There was a huge amount of energy and enthusiasm in the room when the ACI Committee 117, Tolerances, met in Puerto Rico in October. One interesting item discussed was a change in the measurement protocol for slab thickness testing.
After some debate, the committee voted to require four sample cores per 5000 square feet of floor area. This was accepted in place of the previously suggested proposal of three samples for the first 5000 square feet and then one sample per each additional 5000 square feet or part thereof.
That new requirement will become a part of the ACI 117, "Specifications for Tolerances for Concrete Construction and Materials,"slated for publication in 2008. Even though a new update was just published in 2006, its rapid updating is specifically to get it on the same three-year publishing cycle as the ACI 301, "Specifications for Structural Concrete."
The vote on how many core samples to require quite naturally led into a discussion of the coming change to statistical analysis as the basis for acceptance of concrete construction.
"The members of the committee are committed to making every tolerance in the 117 document a statistically valid tolerance," says Allen Face, The Allen Face Companies, Wilmington, N.C., in a conversation subsequent to the meeting. Face, who developed the F-Number System for specifying and measuring floor flatness, used the example of concrete floors to give those at the meeting an overview of what the change to statistics-based acceptance criteria will mean. Among his observations, two in particular stood out.
First, once you decide to go with a statistical approach and how confident you want to be in your conclusions, the math is fixed. The key here is that statistical analysis is a science. Deciding beforehand how the dimensions to be controlled are distributed on the physical entity, and how certain you want to be that the testing reflects the entity's true condition, dictates the entire sampling and analysis protocol.
Second, adopting a statistical approach will accommodate any number of samples, provided that they are randomly located. Of course, as the sample size increases, the ability to reach definitive conclusions improves. The analysis automatically will convert the sample statistics into a formal, ACI-sanctioned conclusion as to whether the subject entity meets or does not meet the specified tolerances. It is the committee's firm belief that this approach will eliminate the arbitrariness and bad science that all too often find their way into present tolerance disputes.
"This doesn't mean that contractors are going to have to become statisticians," Face says. "The tester's role will simply be to collect the measurements in accordance with the rules." Going forward, the plan would be for the committee to provide easy-to-use software, probably based on a simple spreadsheet, with all the tools needed to evaluate the data. Then it would be a simple matter of entering the data into that evaluation software, which would compare the data to the acceptance criteria and render the conclusion.
Future acceptance criteria will be based on comparing the mean and standard deviation of the data to their specified values. In everyday terms, the mean is the average value of the readings, while the standard deviation is just a reflection of how spread out the readings are around the mean. As an example, a future slab thickness specification may simply require that if the mean thickness is less than the nominal thickness it must still be within 1/4 inch, and that the standard deviation of the thickness not exceed 1/2 inch.
If the industry's experience with the F-Number System is any indication, the establishment of reasonable, rational, statistics-based acceptance criteria for the various aspects of concrete construction should benefit all involved. Providing such objective criteria also should go a long way toward helping to reduce litigation related to construction tolerances.
The 117 committee's next meeting is scheduled for April 1, 2008, in Los Angeles. It should include animated discussion on the movement toward statistically based tolerances, an update on Colin Milberg's data collection project to help establish rational tolerances, and more on the tolerance compatibility and measuring protocol documents. Milberg also is hoping to arrange a demonstration of his laser scanning equipment for the committee. So mark your calendars!