**3. Results**

After the cracks and internal voids were identified within each of the aggregates, their characteristics could be quantified and compared. Of primary interest for this analysis was the determination of volume and surface area characteristics. In particular, in order to compare the results of the CT-analysis with those of other non-destructive measurement techniques, it was important to separate surface-connected cracks and voids (referred here to as "open voids") from internally isolated cracks and voids (referred to here as "closed voids"). This is because measurements of internal surface area using the Brunauer–Emmett–Teller (BET) method [53,54] only account for internal voids accessible from outside of the sample. Similarly, the measurement of void volume through mercury porosimetry is thought to depend primarily on the saturation of internal voids that are connected to the stone surface. Such a separation of surface-connected cracks could be completed using a connected

components analysis in which only cracks and voids containing voxels that touched the stone surface were retained.

The crack/pore surface area measurements obtained using this approach are provided for all of the river-gravel type aggregates in Figure 13 and for both river-gravel and quarried-stone aggregates of the minerals rhyolite and greywacke in Figure 14. Note that the measurements for the greywacke (GK4) and rhyolite (GK1) river gravels appear in both figures for comparison purposes. Figure 15 also provides 3D images of cracking distributions for two selected individual grains of the same mineral (greywacke), where one grain has been extracted from a quarry and the other has been taken from river gravel. For images of all analysed aggregate types, see Appendix A. Tabulated values of the surface area measurements are also provided for each individual sample in Appendix B.

**Figure 13.** Measured surface area for river-gravel type aggregates (based on figure in [10]).

**Figure 14.** Surface area measurement comparison between quarried-stone (GK2 and GK3) and river-gravel (GK1 and GK4) aggregates of the same mineral types (based on figures in [10,11]).

**Figure 15.** Example images of cracking within a quarried greywacke aggregate (**a**) and a river-gravel type greywacke aggregate (**b**). Red pixels denote externally accessible cracks/pores and yellow pixels denote externally inaccessible (closed) cracks/pores (reproduction of figures from [10,11]).

Error bars in Figures 13 and 14 could not be calculated. This is because there is still no universally accepted method for estimating the combined error introduced by CT measurement systems and image processing algorithms. Although numerous approaches for estimating such error bars have been proposed [55,56], these tend to be rather computationally intensive and time consuming and remain an active area of research. For pure dimensional measurements, it is common to use either the voxel size or the focal spot size of the X-ray tube as an estimation of error. Given that the focal spot size of this scanning system was significantly smaller than the voxel sizes obtained during these investigations, the voxel sizes for each scan (also listed in Figures 13 and 14) can be taken as an estimation of possible error in the dimensional measurements.

From Figure 14 it is clear that the amount of internal cracking (including both surface-connected cracking and non-surface-connected cracking) in the river gravel aggregates was much higher than that in the quarried aggregates, even when their mineralogical characteristics were similar. The magnitude of this effect is also much too large to be attributed to variations in CT resolution. The underlying basis for these differences in quantitative crack measurements can also be estimated through visual observation of CT images, such as those displayed in Figure 15. Clear, layered cracking is visible within river gravel greywacke stones; this is not present within the greywacke stones extracted from quarries. This is thought to result from the aggressive weathering process that river gravel is subjected to during its lifecycle prior to construction use. This indicates that the selection of high-quality aggregate based on mineralogical characteristics alone may be insufficient.

It is also clear from Figure 13 that even for stones from a single source and with a single mineral composition, a significant amount of variability in internal porosity and cracking is present. The variation between individual stones of a single type (such as rhyolite (GK1)) is often much larger than the average difference between two entirely different stone types (such as between rhyolite (GK1) and granite (GK4)). Thus, we recommend the use of large statistical samples to properly characterize each stone type for ASR sensitivity.

### **4. Discussion and Conclusions**

This research clearly demonstrates the need for universal, automated, and consistent crack detection methods that allow the cross comparison of results from large quantities of CT-scan data from di fferent sample types. A framework, called "virtual data fusion", was developed that has the potential to successfully provide such a method. A partial implementation of this method in a custom program was developed for use in research focused on crack measurement in ASR-sensitive aggregates. Our results demonstrated the success of the program in e ffectively identifying crack-like structures and measuring their characteristics such as crack extension (relative surface area) and surface connectivity.

These results demonstrate the significant impact that the source of extraction can have on the characteristics of aggregates. Even for aggregates of the same mineral type, river gravels contain significantly higher levels of internal porosity and cracking than quarried stone. This is thought to result from the aggressive weathering process that river gravel is subjected to prior to its selection and use for construction. This indicates that the selection of high-quality aggregate based on mineralogical characteristics alone may be insu fficient. It is also clear from these results that there is a significant amount of variability in internal porosity and cracking even for stones with the same mineralogical characteristics and extraction source. Thus, large statistical samples will be necessary to properly characterize each stone type for ASR sensitivity.

**Author Contributions:** Conceptualization, T.O.; methodology, T.O. and F.W.; software, T.O.; investigation, T.O. and F.W.; resources, T.O., F.W., and G.B.; writing—original draft preparation, T.O. and F.W.; writing—review and editing, T.O., F.W., and G.B.; visualization, T.O.; supervision, F.W. and G.B.; project administration, F.W.; funding acquisition, F.W. All authors have read and agreed to the published version of the manuscript.

**Funding:** This presentation is based on parts of the research project carried out at the request of the Federal Ministry of Transport and Digital Infrastructure, represented by the Federal Highway Research Institute, under research project No. 06.0108/2014/BRB. The author is solely responsible for the content. (Dieser Präsentation liegen Teile der im Auftrag des Bundesministeriums für Verkehr und digitale Infrastruktur, vertreten durch die Bundesanstalt für Straßenwesen, unter FE-Nr. 06.0108/2014/BRB durchgeführten Forschungsarbeit zugrunde. Die Verantwortung für den Inhalt liegt allein beim Autor).

**Acknowledgments:** The authors would like to thank Dietmar Meinel from the Bundesanstalt für Materialforschung und—prüfung (BAM, Federal Institute for Materials Research and Testing) for his guidance and support during the CT scanning and data analysis.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

### **Appendix A. Visualization of Individual Grains**

A visual impression of the internal microstructure of the individual grains is provided by selected CT-based visualizations (Figures A1–A4). In these images, the solid material of the individual grains is shown semi-transparently. This allows a better spatial visualization of the cracks and pores emanating from the outer surface (i.e., open voids—colored red) and the cracks and pores not accessible from the outside (i.e., closed voids—colored yellow).

**Figure A1.** *Cont.*

**Figure A1.** CT-visualizations for individual river gravel aggregate of type GK1 (the left and right images for each aggregate show views from the 0 and 90 degrees, respectively) (based on figures in [10,11]).

**Figure A2.** *Cont.*

**Figure A2.** CT-visualizations for individual river gravel aggregate of type GK4 (the left and right images for each aggregate show views from the 0 and 90 degrees, respectively) (based on figures in [10,11]).


**Figure A3.** CT-visualizations for individual greywacke aggregate from quarried stone (GK2) and river gravel (GK4) (based on figures in [10,11]).


**Figure A4.** CT-visualizations for individual rhyolite aggregate from quarried stone (GK3) and river gravel (GK1) (based on figures in [10,11]).

The strong fluctuation in the amount of open and closed voids in the individual quartz/quartzite grains of aggregate type GK1 is clearly visible in Figure A1. For the rhyolite of aggregate type GK1, Figure A1 includes a CT-based visualization of the single grain with the highest surface areas of open and closed voids. The broad spectrum of open and closed void surface characteristics for plutonite is demonstrated by the CT-based visualizations of the two individual grains shown in Figure A1.

In contrast to the GK1 aggregate, the surface characteristics of the individual grains of aggregate type GK4 tend to vary much less. In Figure A2, a CT-based visualization of the individual sandstone grain with the highest surface area of open voids of all examined sandstone grains is provided. As expected, this image shows a relatively high content of open voids and a low content of closed voids. The spatial arrangemen<sup>t</sup> of the open voids suggests a layered structure. The examined single grains of mudstone and greywacke also show a stratification due to the similar formation history. In the latter case, however, the closed voids predominate over open voids. The final CT-based visualization in Figure A2 is a single grain of granite and demonstrates that this type of rock can also have high contents of open voids.

As mentioned in the paper, the individual grains of the quarried stone have very small open and closed void surface areas compared to the river gravel grains. This is impressively documented by the CT-based visualizations for greywacke and rhyolite, which are shown in Figures A3 and A4 and which compare individual grains originating from quarried stone and river gravel. Aside from those shown in Figures A3 and A4, further visualizations of the quarried stone aggregate (GK2 and GK3) have not been included in the appendix. This is because these stone types have little to no visible porosity in the CT-based visualizations.



**Table A1.** Individual grain results from the cracking analysis of stone category GK1 (river gravel; alkali sensitivity EIII-S).












