Simulating the failure of Safe Harbor Dam during “sunny day” and probable maximum flood conditions allowed the determination of the depth, duration and extent of flooding, as well as the elapsed time from dam failure to flood wave arrival at selected locations downstream.
By Jay Greska, Bryce Mochrie, Chii-Ell Tsai and Christopher Godwin
The September 2011 flooding along the Susquehanna River was a stark reminder of our susceptibility to extreme events and the need for advance planning and preparedness. Images of flooded homes and sandbagged levees in Wilkes-Barre, Pa., focused the nation’s attention on the loss of life and property associated with flooding.
For dam owners and engineers, advance planning and preparedness is an everyday necessity. The Federal Energy Regulatory Commission (FERC) requires owners of dams classified as “significant” or “high” hazard to prepare an emergency action plan (EAP), including maps showing the limits of inundation should the dam fail. Most EAPs include inundation limits that assume dam failure at the reservoir’s full pool elevation and normal river flow (sunny day failure) and at the inflow design flood, which for many large dams is the probable maximum flood (PMF).
For Safe Harbor Dam in Pennsylvania, preparation of the EAP involved creation of an unsteady-flow model using HEC-RAS software from the U.S. Army Corps of Engineers; calibration based on observed water surface elevations; and unsteady-flow simulations to establish stages, discharges and travel times for the resulting flood wave from failure of the dam under sunny day and PMF conditions.
Understanding the area
Safe Harbor Dam is about 32 miles above the mouth of the Susquehanna River at Chesapeake Bay, with its east end in Lancaster County and its west end in York County. Owned and operated by Safe Harbor Water Power Corporation (SHWPC), the concrete gravity dam is 4,869 feet long and 70 feet high, and the powerhouse has a capacity of 417.5 MW at 110,000 cubic feet per second (cfs). Lake Clarke extends about 10 miles upstream and has a surface area of 11.5 square miles.
The 31 spillway openings at Safe Harbor Dam have a combined capacity of 1,375,000 cfs at the 232-foot maximum forebay elevation. Each opening is 48 feet wide and features an ogee-shaped spillway at a crest elevation of 195 feet. The spillway is divided into two sections, with seven east and 24 west openings separated by a 1,356-foot-long bulkhead at Else Island. Through operation of the powerhouse and 31 spillway gates, a normal pool elevation of 227.2 feet is maintained.
Between Safe Harbor Dam and Holtwood Dam 7.6 miles downstream, the Susquehanna River averages about 3,000 feet wide, with depths varying from less than 10 feet just below Safe Harbor Dam to more than 50 feet at Holtwood Dam. There are no bridges along this portion of the river; the nearest crossings are 1 mile below Holtwood and 10.7 miles above Safe Harbor.
PPL Holtwood LLC’s Holtwood Dam, at the downstream end of the study reach (River Station 25), is 2,392 feet long and 55 feet high. Its 10-unit powerhouse has a discharge capacity of 31,000 cfs, with a generating capacity of 107.2 MW.
The ogee-shaped spillway is 2,368 feet long, with a concrete crest elevation of 165 feet, and is equipped with 4.75-foot-high flashboards along its entire length. The spillway controls water surface elevations upstream to the Safe Harbor Dam tailrace. The dam lacks spillway gates, so rubber dams designed to deflate at a water surface elevation of 173 feet were used to maintain the 171-foot normal pool elevation. Over the years, the rubber dams were replaced with flashboards. At water surface elevations above 174.2 feet (about 102,000 cfs), the flashboard pins fail and the flashboards deflect.
Potential downstream impacts
Under FERC’s Dam Safety Performance Monitoring Program, owners of dams that impound water for hydro powerhouses must evaluate facility safety under all potential failure modes. As such, the hydraulic performance of large dams such as Safe Harbor whose failure could endanger lives or property must be evaluated to the PMF level. During a series of visits in the summer of 2002, a FERC inspector observed fishermen, swimmers and boaters along the river between the two dams. In a November 2003 letter to SHWPC, FERC upgraded the dam’s hazard potential from low to high, citing the risk to recreational users should the dam fail.
The riverbanks between the two dams are sparsely populated, with recreational uses predominating. The lowest inhabited structure, a cottage just above Holtwood Dam, is not impacted until discharge exceeds 600,000 cfs (between a 25- and 50-year recurrence interval). Pequa Boat Club, 2.6 miles below Safe Harbor Dam, provides ramps for launching boats, with the lowest permanent building at elevation 187.2 feet. Holtwood Dam also operates public boat launches, with no permanent buildings, at Otter Creek, 3.1 miles below Safe Harbor Dam, and at York Furnace, 3.5 miles downstream. According to a field survey by SHWPC, the lowest impact elevation is at Camp Minqua, 4.2 miles below Safe Harbor, where about 30 trailer sites at a floor elevation of 176 feet are impacted at a discharge of 297,000 cfs (below the 445,000 cfs 10-year discharge).
The flood of record at Safe Harbor Dam occurred during Hurricane Agnes in June 1972. With a return frequency of 100 to 500 years, this event generated up to 18 inches of rain, resulted in 117 deaths, and caused damage from Georgia to New York. At Safe Harbor Dam, operations staff measured water surface elevations every hour on the up- and downstream sides of the dam, with a maximum elevation of 195.5 feet on the downstream side. This data would form the basis of the HEC-RAS model calibration. SHWPC’s Hurricane Agnes data also includes hourly discharges from the USGS gage station at Harrisburg and peak daily values from the Marietta gage.
By comparing the 990,000 cfs peak discharge at Harrisburg to the corresponding 1,040,000 cfs value at Marietta, the authors believed a more accurate estimate of inflows to Safe Harbor Dam would be obtained by multiplying the Harrisburg hydrograph, ordinate by ordinate, by 1.06 (the approximate ratio of peak discharges). Although the Marietta gage is closer to the site, no hourly observations were available.
The resulting inflow hydrograph – with an initial discharge of 859,000 cfs, peak of 1,050,000 cfs, and 66-hour duration – was used to simulate operations at Safe Harbor Dam during Hurricane Agnes and to calibrate the model between the two dams.
Sunny day hydrograph
Although Safe Harbor’s capacity is 417.5 MW at 110,000 cfs, SHWPC says a 100,000 cfs powerhouse discharge is typical. The project is operated as a daily peaking facility, with water stored at night when demand is low and released during the day. Because the 100,000 cfs powerhouse discharge is considerably greater than the 37,000 cfs average annual flow, spills at Safe Harbor Dam are relatively rare. Barring a flood or gate opening, discharges on the Lower Susquehanna River seldom exceed 100,000 cfs.
Consequently, a 66-hour-long inflow hydrograph with a pre-breach discharge of 100,000 cfs was used. To minimize oscillations at the start of the model runs and ensure downstream peak stages are associated with the breach, the sunny day simulations begin at 70,000 cfs and gradually increase to 100,000 cfs 10 hours into the simulation. Flow continues at 100,000 cfs until the end. For the sunny day breaches, dam failure begins nine hours after the 100,000 cfs hydrograph levels off, thereby ensuring steady flow prior to the breach. All inflow hydrographs are applied at the upstream end of the reservoir, 10.7 miles above Safe Harbor Dam, with no inflow or tributary flow between the two dams.
For the PMF simulations, an inflow hydrograph developed in the 1960s by Chas. T. Main Inc. was used. This hydrograph has an initial discharge of 900,000 cfs, peak of 1,401,000 cfs, and 42-hour duration. Like the sunny day hydrograph, the PMF is purely theoretical; therefore, the starting date and time for each simulation have no effect on peak stages, discharges or travel times for the flood wave, which depend on downstream distance and elapsed time since the breach.
In September 2006, PB Power (now called Parsons Brinckerhoff) was retained to perform a dam breach simulation and provide other engineering services as required to determine the dam’s hazard classification. In November 2006, the authors obtained the Federal Emergency Management Agency’s HEC-2 input data set for the Susquehanna River. FEMA used this all-encompassing numerical model to prepare the Flood Insurance Study for Lancaster County. It includes more than 100 field-surveyed hydraulic cross-sections based on more than 2,000 individual ground points above and below the water surface.
In 2008, this cross-section data was imported into HEC-RAS and the elevations compared with streambed elevations obtained by SHWPC in 1994. The FEMA cross-section elevations, which date back to the 1980s, were adjusted accordingly. Because FEMA modeled the entire Susquehanna River within Lancaster County, the model was truncated to include only those cross-sections between Holtwood Dam (R/S 25) and the upstream end of the Safe Harbor impoundment (R/S 86).
The resulting HEC-RAS model contains 116 cross sections – 64 imported from HEC-2, the remainder interpolated by the HEC-RAS program. Of the 64 cross sections, 35 are located above Safe Harbor Dam, and 29 are between Safe Harbor and Holtwood dams, with eight of these at or near sites that are inhabited year-round or seasonally. These eight sections are included in the data tables, with peak stages, discharges and travel times calculated at each.
Additional adjustments to the FEMA HEC-2 model were necessary to simulate the hydraulic characteristics of the two dams. Using as-built plans of Safe Harbor Dam, an inline structure was added at R/S 52.5 to model the dam, spillway and gates. Three spillway gate groups with a total of 31 gates were added, with each gate 48 feet wide at an invert elevation of 195 feet. A lateral structure was then added at R/S 53.1 to represent flow through the Safe Harbor powerhouse.
Because Holtwood Dam represents the downstream study limit, a stage-discharge curve based on the hydraulic characteristics of the spillway, flashboards and powerhouse was used in lieu of an inline structure to model it.
In April 2004, SHWPC plotted a series of water surface profiles for the Susquehanna River, beginning at Holtwood Dam and extending upstream to Safe Harbor. These profiles are based on the Holtwood Dam stage-discharge curve, the tailwater curve developed for Safe Harbor Dam, and high-water marks recorded between the two dams during Hurricane Agnes and lesser floods. In addition to the 37,000 cfs average annual flow, profiles were plotted for discharges between 100,000 and 1,000,000 cfs, in 100,000 cfs increments. An additional profile, based on stage-discharge relationships at the two dams, was plotted for the 1,400,000 cfs PMF. These profiles were used to calibrate the HEC-RAS model.
Using the inflow hydrograph developed for Hurricane Agnes, the HEC-RAS model was run and the calculated water surface elevation on the downstream side of the dam compared to the 195.5 foot observation. The Manning’s “n” values were adjusted as necessary and the model re-run. This iterative process of Manning’s “n” adjustment, model runs and comparison of results was repeated until a water surface profile was generated with an elevation closely matching the 195.5 value on the downstream side of the dam. Table 1 provides results of the Hurricane Agnes model calibration.
To ensure accuracy under sunny day conditions, two additional calibrations were performed, with the resulting water surface profiles closely matching the 177.92 and 171.5 foot observations on the downstream side of Safe Harbor Dam at 200,000 and 37,000 cfs, respectively.
Dam breach analysis
In accordance with the FERC Engineering Guidelines, a 104.5-foot-wide, top-to-bottom failure of the dam encompassing three monoliths was investigated. For purposes of this investigation, a monolith consists of a single 48-foot-wide spillway or a single 8.5-foot-wide pier. Drawings provided by SHWPC indicate foundation elevations at or above 162 feet, so a base-of-breach elevation of 162 feet was used.
For the sunny day breaches, a 227.2-foot reservoir elevation and 0.2 hour time to failure were assumed. For the sunny day breach and non-breach runs, the dam’s spillway gates are closed, with all flow directed through the Safe Harbor powerhouse before the breach. Because the powerhouse is manned around the clock, peak stages, discharges and travel times for the breach wave are based on the assumption that the powerhouse is shut down promptly once the breach occurs.
The PMF breach was also based on a 0.2 hour time to failure. Unlike the sunny day simulations where pre-breach flows and stages are constant, for the PMF dam failure does not occur until the reservoir rises to the 231.6 foot elevation indicated on drawings provided by SHWPC. Unlike the sunny day simulations where the Safe Harbor Dam spillway gates remain closed, all 31 spillway gates are assumed open the full 35 feet at the time of the breach. For the sunny day simulations, the 2,368-foot-long flashboard section at Holtwood Dam is assumed to fail at a water surface elevation of 174.2 feet. For the Hurricane Agnes and PMF simulations, it is assumed the Holtwood Dam flashboards have already failed, resulting in a bare dam.
As stated in its 2005 letter, FERC EAP guidelines suggest failure of two or more monoliths at the foundation contact. Thus, a 104.5-foot-wide breach encompassing two 48-foot-wide spillway gates and one 8.5-foot-wide pier was modeled. For a breach this size, the 100,000 cfs river flow maximizes the depth (and area) of inundation between the two dams and is therefore appropriate for the preparation of inundation maps for the sunny day event (see Table 2 on page 52).
As expected, the elapsed times to breach wave arrival increase in the downstream direction. However, the elapsed times to peak stage decrease. This is due to the flashboards at Holtwood Dam, which are designed to fail when the water surface elevation exceeds 174.2 feet.
As Table 2 shows, the breach wave reaches Holtwood Dam in about 30 minutes. Six minutes later, the river has risen to elevation 174.2 feet, causing the flashboards to fail. This lowers the river, creating a negative surge that propagates upstream. At Holtwood Dam, the sudden drop pre-empts the breach wave, with the peak stage occurring shortly thereafter. Because the negative surge takes 20 to 25 minutes to reach Safe Harbor Dam, pre-emption of the breach wave (and resultant peak stage) at each upstream cross section does not occur until the negative surge arrives.
At a forebay elevation of 227.2 feet, the breach wave resulting from a 104.5-foot-wide breach at Safe Harbor Dam is less than 1 foot high throughout the study reach. At Camp Minqua, the 174.7-foot peak stage is 1.3 feet below the 176-foot impact elevation, indicating no hazard to lives or property (see Figure 1).
FERC guidelines stipulate that incremental hazards for the inflow design flood (in this case the PMF) be evaluated based on the assumption that the dam fails at the peak of the flood hydrograph. The PMF hydrograph for Safe Harbor Dam has a peak inflow of 1,401,000 cfs and a peak stage, on the upstream side of the dam, of 231.6 feet. Using this hydrograph, the HEC-RAS model was run assuming no failure of the dam, to establish baseline stages and discharges throughout the study reach. Assuming no breach, the HEC-RAS PMF water surface elevations are 201.8 feet at Safe Harbor Dam, 200.6 feet at Shenks Ferry, 198.9 feet at Pequea, and 196.1 feet at Camp Minqua.
To assess the downstream impacts associated with a three-monolith breach at Safe Harbor Dam during a PMF event, the HEC-RAS model was run and the peak breach and non-breach stages compared. In Table 3, peak breach flow at R/S 52 does not include powerhouse flow, which re-enters the river immediately downstream. Therefore, its value is about 100,000 cfs lower than corresponding values at the downstream cross sections, which include powerhouse flow. For the 104.5-foot-wide breach, the 1,499,400 cfs peak discharge exceeds the 1,401,000 cfs PMF peak discharge by less than 100,000 cfs. The result is a breach wave less than 2 feet high throughout the study reach, including Camp Minqua where it is less than 1 foot high (see Figure 2).
Summary of findings
The sunny day breach at Safe Harbor Dam is 104.5 feet wide and represents a top-to-bottom failure of two spillway gates and one pier. At 100,000 cfs, the resulting breach wave varies from 0.1 to 0.9 foot in height, with a 0.5-foot incremental rise and 1.3 feet of freeboard at Camp Minqua. Because the 104.5-foot-wide-breach at 100,000 cfs involves two or more monoliths, peak stages, discharges and travel times associated with this scenario (see Table 2) are appropriate for the preparation of inundation maps and other EAP documents.
The PMF breach is a 104.5-foot-wide, top-to-bottom failure of two spillway gates and one pier. The breach wave varies from 0.6 to 1.8 foot high, with a 0.9 foot incremental rise at Camp Minqua. According to SHWPC, the PMF discharge at Holtwood Dam is 1,700,000 cfs, or about 245,000 cfs more than the peak discharge there resulting from a 104.5-foot breach at Safe Harbor. Stability analyses for Holtwood Dam indicate it would not fail during the 1,700,000 cfs PMF event; therefore, it is stable during the 150,000± cfs sunny day breach and 1,455,000± cfs PMF breach. The top-to-bottom failure of two spillway gates and one pier at Safe Harbor Dam during a PMF event would not result in a similar failure at Holtwood Dam, and the PMF peak stages, discharges and travel times in Table 3 are appropriate for preparing inundation maps and other documents.
Dam breach analyses are typically undertaken to estimate the magnitude and extent of flooding likely to occur should a dam fail and to determine the associated impacts. For the Hurricane Agnes simulation, HEC-RAS results indicate a water surface elevation of 195.46 feet just below Safe Harbor Dam, closely matching the observed high water mark. For the sunny day breach simulation, HEC-RAS results indicate a flood wave less than 1 foot high between the two dams. Therefore, no hazards to lives or property are likely to result from the top-to-bottom failure of two spillway gates and one pier.
For the PMF breach simulation, HEC-RAS results indicate that the flood wave resulting from the failure of two spillway gates and one pier will be less than 2 feet high between Safe Harbor and Holtwood dams, with inundation depths that vary from 10 to more than 21 feet. This indicates that any impacts associated with a PMF event are likely to occur well before the dam fails, allowing individuals time to evacuate.
Parsons Brinckerhoff thanks Juan Kim-ble, P.E., president and CEO, and Ted Rineer, P.E., project engineer, of SHWPC for assistance in preparing this article.
Jay Greska is lead engineer, Bryce Mochrie is senior project engineer, Chii-Ell Tsai is supervising engineer and Christopher Godwin is civil engineer with Parsons Brinckerhoff.
This article has been evaluated and edited in accordance with reviews conducted by two or more professionals who have relevant expertise. These peer reviewers judge manuscripts for technical accuracy, usefulness, and overall importance within the hydroelectric industry.