The objective of this research was to investigate the stripping problem in bituminous mixtures using both conventional and image processing techniques (IPT). Partial factorial combinations of asphalt mixture slabs were fabricated. A Marshall mold was modified to produce slabs of 300 mm × 300 mm × 60 mm. The mixtures were prepared using different combinations: two types of aggregates (limestone and valley gravel), two asphalt penetration grades (80/100 and 60/70), three types of additives (lime, cement, and a mixture of lime and cement), and one aggregate gradation (average of ASTM boundaries). The slabs were exposed to one of four environmental conditioning techniques: wet-dry (W-D), freezing-thawing (F-T), water immersion (I), and surface bonding (S-B). Eighty-one slabs were prepared, 19 were tested without conditioning (control slabs), and the rest were exposed to different weathering conditioning. Four cores were prepared from each slab, two for the indirect tensile strength test and the others for the Marshall stability test.
A normal-based camera was used to map the surfaces of the control and conditioned slabs as well as cross-sections generated from cores tested under the indirect tensile strength test. The Silicon Video Image Processing (SVIP) software was used to digitize images for the surfaces of slabs and cores. Images were analyzed to prepare statistical parameters, intensity diagrams, and histograms for intensities based on two approaches: area-based analysis and line-based analysis.
Results of the analysis indicated that 1) conventional and image processing techniques proved that surface bonding increased stability and tensile strength for conditioned slabs more than for unconditioned slabs; 2) IPT predicted stripping in limestone mixtures with AC 80/100 using the retained intensity concept and this result agreed with that of retained Marshall stability (RMS); however, it did not consistently predict the stripping potential of mixtures consisting of valley gravel due to the tone intensity effect of aggregate; 3) lime additive demonstrated the best indirect tensile strength and Marshall stability; however, the additive mix consisting of lime and cement exhibited reduction in water damage for mixtures with AC 60/70; 4) slab mixtures resisted weather conditioning more than conventional cylindrical Marshall specimens; and 5) an area-based analysis scheme using peaks, not averages of intensities, was more reliable in predicting stripping than the line-based analysis, and it was more reliable for core cross-sections than slab surfaces.