Humboldt H-4140 GeoGauge - Manual (Page 29)

29
The modulus cure rates for the individual sections generally follow that of the
average (see Figure 2). The correlation coefficients are not as good, but do they
need to be for QC/QA purposes? All data provided was used for the analysis
with the exception of the following:
• tsg01_VA288, 10/24 to 10/29 (discarded for poor measurement
quality [precision])
• Main Line_ NBL_…2 to 196+57, 7/18 to 7/25, Day 8 (questionable data
quality [missing information])
Additionally, the data from loop D_ Sec E _…Interchange, 7/18 to 7/22 was used
because of good measurement quality, but the data indicates that changes in
the site conditions may have influenced the cure rate in an unexpected way.
Figure 3 shows the number of data points used in the analysis by day in the cure
cycle, almost 1,000 points. This is many more that the 84 data points used to
determine the NM 44 cure rate. The fact that the correlation coefficients for
the 11 sets of cure rate data are typically not as good as that for the one data set
from NM 44 leads me to believe that the cure rates in the VA288 data may be
masked by changing site conditions (e.g., temperature, moisture, traffic, material
differences, installation practice, …). Based on first hand observations, these
types of conditions changed very little on NM 44. I suggest that minimizing the
effect of these variables should influence future section and test date selection.
Tables 1 and 2 contain the summary data used in the figures.
In my opinion, Figure 1 embodies the basis for a trial specification. Effective
modulus should be within +/- 2σ of the target modulus (average) 95% of the
time on any given day after installation. This specification may be unique to the
general materials and construction methods used. Also, such a specification
should include a practical spatial sample for QC testing (e.g, every 500 ft.).
Figure 1 also suggests that such a specification may be largely independent of
site conditions or even job site.