The joint venture won the best-value bid in 2002 and is responsible for the dam portion that will complete the project. The Corps contract requires the dam be built "in the wet," instead of using cofferdams, because of a 1997 Corps study that cited the method as faster and less costly, says Gilmour.

The dam will have five, 110-ft-long radial tainter gates that will close on sills fitted between six piers with downstream stilling basins. The shells are linked by pintles and receivers—like 3,500-ton Lego blocks.

The stilling-basin shells have energy-dissipating blocks to break up the water cascading down the incline below the gates. When cast, the shells are only two feet thick in most parts. A waffle pattern creates hollow spaces underneath the shells after they are placed. Crews solidify the hollows with tremie concrete; the finished shells will form both the foundation and the finished surface of the dam.

Once the gate section is done, smaller shells will be placed farther along the dam line in the navigable- pass section. The shells will be used to mount 140 30-ft-tall wicket gates.

When the river rises, gates are laid flat on the bottom of the riverbed so that vessels can pass. When the river is low, tainter gates are closed to force more water over the wickets. Cranes raise the wickets in sections to increase the impoundment level upstream. If all the gates are raised, the navigable pass is no longer used, and vessels have to pass through adjacent locks.

Building in the wet with off-site modular fabrication is challenging, says Gilmour. "I think we have been very successful at doing what we are doing," he says, but he has reservations. "I think this heavy-lift concept for in-the-wet construction has its place, but it may or may not be a good choice for Olmsted Dam."

However, the team is deep in the throes of the heavy-lift method and not about to change course. The solutions achieved and the innovations developed here may find use in future dam construction, says Gilmour. But tackling the project brought risk. "Even when we put the [bid] proposal together, there were some issues that we used a magic pencil on because we had no idea how they were going to be done," says URS's Bennett. "We knew we were going to have to figure out a way to set these shells to these kind of tolerances, but we didn't know how."

The contractor's goal is to meet the 2016 completion deadline. Success depends on funding and the river itself. This construction season, the immediate goal is to place six tainter- gate shells—a goal the river threatens to squelch.

Waterworks

"The contract says shell-placing season is between June 15 and Nov. 30, but we place them any time the river allows, even in February. This year, the river was too fast to start until mid-July. Water levels were higher than they had been since 1937," says Bennett.

One of the challenges was grading the riverbed. Workers crafted a machine they call "the Aquadigger," which is a Komatsu PC 3000 hydraulic excavator, ring- mounted on a 50-ft-wide by 180-ft-long barge that dredges to a depth of 85 ft. An "autodig" feature reads 3D excavation plans of the bottom's finished profile and screeds automatically to a +0.0 to -6.0-in. tolerance. Autodig prevents cuts that are too deep.

Compacting was another challenge. The original strategy was to subcontract vibro compaction, but URS decided to self-perform the work, thinking it would be more economical. It modified the hydraulic hammer used to drive piles underwater, equipping it with a big, flat head. "It's like a dirt compactor used to make sidewalks, only 100 times stronger," says Bennett.

Once the bed is prepared, grout mats are placed to protect it until the shells are laid. The mats must weather the winter and spring, when the river runs too fast and high for placement. But some work can continue, such as driving 400 24-in.-dia pipe piles through the mat. The contract calls for minimizing diving, so URS relies heavily on robotics.

The crew needed to drive piles efficiently and accurately 120 ft below the river surface in fast water. "We went through so many scenarios," says Bob Goreing, URS's construction area manager. "We thought about using 120-ft piles, driving them, cutting them off where we needed and then welding those cut pieces together to make more."

Realizing how expensive and slow that would be, Goreing came up with an idea for a barge-mounted, hydraulically actuated, pile-driving template. A younger co-worker, marine superintendent Joel Ayers, used Dassault System's SolidWorks 3D design software to develop the idea into a design for a hammer and template that could hold up to seven piles per setup.