Using waste glass as an aggregate in concrete can cause severe damage becau
se of the alkali-silica reaction (ASR) between the alkali in the cement pas
te and the silica in the glass. Recent accelerated 2-week tests, conducted
according to ASTM C 1260, revealed that the damage to concrete caused by ex
pansion of the ASR gel, which is manifested by strength reduction, depends
in these tests strongly on the size of the glass particles. As the particle
size decreases, the tensile strength first also decreases, which is expect
ed because of the surface-to-volume ratio of the particles, and thus their
chemical reactivity increases. However, there exists a certain worst (pessi
mum) size below which any further decrease of particle size improves the st
rength, and the damage becomes virtually nonexistent if the particles are s
mall enough. The volume dilatation due to ASR is maximum for the pessimum p
article size and decreases with a further decrease of size. These experimen
tal findings seem contrary to intuition. This paper proposes a micromechani
cal fracture theory that explains the reversal of particle size effect in t
he accelerated 2-week test by two opposing mechanisms: (1) The extent of ch
emical reaction as a function of surface area, which causes the strength to
decrease with a decreasing particle size; and (2) the size effect of the c
racks produced by expansion of the ASR gel, which causes the opposite. The
pessimum size, which is about 1.5 mm, corresponds to the case where the eff
ects of both mechanisms are balanced. For smaller sizes the second mechanis
m prevails, and for sizes <0.15 mm no adverse effects are detectable. Extra
polation of the accelerated test (ASTM C 1260) to real structures and full
lifetimes will require coupling the present model with the modeling of the
reaction kinetics and diffusion processes involved.