External Sulfate Attack on Cementitious Binders: Limitations and Effects of Sample Geometry on the Quantification of Expansion Stress

The hollow cylinder method was used to estimate the expansion stress that can occur in concrete due to the crystallisation pressure caused by the formation of ettringite and/or gypsum during external sulphate attack. Hardened cement paste hollow cylinders prepared with Portland cement were mounted in stress cells and exposed to sodium sulphate solutions with two different concentrations (3.0 g L SO₄^2- and 30.0 g L SO₄^2-). Microstructural analysis and finite element modelling was used to evaluate the experimental observations. The expansion stress calculation was verified for a range of diameter/length ratios (0.43-0.60). Thermodynamically predicted maximum expansion stresses are larger than expansion stresses observed in experiments because the latter are affected by the sample geometry, degree of restraint, pore size distribution and relaxation processes. The results indicate that differences in self-constraint at the concave inner and convex outer surfaces of the hollow cylinder lead to an asymmetric expansion stress when ettringite is formed. This leads to macroscopic longitudinal cracks and ultimately failure. Heavy structural components made of concrete are likely to support larger maximum expansion stresses than observed by the hollow cylinder method due to their self-constraint.