Court determines that spillway redesign lacked sound engineering
|The redesigned 500-foot-long, horseshoe-shaped spillway contributed to flooding of a stream in Northwest Ohio.|
The flooded property was a 87,500-square-foot racquet and health club known as the RecPlex that was built on 21 acres of land south of the city of Celina, Ohio, a few hundred yards west of Grand Lake St. Marys. The state of Ohio constructed Grand Lake between 1835 and 1845 to serve as a source of water for the Miami-Erie Canal System. In 1997, the state replaced the lake’s 39.4-foot-long western spillway that emptied into Beaver Creek with a horseshoe-shaped spillway that was 500 feet long. When heavy rains passed through Northwest Ohio in early July 2003, rainwater overflow exiting Grand Lake from the west bank spillway flooded Beaver Creek, which lies just to the north of the RecPlex. The entire length of Beaver Creek overflowed its banks, flooding the RecPlex and closing the facility for months for extensive repairs.
Was the flood an unavoidable natural disaster or was the new spillway or some other human activity the cause of the flood? Could the flooding have been prevented? Should it have been prevented?
Like a hose in the kitchen sink
In response to the 1977 Johnstown, Pa., flood, where the failure of six dams caused 77 deaths, the U.S. Army Corps of Engineers (Corps) inspected dams around the country to determine if they were safe. The Corps defined a safe dam as one that would survive the Potential Maximum Flood (PMF), defined as "a flood that may be expected from the most severe combination of critical meteorological and hydraulic conditions that are reasonably possible." In 1978, the Corps inspected Grand Lake and concluded that Celina would flood if the PMF occurred. In response, the Ohio Department of Natural Resources (ODNR), the owner and operator of Grand Lake, substantially increased the length of the western spillway at Grand Lake so that the PMF would pass through the lake without overtopping the lake’s embankments in the Celina area. Although this change protected Celina, it also had unintended consequences.
The first and most complete analysis of the law of unintended consequences was done in 1936 by the American sociologist Robert K. Merton. In an influential article titled "The Unanticipated Consequences of Purposive Social Action," Merton identified five sources of unanticipated consequences. The first two—and the most pervasive—were ignorance and error. Analysis would show that the state of Ohio was guilty of both.
When the inevitable heavy storms passed through the Grand Lake area in 2003, a vastly greater amount of water discharged over the spillway into Beaver Creek. However, Beaver Creek did not have the capacity to handle the increased flow. The effect was like replacing the faucet in the kitchen sink with a fire hose.
When the proposed spillway redesign was made public in 1990, local agencies and landowners, as well as the Corps, warned ODNR of the likelihood of greater flooding along Beaver Creek. The county engineer specifically identified the athletic club as a property that would be endangered by the new spillway. The state ignored the warnings and entreaties.
When the new spillway was completed in 1997, it was 500 feet long—450 feet at 871.5 feet above mean sea level (msl) with a 50-foot notch at 870.6 feet msl. It had two, 60-inch-diameter outlets near the bottom of the structure that could be opened manually to release water and lower the level of the lake.
With the new spillway, the Ohio Department of Natural Resources deemed the lake level to be self-regulating and ceased manual drawdowns, leaving little capacity for stormwater.
Expert analysis of the circumstances concluded that the state had radically departed from accepted engineering practices. In its single-minded focus on preventing the lake from overtopping its western embankments, ODNR had failed to obtain, or properly evaluate, much of the geological information regarding Grand Lake and Beaver Creek or the historical precipitation, streamflow, lake level, and other available hydrologic data; had incorrectly and incompletely modeled the data; had failed to undertake a sensitivity analysis to evaluate other spillway designs; had ignored other, feasible flood control measures; and had adopted a disastrous lake management policy.
The cause of the flooding was determined by an analysis of the rainfall data (recorded since 1913); lake level information (available beginning in 1927); streamflow records; drainage basin characteristics for Grand Lake (including area, shape, slope, land use, topography, and soil types); the characteristics of Beaver Creek (such as geometry, slope, roughness, and constrictions); the basis of ODNR’s design of the replacement spillway (including the calculations and evaluations that ODNR had performed); and the statutes, regulations, policy, guidance, and other documents applicable to dams, spillways, and floodplains.
With this information, legal counsel and hydrological consultants Conestoga-Rovers & Associates (CRA) performed a number of evaluations.
Using accurate lake level measurements (discovery revealed that the ODNR lake level gauges were not accurate and had to be recalibrated after the fact) CRA calculated the amount of water that was (or would have) discharged over the 39.4-foot spillway and the 500-foot spillway into Beaver Creek. This analysis had not been performed by ODNR during spillway design. Had the state done so, it would have learned that the 500-foot spillway would have caused Beaver Creek to flood, on average, every five years over the 70-year period of record. The analysis also showed that the 39.4-foot spillway would never have caused Beaver Creek to overflow its banks in the vicinity of the RecPlex.
Using the (corrected) historical lake levels, rainfall, and streamflow data, CRA determined the precipitation events that would have potentially resulted in flooding in the area (i.e., severe storms), another analysis that the state failed to undertake. The purpose of determining and analyzing the significant storm events is to understand how rainfall in the area has previously caused flooding. The omission of this analysis is akin to redesigning a highway without reviewing the amount of traffic that previously traveled on the road during rush hour.
CRA ran a series of HEC-1 and HEC-2 models/simulations to analyze the historical rainfall and lake level data. HEC-1 programs use rainfall data and hydrologic information (such as the physical/geologic make-up of the local drainage basin and various climatological parameters) to determine the discharge (amount of flow) at a given point. HEC-1 models are commonly used to simulate the effects of precipitation events in a given area.
HEC-2 programs use the HEC-1 discharge values together with cross-section information (such as the width, depth, and slope at given points in a channel or stream) to calculate the water elevation at those selected points. The purpose of the HEC-2 model is to simulate water surface elevations resulting from a discharge of water into a specific stream, such as Beaver Creek. CRA used these modeling techniques to review the 16 largest storms on record, dating back to 1913, and to analyze the flooding effects of those storms for both the 39.4-foot and 500-foot spillways.
Although the state had performed some modeling prior to designing the spillway, its work had three significant oversights. First, ODNR had modeled only hypothetical, and not the historical, precipitation events. Secondly, it only simulated 72-hour events. However, our investigation revealed that, in this drainage basin, the flood events resulting in flooding of Beaver Creek were 120-hour or longer events. Most critically, although the state did model the discharge from Grand Lake into Beaver Creek, it failed to take the critical next step: modeling the flows into Beaver Creek to determine the resultant increase in surface water elevation. Had it done so, as sound engineering practice dictates, the state would have learned that even a 72-hour storm event would cause an unacceptable increase in the water surface elevation.
Perhaps the most damning conclusion was that the RecPlex would not have flooded even once during the 16 most severe storm events with the 39.4-foot spillway. By contrast, the 500-foot spillway would have caused Beaver Creek to overflow its banks and flood the business (and surrounding properties) no fewer than 10 times out of the 16 storms.
Depositions of ODNR personnel responsible for managing the lake produced a startling discovery regarding the state’s lake level management policy, or lack thereof. Prior to 1997, ODNR had engaged in manual drawdowns of the lake’s water level. However, once the new spillway was constructed, ODNR ceased doing so, deeming Grand Lake to be a self-regulating lake, meaning that the state would not intercede to maintain water at a safe level. The 60-inch outlets went essentially unused.
Analysis of the lake level data revealed the danger that this policy created. Since 1997, lake levels exceeded the height of the spillway notch (normal pool level) almost 75 percent of the time; exceeded the height at which water overflows the entire 500-foot length of the spillway more than 25 percent of the time; and exceeded the level at which water overflowing the spillway would cause Beaver Creek to overtop its banks 10 percent of the time.
The obvious problem with ODNR’s policy is that more water stored in the lake increases the likelihood that a storm will cause Beaver Creek to flood and that the flooding will be significant. A sink that is consistently filled to the rim is more likely to flood the kitchen. Just days before the Independence Day storm that devastated the area, the water in Grand Lake was more than 7 inches above normal pool level.
Beaver Creek is shown here at its normal level at a point just below the Grand Lake spillway.
ODNR only considered two potential spillway designs. However, expert analysis showed that even a cursory analysis by the state would have disclosed other designs that could successfully convey the PMF through the lake while also reducing the frequency and severity of flooding of local properties. Selecting a less destructive spillway design was not the only flood control measure that was available. The state had failed to investigate other options, such as modestly raising the height of the western embankment in the Celina area, maintaining a manual lake level drawdown policy, widening or deepening Beaver Creek in the area of the spillway, discharging a portion of the flood waters through the eastern outlet structure (which, unaccountably had not been designed to do so, despite the recommendation of the state’s consultant) instead of disgorging all of the water into Beaver Creek, or some combination of these measures.
The court’s decision
In its June 2008 decision in Case Leasing & Rental, Inc. v. Ohio Department of Natural Resources, the Court of Claims of Ohio blasted the state for failing to comply with accepted hydrologic engineering practices when it redesigned the spillway. The state’s failure to investigate and evaluate the available data, the shortcomings of the modeling of hypothetical storms that it did perform, its poorly considered lake management policy, and its failure to consider and implement other flood management options were cited by the court as the basis for its finding that the state had been negligent. Based on the data that was available to it at the time, ODNR knew or should have known that installation of the replacement spillway as designed would result in more frequent and more severe flooding to downstream landowners.
The court was not persuaded by the state’s claim that it should not be held liable because its primary objective in building the new spillway was to keep the west bank of the lake from overtopping and flooding the city of Celina. "The court does not dispute the utility of ODNR’s dam safety objective; however, balanced against the gravity of the foreseeable and avoidable harm caused, the court finds that the manner in which ODNR implemented its objective was unreasonable and negligent."
In light of ODNR’s statutory role as watchdog over dam safety in Ohio, and the rigorous proofs it requires of private dam owners, some may find it surprising that the agency didn’t "practice what it preaches." But the result in Ohio is hardly unique. Witness the failure of the government to properly construct and maintain the levees around New Orleans before Katrina, and numerous other avoidable flooding disasters. It may be that some of these errors and omissions are inevitable, as Robert Merton suggests. But it may also be that the frequency and seriousness of these mistakes are exacerbated because we have entrusted the fox to guard the hen house.
The authors gratefully acknowledge the many contributions of Pressley Campbell and Charles Munce of Conestoga-Rovers & Associates—from the education of their lawyers in the science of hydrology to their tireless and imaginative devotion to the pursuit of truth.
Steve Samuels, a partner and coordinator of the Environmental Practice area at Schottenstein Zox & Dunn (SZD), Columbus, Ohio, has extensive experience regarding environmental and natural resource permits, compliance and enforcement proceedings, and zoning matters. He can be contacted at firstname.lastname@example.org. Linda Mindrutiu is an associate in SZD’s Environmental, Business Restructuring and Reorganization, Creditors’ Rights and Corporate Trust, and Commercial Litigation Practice areas. She can be contacted at email@example.com.