Field Testing of an Ultra-High Performance Concrete Overlay

CHAPTER 5. RESULTS AND DISCUSSION

A summary of results from bond testing is shown in table 5. Fourteen bond tests were completed. Specimens exhibited 1 of 2 failure modes: Mode 1, which is failure in the substrate, or Mode 4, which is failure in the adhesive layer between the test disc and the UHPC overlay. Results are presented in three groups: (1) test locations G1 and G2; (2); test location G3; and (3) test locations B1 and B7.

Table 5. Summary of results from bond testing.

TEST LOCATIONS G1 AND G2

Specimens tested from region G1 and G2, which represented a test location with no indication of delamination and scarification prior to placement of the UHPC overlay, all failed in the adhesive between the test disc and the UHPC overlay (Mode 4). Peak stresses recorded prior to failure were between 182.9 psi (1.26 MPa) and 420.3 psi (2.90 MPa). Figure 30 shows a set of test discs removed from G1 and G2 samples. The figure shows the face of the disc that was bonded to the UHPC overlay. Portions of the failure are seen in the UHPC and at the interface between the adhesive and the UHPC.

Specimens failing in Mode 4 were subsequently removed from the test location by applying a prying action to the sample using a hand tool. In each case, the prying action resulted in fracture of the substrate concrete. Figure 30 and Figure 31 show a representative sample (G2-9) from the G1 and G2 test locations. The bond between the UHPC overlay and the substrate concrete appears intact.

Figure 30. Photo. Test disc removed from samples from G1 and G2 locations.

A. Front of G2-9 specimen after testing.	B. Back of G2-9 specimen after testing.	C. UHPC-concrete interface of G2-9 specimen after testing.

Figure 31. Photos. Specimen G2-9 after testing.

Figure 32 shows the core location from which sample G2-9 was removed. As expected, apparent damage or delamination in the deck concrete or at the UHPC-concrete interface was not present.

Figure 32. Photo. Core location for specimen G2-9.

The microstructural analysis revealed findings that were complementary to data gathered by pull-off testing and visual inspection of test samples. The initial observations from the General map of the G2-8 specimen (shown in Figure 33) revealed the absence of steel fibers in direct contact with the concrete substrate. The width of the UHPC paste band in proximity to the interface exhibiting an absence of steel fibers varied from 0.015 in (0.4 mm) up to 0.09 in (2.4 mm). Roughness induced by the scarification is also visible on the General map by the “valley-hill” configuration of the surface of the concrete deck where the UHPC overlay was deposited.

The General map shown in figure 33 displays a sound interface with close physical contact between hydrated phase of UHPC and the existing concrete deck substrate. This observation is justified by the absence of a gap dividing the paste from the concrete substrate and the presence of hydrated cement products—most likely calcium silicate hydrated (C-S-H) and portlandite—on the interface. The high degree of direct contact between the UHPC overlay and the concrete substrates justifies the relatively high tensile stresses sustained during bond strength testing and the absence of failure at the interface. However, this direct contact between hydrated products of the UHPC overlay and the concrete substrate was locally disrupted by the presence of entrapped air, (shown in figure 34) or poor consolidation or accumulation of debris, which is shown in figure 35. The presence of these features near the UHPC-concrete interface was marginal and did not compromise the overall bond performance.

Figure 33. Electron Microscope Image. General map of the UHPC-concrete interface for specimen G2-8.

Figure 34. Electron Microscope Image. BSE image of a void between the UHPC overlay and the substrate concrete.

Figure 35. Electron Microscope Image. BSE image of poor consolidation or debris accumulation at the interface.

Distribution of the porosity, aggregate, unhydrated, and hydrated cement particles on the UHPC overlay as a function of the distance from the concrete surface is presented in figure 36. This graph shows a high content of hydrated products, above 80 percent, at the interface (from 0- to 10-μm distance). The percentage of hydrated products was reduced as the distance from the interface increased. This diminishing percent of hydrated products was caused by an increase of the aggregate content in the UHPC paste. At approximately 90-μm distance from the interface, the content of both hydrated and aggregate stabilized at from 46 and 40 percent, respectively. This effect has been observed in other types of overlays, and it is associated with reduction in the packing efficiency of the UHPC paste in the vicinity of the concrete surface, also known as “wall-effect.” (Garboczi and Bentz 1991; Ollivier, Maso, and Bourdette 1995; Scrivener, Crumbie, and Laugesen 2004; H Beushausen and Gillmer 2014) The inflexion point on the content of phases, hydrated product, and aggregates at 90-μm distance from the surface is depicted in figure 36 with a vertical dashed red line that marks the extent of wall effect on the UHPC paste.

Figure 36. Graph. Overall distribution at the interface as a function of distance from the concrete surface of specimen G2-8 phases.

The content of unhydrated cement was below 8 percent in the 10-μm gap close to the surface. The value reached a maximum of approximately 17 percent at 30 μm away of the surface and progressively decreased to 10 percent at 100-μm distance. The porosity also showed a maximum of 7 percent close to the interface in the 10-μm-wide gap close to the surface. The high packing efficiency of the UHPC and low water-to-cement ratios explained the comparative low values of porosity relative to other overlay configurations where ordinary portland cement or grout materials were used. These systems typically showed porosity values on the interface on the order of from 30 to 20 percent. (De La Varga et al. 2017; Hans Beushausen, Höhlig, and Talotti 2017)

TEST LOCATION G3

Specimens tested from region G3, which represented an apparent region of good bond without scarification prior to placement of the UHPC overlay, all failed in the concrete substrate (Mode 1). Peak stresses recorded prior to failure were between 344.6 psi (2.37 MPa) and 474.6 psi (3.27 MPa). Figure 37 shows a representative sample (G3-13) from the G3 test location. The bond between the UHPC overlay and the substrate concrete appears intact. Further, the difference in the UHPC-concrete interface texture can be observed when comparing the photos in Figure 31-C and Figure 37-C. Figure 38 shows the core location from which sample G3-13 was removed. As expected, apparent damage or delamination in the deck concrete or at the UHPC-concrete interface was not present.

A. Front of G3-13 specimen after testing.	B. Back of G3-13 specimen after testing.	C. UHPC-concrete interface of G3-13 specimen after testing.

Figure 37. Photos. Specimen G3-13 after testing.

Figure 38. Photo. Core location for specimen G3-13.

The UHPC-concrete substrate interface of the G3-14 specimen had characteristics similar to those identified on the G2-8 sample. As shown in figure 39, the width of the UHPC paste adjacent to the concrete surface free of steel fibers varied between 0.024 in (0.6 mm) and 0.1 in (2.6 mm). This band was also characterized by its high degree of contact with the concrete surface. This was also motivated by high content of hydrated products of the UHPC paste in the vicinity of the concrete surface.

Figure 39. Electron Microscope Image. General map of the UHPC-concrete interface for specimen G3-14.

The main difference of this specimen relative to the two analyzed under the SEM is the absence of scarification on the concrete surface. This can be observed in the General map shown in figure 40, which shows that the interface line is relatively flat and lacks the “valley-hill” topography observed in the G2-8 sample shown in figure 33. This flat configuration contributed to improving the packing efficiency of the overlay material in direct contact with the substrate. (De La Varga et al. 2017) The direct consequence was a reduction in the wall effect on this interface compared with those having a valley-hill topography (surfaces previously subjected to scarification).

The trends in the distribution of phases on the interface for the G3-14 specimen, displayed in figure 40, were similar to those in the previous specimen, G2-8. Hydrated products of the UHPC paste were the principal phase in the first 100 μm of the interface microstructure. Its value progressively decreased from 80 percent in the first 0 to 10 μm to approximately 55 percent at 80 μm away from the concrete surface. In parallel to the decrease in hydrated products, the aggregate content increased from 0 percent right at the interface to a value of approximately 25 percent at 80 μm. The content of both unhydrated particles and porosity evolved following the same trend as the one observed in specimen G2-8. In the case of the unhydrated particles, the initial 8 percent content right at the surface increased to approximately 20 percent at 30 μm away of the surface and progressively decreased to 10 percent at 100-μm distance. While at initial 10 percent value of porosity, the interface constantly diminished to below 2 percent at 100-μm distance from the surface.

Finally, the distribution graph in figure 40 also confirms a reduction in the extent of the UHPC paste affected by the wall effect induced by the “plateau” configuration of the concrete surface. The inflexion point in this G3-14 specimen was observed at 80-μm distance from the surface instead of the 90-μm mark registered in the previous specimen (G2-8).

Figure 40. Graph. Overall distribution at the interface as a function of distance from the concrete surface of specimen G3-14 phases.

TEST LOCATIONS B1 AND B7

Specimens tested from regions B1 and B7, which represented regions of potential delamination with scarification prior to placement of the UHPC overlay, all failed in the concrete substrate (Mode 1). Peak stresses recorded prior to failure were between 0 psi (0 MPa) and 114 psi (0.79 MPa), which were relatively low compared with other regions. Two of the four samples (B1-3 and B7-12) were loose after completing the coring procedure. These two samples were removed without applying load (i.e., they were loose after completing the particle coring). Further examination of the four test samples and their respective core locations indicated that pre-existing delamination was present in the conventional concrete deck. Figure 41 shows a representative sample (B7-11) from the B1 and B7 test locations. Although the bond between the UHPC overlay and the substrate concrete appears intact, pre-existing damage can be observed within the concrete substrate layer.

A. Front of B7-11 specimen after testing.	B. Back of B7-11 specimen after testing.	C. UHPC-concrete interface of B7-11 specimen after testing.

Figure 41. Photos. Specimen B7-11 after testing.

This can be most clearly observed in figure 42, which shows the core location from which sample B7-11 was removed. A horizontal crack plane can be observed within the concrete deck layer approximately 0.5 in (12 mm) below the UHPC-concrete interface. Further, it is evident that when sample B7-11 failed, failure was initiated from this existing crack plane.

Figure 42. Photo. Core location for specimen B7-11.

The microstructural analysis of specimen B7-12 also revealed a dense interface dominated by the presence of hydrated products, as in the previous two samples. In this case, the width of the UHPC band without steel fibers ranged from 0.024 in (0.6 mm) to 0.12 in (3.0 mm). The typical valley-hill topography was observed on the concrete surface as illustrated in figure 43. Additionally, as in the case of the G2-8 specimen, entrapped air and accumulated debris were visualized disrupting the contact between UHPC hydrated products and concrete surface. However, the extent of this disruption was limited, and it was not expected to have a deleterious effect on the overall bond strength of the system.

Figure 43. Electron Microscope Image. General map of the UHPC-concrete interface for specimen B7-12.

As shown in figure 44, the results of quantitative analysis for this specimen are similar to those measured in G2-8. At the interface, the hydrated products were above 80 percent, and porosity was around 7 percent. The aggregate content showed a similar evolution, reaching 40 percent at 100-μm distance from the interface. This similarity with the G2-8 specimens also extended to the distance where the inflexion point in the trend of hydrated particles and aggregate was located. Based on the data shown in figure 44, it was inferred that this limit was located close to 100-μm distance from the surface. Significantly, this sample came from a location where the concrete surface was subjected to scarification. Thus, as in specimen G2-8, the distance from where the inflexion point occurred was significantly higher than the distance observed for the nonscarified sample (G3-12).

Figure 44. Graph. Overall distribution at the interface as a function of distance from the
concrete surface of the specimen B7-12 phases.

SUMMARY OF RESULTS

Generally speaking, results indicated good bonding between the UHPC overlay and the substrate concrete deck, with and without scarification prior to overlay placement. Peak tensile stresses sustained by specimens in regions G1, G2, and G3 prior to failure were comparable if not higher than the bond strengths of other UHPC-class materials (nonthixotropic) bonded to a roughened concrete substrate. (Haber and Graybeal 2016)

Furthermore, peak tensile stresses sustained by these specimens were, in most cases, higher than those exhibited by conventional grout-like materials bonded to a roughened concrete substrate. (De La Varga, Haber, and Graybeal 2017)

As for regions of potential delamination, it was found that bond between the UHPC overlay and the substrate concrete was intact, based on visual inspection. The UHPC-concrete interface appeared similar to that observed in samples taken from locations G1, G2, and G3. It was evident that failure of specimens from regions B1 and B7 was the result of pre-existing concrete deck delamination. This caused specimens to carry little-to-no tensile load prior to failure. Notably, other regions of potential delamination (B2, B6, and B8) were not tested, and findings from tests completed at locations B1 and B7 may not extrapolate to these locations. Many of these untested regions were small and located near the boundaries of the two overlay stages.