERTIARY D PIIOoene D Paleocene, Eocene, and Miocene CRETICEOUS D Upper Cretaceous TRIASSIC D THasslc basins SIGNIACANT STRUCTRAL BLUE RIDGE AND PIED~IONT FEATURES AZ Augusta zone D BlueRldge

[ ] [ ] Brevard zone D Chauga belt BRSZ Buzzards Roost shear zone D Walhalla thrust sheet CA Cross Anchor laul - Slxmlle thrust shfft RR Re8dy Rlwr lautzone D Laurens thrust stack IGWSHI Gold HIii / Sllwr H~I shear zone - Kings Mountain terrane KMSZ Kings Mountan &hear zone ,._____., '?".. Charlotte terrane BTSZ Boogertown shear zone 1ouu"°~'-- D carollna terrane (slate belt)

- Lowndesvlle 111ear zone D savann0h River teNMe S

Modoc shear zone Seneca thrust South Carolin!\

Dtparlmt nl of ::\:IUUl' Al R t.SOUttfS D Augusta terrane Gt-olog:kal Surny IGNEOUS ROCKS SIGNIRCANT WAVE-CUT SCARPS

~ - D Gabbro OS Orangeburg Scarp D Granite SS Surry Scarp

,...3" .

Regional geologic map (Willoughby et al., 2005)

A more detailed geologic map of the region where Jocassee and Keowee Dams are located is available from the State of South Carolina, i.e. Geologic Map of the Salem and Reid quadrangles, Oconee and Pickens Counties, S.C. (2007) (1 :24,000) by C. W. Clendenin, Jr. & J. M. Garihan; http://www.dnr.sc.gov/geology/publications.htm#charts.

The figure below was obtained during the site visit and shows that the quarry (above water in this image) was located adjacent to the dam, which minimized the cost of transporting materials from quarry to dam site. The intake structures are visible at bottom center of the image.

14

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l t-tistoric photograph of Jocassee Dam during construction (south is up).

Left panel : Biotite gneiss (metamorphic rock) that serves as rip rap at Joca ssee Dam. This rock type is very durable and not subject to breakdown over time under saturated conditions (NRC photo).

Right panel: Arrow points to lake embayment that Duke representatives identified as the submerged main quarry site used to obtain Jocassee Dam rock fill materials. Engineered feature at right is one of two Jocassee intake structures (NRC photo).

In considering the likelihood of a hypothetical seismically induced "sunny-day" failure, reviewers should note that the seismic hazard for far western South Carolina, where Jocassee Dam is located, has been reduced. The 2002 mapping data suggested that for a probability of recurrence of 2% in 50 years, the peak ground acceleration at the dam site would be 0.197 g. The latest update (Petersen et al. , 2008, Fig. 28) lowers this estimate to <0.1 g (see figure below), which significantly reduces the risk of a seismic event inducing a piping failure in the dam face. Thus, the 15

Offieial LJ3e Ol'lly Geeu~ity Relateel ll'lfo~l'l"lfttiol'l risk of a seismically-induced "sunny-day failure for Jocassee Dam, which was very low to start with, is now substantially reduced.

1.20 0.93 0.72 0.56 3S-N

• 0.43 0.33 0.26 S 0.20 A 0.15 0.12 g 0.09 0.07 0.06 0.04 km 0.03 0.02 a 500

,irw Map of 1-hertz spectral acceleration for 2-percent probability of exceedance in 50 years in the Central and Eastern U.S. in standard gravity {g).

From Figure 28 of Peterson et al. (2008).

Geotechnical Data A letter dated January 15, 1973 (to W. S. Lee from R. L. Dick) identifies the volumes of materials used for construction of Jocassee Dam. Of a total dam volume of 11 .03 million cubic yards, the rock volume was estimated at 8.76 million cubic yards. The volume estimate for the low-permeability core material was 1.57 million cubic yards. The remaining volume was earth material c lass ified as "stripping." The letter provides evidence that most of the volume of this dam is indeed rockfill, per design, with a low-permeability core. An example of quality control records for test results on the core material at Jocassee Dam is provided by fvbore (1972).

The 2008 dam surveillance and monitoring report for Jocassee (Duke Energy, 2009) provides current data on dam deformation surveys. Gradual and predictable deformation of dams is to be expected as they settle after construction, especially for very large dams. At Jocassee Dam, 2008 was the 1th year of deformation surveys. The graphs indicate continued gradual settlement of the nine crest survey monuments. Since 1993 the monuments have settled 0.05 to 0.25 ft on a linear trend of continued vertical displacement. Five of the monuments show a gradual settling rate of

~0.016 ft/year. Th e crest monuments also show a small amount of horizontal displacement in the downstream direction, varying from near zero to 0.3 ft. The "waveform" pattern seen in the last 15 16

Offieial LJ3e Ol'lly Geeu~ity Relateel ll'lfo~l'l"lfttiol'l years of data is consistent with the patterns of reservoir level. Reservoir levels in early 2008 were relative ly low, as was also the case in 2001 and 2002. The result was reduced water pressure on the dam face and reduced horizontal deformation in the downstream direction. Water levels for observation wells were lower, consistent with the lower water levels in the Jocassee Reservoir.

Duke Energy (2009) did not consider these lower water levels to indicate any adverse trends. But they advised reexamination during 2009 to see if trends continue to follow the reservoir level when the drought ends, when the groundwater levels should rise.

The Jocassee Dam and reservoir were recently inspected by the Federal Energy Regulatory Commission (FERG, 2008). The results of this inspection are summarized below:

Based on observations during the field inspection and review of data submitted by the licensee, the Jocassee project appears to be in satisfactory condition and adequately maintained ... .No conditions were found that should adversely affect the safety, permanence, and operational reliability of the project works. A review of the potential failure modes (PFM) developed on June 3, 2004 was performed and no changes or additions were necessary....The licensee appears to be complying with the license articles and Commission Regulations. There are no outstanding deficiencies from previous ARO [Atlanta Reg ional Office] or consultant's inspections. ARO letter to the licensee dated April 2, 2008 presented no new recommendations as a result of the operation inspection.

The project is well maintained, and the project components were found to be in good order. The project appeared to be in good condition, with no immediate dam safety concerns. Review of project records indicates that the project has generally performed well since construction, except for some remediation projects implemented to control/decrease abutment seepage.

The PFMA [Potential Failure Modes Analysis] Core Team reviewed the most current stability calculations/reports and concluded that they were made using methods in accordance with current engineering practice and indicate that all structu res are stable under the pertinent design loading conditions.

No signs of instability such as sliding, sloughing, cracking, or settling were noted. No unusual seepage or discoloration off seepage was noted. No potential slide areas along the reservoir slopes that could affect the safety of th e dam were observed. The licensee is controlling vegetation near the toe and along the groins. No significant movements of materials at the abutments/toes, dampness on the downstream slopes , uncontrolled vegetation, erosion, or significant cracks were observed at the earth structures. No slumping or beaching of the upstream materials or unprotected or exposed areas were observed.

The Core Team feels that the bi-weekly visual monitoring of the abutments should continue; however, the frequency of the monitoring of the weirs and pipes could be reduced to monthly, if desired by the owner. One exception is that the Core Team felt that the Parshall Flume on the right abutment should continue to be monitored bi-weekly.

Independent Dam Breach Assessment for Jocassee Dam Various analytical methods are available to estimate the magnitude of flows that could result from a hypothetical breach of the Jocassee Dam. The most recent technical paper on this topic, by Froehlich (2008), provides estimates of expected values of the final width and side slope of a trapezoidal dam breach, along with its formation time. He used data collected from 74 embankment dam failures to develop mathematical expressions based on the empirical data base.

17

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l Froehlich provides the following equations which are based on a regression analysis of his selected database of dam failures:

Where /3 = average width of final trapezoidal breach Ko = ov.ertopping coefficient used~ alculate average breach width 7

~ for overtopping failure (b)~ > for other failure modes)

Vw ~ oir volume at time of fa1 ure (m 3 )

Hb = maximum height of the final trapezoidal breach (m) t1 = time needed for complete breach development from the initial breakthrough at the crest to the end of significant lateral enlargement g = gravitational acceleration (9.8 m/s 2 )

For Jocassee Dam, the following numbers are obtained for the case of hypothetical failure without overtopping. These are expected values, and would normally be expressed as part of a range of values to consider uncertainty:

(bX3)*16USC § 7

I! s~.io-J(d>. <bX XE> (average width of final trapezoidal breach) ft ). (time needed for complete breach development)

'b 3 : 6 U Several things must be considered in evaluating these analytical results with Froehlich's method.

First, most of the failed dams in his database are earthen dams, not rockfill dams like Jocassee. Of the 74 dams, 56 were homogeneous earthen embankments, which have much lower integrity than width of the breach calculated above (i.e.,

SRlfGEII upper value rather than a mean value. HD

~z~\~gbg the rockfill Jocasse Dam, which also includes a low-permeability clay core. Therefore, the average A

!) more likely represents a conservative

9) used the following final breach I

width Breach bottom elev. 1~~;;~~* IDam crest = ,~~;m* I dimensions to analyze a hypothetical break of the Jocassee Dam: Top width=!~~;;<;'!* Bottom

=1~8;&~* I

~EN~ITl~1E SECURITY RELATED l~ffORMATIO~I GRITIGAL D~ERO>,','ELEOTRIOiS<L ltWRiS<STRUOTURE INFOR~b\"FIOI~

18

OHieial LJ3e Ol'lly Seeufity Relateel ll'ltOfl'l'IBtiol'l l.S'TIMATLD "1PAA11> f(U()A'OON AT OAM C£H'fUL,tl(

IISO DJU l'ICWE"- Ct:IWNN JOCASSa PU>FlD S'IOAAGC ~JECT JOCASSEE DAM BREACH PROFILE Figure 6.1 of HDR/DTA(2008), Appendix B.

(bX7)(F), (bX3)-J6 U.S C. § 8240-l(d)

~F\IJCEII (bX7)(F). (bX3):16 USC. § 8240-l(d) 9RlfGEII Laraer breach size considered by Duke Energy._

9Et~91Tl'o'E OEOURITV REb\TED l~ffORMATIO~l 19

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l (bX7XF), (bX3):16 U S.C. § 8240-l(d)

SRltCEII Final~each size considered bv Duke for flood orotection studies.

(bX7)(F), (b)(3)J6 U.S.C. § 8240-l (d)

SRlteEII (bX7XF), (bXJ):16 USC § 8240-l (d)

Peak Discharge from a Hypothetical Breach of Jocassee Dam Various empirical equations exist that can be used to obtain approximate estimates of the peak discharge rate from a hypothetical breach of Jocasse Dam. These methods use two values: the volume of water impounded by the dam and the height of the lake water column at the dam itself (or depth of breach). The results are given below.

Based on the breaching parameters of the Jocassee Dam, the peak outflow is computed. The Jocassee Dam breach peak outflow was computed by Duke using the HEC-RAS model. The peak 20 ORAlOtlL EtERO,'rCLEOffilOtlL ltFR,lt!3ffiUOfURE ltFORfnv'oArFIO~

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l outflow was determined to be t~:1~~?X 3 6 P u s.c. § per second (cfs}, which was greater than the empirically determined peak ou ows, using several available models, as listed in Table 2, below.

Table 2:

tvbdel Peak Outflow MacDonald & Langridge-Monopolis, 1984 ib)(7)(F), (6)(3) 16 U.S C. § 8240-l (d)

MacDonald & Langridge- Monooo lis 1984 Costa, 1985 Bureau of Reclamation, 1982 Evans, 1986 Based on a comparison with the values determined from empirical models (Table 2), , the staff determined that the HEC-RAS model results for peak outflow are conservative.

(bX7)(F), (6)(3):16 U S.C § 8240-l (d)

SAl~GEII (bX7)(F), (bX3)1 6 U.S C. § 8240-l(d)

\ The licensee used these methods in m_e_ir_e_s_rn_m_a_ rn_o _

n_- - - - - - - - - - - - - - - -

~R11ce11 (b)(7)(F), (bX3):1 6 U S.C. § 8240-l (d)

SDJSITIVE SECURITY RELATED l~JFOAM,A,TIO~J GRITIGAb e~u;;RcY/ebeGTRIG,Aib l~IFRAeTRbJGTbJRe IMFORl'i.4ATJON 21

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l (bX7)(F), (bX3)16 U.S C § 8240-l (d)

SRI/Cell

~ atelllte 1maae ot Jocassee Dam snow1na sut:>meraea auarry_i;1te and ernt:>avment area tie _ow tne aam wn1c.n__ru.s a restrictea outlet connection to tne rest ot Keowee LaKe Uncertainty in Dam Breach Assessments Wahl (1998) pointed out there is a need to achieve significant improvements in technology used to analyze embankment dam breach processes. The distinction between breach initiation time and breach formation time has not been clearly made in the literature or the available case study data.

Although breach initiation time is a critical parameter, there is little guidance in the literature for its prediction. Numerical dam breach models have the potential to predict breach initiation times, but are not widely used and are not based on observed breach erosion mechanisms. Breach parameter prediction equations based on analyses of dam failure case studies have significant uncertainty, and breach formation time is especially hard to predict. The case study database used to develop most existing breach parameter prediction equations contains a disproportionately small number of examples of high dams and large reservoirs compared to the population of embankment dams to which the equations are being applied. The primary mechanism of embankment dam failure is headcut erosion that initiates at the toe of the downstream slope and advances headward until it breaches the crest of the dam . This mechanism is not modeled in any of the available dam breach simulation models. Erosion models based on relations between hydraulic energy dissipation rate and erodibility indices based on excavability show promise for simulation of high energy erosion processes in widely variable materials. Recent research has improved the technology for modeling the stability of riprap and vegetated surfaces. Combining this technology with recent improvements in headcut erosion modeling may eventually yield better tools for determining if a dam will breach (Wahl, 1998).

22 eli)JSITl>.<E SECURITY RELATES l ffORMAT I O J CRITICAi.. li:t>lliiR'9¥,,liil..liiCTRICA L lt>IFRA 5TRI !CTI IR!ii l!>IFOR~4AJIO!>I

Offieial LJ3e Ol'lly Geeu~ity Relateel ll'lfo~l'l"lfttiol'l Wahl (2004) analyzed the uncertainty of numerous dam breach parameter and peak flow prediction methods. He found that four available methods for predicting breach width or volume of material eroded, from which breach width can be estimated, all had absolute mean prediction errors less than one-tenth of an order of magnitude, indicating that on average their predictions are on target.

The uncertainty bands were s,imilar (+/-0.3 to +/-0.4 log cycles) for all of the equations except the MacDonald and Langridge-tvbnopolis equation, which had an uncertainty of +/-0.82 log cycles.

Five methods for predicting failure time al l underpredict the failure time on average, by amounts ranging from about one-fifth to two-thirds of an order of magnitude (Wahl, 2004). This is consistent with the observation that these equations are designed to conservatively predict fast breaches, which will cause large peak outflows. The uncertainty bands on all of the failure time equations are very large, ranging from about +/-0.6 to +/-1 order of magnitude, with the Froehlich (1995a) equation having the smallest uncertainty. tvbst of the peak flow prediction equations tend to overpredict observed peak flows , with most of the "envelope" equations overpredicting by about two-thirds to three-quarters of an order of magnitude. The uncertainty bands on the peak f low prediction equations are about +/-0.5 to +/-1 order of magnitude, except the Froehlich (1995b) relation which has an uncertainty of +/-0.32 order of magnitude. The Froehlich equation has both the lowest prediction error and smallest uncertainty of all the peak flow prediction equations.

Froehlich (2008) wrote a paper titled "Embankment Dam Breach Parameters and their Uncertainties." He provides a table of 73 dam failures and develops empirical equations for dam breach parameters based on them . What needs to be recognized is that this database represents primarily a group of poorly designed or built dams. Of the 73 dams, 56 were of homogenous (random) earthfill design. Six other dams were also of homogeneous earthfill construction, but with the addition of a corewall. Five additional dams were built using zoned earthfill. Only five of the dams included significant amounts of rockfill. Cougar Creek was a homogeneous rockfill dam, the Oros dam in Brazil was of zoned earth and rockfill, the Lower Otay dam was of rockfill construction with a corewall, and the Lake Avalon dam was made of bot h earthfill and rockfill. The dam failures at Cougar Creek, Oros, and the Lower Otay were caused by overtopping. The Lake Avalon dam in New tv1exico reportedly failed in 1905 due to piping. Its breach height was 14.6 m , the breach was 130 m wide, and the embankment was 42.7 m wide.

Data on Large Historical Dam Breaches There is considerable confusion about how many rockfill dams have failed. In fact almost none have failed, and those that have were damaged or operated under unusual conditions. There are numerous inaccuracies in the available databases on dam failures that may mislead some reviewers about the incidence of failure. Examples are given below.

We have exam ined the database on dam breaches provided by Wahl (1998) , who summarized 108 dam breaks. Three of these involved sizeable dams for which the entire embankment was lost.

The Frias dam (Argentina), built in 1940, was described as a homogeneous rockfill dam 15 m high with a crest length of 62 m. Singh (1996) gives further details, noting that "At the dam s ite, sandstone and conglomerate beds varied considerably in thickness , both of which when saturated had constituents tending to soften and disintegrate under pressure." This dam overtopped by 0.9 m and the entire embankment was lost. Singh (1996) notes there were design problems related to flash floods and consequent erosion and sedimentation. Singh comments that " ... undermining of the downstream foundation and erosion of the fill were the principal causes for sudden collapse of the structure."

The Lower Otay Dam (California) was built in 1897 with a crest length of 172 m (Wahl, 1998).

This was a dumped rockfill and earthfill structure that had a steel plate embedded in cement mortar 23

Offieial LJ3e Ol'lly Geeu~ity Relateel ll'lfo~l'l"lfttiol'l to provide water tightness. As described by Singh (1996), a 9- inch rainfall caused rapid erosion, loosening of the rockfill, and eventual overtopping of the dam. The steel diaphragm was torn off

" ... and the remainder of the dam gave way like a pair of swinging gates. Within two and one-half hours, the reservoir got emptied. The foundation was shattered and seamy."

The Swift Dam ( Montana) was built in 1914 and had a crest length of 226 m (Wahl, 1998). This was a rockfill dam with a concrete slab facing the upstream slope and a compacted earthfill facing the downstream slope. Singh (1996) noted that, after its spillway capacity was exceeded, "the dam was overtopped and had given way, rapidly washing away the downstream rock and the compacted earthfill facing. The dam was eroded completely within minutes." Wahl (1998) provides data to suggest the breach had a "trapezoid" form with a top, bottom, and average width all equal to 225 m (which is not a trapezoid). However, Singh (1996, p. 97) gives much smaller breach dimensions of 21 m for the top width and 18.2 m for the bottom width. So there is uncertainty about the actual breach characteristics of this dam failure.

None of the three dams described above are analogous to the design, construction , foundation characteristk:s, and safety features of Jocassee Dam. But certainly these examples show why overtopping of a dam crest must be avoided. Additional examples of failed dams and the circumstances related to each are given below. These dams include the Oros Dam, Hellhole Dam ,

and Teton Dam.

Hell Hole Dam - Only one failed rockfill dam in the Froehlich (2008) database is analogous in construction method to Jocassee Dam in South Carolina. That was Hell Hole in California, which failed during construction in 1964. Froehlich (2008) and Wahl (1998) both list piping as the cause, which is surprising given its rockfill construction. This dam was 67 m high compared to Jocassee Dam at 117 m, was also built using zoned rockfill, and was built at around the same time. The dam construction was not yet completed, and that fact, along with heavy water inflow, caused the failure.

A major precipitation event occurred that produced more than 20 inches of rain over several days.

The water level in the lake rose and could not be lowered because the discharge tunnel that had been provided during construction was overwhelmed, its discharge capacity being exceeded. The outer rockfill portion of the dam was complete or nearly so. But the inner clay barrier was not yet at full height. The failure was due more to an overtopping mechanism rather than piping. The clay barrier was overtopped, not the rockfill section. Froehlich (2008) lists the breach formation time as only3/4 of an hour. From the time the clay barrier was overtopped and discharge began at the toe of the dam, the embankment resisted failure for 18 hours2.083333e-4 days <br />0.005 hours <br />2.97619e-5 weeks <br />6.849e-6 months <br /> even under these extreme conditions.

This period provided substantial time to warn people downstream of an impending failure.

Oros Dam, Brazil - As in the case of Hellhole Dam, this rockfill dam failed during construction.

The dam was planned to be 54 m high when completed, but had only reached a height of 35 m when a week of bad weather ensued, producing 26 inches of rainfall in 7 days of what was normally a relatively dry season. Unfortunately some dams have minimal discharge capacity when they're being built, and the spillways are not yet in play, and for both Oros and Hellhole dams the builders had bad luck in experiencing very heavy rains before completing the projects.

Teton Dam - This was a 305-foot high earthfill dam across the Teton River in Madison County, southeast Idaho, failed completely and released the contents of its reservoir at 11 :57 AM on June 5, 1976 (see http://www.geol.ucsb.edu/faculty/sylvester/Teton_Dam/narrative.html). The Teton Dam was designed as a multi-purpose facility that would provide irrigation water, flood protection, electrical power, and water-based recreation. It was built across a deep canyon on the Teton River in the watershed of the Snake River about 12 miles northeast of Rexburg in southeastern Idaho.

The dam rose ~305 feet above the stream bed and was ~3200 feet long at its crest. The reservoir 24

Offieial LJ3e Ol'lly Geeu~ity Relateel ll'lfo~l'l"lfttiol'l behind the dam was 17 miles long with 200,00 acre feet of active capacity and 270,780 feet of inactive acre feet capacity, and 470 acre-ft of dead storage, for a total capacity of 288,250 acre-ft.

Teton Dam Failure - June 5, 1976 (Image credit: Eunice Olson, St. Anthony, Idaho)

The Teton Dam failure was initiated by a large leak near the right (northwest) abutment of the dam, about 130 feet below the crest. The dam , designed by the U.S. Bureau of Reclamation, failed just as it was being completed and filled for the first time, and thus represents a juvenile fai lure of an earthen dam (not rockfill). The resulting flood inundated farmland and towns downstream with the eventual loss of 14 lives, directly or indirectly, and with a cost estimated to be nearly $1 billion.

Eyewitnesses noticed the first major leak between 7:30 and 8:AM, June 5, although two days earlier engineers at the dam observed small springs in the right abutment downstream from the toe of the dam. The main leak was flowing about 20-30 cfs from rock in the right abutment near the toe of the dam and above the abutment-embankment contact. The flow increased to 40-50 cfs by 9 AM. At about the same time, 2 cfs seepage issued from the rock in the right abutment, approximately 130 feet below the crest of the dam at the abutment-embankment contact. Between 9:30 and 10 AM, a wet spot developed on the downstream face of the dam, 15 to 20 feet out from the right abutment at about the same elevation as the seepage coming from the right abutment rock. This wet spot developed rapidly into seepage, and material soon began to slough, and erosion proceeded back into the dam embankment. The water quantity increased continually as the hole grew. Efforts to fill the increas ing hole in the embankment were futile during the fol lowing 2 to 2 1/2 hour period until failure. The dam breached at 11 :57 AM when the crest of the embankment fell into the enlarging hole and a wall of water surged through the opening. By 8 PM the flow of water through the breach had nearly stabilized. Downstream the channel was filled at least to a depth of 30 feet for a long distance. About 40 percent of the dam embankment was lost, and the powerhouse and warehouse structure were submerged completely in debris (see http://www.geol.ucsb.edu/faculty/sylvester/Teton_Dam/narrative.html).

25

OHieial LJ3e Ol'lly Geeufity Relateel ll'ltml'l'lf!ltiel'l Catastrophic Dam Failures in August 1975 in Zhumadian, China - During August 4-8, 1975, an extreme storm occurred in Henan Province, China (Xu et al., 2008). The maximum 5-day rainfall reached a record of 1,631 mm. Two large dams (Banqiao Dam and Shimandan Dam), two medium dams (Tiangang Dam and Zhugou Dam), and 58 small dams failed from overtopping in the storm event. The breach peak flow rate was as high as 78,100 m 3/s from the Banqiao reservoir and 30,000 m3/s from the Shimantan reservoir. The breaching of these dams caused an inundated area of 12,000 km 2 , a death toll of over 26,000, and economic loss of more than RMB10 billion. This paper introduces 26 dam failures in a smaller region of Zhumadian, i.e., Banqiao Dam, Zhugou Dam, and 24 small dams. The catastrophic event is described in detail especially for Banqiao Dam and Zhugou Dam. The causes and mechanisms of the principal failures are discussed, as well as the influence of an upstream dam failure on the downstream dams.

National Performance of Dams Program We have conducted searches for information about rockfill dams on the website of the National Performance of Dams Program (NPDP), maintained by the Dept. of Civil and Environmental Engineering at Stanford University. Formally launched in 1994, the NPDP is an effort to establish within the dam engineering and safety community the ability to learn from the in-service performance of dams, supporting improvements in dam design, operation, engineering, and public policy. The mission of NPDP is to be the leading technical resource for information on the performance of dams, supporting dam safety, engineering, and public policy.

For the period from 1940 to 2008, 24 "events" related to rockfill dams were documented in the NPDP database. Of these 24 events, 1Owere precautionary related to inspections in the aftermath of earthquakes in the region of the dams. No damage was found and the dams were reported to be in good condition.

~:~~~m 1 l!l!

NPDP Dam Incident Summary http :// npdp.s tanford.ed u/Da mD irecto ry/Da mlnc ide ntQ uery/Q uicklncide ntQuery.jsp Time Period: 1940 to 2008 lnc.ident Type: All Dam Type: Rockfill Dam Failure: All State: All NPDPID +

Dam Height (ft) r Dam Name Incident Date

-~

Incident Type Dam Failure Unknown. No info CA10268 5 ft Cascade 1943 Not Known available for this failure.

- I

~

Frenchman Yes. Runoff from melting Dam Inflow Flood - snow. A dike section was MTil0003 63 ft (Wahl and 4/15/1952 ove rtopped early morn ing Hydrologic Event April 15, 1952 . La ter t hat Froehlich lis t this day, dam breached .

as homogeneom I 26

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l earthen dam2 NOT rockfill}

Yes. Dam failed during construction. Overtoi;u;ied bl'.

CA00857 410 Hell Hole Not Known 100 feet [NO - see (zoned rockfill 1964 discuss ion in t e xt ft above] - washing out most dam) of the fill. Construction to continue.

I Inflow Flood - Yes. The dam f ailed CO00481 79 ft Skagway, CO 1965 Hydrologic Event - dur ing a flood i n 1965.

2 fa talities ~ittl~ intQ sl~slils}QI~.

Unknown - increased WA00300 425 Mud Mountain 7/15/1976 Not Known flows from dam drowned ft Dam 2 girls Monashka Creek Dam Yes. Three inches of ra in Kodiak Island, had fallen on October 16, Structural Failure; AK00073 47 ft AK 10/17/1978 1978 in the drainage (area),

Listed else where Erosion resulting in the dam's a<, earthen dam failure.

with s heet steel core Yes. Seismic event CAl0268 5 ft Cascade 1981 Earthquake ruptured the penstock.

South Truckee NV I0384 135 ft Meadows 9/12/1994 Earthquake No report of damage Effluent No damage. L1.ke can~

Inflow Flood -

WV033 19 25 ft Lake Floyd Dam 5/16/1996 within 3 inches of Hydrologic Event overtopp ing.

No breach. Sinkhole developed at shoreline on dam, with vortex noted in reservoir. No damage to the 5/29/1997 dam or appurtenant structures. No downst ream CO01241 20 ft Clear Lake Seepage;Piping damage.

1998 Sinkholes related to t he outlet works. Reservo ir drawndown, sinkholes repaired, and reservoir refilled.

No damage. Dam found CA00337 145 ft Crane Valley 6/8/1998 Earthquake to be in good condition.

CA004 I I 265 ft Wishon Main 6/8/1998 Earthquake No damage. Darn found 27

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l to be in good condition.

No damage. Dam found CA00446 82 ft Hillside 6/8/1998 Earthquake to be in good co ndition.

South Truckee No apparent damage to NVI0384 135 ft Meadows 10/30/1998 Earthquake da m Effluent I No. A sinkhole developed in the crest of the South Dam. Flow through bottom of sinkhole was collected by a screened Milner embankment drain in accordance with design.

1000223 73 ft 9/24/ 1998 Seepage;Piping The sinkhole was filled on Snake Rive r Sep. 24, 1998, with 20 cubic yards of bentonite

& 485 cubic yards of stockpiled fill. Reservoir lowered 1.4 feet as a precaution.

No damage. Dam found CA00455 45 ft Saddlebag Lake 6/8/1998 Earthquake in good condition.

No damage. Dam fo und C A00457 17 ft Rhinedo llar 6/8/1998 Earthquake in good condition.

No damage. Dam found CA00456 27 ft Tioga Lake Main 6/8/1998 Earthquake in good cond ition.

No damage. Da m found CA00448 70 ft Sabrina 6/8/1998 Earthquake in good condition.

No damage. Dam found CA00412 315ft Courtright 6/8/1998 Earthquake in good condition.

No failure. Dam overtopped due to Shotgun Creek Inflow Flood - sp illway a nd diversio n AK83009 40 ft 6/8/1999 Div Hydrologic Event being plugged with ice and snow. No da mage downstream Duke Powe r da m. 4 tainter gates and 2 fusep lug sections. Leak in NC00371 250 ft Nantahala (NC) 4/21/2001 Seepage Penstock that carries water to powerhouse. Does not constitute a dam safety issue, and was repaired.

WA00208 262 Culmback (near 11/2/2006 Inflow Flood - No damage. Incident was 28

Offieial LJ3e Ol'lly Geeu~ity Relateel ll'lfo~l'l"lfttiol'l ft Sultan, WA)

-- Hydrologic Event from extreme precip event and inflows to reservoirs.

It was reported that FERC licensed darm performed well.

Number of Events Found:

24 As noted above in the table, some discrepancies in data exist in the NPDP database, when it is compared to other databases. Examples: classification of Frenchman Dam and the failure mode of Hellhole Dam.

Another database of dam failures and incidents in the U.S. is maintained by the Association of State Dam Safety Officials. The database can be found online at this address:

http://www.damsafety.org/media/Documents/PRESS/US_Failureslncidents.pdf.

Insights About Probability of Dam Breach As reported by the Association of State Dam Safety Officials (http://www.damsafety.org/news/?p=412f29c8-3fd8-4529-b5c9-8d47364c1f3e), overtopping of a dam is often a precursor of dam failure. National statistics show that overtopping due to inadequate spillway design, debris blockage of spillways, or settlement of the dam crest account for

~34% of all U.S. dam failures. Foundation defects, including settlement and slope instability, cause

~30% of all dam failures. Another 20% of U.S. dam failures have been caused by piping (internal erosion caused by seepage). Seepage often occurs around hydraulic structures, such as pipes and spillways ; through animal burrows; around roots of woody vegetation; and through cracks in dams, dam appurtenances, and dam foundations. Other causes of dam failures include structural failure of the materials used in dam construction and inadequate maintenance.

Several efforts have been made to evaluate the probability of dam failures. These estimates are usually based on statistics of the past and present dam population. In the period from 1993 to 1999, 421 dam failures of varying degrees of severity have occurred (Hydro Review, 1999) which translates to an overall probability of fai lure in the order of 6 x 10*4 failures per dam year. See the detailed discussion in Appendix II. What needs to be understood about this number is that the population of dams largely consists of homogeneous earthen dams, a type of construction far inferior to the zoned rockfill (with clay core) design of Jocassee Dam. A rockfill dam with a clay core is far more resistant to piping failures, in which water works its way through the dam, mobilizes and transports fine grained materials out of the dam, leading to localized enhanced flow through the dam, loss of embankment, collapse, and ultimately dam breach, unless the piping is found early enough and rectified. So few rockfill dams have failed that it is not practicable to derive useful statistics based solely on that limited dataset. We recommend consideration of a dam failure rate for rockfill dams that is at least an order of magnitude lower than that determined for the overall dam population.

A further consideration that supports this approach is that, as noted above, one-third of dams fail by overtopping. tt overtopping failure can be reasonably eliminated, the failure rate for a given dam should be significantly less than the failure rate for the overall dam population. Another 29

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l consideration that is specific to Jocassee Dam - it has already operated safely for ~37 years through a variety of weather conditions and through water level fluctuations in response to pumped storage for hydroelectric purposes.

NRC Evaluation of Failure Rate for Jocassee Dam The NRG staff recently evaluated the generic failure rate for Jocassee Dam (NRG Division of Risk Assessment, 2009). This report relies on two databases to obtain information about the population of dams in the U.S. These are the National Inventory of Dams, maintained by the U.S. Army Corps of Engineers, and the National Performance of Dams Program, developed by Stanford Un iversity.

The staff considered two time spans in their analysis: 1900-2008 and 1940-2008. The following conclusions are reached based on the longer timeframe (NRG Division of Risk Assessment, 2009):

The staff estimated generic dam failure rates which it considers applicable to the Jocassee Dam. The point estimate fluctuates around a 2E-4/dam-year value. Given the nature of the data and the assumptions involved in narrowing the applicable failure events and subset of the U.S. dam population comparable to this specific dam, the staff performed a Bayesian analysis.

This yielded a range between 1.4E-4 per dam-year and 4.3E-4 per dam-year (5th-951h percentile) around a mean of 2.?E-4 per dam-year. In other words, while a value between 1E-4/dam-year and 5E-4/dam-year could be possible under the set of assumptions and criteria used above, the staff concludes that results lower than 1E-4 per dam-year are not presently justifiable based on available information.

The NRG evaluation overestimates the generic failure rate because it is based on incidents at dams that are not representative of Jocassee Dam. The table below shows 13 dams that were considered in the NRC staff's analysis. Of these, the six dams highlighted in blue were used to estimate the generic failure rate for Jocassee.

Tabl* 1. Initial List of dam failure events aoclicable to lhe J ocassee Dam Incident Completion Height Dam Name Yea, (Est)

Incident Type Dam Type Descriplioo From NPDP Database (Except Taum Sauk)

Year (fl)

OVertopped due to oveipumping of reservoir Independent anatysis TaumSauk 2005 1963 Overlopping Rocklin 94 Indicate<! several root causes<* o.. la<:k or monitoring, spillway)

DresserNo.4 Unknown Piping Eanh Rockfill cai.asttol)h<c failure that create<! a breacll 300 feet Wide In me levee Dam 1975 Olller 105 See case Ne Inflow Flood

• Skagway 1965 1925 Hyctotoglc Event Rocknll 79 Tho dam la/18<1 duMO a flood In 1965 Dam failed during construction Overtopped by 100 foot - washing Hell Hole l9e4 1954 Not Known Rocknll 410 out most or tne fill Penn F01est Piping COncrete Earth 151 Partial failure SinkhOle occurrea In upstream slope of dam.

1960 1960 Rocknll Frenchfoon 1952 1951 Inflow Flood* Rocknll 63 Runoff from melt,ng snow A dike section was overtoPl)ed early Dam Hy<i"ologic Event mcm1no April 15, 1952. Later tnat day, dam breached.

Ka,n Brothers 1949 UnknOwn Sel1fement Earth Rockftll 54 Fatfllre due to excessive setUement of ffl:l Reservoir Btov.out foiure unda, concrete SPIiiway weir sll\lcture dun110 period Lake Francis 1899 1899 Piping Earth Rockftll 79 or heavy spillway now Spltlway failure thought to be due 10 J)llllng In soft saturated foundation Foundation slide dunno constructl0<1 (a1 120 feel). Helgnt raised to Lafayette 1928 1928 Embankmefll Slide Earth Rockfill 132 170 feet irt 1932 Not sure if this is considered a failure.

Manltoo 19'24 1917 Seepage Earth Rockfill 123 Panial fa111Ke was d1s1megrat1no and converted Into or.,vei fit!

FailUre by piping mrough abutment; undermined by passage 0: water Lyman 1915 1912 Piping Eanh Rockfill 76 4 under cap or lava rock which nan1<ed dam and extended 1Janeat11 spJtw3y. Main pan. of dam uninlured.

Fooodalioo slide during construcilon (al 120 feel). Height raised to Lower Olay 1916 1897 Spltway Earth Rockfill 154 170 foot In 1932 Not sured this IS considered* failure.

Failure l>y piping through abutment; undermined by passage of water unda, cap of lava rock which nanl<ed dam and extended IJeneath Black Rock 1909 1908 Piping Earth Rockflll 70 spillway. Portlon of spiDway dropped 7 feet; some fill at SOlllh end Waslle<I out Main part ol dam uniniured Comments are provided below for each of the six highlighted dams used to compute the generic failure rate.

30

Offieial LJ3e Ol'lly Geeu~ity Relateel ll'lfo~l'l"lfttiol'l Taum Sauk- This was a pumped storage facility located in the St. Francois mountain region of Missouri ~ 90 miles south of St. Louis near Lesterville, Missouri. The pumped-storage hydroelectric plant, operated by the AmerenUE electric company, was designed to help meet peak power demands during the day. Electrical generators are powered by water flowing from a reservoir on Proffit Mountain into a lower reservoir on the East Fork of the Black River. The generators and turbines at river level are reversible, and at night the excess electricity on the power grid is used to pump water back to the mountaintop. The plant ceased operation when the upper reservoir catastrophically failed on December 14, 2005. The upper reservoir dam is being replaced with a roller-compacted concrete dam and construction is expected to be completed by May, 2010.

This dam failure has been extensively investigated and documented [FERG, 2006). This dam was a zoned rockfill dam like Jocassee that was built in the same decade. What should be recognized about this dam failure is that, just as TMI improved safety in the nuclear industry, Taum Sauk improved safety in the hydroelectric industry. We believe that the overtopping failure that occurred at Taum Sauk is extremely unlikely to occur at Jocassee Dam, given the multiple checks and safeguards that exist to detect and avert this scenario. The material below is from the executive summary of FERG (2006).

The Upper Reservoir of the Taum Sauk Pumped Storage Project 2277-fv'O was overtopped during the final minutes of the pumping cycle on the morning of December 14, 2005. Reservoir data indicate that pumping stopped at 5: 15 AM with the initial breach forming at approximately the same time. Once overtopping began, erosion started at the downstream toe of the 10-foot-high parapet wall. Erosion progressed below the parapet wall, likely causing instability and resulting in the initial loss of one or two parapet wall sections. Subsequent erosion and breach of the rockfill embankment formed a breach about 656 feet wide at the top of the rockfill dam and 496 feet at the base of the dam. The peak discharge from the breach was about 273,000 cfs which occurred within 1 O minutes of the initial breach. The complete evacuation of the reservoir occurred within 25 minutes.

The breach flows traveled down the west side of Proffit Mountain into the East Fork of the Black River. Flows destroyed the home of the Johnson's Shut-Ins State Park superintendent, flooded motorists on Highway N, significantly damaged the park, campground, and adjacent properties, and entered the Lower Taum Sauk Reservoir. The Lower Dam stored most of the releases and had a peak spillway discharge of approximately 1,600 cfs. This equates to about 1.1 feet over the spillway crest which is well within the capacity of the lower reservoir spillway. Upon leaving the Lower Dam area, flows proceeded downstream of the Black River to the town of Lesterville, fv'O, located about 3.5 miles downstream from the Lower Dam. The incremental rise in the river level at Lesterville was about two feet which remained within the banks of the river.

Post-breach inspections and evaluations revealed the following information:

1. The project had historically operated with a minimum of two feet of freeboard on the lowest section of the parapet wall. Following installation of a geomembrane liner in 2004, AmerenUE operated the project to fill the upper reservoir within one foot of the lowest section of the parapet wall. Post breach evidence shows the reservoir may have been routinely filled to within 0.25 foot of the lowest section of the parapet wall.

2. The December 14, 2005 breach was preceded by significant wave overtopping that occurred on September 25, 2005. Factors involved with this event were waves due to winds from the rem nants 31

Offieial LJ3e Ol'lly Geeu~ity Relateel ll'lfo~l'l"lfttiol'l of Hurricane Rita combined with a reservoir level pumped to within 0.4 foot of the top of the parapet wall.

3. On September 27, 2005, AmerenUE adjusted the reservoir control programming to account for the difference between the actual reservoir levels and the readings from the reservoir level instrumentation.

4. On October 3-4, 2005, AmerenUE personnel discovered that the conduit which housed the instrumentation for monitoring reservoir levels was not properly secured to the dam. Deterioration of the instrumentation tie-down allowed the conduits to move adversely impacting the reservoir level readings. The instrumentation readings showed reservoir levels that were lower than actual levels. /ls a safety measure, AmerenUE adjusted the reservoir level control programming to shut down the pumps when the instruments showed the reservoir levels were two feet lower than normal settings.

5. Two Warrick Conductivity Sensors were used as a safety system for shutting down the units in case of high water levels. The sensors would send a signal to shut down the units when they became wet. The sensors were physically relocated to a height that was higher than the lowest point on the parapet wall. Therefore, if th e Warrick Sensors were contacted by water, the Upper Dam would already be in an "overtopping" condition.

6. rvbdifications made to the reservoir control programming adversely affected how the signals from the Warrick Sensors were managed and reported. The modifications required that both sensors make contact with water to initiate shutdown. This removed a layer of redundancy to the safety system.

Dresser No. 4 Dam- [Source: National Dam Safety Program, Phase I Inspection Report, June '79; http://oai.dtic.mil/oai/oai?verb=getRecord&metadataPrefix=html&identifier= ADA106598]. Th e purpose of this dam was to impound tailings from a barite separation and beneficiation operation.

The dam completely encircles the impoundment except for a 300-foot wide breach at the north (up-stream) end where the structure failed in 1975. The dam cannot retain water in its present state because of this breach. This dam is classified as large. The U.S. Corps of Engineers has classified it as having a high downstream hazard potential to indicate that failure of this dam could threaten life and property. The damage zone, estimated by the U.S. Corps of Engineers, extends about 6 miles downstream of the dam. However, since the embankment cannot retain water, the overtopping potential of the embankment was not analyzed. Instead, studies were made to estimate the amount of flow that could occur in the diversion ditch (located to the west and south of the impoundment) under different flood conditions. It was calculated that the diversion ditch could pass a 100-year flood probably without significant erosion of the ditch or adjacent embankment toe.

However, it was estimated that the ditch cannot pass 50 percent of the PMP without significant erosion of the ditch and adjacent embankment materials. [Accession number: ADA106598]

Dresser No. 4 "Dam" is in reality a tailings impoundment, not a dam on a river like Jocassee.

Although we have not located detailed construction data for this impoundment, given its purpose, it was not a zoned rockfill dam and would not be analogous to Jocassee Dam .

Skagway Little information has been found on this dam breach. Two fatalities resulted from the breach. According to HMPC member input from Cripple Creek a dam failed on the southwest slopes of Pikes Peak in June of 1965 (Skagway Reservoir). Some reports refer to two fatalities - one says one person was killed. One house and a water transmission line to Cripp le Creek were damaged. Repairs were made to the dam by the Army Corp of Engineers.

32

Offieial LJ3e Ol'lly Geeu~ity Relateel ll'lfo~l'l"lfttiol'l J Skagway Reservoir JJ Rese rvoir JJ Teller JJ Big Bull Mountain JJ 8876 feet JJ 38.688°N JJ 105.051°W Manitou I Manitou Park Lake II Reservoir JI Teller JI Mount Deception JJ 7734 fee t 11 39.090°N JI 105.096°W Manitou, Colorado This concrete arch dam was 50 feet tall with a crest length of 300 feet. A portion of the dam failed in 1924 due to deterioratio n of the concrete.

Kern Brothers Reservoir - This was not a rockfill dam like Jocassee. The State of Oregon approved a permit in 1948 for a ranch to expand a small existing irrigation dam. The reservoir permit number is R-959. The reason for expanding the dam was not just to store more water but to raise the crest so that a bedrock-floored spillway would become available. It was to be of earthfill construction based on language in the permit. It is possible that some rocky material was included in the fill. Although the height of the dam was to be about 52 ft, the bottom width was only 80 ft, indicating the dam base filled a narrow canyon. The top width of the dam was 315 ft. The maximum amount of water to be stored under the permit was 1220 acre-feet. The water-covered area of the reservoir was only 104 acres, making this a relatively small lake but larger than most ranch irrigation lakes. The web page with info about the Kern Brothers Reservoir is:

http://apps2.wrd.state.or.us/apps/misc/vault/vault.aspx?Type=Permit&permit_char=R&permit_nbr=

959. There appears to be a newer dam at this site because the Oregon Water Resources Department lists it as an OWRD regulated dam without hydropower as beneficial use. A survey map dated 1963 is shown below. However, the information from the old permit appears the best available to describe the 1949 dam. The dam is located in Harney County on Dry Krumbo Creek.

We have not located details about the incident or failure of the 1949 dam, but the fact that it was an earthen dam to supply ranch irrigation confirms that it is not analogous to a zoned-rockfill dam like Jocassee, that is regulated by FERG.

T.30 S 31S. R.32S32 112E.W.M.

32-- -sort .-~ 33 Surveyed Sept. 16, 8 17, 1963 by T. Louer SEPT. 16, 196, by L . COLEBANK 33

Offieial LJ3e Ol'lly Geeu~ity Relateel ll'lfo~l'l"lfttiol'l Lower Otay- Listed as an overtopping failure by Wahl (1998), Froehlich (2008), and MacDonald and Langridge-Monopolis (1984). That is consistent with information on the City of San Diego website nfo/history.shtml, which states that in 1916: "A series of very heavy rain storms ... hit the county. The San Diego River floods Mission Valley from cliff to c liff cutting all highways to the north. Lower Otay Dam, built without a spillway, tops out [overtops) and bursts, flooding the Otay Valley."

The embankment was 172 m long and the average breach width was 133 m (Wahl, 1998). Most (77%) of the embankment was lost, which is a high percentage. Overtopping tends to cause larger breaches than other failure mechanisms. Nine inches of rain occurred before the dam breach (Singh, 1996). Various authors give different heights for this dam: 11 m (Singh, 2008); 41 m [134 ft] (Wahl, 1998); 41 m (MacDonald and Langridge-Monopolis, 1984); breach height 40 m (Froehlich, 2008). Thirty fataMies resulted from the dam breach.

Conclusion - Based on available information, none of the dams listed above and their failure scenarios are representative of Jocassee Dam and the "sunny day" random failure that is being evaluated for inundation studies. We consider the overtopping scenario as inapplicable to Jocassee and therefore Taum Sauk and Lower Otay dams are not relevant. Dresser #4 is a breached tailings embankment that was used to impound tailings from a mining operation, specifically separation of the mineral barite. Tailings dams are not built to the rigorous design and inspection specifications of a dam like Jocassee. There is no evidence that this was truly a "rockfill" dam. And the ranch dam known as the Kern Brothers Reservoir was an earthen dam, not a rockfill dam, based on language in the original permit from the state of Oregon. This also makes sense given the incident description in the NPDP database which says "Failure due to excess ive settlement of fill." The instability of the fill suggests that fine grained material was indeed the major component rather than rock.

Independent Assessment of Random Failure Probability In the discussion above we present evidence that disputes that fact that six dam failures in the U.S.

are representative of conditions for random failure at Jocassee Dam. On this basis we provide independent insights about the random failure probability of Jocassee Dam. If we assume that at least one failure may be representative, that would reduce the estimated random failure rate from 2.8E-4 to 4.6E-5 (for the period from 1900-2008) and from 2.88E-4/yr to 7.2E-5/yr (for the period from 1940-2008). These are point estimates. But we are not aware of any dam failure that was truly analogous to the physical and operational conditions at Jocassee, and conclude that no failures have yet occurred in this category. Therefore we can use a Poisson distribution to estimate an upper bound on the failure rate for the case of zero occurrences in 21,490 dam years. For the period from 1900-2008, we derive an upper 95% confidence bound on the fai lure rate of 1.4E-4/yr.

This failure rate is similar to that developed by the NRG staff, except here it is an extreme upper bound. This number is constrained by the number of calculated dam years for the indicated time span and would probably be significantly smaller given a longer time span. The fact that Jocassee Dam and reservoir have operated safely for ~37 years is a key observation with respect to future expectations of reliability. Decades of geotechnical monitoring of the dam have not revealed any anomalous settlement of the embankment structure, further enhancing confidence in its long-term stability.

We can gain further insights about the probabil ity of dam failure by reviewing the analysis by Baecher et al. (1980). We updated their 1980 calculation (see their page 454, upper left) through 2009. The updated numbers yield 15 failures (increased from their 12) out of 6190 dams (increased from their 3200), with an average operational period of 69/2 = 34.5 (updated from their 32/2 = 16). The annual failure rate per dam year (dams higher than 15 m) from Baecher et al.

34

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l (1980) (i.e., 2E-4) is updated to 7E-5. The population mainly consists (~85%) of homogeneous earthen dams, and so best represents that class of dam. It is suggested that the updated number better represents an upper failure probability for a continuously monitored, well built, zoned rockfill dam like Jocassee. Additional analysis of this problem may be possible to yield a better expected value or range (rather than just an upper failure probability). Until doing further analysis, a useful estimate is that an expected value for Jocassee may approach ~2E-51 or perhaps less. We speculate that the updated failure rate (7E-5) may continue to fall as the period of record continues to lengthen. Alternatively, it may level off as dams continue to age.

Baecher et al. (1980) state (page 454) that "Decrepitude, inadequate maintenance, and poor operational procedures seem to account for few failures of dams of moderate to large size." The Taum Sauk failure (which post-dates their paper) was caused by human error and poor maintenance (or poor design) of stage monitoring systems , and it should have had an uncontrolled spillway to prevent the failure scenario. This is the only recent failure of a large U.S. dam with documented construction quality like that of Jocassee.

Based on the above discussion and analysis, for Jocassee Dam we advise using a random failure frequency of ~2 to 7 x 10*5 per year for use in probabilistic risk assessments.

(b)(7)(F), (bX3)J6 U.S C. § 8240-l (d)

Probable Maximum Precipitation and Probable Maximum Flood When dealing with dam design or evaluation, the Probable Maximum Flood (PMF) is a familiar extreme flood. It is based on the assumption of the most severe hydrologic and meteorologic conditions considered to be reasonable at a site. The safety evaluation examines whether a dam can safely capture and release a probable maximum flood (PMF) resulting from a probable maximum precipitation (PMP). The PMP can be defined as, theoretically, the greatest depth of precipitation for a given duration that is physically possible over a given size storm area at a particular geographic location at a certain time of the year.

The National Weather Service has published Probable Maximum Precipitation reports on which PMF estimates are based. For other design projects, the U.S. Army Corps of Engineers Standard 35

Offieial LJ3e Ol'lly Geeu~ity Relateel ll'lfo~l'l"lfttiol'l Project Flood (SPF) is also a framiliar extreme flood. Neither the PMF nor SPF have any associated probability of exceedance associated with them.

NOAA's National Weather Service has provided PMP guidance and studies since the late 1940s at the request of various federal agencies. In recent years funding for this work has ceased and NOAA can no longer continue PMP activities. However, they provide copies of related documents on their website, many of which need updating. The Federal Advisory Committee on Water Information's Subcommittee on Hydrology is examining this issue.

(bX7)(F), (bX3):16 U.S.C. § 8240-l(d)

The Probable Maximum Precipitation (PMP) which serves as an input to the flood analysis was determined through Hydrometeorological Report Number 33. Hydrometeorological Report 33 was superseded by Hydrometeoro logical Report Nos 51 and 52, which resulted in higher design rainfall amounts in the Carolinas due to the revised transposition limits of the Yankee Town Florida Storm.

From a comparison of 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> (200 sq. mi.) all season PMP from HMR 33 to HMR 51 for the Jocassee Dam drainage area, an increase of about 10% can be determined.

Duke Power Company contracted with Law Environmental Inc. to provide an independent review of the PMF watershed model and methodology as part of a response to a request from the Federal Energy Regulatory Commission (FERG) in 1990. The report from LAW Engineering presented a computation using the U.S. Army HEC-1 computer program and included 4 upstream reservoirs and over 40 sub-basins including direct precipitation on the lake surface and runoff from areas adjacent to the lake shore. Th e model employed the Soil Conservation Service method for soil infiltration and unit hydrograph parameters. The resulting flood had a peak inflow of (b)(7)(F). (bX3)16 U.S.C. § 8240-l (d) 36

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l l<<A,..,,umped SlOr,c* °'"" No. 2SOJ PMF An*ly,b Flood Hydrographs at Jocassee Dam PIIFll~<l/tlt_,.,._,,._

...... J"""1 IWJDoME--Jod.wl dJ/ C,, f.i>

IIU IIU

'"'"° 1110

<toOOO 111*

"'°°°'

i\4eot0 i~*H0OOO l to 000

/ EJ un uu ll lt I

..... i 1

uot

'" 1000 1\00 loot 1'>00 1000 \\ff 1000 .-~

-110*

Inflow, Outflow, and Reservoir Level for the PMF at Jocassee Lake The infiltration in the LAW model was originally modeled using an SGS Curve number (CN) of 55.

In response to further questions from FERG, the PMF was recomputed using a Curve number of 60 for all sub-basins. This curve number was justified by the licensee on the basis of constant infiltration rates used by the Corps of Engineers for Hartwell Dam and North Georgia Dams. - - -....

3 :16 U.S.C. 8240-I d Evaluation of Curve Number (b)(7)(F), (bX3):16 U.S.C. § 8240-l (d)

The significance of using a higher curve number The most direct way to evaluate the effect of an increased runoff condition on the basin above Jocassee Dam is to run the HEC-1 model with the revised CN as an input. Do to the size of the input and lack of an electronic copy, re-running the model was considered to be a prohibitively labor intensive task. To approximate the effect of increased runoff, the inflow hydrograph to Jocassee Reservoir from the 111 miles upstream was taken from the outputs of the LAW 37

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l Engineering computations for curve numbers of 55 and 60. A comparison with NRR calculated runoff amounts was then made as follows:

Runoff over 111 sq. mi (in)

Curve number I Max 6 hour6.944444e-5 days <br />0.00167 hours <br />9.920635e-6 weeks <br />2.283e-6 months <br /> I Max 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> I Max 72 hour8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br /> (bX7)(F), (bX3)16 U.S C. § 8240- l(d)

Ratios were then computed from the above runoffs for comparison with the computed flood flows:

Ratios CN ratio I Max 6 hour6.944444e-5 days <br />0.00167 hours <br />9.920635e-6 weeks <br />2.283e-6 months <br /> I Max 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> I Max 72 hour8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br /> (b)(7)(F), (b)(3):16 U S.C. § 8240-l(d)

The peak flood flows from the LAW engineering model were then compared for the 111 sq mi area:

Inflow, cfs Curve number I Peak Discharge I Max6hour I Max 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> I (b)(7)(F), (bX3):16 USC. § 8240-l(d)

IO>X7l(F). 0,)(3) 16 u.s.c. § 82*0-l(d}

r XWJ, Q>X3}16 U.S C. § 82,o-l(d}

l(b)(7)(F), (b)(3):16 U.S.C. § 8240-l(d) l(bX7)(F), (bX3):16 U.S.C. § 8240-l(d) 1 The computation of the hydrographs from the LAW HEC runs ended prior to 72 hours8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br />.

38

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l

,..= (),)(3) " u.s.c. § .,,,.J(d)

The Effect of an Antecedent Storm ANS 2.8 (1992) Section 5.2.7.1 , states that a se uential storm could recede or follow the Probable Maximum Preci itation b 3 to 5 days.

(b)(7XF), (b)(3):16 U.S.C. § 8240-l(d) e e ect o an antece ent storm on soi mo1s ure 1s a e eva ua 10n o curve number.

Wind Wave Effects (bX7)(F). (bX3):16 U.S.C. § 8240-l(d)

Conclusion regarding Spillway Design Flood (PMF) here are some significant differences between the PMF study evaluated by the AEC and that performed by LAW Engineering for FERG review:

• The greater PMP values from HMR 51 over those determined from HMR 33

r X7)(F). (bX3):16 U.S.C. § 8240-l(d) 1

• Use of Lower infiltration rates in the LAW study.

r Xl:<>J. (),)(3) 16 U.S C. § 8240-Hd)

References for Probable Maximum Precipitation and Probable Maximum Flood LAW Environmental Inc, Jocassee Dam Hydrologic Analysis 1991 and 1993 U.S. Army, Coastal Engineering Research Center, Shore Protection Manual, 1977 NOAA and US Army, Probable Maximum Precipitation Estimates, United States East of the 105 th Meridian, Hydrometorological Report No 51 , June 1978 39

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l Upstream Reservoirs Other reservoirs in the waters hed upstream of Jocassee Dam include the Bad Creek facility, La ke Toxaway, Fairfield Lake, and Sapphire Lake (Law Environmental, 1991). /

(b)(7)(F), (bX3):16 U.S.C. § 8240-l(d)

I (bX7)(F), (bX3):16 U.S C § 8240-l(d)

(b)(7)(F), (b)(3)*16 USC. § 82-lo-l(d) 8Rh'CEII (b)(7)(F), (b)(3) 16 U.S.C. § 8240-l(d)

Bad Creek Pumped Storage Project: Diagram (top) and Map View 40 SD~SITIVE SECURITY RELATED l~ffORMATIOPJ CRITICAL DJERGYfELEGTRIGAL l~JFRAST RUGTURI!! l~li;GRMAT IQ~I

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l (b)(7)(F), (bX3):16 U.S.C. § 8240-l(d)

(bX7)(F), (b)(3):16 U S.C § 8240-l (d)

(bX7)(F), (bX3):16 U.S.C. § 82-lo-l(d)

'.b)(7)(F), (bX3):16 U S.C. § 8240- l(d) 41

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l (b)(7)(F), (b)(3):16 USC § 8240-l(d)

Overtonnina failures are the most hazardous for any embankment dam and must be avoided.\

(b)(7)(F), (b)(3):16 U.S.C. § 8240-l(d) e~ltCliill I

SENSITl~E SECURITY RELATED l~ffOAMATIO~~

CRITICAL D KAGWELEGTRIGAL INFR,0.6TRlt,lGTY~~ l~Ji;OR~ 4ATIO~I 42

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l (b)(7)(F), (b)(3):16 U .S.C. § 8240-l(d)

(bX7XF). (b)(3):16 U.S C. § 8240- l(d)

(b)(7)(F). (b)(3):16 U.S.C. § 8240-l(d)

Map (above) and aerial (below) views of Jocassee Dam, its saddle dikes, and spillway.

43

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l Saddle dikes at Jocassee Lake are protected from wave action by rip rap on sides facing lake. Left: Saddle Dike# 2. Middle: rip rap on lake side of Dike# 1. Right: downstream side of Dike no. 1 has grass but minimal rip rap. Rock covered drain is visible along toe of dam (bottom 3 images are NRC photos}.

(bX7)(F), (bX3):16 U.S.C. § 8240-l(d) 44

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l (bX7)(F), (bXJ):16 U.S.C. § 8240-l(d)

Topographic map, illustrating that overtopping of saddle dike #1 would initiate flow in the same stream valley that drains the Jocassee spillway. This stream empties into Lake Keo\We a short distance to the east.

(bX7)(F), (bX3)16 U S.C. § 82-lo-l(d)

(b)(7)(F), (bXJ):16 U.S.C. § 8240-l(d) 45

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l r )(l)(F), (b)(3) 16 U S. C § 82,o-l (d)

Notes about Keowee Lake and Dam During the afternoon of June 15, 2009, the NRC review team visited the Keowee dam and its appurtenant structures, and found them to be in good condition. Lake Keowee was constructed in the late 1960's primarily to provide cooling water for the Oconee Nuclear Station. Water from Lake Keowee never comes in direct contact with radioactive material, but rather supplies a secondary flow path, which circulates around the primary system to cool the steam back into water for re-heating. This water also removes heat from the reactor itself to prevent the unit from overheating.

Colder water is brought in to the reactor building from the bottom of Lake Keowee. After absorbing heat from the three reactors at the Oconee Nuclear Station, the water is warmer and therefore less dense. It is therefore released at surface levels of the lake, which are usually warmer than bottom waters, to minimize temperature differences in the lake water and any subsequent threat to aquatic organisms in the lake.

(bX7)(F), (bX3) I 6 U .S C § 8240-l ( d)

(bX7)(F), (bX3) 16 U.S.C. § 8240-l(d) eR.ltCl!ill Aerial photo of Oconee Nuclear Station and Keowee Dam. Note narrow waterway at upper left (marked as "Hwy 130/183") which connects the two large parts of Keowee Lake.

46 9Et491=t"P1,'E OEOURIT'1'11~l::MED IP4FORftJ1'At"flOt4 ORITI081L D4EROV,ELEGfRIOi'td... IP4FRACTRU0R:JRE INFORPAAl:i~~

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l Keo~e Dam spillway with four radial gates (left), and view of outflow channel (right)

(N~C photos).

Comments on Bureau of Reclamation Technical Letter Report (Feinberg, 2009)

We are in general agreement with the comments provided by the Bureau of Reclamation (Feinberg ,

2009), except for the following:

(1) A recommendation that NRC revise its standards to include a provision to require 2D hydrologic modeling where appropriate - we note that it would be very difficult to generically describe circumstances and scenarios where 2-D modeling would be required. If NRR chooses to require 2-D modeling for some circumstances, then licensees may feel compelled to use 2-D methods even if 1-D modeling may be adequate for their situation.

(b)(7XF), (b)(3) 16 U.S.C. § 8240-l(d)

,....r"'\ ,__

I Review Conclusions

• Based on the NRC staff visit to Jocassee Dam and review of materials provided by Duke Energy, the dam appears to be a well-constructed, relatively stable, well-maintained and closely monitored engineered structure. Past work to minimize and control seepage appears to have been successful.

• Duke Energy has performed several dam break analyses for Jocassee Dam to support inundation studies and emergency planning efforts by FERG. Their main anal sis is of a hypothetical "sunny day" failure in which the dam would fai I (bX7)CF). (bX3): 16 u.s c. § 824°-l(d) 47 ee~Je lTl'.'e eeC61RITY RebATer;l l~JlaGRP.4ATIG~J GAITIGAL ENEACWeLEGTRICAL IMlaR,'\eT~UCTU~li. l~lF0~~4A TIO~I

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiOl'l (b)(, ):16 The lake level for such an event should be ~<~~~x7; (bX3) 16 U SC § 8240-l(d), (b)(7)(F) 8 40 Cb"'"

X3'""

):""'

!6""" u,...s ....

c """§....

82,...

4o - (d-), ....X7...,

(b.... )(E

.,..)_ _ _ _ _ _ _ _ _ _ __

(b)(7)(F), (bX3)-16 U .S C. § 8240-l(d)

• fvbst dam breach predictive methods are based on an empirical database of failed dams, most of which were earthen dams, not high-quality rockfill dams like Jocassee. Available predictive methods will tend to overestimate breach widths and underestimate breach formation times.

• We are not aware of any large, well-designed, zoned rockfill dams, like Jocassee Dam, that have failed and lost their entire embankment.

• The normal pool level of Lake Keowee will produce flooding tailwater effects downstream of Jocassee Dam that would limit the depth to which a hypothetical breach could form (i.e.,

would limit the breach base elevation to ~800 ft)

• Reviewers should note that the seismic hazard for western South Carolina, where Jocassee Dam is located, has been reduced by about half. Thus, the risk of a seismically-induced "sunny-day" failure for Jocassee Dam, whch was low to start with, is substantially reduced.

This is a key observation .

(b)(7)(F), (b)(3):16 U.S.C . § 8240-l(d) s 1ma es o e m ux o ocassee .

""="

n ....,....,,.....-:;t~e ~ MF hydrograph for inflow to Jocassee had a peak of ~245,000 ft3/s. The PMF peak that is now reported is more than twice that rate. It is important to recognize that the PMF is not a risk-informed concept based on PRA results. The PMF is the theoretical maximum flood that can happen and represents an extreme upper bound.

l(b)(7XF), (b)(3)16 U s.c. § 8240-l(d)

• We are in general agreement with the comments provided by the Bureau of Rec lamation (Feinberg, 2009), except for the two points discussed in the above report, regarding (1) a recommendation that NRC revise its standards to include a provision to require 2D h drolo ic modelin

• and 2 (bX7)(F), (bX3):16 U.S.C. § 8240-J(d)

This L------,

scenario wou

. ,. . ,.- --,-----------.,........,---.,...,....-,--....,....,---~

e over y conservative as 1scussed above, and is inconsistent with the specific rocommendation made by Feinberg (2009) for breach parameters.

48

Offieial LJ3e Ol'lly Geeu~ity Relateel ll'lfo~l'l"lfttiol'l

• Regarding the key role of Jocassee Dam in the region, in addition to providing recreation and electrical power (with minimal CO2 release) , Jocassee Dam provides flood protection for downstream dams and for the Oconee Nuclear Station.

References Bowles, D.S., L.R. Anderson, T.F. Glover, and S.S. Chauhan (2003). Dam safety decision-making:

Combining engineering assessments with risk information. In Proceedings of the 2003 USSD Annual Lecture, Charleston, SC. [Online at: http://uwrl.usu.edu/people/faculty/DSB/ussd2003.pdf]

Bureau of Reclamation, 1987. Guidelines for Using Fuse plug Embankments in Auxili ary Spillways.

ACER Technical Memorandum No. 10, U.S. Dept. of the Interior. July 1987.

Chauhan, S. S., D. S. Bowles, and L. R. Anderson, 2004. Do current breach parameter estimation techniques provide reasonable estimates for use in breach modeling? Dam Safety 2004, ASDSO 2004 Annual Conference, Phoenix, Arizona. [Online at:

http://uwrl.usu.edu/people/faculty/ DSB/breachparameters.pdf]

Clendenin, C. W . and J. M. Garihan, 2007. Geologic Map of the Salem and Reid quadrangles, Oconee and Pickens Counties, S.C. (2007) (1 :24,000). Geologic publication of South Carolina's Dept. of Natural Resources.

Cochran, 1972. Letter from A. L. Cochran (LBC&W Associates of South Carolina) to L. G. Hulman (Atomic Energy Commission), Nov. 15, 1972 [letter about hydrologic engineering studies pertaining to the Keowee-Toxaway development of Duke Power].

Costa, J. E., 1985, Floods from Dam Failures, U.S. Geological Survey Open-File Report 85-560, Denver, Colorado, 54 p.

NRC Division of Risk Assessment, 2009. Generic failure rate evaluation for Jocassee Dam (Sensitive Information document - official use only).

Duke Energy, 2009. Dam Surveillance and Monitoring Report - 2008. Duke Energy Carol inas, LLC, April 1, 2009.

FEI, 2004. Potential Failure fvbdes Analysis - Jocassee Development, Keowee-Toxaway Project.

FERC Project No. 2503-SC. Prepared by Findlay Engineering, Inc. for: Duke Power, Charlotte, NC, December, 2004.

Feinberg, B. D., 2009. Jocassee Dam, South Carolina Inundation Study Review Comments.

Technical Letter Report dated July 6, 2009, Technical Service Center, Denver, CO, Flood Hydrology and Emergency Management Group, 86-68250, Bureau of Reclamation.

FERC, 2006. Taum Sauk Pumped Storage Project (No. P-2277), Dam Breach Incident FERC Staff Report, April 28, 2006, available online at:

http://www.fere.gov/industries/hyd ropower/safety/projects/tau m -s au k/s taff-rpt.as p.

FERC, 2008. Dam Safety Inspection Report. Federal Energy Regulatory Commission, Division of Dam Safety and Inspections. Project name: Jocassee Development of Keowee-Toxaway Project.

Submitted by W . H. Duke, June 9, 2008.

49

Offieial LJ3e Ol'lly Geeu~ity Relateel ll'lfo~l'l"lfttiol'l Froehlich, D. C., 2008. Embankment Dam Breach Parameters and Their Uncertainties , Journal of Hydraulic Engineering, Vol. 134, No. 12, Dec. 1.

Froehlich, D. C., 1995a. Embankment dam breach parameters revisited. Water Resources Engineering, Proc. 1995 ASCE Cont. on Water Resources Engineering, New York, 887-891.

Froehlich, D. C., 1995b. Peak outflow from breached embankment dam. J. Water Resour. Plan.

Manage. Div., Am. Soc. Civ. Eng., 121 (1 ), 90-97.

HDR/DTA, 2008. Oconee Nuclear Station - Jocassee-Keowee Dam Breach lvlodel Report.

Prepared for Duke Energy by HOR Engineering, Inc. of the Carolinas, 579 pp.

Hydro Review, Dam Safety; Vol XVIII, No. 5; pp 84-85; August, 1999.

Law Environmental, Inc., 1991. Jocassee Dam Hydrologic Analyses. Prepared for Duke Power Co. by Law Environmental, Inc. of Kennesaw, GA (Jan. 18, 1991 ).

Lecointe, G., 1998. Breaching mechanisms of embankments -An overview of previous studies and the models produced. CAD AM (European Un ion: Concerted Action on Dam Break lvlodeling) meeting at the Universitat der Bunderswehr M0nchen, October 8-9, 1998. [Online at:

http://www.hrwallingford.co.uk/projects/CADAM/CADAM/Munich/MU7.pdf]

MacDonald, T. C., and J. Langridge-lvlonopolis, 1984, "Breaching Characteristics of Dam Failures," Journal of Hydraulic Engineering, vol. 110, no. 5, p. 567-586.

tv1oore, J. T., 1972. Quality Control Records - Earthwork. Letter from tv1oore (Duke Power Co.,

Construction Department) to R. L. Dick, September 27, 1972.

NRG, 2009. NRC site visit to the Oconee Nuclear Station on June 15, 2009. Memorandum (trip report with security related information) from R. Pichumani to M. Khanna, dated June 25, 2009.

NRG Accession No. ML091760072.

Peterson, M. D., et al., 2008. Documentation for the 2008 Update of the United States National Seismic Hazard Maps. U.S. Geological Survey Open-File Report 2008-1128, 61 p.

Pugh, C. A., 1985. Hydraulic model studies of fuse plug embankments. REC-ERC-85-7, US DOI, Bureau of Reclaimation, Engineering and Research Center, Dec. 1985.

Simler, H. and L. Samet, 1982. Dam failure from overtopping studied on a hydraulic model.

Transactions -14th Intl. Congress on Large Dams, Rio de Janeiro, Brazil, May 3-7, 1982.

Singh, 1996. Dam Breach M'.:>deling Technology. Kluwer Academic Publishers, 260 pp.

Wahl, T. L., 2004. Uncertainty of predictions of embankment dam breach parameters . J. of Hydraulic Engineering, doi: 10.1061/(ASCE)0733-9429(2004)130:5(389).

Wahl, T. L., 1998. Prediction of Embankment Dam Breach Parameters -A Literature Review and Needs Assessment, DSO-98-004, Dam Safety Research Report. Dam Safety Office, Water Resources Research Laboratory.

Walder and O'Connor, 1997. Methods for predicting peak discharge of floods caused 50

Offieial LJ3e Ol'lly Geeu~ity Relateel ll'lfo~l'l"lfttiol'l by failure of natural and constructed earthen dams. Water Resources Research, Vol. 33, No. 10, pp. 2337-2348.

Willoughby, Howard, and Nystrom , 2005. Generalized Geologic Map of South Carolina. South Carolina Department of Natural Resources, Geological Survey.

Xu, Y., L. Zhang, and J. Jia, 2008. Lessons from Catastrophic Dam Failures in August 1975 in Zhumadian, China. GeoCongress 2008: Geosustainability and Geohazard Mitigation, pp. 162-169.

Acknowedgment Comments related to seismicity benefitted from discussions with William Hinze, Professor Emeritus, Purdue University.

Appendix 1- Dam Facts for South Carolina (source: South Carolina Emergency Management Division; http://www.scemd.org/News/publications/brochures/dam 1.pdf)

Thirty-four dams in South Carolina are hydro-electric and are regulated by the Federal Energy Regulatory Commission in combination with various power companies and municipalities. These dams are required to have emergency action plans that must be updated annually and exercised periodically.

Of the more than 50,000 dams located in South Carolina, only 2,200 are large enough to be regulated under state law. IVost dams are privately owned, although a few are owned by public agencies and are regulated by the S.C. Department of Health and Environmental Control (DEHC).

State-regulated dams are divided into three classes, depending on what is located below the dams:

High Hazard, Significant Hazard, and Low Hazard. DHEC inspects High Hazard dams annually and Significant Hazard dams every three years. Low Hazard dams are not inspected, but areas below the dams are checked every three years for new development and possible reclassification.

Under state law, owners of "High" and "Significant Hazard" dams are required to maintain emergency notification plans. Both Jocassee and Keowee dams are classified as High Hazard dams because of sizeable downstream populations . The classification does not mean that engineering conditions at the dams are hazardous.

Appendix II - Potential for Dam Failure The following discussion about the potential for dam failure is quoted from part one of a two-part article in which C. Richard Donnelly discusses the use of risk-based techniques for assessing dam safety. The web link to this material is:

http://www.waterpowermagazine.com/story.asp?storyCode=2040340 Potential for dam failure Traditionally, dam safety has been, and still is, achieved using an engineering Standards Based Approach that requires dams to withstand certain defined loads. This is a transparent and readily understood methodology that has wide pubic acceptance. However, in the early years of scientific dam design, and indeed up to the middle of the 20th century, there were really no formal and 51

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l accepted methods for selecting the standards that a dam must satisfy to be 'safe'. For this reason, dam design standards often varied depending on the judgment and experience of the practitioner.

The risk that a dam posed to the downstream public was not explicitly assessed in the selection of the design parameters, although the experienced engineer would generally account for it implicitly in his choice of 'safety factors' and magnitude of loads.

Despite these shortcomings, the dam industry has had an admirable record with major dam failures being a very infrequent occurrence. However, dam failure does occur in the United States, a process of reporting dam failures and dam safety incidents is an integral part of dam safety practice. The result of this monitoring program has shown that, in the period from 1993 to 1999, 421 dam failures of varying degrees of severity have occurred (Hydro Review 1999) which translates to an overall probability of fai lure in the order of 6 x 10*4 failures per dam year (Figure 1).

As at 2005, the total number of dam failures has reached 626 continuing to follow this long established trend (Hydro Review, 2005).

M 8: 2500

....£ 2000 f, 1500

-8' US Dom solely E legislolion onoctod 1000 0

.8 500 0 0

§ 0 0 e

o 0 l 994 1995 l 996 1997 1998

~ Yeor Figure 1. Cumulative occurrence of dam failures and dam incidents in the US (1993-1999).

As shown in Table 1, numerous other researchers throughout the world have reported similar dam failure rates with the overall average equivalent to about one dam failure every 2500 years.

Clearly then, there are compelling statistics that dam failure can be an expected consequence of inadequate or improper maintenance of a dam. As well, there are failure statistics that indicate certain types of dams might be more likely to experience problems. As detailed below, at first glance, it would seem that the most vulnerable dam is an embankment structure which, according the FEMA'ICOLD, represent nearly three quarters of the total number of dams that have failed (Figure 2).

Table 1: Frequency of occurrence of dam failures reported in literature (after www .ence.umd.edu, August 1999)

No. of Total dam years Failure Area Reference failures (x10-3) Rate us Gruner (1963, 1967) 33 71 .0 5 X 10-4 Babb & rvlermel (1968) 12 43.0 3 X 10-4 USCOLD (1975) 74 113.0 7 X 10-4 Mark & Stuart-Alexander (1977) 4.5 2 X 10-4 World Mark & Stuart-Alexander (1977) 125 300.0 4 X 10-4 Middlebrooks (1953) and Mark & Stuart-9 47.0 2 X 10-4 Alexander (1977)

Japan Takase (1967) 1046 30 000.0 4 X 10-5 Spain Gruner (1967) 150 235.0 6 X 10-4 52

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l Overall Average Dam 14 x 10-4 I_Failure Rate _

Conc,efo ord1):is Con,,ete b111t1essJ= 12 Concret~ gro, ,t ]=:= 9 Eorthf.11 rockltll} 74 0 10 2cr- 30 40 50 0 70 0 Percentage Figure 2. Percentage of dam failure.

However, in the case of dams, as with most issues, simple statistics alone can be quite misleading.

The actual risks presented by an embankment dam are considerably less when one accounts for the fact that they represent the oldest and most common type of water retaining structure.

Adding to the complexity of assessing potential for failure is the fact that each dam represents a unique structure which, as is shavvn in Figure 3, can fail for a variety of reasons. This complex behaviour stems from the fact that each dam has site specific and complex foundation characteristics that cannot be easily represented in physical or mathematical models. Finally, it is possible for a dam to fail for reasons other than design deficiencies of unknown geological defects.

Human error in operation of the flow control equipment or deliberate acts of sabotage could cause an uncontrolled release of the impounded reservoir.

Concrete doms 29 53

ljj 1t= ~~ "30 4 50 oO P~rcentoge Eorthlill/rockfill clomJ Ovettopp,n'-'J 35 Foundolron] 21 Pipmg & $eepog,-j 38 Olht-, = 6 0 10 20 30 75"' 50 0

~w* IIMA/IC.OUI 191-1 Pere en toge-Figure 3. Causes of dam failure .

It is evident therefore, that evaluation of the potential for a dam to fail due to physical reasons is a difficult task. Indeed, a strict quantifiable evaluation of risk is, in general, not feasible within the current state of knowledge. In this regard, in a discussion on Risk assessment in dam safety management, the International Congress on Large Dams (ICOLD) concurred with this sentiment 53

Offieial LJ3e Ol'lly Geeu~ity Relateel ll'lfo~l'l"lfttiol'l noting that; 'Risk assessment is in the development stage and this is especially true of its application to dams' (ICOLD Bulletin 130).

References Hydro Review, Dam Safety; Vol XVIII, No. 5; pp 84-85; August, 1999.

ICOLD (International Congress on Large Dams), Risk Assessment in Dam Safety, Bulletin 130, 2004.

Appendix Ill Inflow Design Flood Probability As stated earlier in this report, the Probable Maximum Precipitation (PMP) which serves as the basis for the Probable Maximum Flood (PMF) is defined by the American Meteorological Society as "the theoretically greatest depth of precipitation for a given duration that is physically possible over a particular drainage area at a certain time of year." I (b)(7)(F), (bX3):16 U .S.C. § 8240- l(d)

I In order to evaluate the actual effects of these differences on plant safety, the exceedance probabilities of the inflow design floods should be determined. The most direct way to do this is to look at the probabilities of the rainfall, recognizing that antecedent moisture, centering over the basin, and critical arrangement of rainfall increments decrease the probability of the design storm over point rainfall amount for the critical duration. The critical storm durations are 48 hour5.555556e-4 days <br />0.0133 hours <br />7.936508e-5 weeks <br />1.8264e-5 months <br /> for the storm determined from HMR 33 and 72 hour8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br /> for the storm determined from HMR 51. The total rainfall depths for these two storms were 33.6 inches and 37 inches respectively over approximately 200 sq. miles.

The frequency relationship for point rainfall depths over the Jocassee dam drainage basin from NOAA Atlas 14 (Bonnin et al, 2006) is shown in Figure 111-1 .

54

Offieial LJ3e Ol'lly Seeu~ity Related ll'lto~l'l"IB:tiol'l Part i a l d uration based Point P recipitat i on Freq uency Es t imat es - Version: 3 3 4. 726 N 83 . 007 W 935 f t 30 29 28

• f-

- __,...- ~ - ---l 27 26 25 24 23

---

-- ~

_ _. -- ~

~

C 22 21 - -

~ ...- --

20

.c 19 -

~

- ~

-- ~---- .,.,..

0. 18 ~ ~

~

Qj A

17 16 ~

- ---- ------- -----.----- ~ '

C 15 0

14 -

13 -

~

~ ~~ ~

~ 12

~

~ ~ ~;:::-- -~

11 -*

0.

10 -=--

--- ------~ ----

~ ~

0 9 ...-

--.- ~- ~~ ~ _- .,,

Qj

---

------ ~~

t. 8 ~

~

0.. 7 6

5 4 ~

.--- .--I I

,- ~

_,J,,...,--

I t

--t -

3 2

~

J. .

1 0

2 5 10 25 50 100 200 500 10 00 Average Re c urren c e I nterval <years)

Wed Oct 2 8 16:52 :51 2 009 Durat ion 5-mi n - 48- hr-->+- 30-day -->+-

10-min-+- 3-hr - - 4-day ....,_

IS- min-+- 7-day-+- 60-day -+-

30- mi n - - 12-hr-+- 10- dt1\j-+-

60-min - 2 4 -hr --e--- 20-dau --e---

Figure 111-1 Point Precipitation Frequency Estimates near Jocassee Dam Unfortunately, 72 hour8.333333e-4 days <br />0.02 hours <br />1.190476e-4 weeks <br />2.7396e-5 months <br /> rainfall is not shown but can be evaluated from 48 hour5.555556e-4 days <br />0.0133 hours <br />7.936508e-5 weeks <br />1.8264e-5 months <br /> and 96 hour0.00111 days <br />0.0267 hours <br />1.587302e-4 weeks <br />3.6528e-5 months <br /> (4 day) values for average recurrence intervals up to 1,000 years. As can be seen, the P MP values for both storms from HMRs 33 or 51 are over twice the 1000 year point values.

Attempts have been made to correlate frequency with PMP values by various investigators, one of the more promising was a method developed by David Hershfield a meteorologist with the Agricultural Research Service. In the method by Hershfield (Hershfield, 1963) the Fisher-Tippet Type 1 distribution is applied to the standard frequency equation:

Xr = ~ + K SN where:

Xr is the rainfall for a specified return period

~ is the mean for a series of N annual maximum events SN is the standard deviation of a series of N annual maximum events K is a multiplier for the desired frequency or return period and may range from 3.5 for a 100 year return period to 15 for a return period of about 108 years.

By calculating the mean and standard deviations for various stations in the United States ,

Herschfield determined that a reasonably good correlation could be determined for the 24 hour2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> 1O square mile PMP and the mean plus 15 standard deviations corresponding to a return period of about 108 years or a probability of about 1o*8 per year.

55

Offieial LJ3e Ol'lly Geeu~ity Relateel ll'lfo~l'l"lfttiol'l Figure 111-2 shows the point precipitation for 48 hour5.555556e-4 days <br />0.0133 hours <br />7.936508e-5 weeks <br />1.8264e-5 months <br /> and 4 day (96 hour0.00111 days <br />0.0267 hours <br />1.587302e-4 weeks <br />3.6528e-5 months <br />) values plotted against Fisher Tippet Type 1 K values.

p r 30 + - - - + - - - - - - , f - - - + - - - - + - - - - + - - - - + -- ----+-----;

e C

p p i 20 0 t i a n - . hr t 15 t j - 4-day 0

n 10 5

n 0 2 4 6 8 10 12 14 16 Fisher Tippet I K-value Figure 111-2 Point Precipitation vs K value For reference, a K-value of 4.9 corresponds to a return period of 1,000 years, 8.5 corresponds to 105 years and 14 corresponds to 108 years. From these analyses it appears reasonable to estimate the PMP exceedance probability as being between 10*5 and 10* 7 per year. In addition, it must be recognized that these frequency analyses are for precipitati on point values. For the same precipitation depths over a larger area ( i.e. 150 square miles), the probability of exceedance will be even lower. Also, for inflow design flood calculations, the PMP is usually critically arranged with the maximum increment of rainfall in the middle and next greater intensities to the sides in a temporal arrangement. This further reduces the probability of the design storm.

References:

Bonnin, Geoffrey M, Deborah Martin, Bingzhang Lin, Tye Parzybok, Mchael Yekta, and David Riley, 2006: "Precipitation-Frequency Atlas of the United States," National Weather Service, U.S.

Department of Commerce.

56

Offieial LJ3e Ol'lly Geeu~ity Related ll'lt0~l'l"IEttiol'l Herschfield, David M, 1963: "Estimating the Probable 1\/laximum Precipitation" Paper No. 3431, Vol.

128, Transactions of the American Society of Civil Engineers Riedel, John T., Appleby, James F., and Schloemer, RobertW., 1956: "Seasonal Variation of the Probable 1\/laximum Precipitation East of the 105 th Meridian for Areas From 1Oto 1,000 square miles and durations of 6, 12, 24 and 48 hours5.555556e-4 days <br />0.0133 hours <br />7.936508e-5 weeks <br />1.8264e-5 months <br />." Hydrometeorological Report No. 33, Weather Bureau, U.S. Department of Commerce, Washington, D.C.

Schreiner, Louis C and Riedel, John T, 1978: "Probable 1\/laximum Precipitation Estimates, United States East of the 105th Meridan." Hydrometeorological Report No. 51, National Weather Service, U.S. Department of Commerce Appendix IV Spillway Discharge Rating Cuive The spillway discharge curve as used in the LAW Engineering analysis could not be confirmed in this evaluation. The staff did calculate a spillway rating curve using information developed by the U.S Army Waterways Experiment Station and presented in Section 14.7 of Open Channel Hydraulics (Chow, 1959). In this method, piers in gated spillways are adjusted for using the relationship L = L0 - K N H0 where:

Lis the effective length of each b, ; :1e gated spillway, ORIJOEII Lo is the clear span between piers (b > for Jocassee Dam)

K is a coefficient (determined to be > >

o~ rl contraction ependent on pier shape, design head, and total head N is the number o s1 e contractions (bX7XF>, (b)(3):16 u.s.c. § He is the total head on the crest of tne sp1 way inc u 1ng the velocity head (bX7XF). (bX3) 16 u.s.c. § 8240-l d Jocassee dam)

With the spillway coefficient ofr discharge equal t (bX;)CF_)* (bX3 )1 6 u sc for each bay of the spillway.

This relationship corresponds approximately to the 1sc arge curve y Duke used by FE I in the FERG report (2004) presented in this report. The discharge determined in the original stud~

evaluated by AEC approximates the value that would be determined from this relationship ~

l(b)(7)(F), (b)(3):16 U.S.C. § 8240-l(d) I The staff also considered the methodology presented in the HEC-1 program manual. Using this methodology, a different shaped curve was determined. Also, the original rating curve ended at el 11,~~fi?- (b6 § ~ ecause the low chord of the bridge is at this elevation and flow out of the reservoir c~anges to orifice flow at this level. Because of the higher water levels to be evaluated, the rating curve was adjusted for orifice flow and added to the program. All three rating curves are shown in Figure IV-1 .

57 OD~OITl\.'E SECURITY RELATED ltffORMATION GRITIG,A,L DJERG>OELEGTRIGAL l J FRAGTRUGTURE l J FORMATIOP~

Offieial LJ3e Ol'lly Geeu~ity Relateel ll'lfo~l'l"lfttiol'l (b)(7)(F), (b)(3):16 U.S.C. § 8240-l (d)

FIGURE IV-1 Jocassee Spillway Rating References for Spillway Discharge Rating Chow, Ven Te, Open Channel Hydraulics, fvtGraw-Hill, 1959 U.S. Army, HEC-1 Flood Hydrograph Package User's fv1anual, September 1990 58