The site extends 39 miles from the outlet of Lake Winnebago into Green Bay. Contamination is found primarily in the first six miles (OU1) and the last 20 miles (OU3 and OU4) of the river, reaching PCB sediment concentrations as high as 3,000 ppm. Contaminated sediment extends to depths of up to 13 feet. Commercial shipping is confined to the last 3.5 miles of OU4, which ends at the mouth of Green Bay. Sediment distribution in the middle 13 miles (OU2) is patchy with intervening bedrock exposures. OU2 contains the site's greatest elevation drop and consequently the majority of the river's locks and dams. OU5 comprises the entire bay, but dredging and capping are anticipated only near the river mouth; monitored natural recovery was selected as the remedy for most of OU2 and OU5.
Based on experience in OU1, dredging did not always remove all sediment above the site-specific cleanup goal of 1 ppm. This goal and the surface weighted-average concentration goal of 0.25 ppm can only be met with a combined approach of dredging, capping, and sand covers. Dredging could not be used in areas with in-water infrastructure such as bulkhead walls or docks, near utility crossings associated with commercial facilities, or near shorelines with steep banks. As a result, a capping approach using four designs (differing in stone and/or sand components) was used to contain residual contamination and non-dredged areas.
Design criteria for dredging included maximum horizontal/vertical slopes of 3:1 for submerged areas and 5:1 for shorelines with nearby infrastructure. A typical dredge management unit covers approximately 6 acres, which involves approximately 1 week of dredging per unit. Prior to full-scale dredging operations, models were used to develop final plans.
Dredging was initiated in OU1 in 2004 followed by the placement of sand covers in areas where residual PCB concentrations remained above 5 ppm; cover thickness ranged from 3 to 6 inches, depending on PCB concentrations. Sand covers also were placed on undredged areas with lower PCB concentrations; caps composed of sand and armor stone were placed over higher PCB concentrations. To date, more than 370,000 yd3 of PCB-contaminated sediment have been removed from OU1, of which approximately 8,000 yd3 contained PCB concentrations above 50 ppm and were classified as TSCA in situ sediment. The sediments were disposed at local state-regulated landfills. Water generated by dredging, sand recovery, and dewatering is treated through bag filters, sand filtration, and granulated activated carbon prior to river discharge. Air monitoring to date has shown no exceedance of thresholds for particulates or PCBs.
Similar operations will continue downstream (mainly in OUs 3 through 4) from 2009 to 2017. Hydraulic dredges and in-water pipelines are used to remove and transport contaminated sediment to a staging area where sand fractions will be washed or otherwise treated for possible beneficial onsite or offsite use.
Engineered caps will be placed at locations where the final cap height allows for at least 3 feet of water and at least 2 feet below the authorized navigational water depth. In order to ensure caps will have long-term stability and effectively contain PCBs, the following conditions were evaluated during cap design:
Frazil ice, particularly in areas of turbulent waters downstream of dams or possible high water velocities due to ice dams;
High flows from tributary flooding; and
Combinations of river flow, due to seiche effects or lake level changes.
To further ensure effective caps, different cap designs will be employed for different river conditions. In navigable areas, nearshore areas, or where sediment contains PCB concentrations exceeding 50 ppm, the armor consists of a 33-inch layer of sand, gravel, and quarry spall. For design and construction purposes, this cap type was designated as a 'Cap C.' (Figure 3). In other areas, the cap consists of 13 inches of sand and gravel (Cap A) covering approximately 200 acres, or 16 inches of sand and gravel (Cap B) covering approximately 70 acres. An estimated 25 additional acres of shoreline caps will be constructed using larger armor stone to be determined case by case. The areal extent of each cap and total acreage of cap types is subject to ongoing refinement. Other areas having sediment profiles with PCB concentrations below 2 ppm or in thin deposits (6 inches or less) are covered with 6 inches of clean sand imported from local sand and gravel pits. Shortly after placement, the thickness of each cap layer is verified to specified engineering standards.
Design of the cap for the navigation channel of OU4 involved site-specific models considering erosive forces such as wave-induced currents, river flows, ice scour, seiche, and vessel propeller wash. This 'JETWASH' model is similar to that recommended by the U.S. EPA and USACE but includes additive velocities accounting for the propeller shaft pitch and reflection relative to the river bottom, which are critical factors for recreational boats. The model includes a momentum-based approach to analyze stability of a bed armoring particle that is subjected to time-dependent propeller wash velocity fluctuations typical of recreational vessels. Due to the wide variety of recreational vessels and modes of operation used in the Fox River, the propwash model relied on Monte Carlo simulations using 2,500 parameter combinations for water depths of 3, 5, 7, and 10 feet each. In addition, two-dimensional hydrodynamic models were evaluated for OUs 1, 3, 4, and 5 to predict bottom shear stresses during maximum flow anticipated in a 100-year flood event. USACE empirical models were used to predict characteristic vessel waves. To consider potential impacts from propwash, a Monte Carlo statistical model was applied, incorporating:
Capping costs and stone size;
Magnitude of likely damage if movement occurs; and
Degree to which a cap can 'self-heal' to re-cover areas where cap materials may have moved due to propwash or other influences such as vessel anchors.
Finally, if regularly scheduled or event-triggered monitoring indicates that caps have been eroded or otherwise adversely impacted, the caps will be repaired as necessary to provide continued containment.
Institutional controls currently include fish consumption advisories throughout the site, establishment of no-wake areas in OUs 3 and 4, and limited public access to waters undergoing active dredging or cap construction. Monitoring during active dredging and cap construction includes routine geophysical surveys and core sampling 2 and 4 years after the initial post-construction survey and every 5 years thereafter. Details of this monitoring plan are being developed during the design phase for OUs 2-5. Long-term monitoring also will include testing of cap integrity and performance with respect to contaminant containment, and analytical sampling of water and fish tissue at least every 5 years to evaluate environmental results.
Approximately 30 acres of caps have been installed to date. Following dredging and cap construction, additional time will be needed for natural recovery, which will result in additional contaminant reductions meeting other cleanup goals. In OU 3, for example, an estimated nine years of natural recovery are needed to reduce PCB fish tissue concentrations to 0.049 ppm, the threshold for unlimited walleye consumption. Construction completion is scheduled for 2018.