Paper USSD Annual Meeting June 24-28, 2002 DESIGN OF THE RCC PORTION OF THE SALUDA DAM REMEDIATION PROJECT

Transcription

1 DESIGN OF THE RCC PORTION OF THE SALUDA DAM REMEDIATION PROJECT By Paul C. Rizzo * William Argentieri ** Jeffrey M. Bair * Kristina Massey ** Harold Moxley ** ABSTRACT Saluda Dam, part of the FERC regulated Saluda Hydroelectric Project located near Columbia, South Carolina, is being upgraded to resist a recurrence of the 1886 Charleston Earthquake. The existing, semi-hydraulic fill embankment Dam was completed in 1930 and, based on current technology, is viewed as being susceptible to liquefaction during the design seismic event. Several remediation options were considered with the selected approach being a back-up berm, essentially a Dry Dam, at the downstream toe of the existing Dam. The Dry Dam will consist of about 5500 feet of Rockfill and about 2300 feet of RCC. This Project is the largest active (year 2002) dam construction project in the United States and the final Project will involve the placement of 1.3 million cubic yards of RCC and 3.5 million cubic yards of Rock Fill. This paper presents an overview of the design aspects of the RCC portion of the Project. We address topics such as the seismic design basis, loading conditions, finite element analysis, mix design parameters, drainage and joint design, facing considerations, thermal stress analysis and joint spacing, and the mix design testing programs. An unusually large number of load cases had to be considered as the fix must consider dry conditions and wet conditions as well as the postulated condition of the existing embankment Dam after a recurrence of the 1886 Earthquake. The mix design program addresses the use of aggregate from an on-site borrow area and the use of landfilled, waste flyash as a constituent of the RCC. *Paul C. Rizzo Associates, 105 Mall Blvd. Suite 270E, Monroeville, Pennsylvania, USA **South Carolina Electric & Gas Company, 2112 North Lake Drive, Columbia, South Carolina, USA

2 INTRODUCTION Saluda Dam owned and operated by South Carolina Electric & Gas Company (SCE&G), impounds the Saluda River to form Lake Murray near Columbia, South Carolina, one of the largest manmade lakes in North America. The Dam is a semi-hydraulic fill structure completed in 1930 following typical puddle dam construction technology popular in the early 1900 s. Table 1 is a summary of the main parameters describing the existing Dam and the impounded Lake. Existing Dam Hydroelectric Plant Lake Murray Table 1. Summary of the Main Parameters Lake Area 78 Square miles Lake Capacity 2,096,000 acre feet Dam Length 7,800 feet Max Dam Height 211 feet Powerhouse Capacity 206 MW Original Construction Semi-Hydraulic Fill Original Completion 1930 The primary purpose of the Dam when originally constructed was for hydroelectric generation by the Saluda Hydroelectric Plant located at the toe of the Dam. As such, the Dam is under the jurisdiction of the Federal Energy Regulatory Commission (FERC). Today, the Lake is a source of cooling water for the McMeekin Steam Electric Plant, drinking water for Columbia and adjacent communities, and a major recreation and residential community with statewide economic benefits. Beginning in 1989, the authors began a series of geotechnical investigations to assess the safety of the existing Dam, particularly under seismic loading. In this part of South

3 Carolina, seismic design bases for critical facilities are, for all practical purposes, governed by a postulated re-occurrence of the 1886 Charleston Earthquake. The Charleston Earthquake is estimated to have had a Magnitude in the range of 7.1 to 7.3 with a recurrence interval in the range of 650 to 1000 years. This event has been established as the Design Seismic Event (DSE) for assessing the integrity of Saluda Dam. A comprehensive liquefaction analysis and a post-earthquake stability analysis were conducted with the DSE yielding typical results as shown on Figure 1. This Figure shows that under certain assumptions, deemed by the authors to be very conservative, a major portion of the embankment will liquefy if the DSE occurs Figure 1. Existing Saluda Dam Factor of Safety Against Liquefaction Should Saluda Dam fail, approximately 120,000 people would be in jeopardy, water supplies for Columbia and surrounding communities would be lost, extreme environmental impacts would be realized and countless millions of dollars would be lost in the local economy. Consequently, a major remediation project has been developed for implementation in the 2002 to 2005 time period. REMEDIATION CONCEPT The Design Basis for the Remediation Concept has two major objectives and a corollary expected behavior criteria. The two major objectives are: Prevent catastrophic flooding downstream of the Dam and, Assure Safe shutdown of the facilities, including lowering of Lake Murray in a controlled manner. From a behavior perspective, the Owner and the FERC accept that some remediation and repair may be required after the Design Earthquake, but catastrophic flooding is not

4 acceptable. These major objectives were translated into loading cases and factors of safety by the authors with major over-riding input from the FERC. After consideration, both technical and financial, the remediation concept was developed as a Dry Dam immediately downstream of the existing earth embankment. The existing Dam will remain in place and function as the primary impounding barrier for Lake Murray. The Dry Dam will become a water retention structure if Saluda Dam ever fails. Hence, the Dry Dam is designed for both dry and wet conditions with a series of loading cases that reflect hydrostatic pressures and uplift pressures in the wet case and neither in the dry case. Both cases consider seismic forces acting both directions horizontally as well as vertical excitation. Additional design criteria to be implemented with the chosen remediation concept include the following: No excavation into the original sluiced Dam. Riprap placed after original dam construction will be removed to facilitate excavation. No excavation or construction into the Saluda River as defined by the maximum normal tailwater at El NAVD. The crest of the new Berm will be El. 372 NAVD 1 (i.e., current minimum crest elevation of the existing Dam). The Dry Dam concept is illustrated on Figures 2 and 3 below. Approximately 5500 feet of the total length of 7,800 feet is a Rockfill Berm (referred to as a berm rather than a dam as it does not retain water under normal conditions) and 2,600 feet of RCC Berm (also referred to as a berm rather than a dam for the same reason.) Figure 2. Typical Section, Rockfill Berm 1 North American Vertical Datum of 1988

5 Figure 3. Typical Section, RCC Berm There is an obvious question of, Why not use a less expensive Rockfill Berm for the entire remediation? In reply we refer the reader to the photo with Table 1 and Figure 3 where it may be seen that the existing Saluda Hydroelectric Powerhouse is at the toe of the existing Dam, thereby precluding a Rockfill Berm because of space limitations. It is in this zone behind (upstream) of the Powerhouse where we chose an RCC Berm as the primary remediation concept. RCC BERM DESIGN The RCC Berm, or Dam depending on one s perspective, is one of the largest in the world and the United States as indicated in Table 2. Table 2. RCC Dams 1999 (Completed and Under Construction) Dimensions Volume Name Country Owner Purpose River Height Length RCC Total (m) (m) (m 3 x10 3 ) (m 3 x10 3 ) 1. Beni Minister de l Hydraulique, de l Algeria Haroun Environment et des Forets WI Kebir Miel I Colombi a Hidromiel S.A. HFI LaMiel Miyagase Japan Ministry of Construction FWH Nakatsu Porce II Colombi a Empresas Publicas de Medellin H Porce Urayama Guanyinge Japan W.R.D.P.C. FWH Urayama

6 6. (Kwan-in- Temple) China Benxi City, Liaoning Province HFI Taizhe Bureya Russia RAO EDD ROSSI FH Bureya Upper Central Utah Water Rock USA WI Stillwater Conservancy District Creek Oliverhain San Diego County Water USA Dam District W N.A Saluda South Carolina Electric & USA Dam Gas FHRW Saluda Jiangya China Lishui Hydropower Corporation HFIWN Loushui Salto Caxias Brazil COPEL H Iguacu It is noted that Olivenhain Dam is slightly larger and is being designed and constructed contemporaneously with the RCC Berm at Saluda. Detailed design criteria for the Saluda RCC Berm include the following: The RCC Berm is founded on competent rock as defined by the commencement of coring operations and verified during construction. Uplift pressures for stability analyses are calculated assuming that the foundation drains will achieve a minimum drainage efficiency of 67 percent. The foundation drainage system consists of four-inch diameter drains installed at ten-foot spacing to a depth into the foundation rock equal to 30 percent of the total head. Usual (normal pool) and Extreme (earthquake) loading conditions include hydrostatic loads based on existing (i.e., pre-dse) groundwater levels. Pseudo-static dynamic analyses are based on a horizontal seismic coefficient of 0.2g. Dynamic finite element analyses are used to determine the distribution of tensile and compressive stresses within the Berm using the DSE as input. Short-term, post-dse loading conditions assume liquefied material (from the failure of the existing dam) on the upstream slope to El. 302 NAVD. For long-term, post-dse loading conditions, the upstream-liquefied material has been assumed to be at El. 274 NAVD. In the following paragraphs, we describe the following design topics. Seismic Design Basis; Loading Conditions; Finite Element Analyses;

7 RCC Mix Design Program; Joint and Drain Design; Facing Options; and Bedding Mix Application. Seismic Design Basis The Design Seismic Event (DSE) at the Saluda Site used to develop the remedial design of the RCC Berm is defined in terms of the Newmark-Hall (NUREG-0098) median-based response spectra anchored to a horizontal Peak Ground Acceleration (PGA) of 0.2g. These response spectra reflect a peak ground velocity of 7.2 in/sec and a peak ground displacement of 4.02 inches. The vertical response spectra are taken to be two-thirds the horizontal across the entire range of frequencies. Thus, the PGA in the vertical direction is 0.133g. The seismic time history applied to the RCC Berm utilizes Artificial Time Histories (ATH) of acceleration. The artificial time histories are developed iteratively on the basis of the Design Response Spectra with consideration of the Power Spectral Density Function. The time envelope function defines a strong ground motion duration of about 16 seconds and a total duration of 30 seconds. Loading Conditions The loading conditions (i.e., load cases) and associated factors of safety used for developing the remedial design were developed by PCRA in conjunction with comments received from the FERC. The FERC Engineering Guidelines (FERC, 1991), Draft Peer Review Chapter III of the FERC Engineering Guidelines (FERC, 2000), and generally accepted design methodologies have been used to develop the loading conditions. However, recognizing that the stability analyses used to develop the remedial design for the Dam present a unique case, engineering judgement and interpretation were utilized in the development of the appropriate loading conditions. The following load cases were used to analyze the stability of the proposed remedial design of the RCC Berm: Case I Usual Loading Conditions (Normal Operating Conditions); Case II Unusual Loading Conditions (Probable Maximum Flood Conditions);

8 Case IIIA Extreme Loading Conditions (Dry Dam with DSE earthquake forces); Case IIIB-1 Post-Seismic Loading Conditions (Short-Term Operating Conditions after the postulated failure of the existing Dam); and Case IIIB-2 Post-Seismic Loading Conditions (Long-Term Operating Conditions after the postulated failure of the existing Dam). As the proposed Berm will be dry prior to the DSE and the existing Dam can pass the PMF without overtopping, the results for Case II (Unusual loading Conditions) will be identical to Case I. For clarity, we have not included these results in the Summary Tables. The acceptable factors of safety for these loading conditions are presented in Table 3. Table 3. Acceptable Safety Factors RCC Berm LOAD CONDITION FS Case I 3.0 Case IIIA 1.1 Case IIIB Case IIIB The critical load case occurs immediately following the DSE (i.e., Case IIIB-1) and after the postulated failure of the existing Saluda Dam. Uncertainties associated with the probable loading for the RCC Berm during this particular load case are a function of the following factors: Height of silt (liquefied soil from the existing Dam against the upstream face of the RCC Berm. Residual shear strength (if any) of the liquefied embankment materials and Rockfill from the Highway Berm (i.e., combination of silt, rockfill, and water) which will be located against the upstream face of the RCC Berm. Buildup of uplift pressure beneath the RCC Berm and, specifically, whether the uplift pressure fully develops before the pore pressure dissipates from the potentially liquefied material on the upstream side of the RCC Berm. To address these uncertainties, we analyzed the RCC Berm for a range of possibilities following the DSE. Our best estimate of the post-seismic loading conditions utilizes partial uplift (i.e., equal to uplift prior the DSE) beneath the RCC Berm and a best estimate strength for the combination of silt, rockfill, and water, which will be located along the upstream face of the Berm. Additionally, we have analyzed the Berm for

9 worst-case post-seismic parameters which include full uplift beneath the foundation of the RCC Berm and a lower bound shear strength for the combination of silt, rockfill, and water which will be located along the upstream face of the RCC Berm. The results of these analyses have been used to design an RCC Berm that is stable under all postulated post-seismic design parameters. Finite Element Analyses The RCC Berm has three basic cross sections - the Powerhouse section, which bears on a mass concrete foundation, and the South Section and the North Section which bear directly on rock. The mass concrete foundation behind the Powerhouse embeds the existing Penstocks and McMeekin Steam Plant cooling water lines. These cross sections were analyzed using conventional Finite Element Analyses (FEA) and we report in this paper the results for the Powerhouse section only. Also, while the FEA model for the Powerhouse section includes the mass concrete foundation, we discuss only the results in the RCC Berm, with particular emphasis on the tensile stresses as they control the RCC mix design. The FEA performed for the RCC Berm assumes that the RCC Berm will be dry prior to the DSE. The evaluation combines the static stress state of the RCC Berm prior to the seismic event with the seismic stresses generated in the Berm due to seismic ground motions. Additionally, the evaluation addresses the stresses in the Berm for the postseismic loads. Both the static and seismic analyses utilize two-dimensional models of Powerhouse Section. Finite Element Models The static analysis of the Powerhouse Section of the RCC Berm assumes plane stress conditions and uses a two-dimensional finite element model of the Planar Section shown in Figure 4. RISA-3D, a general-purpose finite element program suitable for the analysis of frame and panel structures, was used to evaluate the static stresses.

10 111.7 FEET 40 FEET FEET Location of ATH Input to model FEET 200 FEET FEET Figure 4. Finite Element Model (Powerhouse) Without Rockfill The finite element model of the RCC Berm utilized in the static analysis includes panel elements representing the RCC Berm. Assigning a material density to the panels includes the dead load of the RCC. All other conditions existing on the RCC Berm prior to the seismic event are included as applied loads to the finite element model. For example, equivalent lateral loads on the Berm represent the retained rock-fill on the upstream face. Table 4 lists the properties used in the model. Table 4. Material Properties for Static Finite Element Model Material Compressive Elastic modulus strength f c (psi) (ksi) Poisson s ratio Density (kcf) Roller compacted concrete 2,300 2, Conventional concrete facing 3,000 3, Mass concrete base 3,000 3, Existing concrete 4,000 3, Rock foundation Not applicable 1, Not applicable

11 Table 5 shows the material properties for the structural portion of the model. The dynamic elastic modulus for the various types of concrete is calculated by multiplying the static modulus by The soil and rock-fill area retained by the upstream portion of the RCC Berm is treated in the analysis as part of the structure. Table 5. Material Properties for Seismic Finite Element Model Material Description Density (kcf) Shear wave velocity (fps) Compression wave velocity (fps) Shear wave damping ratio Compressio n wave damping ratio Shear modulus (ksf) 1 Existing concrete ,307 11, ,800 2 Mass concrete ,800 11, ,400 3 Conventional concrete ,800 11, ,400 facing 4 Roller compacted ,452 10, ,800 concrete 5 Soil ,099 1, , Rock-fill ,108 to (bottom) to 1,709 to 2,791 11,790 Stress Results in the RCC Table 6 presents the maximum principal tension stresses at the upstream and downstream locations of the RCC Berm Section at the Powerhouse. These stresses are presented for the static loads and the seismic loads resulting from the Artificial Time Histories matching the Design Response Spectra. Table 6. Maximum Principal Tension Stress in RCC (psi) Powerhouse Section Location RCC - downstream face Upstream concrete facing Pre-seismic 1.4*Dead + 1.7*Live Design Response Spectra Without Rockfill With Rockfill Unlimited cohesion at base Post-seismic No cohesion at base

12 As seen from Table 6, the Maximum Principal tension ranges from 5 to 359 psi (based on use of the ground Design Response Spectrum). The stresses with rock-fill on the upstream face are somewhat higher than the stresses without rock-fill. The maximum tension occurs on the downstream face at the interface of the RCC and the underlying mass concrete. Please note that the seismic analysis described herein assumes linear material behavior and allows, as much tension in the RCC and RCC/rock interface as the loads will generate. In reality, the material may sustain cracking at locations where the tension stress exceeds the material tension strength. For example, the tension at the RCC/rock interface will be limited by the cohesion mobilized at this interface. Therefore, the tension stresses reported in Table 6 are an upper bound to the actual tension stresses in the RCC. Limit Equilibrium Analysis and Bedding Mix Requirements The results of a post-dse Limit Equilibrium Analysis (LEA) were used to determine interface friction angle necessary to achieve a sliding factor of safety of 1.1. The analysis assumes that cohesion is zero and that the lift joint considered is fully cracked along the entire width of the Berm. Based on the results of the dynamic analyses, we do not expect appreciable cracking between the RCC lift joints. However, we have conservatively assumed that cracking does occur along the entire length of the RCC lift line and have performed static sliding stability analysis along these planes. A sensitivity analysis was performed to determine the minimum shear strength parameters at the lift joints and within parent RCC. Two limiting cases were considered in each Section, cohesion with no friction and friction with no cohesion. The results indicate bedding mortar should be placed up to El. 215 NAVD to assure a sliding factor of safety in excess of 1.1 along the RCC lift lines. Bedding mix will also be applied along a 22-foot wide zone at the upstream face of the RCC, full-height, for seepage control. At El. 215 NAVD, the interface friction angle to achieve a minimum sliding factor of safety of 1.1 is equal to or less than 35 degrees for both lower bound and best-estimate parameters. RCC Mix Design Program We have established the Minimum Design Strength for the Final RCC mix based on Phase I and II laboratory testing conducted on various RCC mixes. The final RCC mix design will meet or exceed these minimum strengths. A Dynamic Increase Factor (DIF)

13 of 1.5 is applied to the static tensile strength to compute the dynamic tensile strength. A summary of the Minimum Design strengths for the final RCC mix are provided in Table 7. Table 7. Minimum Design Strength Final RCC Mix Property Static Compressive Strength Minimum Design Strength At One Year 2,300 psi Static Direct Tensile Strength Parent RCC and bedded lift joints Unbedded Lift Joints 239 psi 115 psi Dynamic Direct Tensile Strength Parent RCC and bedded lift joints Unbedded Lift Joints 359 psi 173 psi Note: Dynamic Direct Tensile Strengths are calculated utilizing a DIF of 1.5 Direct tensile strength criteria used for design were correlated with a compressive strength of 2,300 psi. The relationship between the compressive strength and the direct tensile strength was extrapolated from the project-specific RCC laboratory test results. Thermal Analyses and Joint Spacing A Simplified Thermal Analysis (STA) or Level I Thermal analysis has been completed to evaluate the potential for thermal cracking of the RCC during placement. A more rigorous thermal analysis, i.e., a Level II Analysis which utilizes detailed finite element analyses and placement schedules was being performed while this paper was in preparation. The actual temperature rise in the RCC Berm depends on the heat generating characteristics of the RCC mix and accompanying thermal properties, environmental conditions, geometry of the section, and construction conditions. Peak temperatures are reached a few days or weeks after placement. For massive structures, the temperature of the RCC slowly decreases until it reaches the average annual temperature of the site. As the RCC cools, the volume of the RCC decreases. Since this volume reduction is constrained by the bedrock foundation, sufficient strain can develop to cause a tensile failure or crack of the RCC. To prevent excessive cracking of the RCC, transverse

14 contraction joints are placed into the RCC mass at a prescribed interval. A thermal analysis is used to justify the selection of the contraction joint spacing. The methodology followed for this Project is described in the American Society of Civil Engineers publication entitled Roller Compacted Concrete III (Tatro and Schrader, 1992). The results of this analysis are summarized in Table 8. Please note that the analysis assumes that aggregate production and stockpiling commences two months before RCC placement. Table 8. Summary of Results Simplified Thermal Analysis Placement Schedule (Month) Placing Temperature ( F) ΔL (in) Number of Cracks (Assuming 0.05 width) Avg. Crack Spacing (ft c/c) February Infinite May August November As shown on Table 8, RCC placement in the summer months (i.e., August) significantly increases the thermal displacements in the RCC mass. The results, of this analysis for the proposed transverse contraction joint spacing of 60 feet center-to-center (c/c), indicate that RCC construction stockpiling and aggregate production should be limited to the fall and winter months. It also indicates that the placement and/or stockpiling of RCC in the spring and summer months may require some form of pre-cooling to minimize cracking. Mix Design Testing Program The Mix Design Testing Program was established in the early project as a four Phase Program summarized in the following Table 9.

15 Table 9. Summary of RCC Mix Design Phases Phase N o. D escription Schedule I II Initial Lab P ro gram (13 M ixes) F inal Lab P ro gram (5 V erification M ixes) S tarted M ay Status: Complete Started December 2000 Status: Complete III O n - S ite, F u ll- S c a le D e s ig n R C C T e s t P a d s w ith Aggregate from O n-site Borrow Area S tarted A ugust Status: O ngoing IV P re-c onstruction Test Pad for V erification of Contactor M ethodology Status:F uture W o rk after R C C B erm C ontractor S election As seen on the table, we began with an exploratory Phase I laboratory-testing program to determine if sluiced, landfilled flyash from the adjacent McMeekin Steam Electric Plant operations could be utilized. Offsite aggregate was used in Phase I as well as in Phase II. The ash is disposed in an on-site landfill located at the toe of the existing Dam and a portion of it has to be moved in any event. Phase I yielded positive results and Phase II was used to refine the mix options and to hone in on a final mix design. We then proceeded with a full-scale test pads using aggregate from our on-site borrow area, fully processed in accordance with a plan for crushing and screening. Phase III consists of three main test pads, each with a different mix with the cement content being the primary differing constituent as shown on Table 10. In addition to the three main pads, a series of smaller pads were constructed to allow for variation of placement parameters, testing procedures and the like. Phase III is currently in the curing stage (April 2002). Table 10. Phase III RCC Mixes Mix Designation a Cement (lbs / yd 3 ) Pozzolan (McMee kin Fly Ash) (lbs / yd 3 ) Water in excess of SSD b (lbs / yd 3 ) Mod. VB Time (sec) Blended Aggregate (Fine + Coarse) (lbs / yd 3 ) Design 3% Air (lbs / ft 3 ) Primary RCC Mix Design Secondary RCC Mix Design Tertiary Mix , , ,

16 Testing of samples (cores and blocks) include the following: Post-cracking (residual) and Intact Direct Shear Tests - 6 inch vertical and inclined cores - 12 inch x 12 inch block samples Controlled and Uncontrolled Direct Tension Tests - Lift Surface - Parent RCC RCC Berm Drainage Design The design of the drainage system for the RCC Berm as shown on Figure 5 consists of three primary elements as follows: A Drainage Gallery 14 feet from the Upstream Face just above Tailwater; Bedrock Drains (4 diameter) from the Gallery drilled to a depth equal to 30% of normal headwater; and Crest-to-Gallery Drains (6 diameter) to intercept drainage through cracks on the upstream face.

17 NORMAL POOL CREST TO GALLERY DRAINS Crest to Gallery RCC BERM SILT / ROCK MIXTURE UPSTREAM FACING DRAINAGE GALLERY GROUND SURFACE RESIDUAL SOIL FOUNDATION DRAINS COMPETENT ROCK Figure 5. Typical RCC Section Upstream and Downstream Facing Upstream and downstream facing considerations include constructability, aesthetics, economics and, in the case of the upstream face, watertightness. Final decisions will be based on cost as quoted by bidders, but options for the downstream face include (a) formed RCC Steps, and (b) precast concrete panels, with the former illustrated on Figure 6.

18 6 FILLET OF BEDDING MIX 3/8 THICK BEDDING MIX RCC FACE OF TEMPORARY FORMWORK 2 LIFT (TYP.) Figure 6. Formed RCC Face For the Upstream face, we are currently considering (a) conventional concrete, (b) grout enriched RCC against forms, and (c) precast panels. Figure 7 illustrates the conventional concrete and the treatment of the top of the Berm. Figure 7. Upstream Facing

19 CLOSING REMARKS The Saluda Dam Remediation Project is one of the largest, active dam remediation projects in the world and will be one of the top ten largest RCC Dams in the world. The Project includes practically all of the complicating factors that a dam engineer can face in remediating an existing dam, including deep toe excavations under near-full head, variable soil and rock conditions and exceptionally difficult space constraints. The use of RCC aided significantly to solving the space constraints, but involved a great deal of upfront testing and verification analysis. We still have to complete testing on the Phase III samples to finalize the mix design and we still have to complete a Phase IV testing program. Final selection of facings will depend on prices provided by bidders.