Preparing for the Worst

By Josh Petersen and Peter Haug

In the early 1900s, a number of cost-efficient dams were being constructed across the U.S. These projects were built to impound reservoirs used by hydropower plants, which in turn provided power for industrialization, or for communities that dreamt of electrification.

In 1911, Henry M. Byllesby shared this vision and decided to provide power to the rural farming communities between Minneapolis and Rochester, Minn., through the utility Northern States Power. Byllesby hired the Boston-based engineering consulting firm Ambursen Hydraulic Co. to design a dam, using its own patented, hollow buttress design known as an “Ambursen dam.”

Although Ambursen’s designs were of high quality and quite reliable, neither Byllesby nor the designers could have predicted that the dam – located on the Cannon River in Dakota County, Minn. – would stand today almost exactly as it was built more than a century ago.

The dam is a 55-foot-tall, 2,000-foot-long hollow concrete structure that houses three original Francis turbine-generator units, giving the project a cumulative capacity of 2.2 MW. Northern States Power, now known as Xcel Energy, sold the plant to Dakota and Goodhue counties during the nuclear boom of the mid-20th Century, at which point the project was considered near the end of its useful service life.

Management by making do

Through the 1970s and 1980s, initiatives from the U.S. Army Corps of Engineers and a number of state agencies required “unregulated” dams like Byllesby to be investigated and inspected. Many, including Byllesby, fell into the pitfall of not having enough overflow or gate flow capacity to accommodate larger flood events. This was due in large part to their original designers, who, in the early 1900s, had very accurate understandings of upstream hydrology. However, they could not account for storms, the frequency interval of large storms, and changes in upstream land use practices.

As such, dam owners with limited budgets were often forced to make the best with what they had in order to accommodate design inflow capacities, leading to a number of creative design practices to manage flooding. At Byllesby, it was determined that a fuse plug structure would be the cheapest and most feasible option to prevent damage to the dam during extreme flood events.

The problem with plugs

While fuse plugs have been cited by a number of entities – including the U.S. Department of Interior’s Bureau of Reclamation and the International Commission on Large Dams (ICOLD) – as being a cost-effective and safe alternative to gated spillways, their performance comes with a level of uncertainty.

Research continues to be published on improvements to design and breach performance, although it is notable that none of the fuse plugs installed on four Reclamation dams since 1985 have been performance tested. They came under further fire in 2003, when a failure at Michigan’s Silver Lake Dam was blamed in part on a fuse plug. And though subsequent investigations showed the incident was a design and construction issue, the failure still led to an acute loss of confidence in fuse plugs by many dam safety regulators.

Byllesby Dam’s fuse plug, installed in 1977, was sufficient to meet dam safety concerns for the next 30 years. However, regulatory hurdles related to passing the full probable maximum flood (PMF) and concerns with the fuse plug’s activation forced Dakota County to reconsider its adequacy in the early 2000s.

Byllesby Dam’s new spillway began operation following the project’s completion in June 2014.
Byllesby Dam’s new spillway began operation following the project’s completion in June 2014.

Because of the very conservative potential failure mode criteria and implementation of the potential failure mode analysis (PFMA) for smaller hydro operators like Dakota County, many owners were suddenly informed that they had to now pass the inflow design flood (IDF), which is commonly the PMF for high-hazard dams. Potential failure modes provided realization and insight into the regulatory realm to create a more structured and timeline-driven process to ensure high-hazard structures were being adequately addressed.

In 2006, a federally-mandated PFMA re-opened the issue of spillway adequacy at Byllesby. Whereas the existing arrangement had for years met regulatory acceptance, the combination of the PFMA process, more severe assumed failure parameters, and a heightened distrust of fuse plugs led to a determination by the Federal Energy Regulatory Commission (FERC) that the flood passage capacity at Byllesby Dam was unacceptable. Furthermore, even if the fuse plug was acceptable from a reliability standpoint, regulators considered that it activated too frequently at the 400-year return event. Thus, FERC demanded a major modification to the structure.

Capacity concerns

Given the magnitude of discrepancy between the project’s flood of record, spill capacity and extremely high PMF, the design team started from scratch and began to re-analyze all components of the problem. Eliminating the fuse plug reduced the facility’s flood capacity to less than a third of the existing PMF.

The last report for Byllesby Dam determining the PMF was released in 1985, during the second Part 12D inspection using a largely un-calibrated watershed and a probable maximum precipitation (PMP) from the National Oceanographic and Atmospheric Administration’s Hydrometeorological Report No. 51, also known as HMR51. These original hydrology models were fairly inaccurate, making the PMP data used in HMR51 unreasonable.

HMRs 51, 52 and 53 were very conservative reports produced by NOAA in conjunction with the Corps. The data used storm events predating 1980 that relied on hurricane-related storm events and other storm distributions taken from southern states to generate a generic rainfall distribution for northern states. In other words, the study was not regionally-specific and therefore was too conservative for southern Minnesota.

Byllesby Dam remained largely unchanged until modern safety standards necessitated an overhaul and addition of a new crest gate spillway in 2014.
Byllesby Dam remained largely unchanged until modern safety standards necessitated an overhaul and addition of a new crest gate spillway in 2014.

Also, since these reports were completed, the Electric Power Research Institute (EPRI) sponsored a study of Wisconsin and Michigan in 1993 that suggested a much lower PMP could be used to determine the PMF than what was demonstrated in HMR51.

With the knowledge of this study – and the fact that the original data was nearing 30 years in age – it was determined more feasible to hire a meteorology consultant to analyze and provide a revised version of the PMF. By completing several weather models based on the 1993 EPRI study, Dakota County’s consultant was effectively able to prove that the PMP for the watershed could be drastically reduced. When analyzing the PMF in a new HEC-HMS model, Dakota County effectively reduced the IDF requirement for Byllesby Dam from 120,000 cfs to 80,800 cfs, or about a 30% reduction, by using the best, most recent data.

Once Dakota County had a firm understanding of what the goal was, it was easy to identify the necessary new capacity by confirming the current spillway capacity and subtracting the lost components, such as the fuse plug and auxiliary structures. Then, the utility could begin to look at the design process to determine which type of structures could fill the gap of remaining spillway capacity necessary to meet the the IDF.

Selecting a solution

The spillway design selection process became the most difficult process in the entire project. Due to the limitations of space on both sides of the facility and the placement of structures, there were only so many configurations that could be used. There were also issues upon evaluation regarding which type of conveyance system could be used with the most reliability.

During the process of selecting a type of conveyance structure, it was determined that modifying the existing fuse plug would be the most fiscally feasible option. The modified fuse plug would have been a dual-stage fuse plug system, with individual chambers and wing walls to protect the surrounding chambers as they “failed” sequentially, thus releasing flood flows downstream without damaging the dam structure.

This proposed design gained regional regulatory approvals but was met with opposition at the federal level because the then-recent experience with Silver Lake Dam had created doubts about any sort of “fusible” spillway.

An overhead view of Dakota County’s Byllesby Dam shows the fuse plug in its original configuration.
An overhead view of Dakota County’s Byllesby Dam shows the fuse plug in its original configuration.

With any type of fuse plug off the table, the project construction costs then increased by as much as three times. Selecting between gate structures – including crest, tainter, obermeyer or rubber dam – became a matter of preference for type of operation and maintenance because preliminary capital costs for all types of gates needed for the project would have been similar.

The final problem wasn’t about the gates at all, but how to fit the gate structure into a tight location while minimizing the downstream tailwater conditions to prevent the necessity of costly retaining walls. The difference between a curved downstream sheet pile retaining wall and a standard straight-cantilevered retaining wall to accommodate the tight space would have been about $1 million. Thus, it was quickly realized that the orientation of the structure was going to be key in reducing costs, failure modes and constructability concerns given Byllesby’s location between two parks and, at that time, an uncooperative land owner.

After completing assessments for each potential design, a crest gate spillway was planned directly downstream of the existing fuse plug structure to save the cost of building a large engineered cofferdam during construction.

Benefits of the PFMA process

The construction PFMA proved to be very beneficial to the owners of Byllesby Dam. FERC had expressed initial concerns that the curved tailrace wall would produce super-elevated flow concentrations and accelerated bedrock erosion, so the design team rotated the spillway to turn subcritical flow and keep subcritical flow going downstream.

FERC also expressed concerns about bedrock erosion, but the thick dolomite on site was eventually approved as a spillway surface, provided that all the cracks and joints were treated and that future bedrock erosion was addressed through adaptive management.

The last outcome of the construction PFMA was an integrated monitoring plan for post-construction performance of the spillway. The design included the addition of piezometers under the apron, pier, walls and embankment, plus redundant underdrains, cutoff walls filter drains and bedrock anchors. FERC also suggested that one tall approach wall’s post- tensioned anchors could be eliminated if the footing were made larger.

In short, to reduce potential failure modes, multiple redundant mitigation measures were included with the design. The design package was finalized shortly thereafter, and FERC had few substantive comments once the construction PFMA concerns were addressed.

Contractor scheduling and construction challenges

As FERC granted conditional approval for the spillway gate design in early 2013, Dakota County realized the schedule’s critical path included the gate installation. To expedite the construction schedule, gates were pre-ordered from Steel-Fab in Fitchburg, Mass., under a separate contract between the dam owner and gate manufacturer before Edward Kraemer & Sons was selected as general contractor. This did complicate the schedule coordination and warranty repair discussions, but, in the end, the decisions saved the county about four months and several hundreds of thousands of dollars.

Construction of the spillway upgrade began in May 2013 and ended in August 2014. The building period was marked with difficult conditions, including more than 600 hours of temperatures below 0 degrees Fahrenheit, minimal average wind gusts of 20 miles per hour, and 70 inches – or more than 30% above normal – of snowfall. Opposite the extreme cold, temperatures also spiked at more than 99 degrees.

This downstream view shows Byllesby’s existing fuse plug wall being removed before the construction of the new crest gate spillway.
This downstream view shows Byllesby’s existing fuse plug wall being removed before the construction of the new crest gate spillway.

By the end of construction, the contractor had removed more than 40,000 cubic yards of earth, 9,000 cubic yards of dolomite and 2,500 cubic yards of existing concrete structure. Two 65-foot-wide by 14-foot-long hydraulic crest gates were installed to pass 37,500 cfs, or 47% of the IDF, and 1,500 feet of perimeter dike was raised to contain the maximum IDF pool.

Looking back at the schedule, Dakota County’s decision to pre-order the gates allowed the new gates to be operational during a June 2014 flood, which occurred less than a month after the gates were wet-tested and the cofferdam removed. The flood demonstrated not only that the new gates work well, but also some weak seams in the dolomite bedrock that were remediated in July 2015.

What price is the probable maximum flood?

The project was a success in terms of meeting an array of regulatory requirements and overcoming design, construction and contracting challenges. Yet, questions linger regarding the high cost of the effort.

Today, Byllesby Dam can pass the full probable maximum flood and is in compliance with federal spillway capacity requirements. It is noteworthy that before the upgrades, the dam was capable of passing 80% of the PMF. To date, the cost to provide the last 20% has been .2 million. And although the cost might seem high, Dakota County has an obligation to balance the decrease in dam safety risk against the sacrifice of other services and facilities.

Josh Petersen is senior water resources engineer for Dakota County Environmental Resources. Pete Haug is a water resources engineer for Ayres Associates.

Previous articleHow Reclamation is Dealing with Biofouling at Lower Colorado River Dams
Next articlePhysical Modeling to Develop Approach Velocities at Trashrack Intakes

No posts to display