Thickness (T), Length (L), Width (W) and size distribution of scaled concrete particles in frost testing were measured. T (mm) increases with particle size surprisingly similarly for different concrete qualities and frost test methods. 2T/(L+W) reduces as function of size and is lowest for the largest particles of the salt scaling test: 0.1 – 0.15 but increases if large aggregate particles scale. Particle size distributions from salt frost testing peak for particles of 1-2 mm. The particles are flakier compared to particles from freeze/thaw in water which also have flatter size distribution no matter type of concrete or degree of damage. Scaling in water is not so efficiently reduced by air voids despite protecting very efficiently against internal damage and scaling in salt frost testing. Comparisons with T predicted by the glue spall model (≈3/4 × ice thickness) and the air void dependent (≈3× critical air void spacing) model proposed by Fagerlund are difficult due to the size dependent flake thickness. Image analysis could well describe shape. Further studies of concrete flake thickness scaled at varying thickness of ice layers are proposed.
