Chemically bonded postinstalled anchors have seen tremendous growth ov
er the past few years for retrofits, as well as new construction. Curr
ently, they are designed from proprietary tables provided by adhesive
manufacturers based on laboratory pullout tests. Recently, Doerr et al
. (1989), Cook (1993), and Eligehausen et al. (1984) have developed eq
uations to predict pullout resistance of anchors. Since chemically bon
ded anchors result in the failure of both the concrete and adhesive-co
ncrete interface, the equations attempt to predict the ultimate resist
ance of the anchor through the sum of the contributions from the concr
ete-failure cone and adhesive-concrete interface. However, this approa
ch requires an estimate of both the average or maximum shear stress wi
thin the adhesive bond layer and the concrete-failure cone depths. To
shed more light on the development of failure for these types of ancho
rs, a state-of-the art elastoplastic finite-element analysis was perfo
rmed and compared to experimental results. Besides being able to predi
ct pullout resistance, concrete-failure cone depths, and orientations,
the analysis revealed that failure initiates as a tension zone below
the concrete surface at the anchor-adhesive interface and propagates w
ith load toward the surface. In the process, both the concrete and adh
esive material dilate increasing the confinement and shear resistance
within the adhesive layer. Once the tension zone reaches the surface,
the confinement is lost, resulting in a diminished shear resistance wi
thin the adhesive layer and anchor failure. After comparing a number o
f proposed methods to predict resistance to the experimental data, it
was found that a simplistic, uniform bond stress applied over the whol
e anchor did an excellent job of predicting pullout capacity.