This past June, thanks to teleconferencing, I was able to sit in on the kickoff meeting for Colin Milberg's “Improved Tolerance Management in Concrete Construction” project. The project has three main objectives: developing construction process capabilities (CPCs) for various aspects of concrete construction based on field measurements; establishing new tolerances for cast-in-place concrete work based on those CPCs; and establishing measurement protocols for meaningful and comparable data acquisition in the field. (For a more complete description, see How Close Are We Getting from the July issue).
Like many meetings, this one had its share of routine, but necessary topics to be covered. But for me, the two most interesting items of discussion were about the realities of taking field measurements.
One issue concerned the windows of opportunity for collecting field data. For example, in the first phase where vertical elements in high-rise construction are being observed, the data collection team will have to get in and out quickly. There is only a short time between stripping the forms and when the area becomes cluttered with materials and equipment for forming and shoring the next level. The contractors participating in the meeting all agreed that some desired data points always will be unavailable, but taking measurements as soon as possible after stripping forms will be the best way to obtain maximum data. Because materials and equipment constantly are shifting around, scanning an entire level may require that the crew come out on two occasions, which will depend on when the area is stripped.
The other reality check was when Milberg asked for feedback on an appropriate grid for scanning, noting that he was thinking of data points measured an inch apart, which provoked several gasps. One of the contractors present said he thought every six inches would be plenty, so Milberg said he would probably go for either a 3-inch grid, or if they had time, a 1-inch grid.
Apparently, even using a sophisticated 3-D laser scanner, measuring a floor on a 1-inch grid takes five or more hours, while a 3-inch grid could be done in less than half the time. It was at that point in the discussion the reality of the project scope began to sink in: there's going to be a lot of data generated in this process.
Meanwhile, I recently visited a new residential high-rise being built in Chicago where the contractor is meeting the floor tolerances in a different way. Rather than building the finished concrete slab at exactly the finished floor elevation, the contractor is leaving each floor an inch or so low with just a darby finish. Later, subcontractor Morton Grove, Ill.-based Barrier Corp. pumps in an engineered gypsum slab cap that produces a beautiful, smooth, and hard surface with minimal labor.
The floor elevation tolerance is ¼ inch in 10 feet, which Barrier president Paul Helmer says is not only easy to meet, but also allows his crew to match the fixed elevation of stair thresholds, window frames, and plumbing components. Using a simple rotating laser, the company's engineers measure and mark on a 5-foot grid how low the floor is compared to the desired finish elevation. That's obviously not enough data density for Milberg's use, but plenty for the Barrier crew to put in good floors on this project.
In one 15x20-foot room, for example, 20 data points had been marked. The required fill went from 1/8 inch in one corner to 2 inches in the opposite corner. An added benefit of this approach is that it covers footprints, gouges, and the occasional piece of tie wire in the as-poured slab surface. In the meantime, the workers installing drywall and mechanical and electrical systems did not have to worry about protecting the surface of the finished floor. Incidentally, by the time I left the site less than two hours later, you could walk on the new floor in that room.
So from San Diego to Chicago, and from measuring on a 1-inch grid to building off a 5-foot spacing, people are looking at different ways to address the tolerance issue, and getting closer all the time.