Concrete is a brittle composite cementitious material that easily fractures under tensile loading. Microcracks can appear throughout the concrete prior to application of any load because of temperature-induced strain and autogenous and drying shrinkage. There is no doubt that these cracks provide preferential access for aggressive agents to penetrate into the concrete, probably causing corrosion of reinforcement steel and degradation of concrete. As a result, the service life of reinforced concrete structures is shortened. To prolong service life of concrete structures after cracking, man-made repair is a typical way. However, it is well known that man-made repair work is very expensive and most repairs can not serve for longer than 5 to 10 years. Self-healing of cracks has a significant potential to extend the service life of reinforced concrete structures and save large amounts of money of man-made repair work. In view of the cost and environmental impact, autogenous self-healing is one of the most compatible and suitable mechanisms of self-healing because it is an intrinsic property of concrete without any “strange” healing agent as additive. However, the physico-chemical process of autogenous self-healing is not completely understood. The potential of autogenous self-healing in cementitious materials is not clear either. The aim of this research is to better understand the physico-chemical process of autogenous self-healing of cracks and to quantify the potential of autogenous self-healing in cementitious materials. The information about the potential of autogenous self-healing is important for service life prediction of underground concrete structures or concrete structures submersed in water. Summary of experimental results 1. Identification of reaction products of autogenous self-healing Techniques including environmental scanning electron microscope equipped with energy-dispersive X-ray spectroscopy, Fourier transform infrared spectroscopy, thermogravimetric analysis and X-ray diffraction were used to characterize the reaction products of autogenous self-healing. The experimental results show that when cracked Portland cement paste (CEM I 42.5N, w/c=0.3) is cured in water or saturated Ca(OH)2 solution under sealed conditions, portlandite is the main component of the reaction products formed in cracks, accounting for about 80% by mass. When cracked slag cement paste (CEM III 42.5N, w/c=0.3) is cured in saturated Ca(OH)2 solution under sealed conditions, C-S-H is the main component of the reaction products formed in cracks, accounting for about 57% by mass. These results help us to better understand the physico-chemical process of autogenous self-healing that is fundamental for investigating the potential of autogenous self-healing in different cement pastes. 2. Quantification of the rate of autogenous self-healing The amount of reaction products formed in a crack as a function of time was quantified by means of image analysis. The reduction of the permeability through cracks after autogenous self-healing was determined. Image analysis and air permeability tests demonstrated that autogenous self-healing of microcracks in cement paste with w/c ratio of 0.3 proceeds fast within the first 50 hours of healing time. After 50 hours, the progress of self-healing slows down. Although cracks are not completely filled with the reaction products, the permeability through the cracks decreases dramatically. In Portland cement paste with w/c ratio of 0.3, a filling fraction of a microcrack of about 28% after curing in water for 200 h decreases the air permeability by about 65%. It demonstrates that self-healing of cracks can reduce the ingress of harmful agents through cracks significantly and thus prolong the service life of concrete structures. The filling fraction of cracks and reduction of air permeability through the cracks due to autogenous self-healing in slag cement (slag content is 66% by mass) paste is much higher than that in Portland cement paste. The information about decrease of permeability of cement paste due to the filling of cracks reveals the potential of autogenous self-healing for prolonging service life of concrete structures. 3. Effect of absorption of water by the bulk paste on autogenous self-healing The effect of migration of water (from cracks into the bulk paste) on the efficiency of self-healing was investigated for the first time. Nuclear Magnetic Resonance (NMR) tests were performed to investigate water migration from cracks into the bulk paste. The changes of water content and water distribution in the bulk paste were quantified by NMR. By knowing the change of water content, further hydration of unhydrated cement was determined and volume of reaction products of further hydration caused by extra water was calculated. In this way, densification of microstructure of the bulk paste was determined. When a crack in cement paste has been exposed to water for 216 hours from the age of 42 days, the capillary porosity of the bulk paste nearby the crack decreases from about 10% to about 3.5%. The densification of microstructure adjacent to the crack, on the one hand, can decrease the ingress of aggressive agents into the bulk concrete matrix and prolong the service life of concrete structures. On the other hand it will decrease the diffusion of reaction products from the bulk paste into cracks and has, therefore, a negative effect on filling of the crack with reaction products. Summary of numerical modeling results A coupled transport-reaction model was developed for simulating autogenous self-healing. The transport model was based on ion diffusion theory and the reaction model was based on thermodynamic theory. The coupled transport-reaction model was validated by experimental results as mentioned earlier. This model is able to describe the basic physico-chemical process of autogenous self-healing, i.e., dissolution of reactive material (cement clinker, slag), diffusion of ions and precipitation of reaction products in cracks. With this coupled transport-reaction model, the potential of autogenous self-healing can be predicted, including the effect of several factors, such as the crack width, the amount of unhydrated cement or slag and the initial ion concentrations of the healing agents. The modeling results show that the reaction rates of cement and slag present at the crack surfaces are faster than those in bulk paste. This is due to the seeding effect of the crack surfaces, dilution effect of the large volume of solution in the crack and the large space in the crack for the growth of reaction products. When the cracks dry out, CO2 can easily penetrate into the cracks. The reaction products formed earlier in the cracks during the healing process are carbonated. The effects of carbonation on autogenous self-healing were investigated by thermodynamic modeling. The modeling results show that portlandite formed in cracks in Portland cement paste is first transformed into calcite, followed by the carbonation of C-S-H forming calcite and gel silica. Because the carbonation of portlandite increases its volume by about 11%, the filling fraction of cracks in Portland cement paste is increased by carbonation. In comparison, in slag cement paste the filling fraction of cracks decreases after the carbonation of reaction products because C-S-H, the main component of the reaction products, decreases its volume after carbonation. Contributions of this research With this research, a better insight is gained into the potential of autogenous self-healing in cementitious materials. Concrete made of coarse cement may have higher potential of autogenous self-healing than that made of fine cement. This is because in the concrete made of coarse cement there is more unhydrated cement left and the more porous microstructure favors the diffusion of ions from the bulk paste into the crack for the formation of reaction products in the crack. Moreover, blending pozzolanic materials, such as slag, can increase the capacity of autogenous self-healing, especially when an alkali solution is available. It is confirmed that autogenous self-healing has a high potential to prolong service life of concrete structures because the permeability through the cracks decreases significantly even if the cracks are only partially filled. More importantly, the microstructure of the bulk paste adjacent to the crack surfaces becomes much denser because of additional hydration of cement caused by the extra water coming from the crack. The densification of microstructure of the bulk paste adjacent to the crack surfaces can decrease the ingress of aggressive agents into the concrete matrix and can, therefore, prolong the service life of concrete structures. This information about the potential of autogenous self-healing is very important for service life prediction of reinforced concrete structures, especially of underground concrete structures or concrete structures submersed in water. The coupled transport-reaction model developed in this research is a powerful tool which can be used directly to predict self-healing in cement-based materials made of cement clinker, slag or fly ash with different fineness. The coupled transport-reaction model can easily be extended to calculate the healing capacity of blended expansive agents, such as calcium sulfoaluminate (Ca4(AlO2)6SO4), free lime (CaO) and anhydrite (CaSO4). Therefore, the coupled transport-reaction model can be used to design self-healing concrete through optimizing the mixture of concrete in the design stage.