The University of Maryland data was summarized in three different testing locations along the floor:Measurement lines parallel to the construction joint and in the middle of the stripMeasurement lines parallel to, and 1 foot away from, the construction jointMeasurement lines perpendicular to the construction joint and crossing all six placements
The box above shows that the FF values at the University of Maryland decreased by about 10 to 20 percent in 7 months.
The initial measurements for the industrial warehouse in Pennsylvania were taken in October 1990 and the final measurements in October 1991. The initial overall FF and FL were 68.7 and 73.4, respectively. One year later, the FF and FL numbers were 41.4 and 36.8, respectively. The floor flatness values had decreased by about 40 percent and the floor levelness values by about 50 percent. A sample floor surface profile illustrating the vertical deviation of the floor at the two different ages is shown in Figure 1. Curling is certainly evident in this figure as shown by the significant peaks at each sawcut joint.
Predicting effects of curling on F-numbers
As mentioned previously, the length of lost contact area as a result of curling is about 20 percent of the joint spacing length at each end of the slab. For a 15-foot joint spacing, the slab curl would be expected to change the profile for about 3 feet from each end. It is possible to take F-number readings from floors with differing profiles, download those values into a spreadsheet, then add a known amount of curl, and calculate new F-numbers. This allows a comparison of F-numbers before and after the curl.
Using this analytical procedure, we recreated the F-number data for the industrial warehouse by starting with a 15-foot panel and working with an analytical F-number profile that was FF 69 and FL 72. The actual industrial warehouse slab had a 15-foot joint spacing and was initially measured as FF 69 and FL 73. The initial analytical and measured surface profiles before curling are nearly identical. Upward curling values of 1/16 inch, 1/8 inch and 1/4 inch were assumed, along with profile variations being represented by a curve starting 3 feet from each end of the panel. Table 1 shows the results.
As stated, the initial measured F-numbers for this slab were FF 69/FL 73 and these had dropped to FF 41/FL 37 a year later. This represented changes of about 40 percent and 50 percent for FF and FL, respectively. These are about the same percentage changes shown in Table 1 that would result from slightly less than 1/8 inch of curl. For the surface profile shown in Figure 1, the average curl is about 1/8 inch. This provides some verification that the analysis method gives a reasonable estimate of as-built curling variations with time.
This analysis method was also used to examine the effect of curling on floors with initial 15-foot joint spacings and considered to be moderately flat, flat and very flat as defined in ACI 117-06. Table 2 summarizes the data. The amount of curling, as measured by upward deflection values ranging from 1/16 inch to 1/4 inch, had a greater effect on the flatness of the floors with higher initial FF numbers. Flatness of floors with an initial FF 25 didn't change much for curling deflections of 1/16 inch and 1/8 inch. A curling deflection of 1/4 inch, however, resulted in a significant reduction in flatness.
Based on this analysis method, the ACI 117-90 statement that curling will not significantly change a floor's FF value is not generally true. As shown by the analysis, floors with an initial FF 25 would experience a flatness change of less than 10 percent when 1/8 inch of curl occurred. At 1/4 inch of curl, however, the percentage change is much larger. For a higher initial FF value, even 1/16 inch of curl can cause a reduction in the FF value of about 10 percent. This analysis indicates that floors with high initial FF values will be more adversely affected by small amounts of curl.
Both analytical and measured values of initial and final surface profile measurements before and after curling have shown that curling can significantly affect F-numbers. To eliminate any curling effect, all F-number measurements must be made within the 72 hours stated in ACI 117-06.