After a number of studies revealed safety deficiencies at Loch Raven dam, spillway and stability improvements were initiated to ensure safe operation of the US project
Loch Raven dam, located just north of Baltimore, Maryland, US, is a 30m high, 198m long concrete gravity dam with an ungated 88m wide spillway. The 93-year-old structure impounds Loch Raven reservoir, a primary water supply source owned and operated by the city of Baltimore. The dam has undergone several modifications during its life, however the structural height and configuration remained unchanged from 1922 until 2002 when recent modifications were initiated. Inadequacies regarding safety of the dam were first noted in the phase I inspection report prepared in 1978.
During the 1980s and 1990s, additional inspections and investigations were performed to evaluate the extent of the safety deficiencies, which consisted primarily of inadequate spillway capacity and structural stability. The studies ultimately led to preparation of final design documents and construction of a US$30M rehabilitation project. The Loch Raven dam rehabilitation construction contract was awarded to a joint venture of ASI RCC, Inc. and Cianbro Corp. on August 2002; construction began in 2002 and was completed in mid-2005.
Identification of dam safety deficiencies
Pursuant to the National dam inspection Act, Public Law 92-367 of 1972, the US Army Corps of Engineers (USACE) commissioned a phase I inspection of Loch Raven dam. The results of the 1978 phase I inspection categorised the dam as ‘large’ in size, and ‘high’ in hazard classification, based upon the potential for downstream damage if the dam should fail. The results presented in the phase I inspection report indicated the dam, when evaluated in context with present dam design standards, was potentially unsafe.
The primary dam safety issues identified in the report with respect to the structure and appurtenances included:
• Concerns over the structural stability of the dam when subjected to extreme event loading conditions.
• The inability of the spillway to pass the required spillway design flood (SDF) without overtopping the structure.
• The need to inspect conditions at the downstream toe of the dam to determine if excessive scouring and undermining of the structure had occurred.
• The need to rehabilitate the internal drains within the dam.
• The advisability of providing a means to drain the reservoir.
Because of the limited site information available at the time of these assessments, it was recommended by USACE that:
• Additional inspections and investigations be undertaken to more fully evaluate the structural stability and spillway capacity issues.
• That further site inspection and investigations be undertaken.
• That a formal Emergency Warning System be established for Loch Raven dam.
Following the recommendations of the phase I report, the City of Baltimore commissioned a phase II inspection and study to collect additional site data and provide more in-depth analyses to confirm the need to implement modifications to Loch Raven dam.
The primary findings of this report, completed in 1985, reaffirmed that the structural stability of the dam was deficient for the mandated design loading conditions which, for a high hazard dam, corresponds to the Probable Maximum Flood (PMF) event. The phase II study also concluded that the spillway capacity at Loch Raven dam was grossly inadequate, capable of passing only 24% of the PMF.
Development of repair concepts
Gannett Fleming was approached in 1992 to evaluate design concepts to bring the dam into compliance with current dam safety regulations. The initial task for this study was to review, refine, and update the prior studies that had been performed. As part of this effort the previously compiled hydrologic and hydraulic studies were updated to incorporate information from current hydrometeriological design reports published in the late 1980s as required by Maryland Department of the Environment (MDE).
The net result of these studies showed that the PMF event at the dam as predicted by current methods increased and that the spillway capacity at the dam was even more inadequate than previously documented. Also as part of the conceptual design studies, limited additional subsurface investigations were undertaken to assign preliminary design parameters for evaluation of conceptual design alternatives. Concrete and foundation material strength parameters, unit weights, and uplift pressures acting on the dam were evaluated as part of these investigations.
The Conceptual Design report and an Environmental Assessment study, completed in 1993, identified environmental permitting issues and presented two alternative design solutions to correct deficiencies at the dam with respect to the dam safety issues of inadequate spillway capacity and structural stability.
The recommended solution included raising the non-overflow sections of the dam to provide additional spillway capacity and increasing structural stability by the addition of a mass concrete buttress section in combination with post-tensioned rock anchors. To reduce the cost of construction, construction of mass concrete sections of roller-compacted concrete (RCC) was recommended. A key outcome of the Conceptual Design study was recognition that the reservoir could not be lowered by the City during construction to facilitate the planned modifications.
The results of the Conceptual Design study demonstrated that sliding stability along potential failure planes within the dam’s foundation presented a controlling influence on the selection of the rehabilitation measures to be installed. In addition, not lowering the reservoir pool during excavation for the RCC buttress would further reduce the effective factor of safety against sliding.
Considering this, it was recommended that inclined post tensioned rock anchors be installed instead of the originally proposed vertical anchors. Adoption of this recommendation enabled a more effective application of the applied forces to counter construction condition lateral loads and reduced the size of post tensioned rock anchors required to achieve suitable construction condition factors of safety against sliding.
In accordance with the original project plan, an expanded subsurface drilling programme was completed in the autumn of 1997 and laboratory testing was conducted to augment previously collected information on subsurface conditions at the dam. Final design parameters for the stability analyses and design of the post-tensioned rock anchor installations were determined from the results of these investigations.
The foundation parameters recommended for final design included a shear friction angle of 26.5° and cohesion of 8 pounds per square inch along planes of foliation within the foundation. These values correspond to the average residual strength of the test specimens and were recommended to account for the discontinuities in the foundation. The orientation of the apparent or effective foliation plane normal to the axis of the dam was characterised as 12° dipping in the upstream direction. This information was incorporated into the stability analysis to establish the probable failure plane when evaluating the effectiveness of alternative remedial measures intended to prevent a sliding failure of the dam.
Preliminary design stability analyses were performed concurrent with the Geotechnical investigations using the then recently released USACE engineering Manual 1110-2-2200, ‘Gravity Dam Design’, dated June 1995. Electronic computation sheets were compiled to evaluate overturning and sliding stability using two-dimensional methods of analysis for existing and proposed conditions of the spillway and non-overflow sections of the dam.
As part of these analyses, the stability of the existing spillway and maximum non-overflow sections were analysed for the conditions that prevailed at the dam during the flood of record (Tropical Storm Agnes-June 1972) to confirm the foundation input parameters and to assess if cracking along the base of the dam had occurred. Although tensile stresses were computed to have occurred at the heel of the structure as a result of the imposed 1972 loading conditions, it was not considered likely that this resulted in the formation of cracks along the base because further analysis demonstrated that modified uplift pressures produced by base cracking would have induced overturning of the dam.
Since the dam did not fail during the 1972 flood, it was concluded that the tensile capacity along the plane of contact of the foundation and the dam was sufficient to prevent crack formation. The computed sliding factor of safety for this historic loading condition was computed to be 1.02 for the spillway section and 1.03 for the maximum non-overflow section, with overturning factors of safety also nominally above unity.
Considering the results of these analyses, preliminary dam modification templates were evaluated and stability computations for the adverse construction case loading condition were analysed. The adverse loading condition adopted in the analysis assumed that the 100-year flood event occurs during the period of time when downstream excavations have been performed but the RCC buttress section is not yet in place. While this represents a highly improbable loading condition, it is similar to what occurred during the 1972 storm event and considering the importance of the dam as a primary source of water for Baltimore and the potential risk to public safety, the assumed adverse construction loading condition was deemed appropriate.
The stability computations based upon this loading condition demonstrated that additional stabilising forces needed to be applied to achieve the recommended sliding factor of safety for unusual loading conditions. The magnitude of the design loads for the post-tensioned rock anchors were then determined to stabilise the spillway and non-overflow sections against construction case loading conditions. The maximum applied loads computed for the spillway section with an inclined anchor arrangement had horizontal and vertical components 141.4 Kips per linear foot.
Upon establishing the design loads for post-tensioned rock anchors to be installed during initial stages of construction, design templates for the RCC modifications were selected to stabilise the structure for the extreme case loading conditions including full PMF and seismic loading at normal reservoir pool. Additional computations were performed to estimate induced shear stresses during the construction stages and to evaluate the stability of the final design templates in the event of relaxation of the anchors during the design life of the structure.
The intake monolith at the dam presented particular problems in the design because of the restrictions placed on positioning the post-tensioned anchors to accommodate internal passageways within these monoliths. It was determined that to achieve suitable factors of safety, installation of the post-tensioned anchors for this monolith would need to be performed in a two-stage operation wherein vertical post tensioned anchors would stabilise the monolith during excavations required to expose and remove the deteriorated prestressed concrete cylinder pipe (PCCP) conduit. Subsequent to this, inclined post-tensioned anchors were installed to provide the remaining stabilising components to permit excavation to foundation grade upon which the replacement conduit and RCC buttress would be constructed.
Hydraulic and hydrologic analyses for the project design were performed to confirm the spillway design head necessary for the spillway to pass the PMF and the corresponding height to raise the non-overflow sections of the dam. Geometry for the reshaped spillway overflow section was computed based upon the results of these computations and standard design curves to assign shape parameters to prevent negative pressures and cavitation from occurring during the passage of the PMF. It also became evident from the H&H analyses that with the reservoir maintained at normal pool throughout the construction period, provisions would need to be incorporated in the flow diversion plans to accommodate the staged removal and reconstruction of the existing spillway crest section.
Key aspects of the design
As previously indicated, reshaping of the crest of the spillway was required to accommodate the design head for the raised dam. This work involved demolishing the top of the existing spillway and constructing a new spillway control section. The upper portion of the replacement section was constructed out of conventional concrete as opposed to RCC because of construction limitations and forming requirements to achieve the design shape. The modifications to the spillway will be performed when the RCC buttress section has advanced to approximately 70m in elevation. A roller bucket and stilling basin was installed within the lower portion of the spillway to dissipate flow energy and reduce potential for scouring and undermining at the toe of the dam.
And as previously discussed, post-tensioned rock anchors were installed to resist potential loads imposed by the adverse construction case loading condition and to act in combination with the RCC buttress section to stabilise the dam for the PMF extreme loading condition. Installation of the stabilising anchors was essential prior to performing excavations for the RCC stilling basin slab and buttress sections in order to maintain required safety factors during construction.
Figure 3 presents a schematic section of the proposed spillway rock anchors. Total anchor lengths, including bonded and unbonded sections, vary from approximately 21m to 40m with the longest anchors situated within the spillway. The capacities of the anchors installed at various locations across the dam range from about 790 Kips (23 strand anchors) to a maximum of 2,039 Kips (58 strand anchors). Anchor lengths were established based upon the requirement that the anchor bonding segment develop the design anchor capacity below the hypothetical plane of sliding within the bedrock below the dam. Post-tensioned rock anchors were installed within the spillway and a portion of the dam’s non-overflow sections. Both vertical and inclined anchors were installed to stabilise the intake monoliths with vertical anchors used near the ends of the existing dam where design loads diminish. The capacity of each installed anchor was verified by performance testing.
Placement of RCC mass concrete at Loch Raven dam was selected to achieve the design objectives of stabilising the structure for the extreme (PMF) case loading condition and to raise the non-overflow sections of the dam to prevent overtopping during the PMF. RCC was placed alternately on one half of the dam and then the other to accommodate diversion of flow during construction. Figure 4 is a construction-phase photo depicting the RCC buttress and related construction activities left of the spillway centerline.
During prior repairs to the dam, the downstream face was repaired with pneumatically applied concrete (shotcrete). The shotcrete surface of the existing spillway was severely deteriorated and this material was removed prior the RCC placement. Removal was performed by a combination of mechanical demolition and hydrojetting to remove debris and limit concrete fracturing that would otherwise diminish the quality of bond between the RCC and the existing dam. Drilled and grouted rebar anchors were installed in the downstream face of the existing dam in advance of RCC placement, and during RCC placement bedding mix was placed on each lift at the contact zone between the existing dam and the RCC buttress to improve the bond along the contact surfaces.
RCC mix design development and associated design considerations
It was recognised that two different RCC mixes would be required for the project. The RCC buttress is a low-stressed structural element that is compatible with a relatively lean RCC mix. Further, it was believed that the lower strength would also develop a lower modulus that would be compatible with the relatively low-strength conventional mass concrete originally used to construct the dam. Conversely, RCC to be used in the stilling basin will be exposed to significant hydraulic loads and abrasion and required a fairly high-strength mix.
The RCC mix design programme was a two-phase programme. The first phase was completed during design of the project. Two preliminary RCC design mixes were developed as part of the final design. Use of a fairly rich mix (high cement content) was planned for the stilling basin slab and a leaner mix for use in the dam buttress and non-overflow raising sections. The design strength for the stilling basin RCC was 3000psi at 90 days. It was estimated that a mix with approximately 240 pounds of cement and 120 pounds of pozzolan per cubic yard would be required to achieve this strength. The target strength for the dam buttress and raising section was 3000psi at one year. The estimated cement and pozzolan contents per cubic yard were 150 pounds and 76 pounds respectively.
The second phase of the RCC mix testing programme was conducted early during construction, after the contractor had identified sources of all constituent materials. The trial RCC batching and testing programme included preparation of approximately 220 cylinders using 18 RCC mix proportions with various water and cement contents. The cylinders were tested at ages of 7, 28, 56, 90, 180, and 365 days. Based on laboratory testing and subsequent evaluation of results from onsite test section placement trials, mix proportions for RCC production were selected as presented in Table 1.
Two key decisions were made at this stage of the design. Firstly, it was decided to use the richer mix in only the top three lifts of the stilling basin to reduce heat generation and effect some cost savings. Secondly, the required RCC strengths were revisited. The buttress mix and lower portion of stilling basin (mix No.1) was selected for an average compressive strength of 2500psi at one year, while the upper 1m of stilling basin mix (mix No.2) was selected for an average compressive strength of 4000psi at 90 days. The gradations of fine and coarse aggregates specified for RCC production are presented in Table 2.
Because of the relatively low cement content of the buttress RCC and its potential susceptibility to deterioration from freeze thaw cycles, it was decided that a conventional reinforced concrete facing would be constructed on all exposed RCC surfaces, with the exception of the stilling basin slab which will normally be submerged.
Challenges during construction
Construction of modifications to Loch Raven dam represented a major undertaking by the City of Baltimore and presented a number of challenges during construction of the project. Some of the most significant of these are described as follows:
Construction in a suburban setting/public involvement
The dam and reservoir are located in a suburban setting, resulting in potential disruption and impacts to the public during construction. To mitigate this, a series of public meetings were held during final design of the project to inform the public of planned modifications and potential short-term impacts. Key topics of discussion at the meetings included potential for closing of local roads, work hours and associated noise impacts, and need for offsite (satellite) staging areas. Involvement of the public revealed concerns that might not have otherwise surfaced without input of the local citizens. For example, use of local roadways by construction traffic represented an area of significant concern. This was mitigated by restricting construction vehicle access to the site via specific roads. Work hours were also an area of concern, however, the citizens saw the need for, and agreed to, multiple shift operations for certain portions of the work such as RCC placement. Public meetings were held periodically throughout the project and contributed significantly to the project’s success. A letter sent to the City by a representative of the Maryland House of Delegates commended the City for it’s citizen-friendly approach to implementation of the project.
Limited construction staging areas
In addition to being located in a suburban setting, the project was also confined by surrounding features that severely limited available staging and work areas. The contractor’s highly efficient use of the available space resulted in a limited need to clear dedicated undisturbed areas surrounding the reservoir. Concrete was delivered entirely from onsite mixing plants with a conveyor system which also minimised the space needed for the work. The limited work areas also made advance delivery of materials a critical path activity. For many RCC projects, it is desirable to stockpile significant quantities of aggregates, cement, and pozzolan onsite to allow for unrestricted concrete placement. The available space permitted only nominal quantities of these materials, resulting in ‘just-in-time’ deliveries of many required materials.
Contractor value engineering/construction schedule modifications
Gannett Fleming’s original design specified RCC placement followed by constructing conventional concrete facing over exposed RCC surfaces in the spillway and non-overflow sections of the dam. The contractor proposed to construct the non-overflow facing concrete sections concurrent with the RCC placement. After reviewing updated thermal analyses and observing a demonstration during the RCC test placement operations, the proposal was recommended for approval by the engineer.
The non-overflow conventional concrete facing was batched on site using Rustler III mobile concrete batch plant and delivered to the placement area using a crane and bucket. The facing concrete was placed against the forms at a nominal width of 38cm.
RCC was also batched on site using a Johnson-Ross batch plant with a capacity of 3m3. Batched RCC was delivered to the placement area via conveyors transporting the material approximately 213m horizontally and 30m vertically. From the conveyor, the RCC was delivered to the lift surface using a CC 200-24 Rotec Creter Crane and telescopic conveyor with 360° of swing and horizontal reach to 60m. The RCC was spread in 30cm lifts and spread using a dozer against the previously placed conventional concrete mix.
Once the RCC was placed, the facing concrete was consolidated using internal vibrators. After the facing concrete was adequately consolidated, the RCC was then compacted using a smooth drum vibratory roller and hand compaction equipment. As the placement operation advanced the facing concrete forms were raised and the sequence repeated.
Within the spillway the facing consisted of 46cm of conventional reinforced concrete constructed after completion of RCC placement as originally designed. This facing concrete was also batched onsite. Spillway RCC was over-placed approximately 15cm beyond the design contact plane of RCC and facing concrete. Excess RCC was trimmed by a combination of mechanical, manual, and high-pressure water jetting prior to placing facing concrete. No. 10 anchor bars were embedded 2.4m into the RCC lifts at approximately 60cm vertical and 1.2m horizontal spacing. The contractor utilised slip-forming techniques to place the facing concrete in continuous vertical sections.
As discussed earlier, the RCC mix proportions were adjusted during the construction-phase mix design programme. The changes resulted in a need for a higher paste content to adequately fill all void space within the aggregate matrix and provide workability needed for complete compaction of the RCC. To minimise the cost impact of this change, the pozzolon contents of the mixes were substantially increased while the cement contents were decreased. At the time of this change, there was also a benefit to be realised later in the project. At the peak of RCC placement in 2004, a major world-wide cement shortage occurred that threatened to impact the project. Having decreased the cement content of the RCC mix, the overall cement demand for the project was lower and the potential impact resulting from limited cement availability was reduced. Even so, cement deliveries often arrived just in time to allow the next shift of RCC placement to occur.
Flow diversion during construction
Because Loch Raven dam is a primary source of raw water, the City required that full use of the reservoir not be impeded. Along with the fact that the existing reservoir drains were no longer operable, this meant that construction would occur without purposely lowering the reservoir. This undoubtedly represented the greatest challenge during both design and construction. Even during a relatively minor renovation to the spillway in 1983, spillway overflows occurred that substantially disrupted the work. To minimise this potential during construction, temporary flow diversion works were installed within the spillway during the installation of the post-tensioned rock anchors and subsequent construction of the RCC buttress and reshaped spillway crest. The plan provided for alternately diverting flow to either side of the spillway by installing a low bulkhead attached to the face of the dam, temporary training walls and a downstream cofferdam to allow construction to be completed in the dry.
Analyses were performed during the design of the project to generate basic information to guide the selection of an appropriate minimum level for these temporary diversion works. Elevation frequency data at the dam was computed using the previously discussed hydrologic and hydraulic models. The stage frequency information predicted by the models was compared to the results of a log-Pearson III frequency distribution based upon annual peak flows at the dam for a period of 42 years. The comparison showed the 100 year flood stages estimated by each method to be with 6cm using the station skew coefficient at the dam. In addition to these analyses, daily flow records measured at the dam at 6hr time intervals were reviewed and plotted for the period 1992 to 1997.
Additional analyses were completed to estimate the duration of time that selected reservoir stages, under existing spillway conditions, have been exceeded at the dam. This information provided the basis for establishing the minimum crest elevation for the flow diversion bulkhead to be installed across either side of the spillway during construction. The selected bulkhead height corresponds to elevation 243.0, which results a maximum of about 2m of hydrostatic pressure against the bulkhead during concrete removal required to construct the modified spillway. Structural design of the bulkhead was performed by the contractor.
The MDE permit notification process stipulated that potential induced increases in the level of the 100-year flood for areas upstream of Loch Raven dam needed to be determined to within 3cm and affected property owners notified. Analyses were prepared to determine multiple frequency peak discharges for each of the tributaries contributing flow to the reservoir. Additionally, peak discharges through the main reservoir were developed based upon the inflow and outflow hydrographs to include attenuation effects on peak rates of flow as the floodwave passes through the reservoir. Step backwater computations for multiple frequency storm events were then executed through the reservoir and continued upstream into each major tributary until the induced increases in flood stage diminished to no increase. This information was combined with the previously surveyed reservoir study data and lists of dwellings and/or commercial units subject to potential increased flooding during the construction period were compiled and the required notification letters distributed.
The recently completed modifications at Loch Raven dam were realised only by cooperation of many parties involved with the project. Although the engineering aspects of the project initially seemed to be the most significant, more challenging and complex issues unfolded as the project evolved. Maintaining the City’s water supply on an uninterrupted basis throughout the three-year duration of the project significantly affected both the design and construction approach. Through advance planning, the impacts of constructing the modifications with a full reservoir were extensively mitigated. Impact on the surrounding population residing in the relatively suburban area also became a significant factor in the overall planning and execution of the project. Through a series of public meetings that encouraged interaction among all parties, the concerns of the local citizens were identified and measures were taken to avoid major inconvenience and disruption to the citizens. With this in mind, the Loch Raven dam project represents a major accomplishment in the field of dam safety engineering.
The authors are Rodney E Holderbaum, PE, Vice President, Gannett Fleming; Donald P. Roarabaugh, PE project engineer, Gannett Fleming and Michel H. Jubran, Construction Manager, Gannett Fleming. For more information contact Rodney E. Holderbaum, 207 Senate Avenue, Camp Hill, PA 17011, US. Tel: (717) 763-7211, Ext. 2539. Fax: (717) 763-1140. Email: firstname.lastname@example.org.
This paper is published with kind permission of the Association of State Dam Safety Officials
|Historical perspective on Loch Raven dam|
The first dam built on the Gunpowder Falls to supply water for Baltimore City was constructed at Loch Raven and completed in 1881. This structure was a stone masonry dam 7.62m in height to the spillway crest with a reservoir storage capacity of about 2.3B litres. As Baltimore continued to grow, recommendations were made as early as 1896 to construct a larger dam to further develop the Gunpowder Falls Watershed as a source of supply. Construction of the first stage of the larger dam, situated about 762m upstream of the original dam, started in 1912. The project included constructing a 3m diameter riveted steel transmission conduit from the old tunnel servicing the original dam to the site of the higher dam.
The dam was completed in 1914 to an elevation of only 57m above sea level, approximately 30m below the design height, due to difficulties securing property within the proposed reservoir pool limits. Figure 1 includes a photograph and schematic of the 1912-14 construction. As shown, the foundation was placed to accommodate the full design height as part of the initial stage thus allowing the dam to be raised at a later date without further excavation within the stream bed. In 1918, Baltimore City annexed additional property and construction started in 1920 to raise Loch Raven dam to the present day spillway elevation of 73m with non-overflow sections at elevation 75m. The dam raising project was completed and put into service in 1922.
The dam that existed prior to the 2002-2005 modifications, although upgraded and modified on several occasions, was essentially the same structure that resulted from the 1922 raising. Figure 1 includes an aerial view of the ‘pre-2002’ dam with a portion of Loch Raven reservoir in the background.