To measure the hardened air content and air void properties of HPC/UHPC, 160
mm x 40 mm x 14 mm beams were used. When performing conventional concrete air void
analysis, one would typically use much larger samples cut to expose the middle portions
(because edges are typically not a good representation of the air void structure). For the
purposes of this study, the smooth edge (edge adjacent to the wall of the mold with 40 mm
x 160 mm dimensions) was lapped to expose the inner air void structure. The entire lapping
process decreased the height of the sample by about 5 to 7 mm or 35% to 50% of 14 mm
height. This, along with the small aggregate size, was considered to provide a good
representation of the inner structure.
The lapping process was performed using an automated grinding/lapping/polishing
mm of material were removed (typically 6 minutes of lapping time). This process then
repeated using No. 120 grit, No. 220 grit, No. 500 grit, and finally No. 1200 grit polishing
disks (typically 3 minutes of lapping time for each). Between lapping the sample on each
grit size, a hardening solution was used to strengthen the surface of the cementitious
material and ensure the rims of the air voids maintain their true shape. This hardening
solution consisted of 10 parts of acetone and 1 part of oil-based lacquer. After each lapping
sequence, the sample was cleaned using a soft brush and allowed to dry. The hardening
solution was then painted onto the surface of the sample and again allowed to dry before
starting the next lapping sequence. After the final lapping using a No. 1200 grit disk, the
samples were briefly (3 to 5 minutes) placed in acetone to remove any leftover hardener.
The samples were then cleaned and dried.
To prepare the polished/lapped sections for air void analysis, their surfaces were
colored black using a broad tip marker pen by marking in parallel lines with slightly
overlapping strokes. This layer was allowed to dry and a second coat of marker was applied
with the strokes 90 degrees from the first. After the second coat was allowed to dry, a layer
of white 99% pure barium sulfate with a typical particle size of 0.7 µm was placed on the
surface of the sample. The barium sulfate was then pressed into the voids using a rubber
stamper with sufficient force to ensure all voids have been filled. Excess powder was then
brushed away using the palm of the hand until a sharp contrast between black paste and
aggregate and white voids was achieved. The sample was then viewed through a
stereomicroscope to ensure an adequate contrast. At this stage, careful attention was
provided to blacken any fibers that appear white using a fine point marker. The samples
ASTM C457 Procedure A – Linear Traverse Method. This method requires the paste
content to calculate the specific surface and spacing factors of the samples. In conventional
linear traverse air void analysis, this would be determined through testing what percentage
of a line was paste, however this is a very time consuming process and Rapid Air machines
require blackened paste and aggregate, therefore the paste content was assumed based on
the mix design of the cementitious composite. While using the Rapid Air machine, a
threshold of 174 was used to distinguish white and black portions of the sample. In this
Rapid Air approach, the air void properties are determined by the chord lengths of air voids
crossing the aforementioned lines. The calculations to determine these are void properties
are further discussed in section 2.4 DEVELOPMENT OF METHODS FOR AIR VOID
ANALYSIS OF HARDENED CONCRETE.
Air void analysis was also performed using flatbed scanner techniques. This
approach used the same samples used for Rapid Air tests with the blackened paste and
aggregate and white powdered air voids. The samples were laid flat on a scanner and
scanned at 4800 dots per inch (DPI) with careful attention paid to cleaning the surface of
the scanner between the scans so that no powder remained on the glass providing false air
voids. The sizes of the samples were often too large for the scanner memory, therefore
these were scanned in pieces and later stitched together using built in stitching software
within Photoshop and then cropped so approximately 5 to 10 mm of the edge of the samples
were disregarded as these portions would not provide an accurate representation of the air
void structure. After this full image was obtained, it was converted to a binary image with
a threshold of 174 to match the threshold used in Rapid Air tests. The image was then
so that the paste and aggregate appeared white and the voids appeared black. The image
was then analyzed using the image processing software to find the number of voids,
percentage of voids space, and area of each individual void. When performing this image
analysis, attention was paid to avoid detection. Some small areas that are only comprised
of a few pixels were not considered as air voids; therefore, in order to be quantified for a
void, the area composed of a minimum of 5 pixels was considered. Additionally, the
circularity of the voids must be considered, because an elongated shape or irregular shape
should not be considered in an analysis. A minimum circularity of 0.2 was applied to
address this restriction.
After completion of the image processing, the area fraction of the voids was
recorded and the area of each individual void was plotted. In order to relate the areas to
conventional air void property calculations, circular air voids with equivalent areas were
generated for each void and their corresponding diameters were calculated. In order to use
methods for calculating the specific surface based on planar methods [125] as discussed in
section 2.4, the expected (taken as the average) value of the distribution of diameters and
the expected value of the distribution of diameters squared was required. The specific
surface was then calculated as follows where Y is the expected value of the distribution of
diameters and Y2 is the expected value of the distribution of diameters squared.
𝛼 =16〈𝑌〉
𝜋〈𝑌2〉 (Eq. 17)
Once the specific surface was determined, the spacing factor was determined using
the same procedure as the Rapid Air. Again, the paste contents had to be assumed based