Abstract
The uncracked semi-circular bend (SCB) test has recently gained attention as a simple and material-efficient method for determining the tensile strength of brittle geomaterials. However, as reported in the literature and confirmed by our experiments, localized damage at the roller supports remains a critical limitation that may compromise measurement accuracy and test validity. This study addresses this limitation through experimental testing on red and gray sandstone, complemented by numerical simulations to provide deeper insight into stress distribution and fracture mechanisms in the SCB test. Experimental results showed that six out of twelve specimens experienced local damage, ranging from slight crushing and surficial cracking at the base roller zones in red sandstone to rock chipping in gray sandstone. The stiffer sandstone exhibited more severe local damage due to its limited deformability. These damages were attributed to minor geometric imperfections introduced during sample preparation. Nevertheless, all tests yielded valid tensile strength values, with SCB results showing good agreement with Brazilian test outcomes and demonstrating significantly lower coefficients of variation. Finite element simulations confirmed that crack initiation consistently occurred at the middle of the flat edge under pure tensile stress, indicating a mode I fracture mechanism. Numerical analyses further revealed pronounced stress concentrations, particularly compressive stresses, at the roller contact zones, induced by the specimen's low span-to-depth ratio, which increased the fracture load required for failure.