Tensile Strength of Embedded Anchor Groups: Tests and Strength Models
- Primary Investigator(s): David Grilli, Amit Kanvinde
- Institution: University of California, Davis
- Year Completed: 2015
Note: This research was conducted in partnership with the Charles Pankow Foundation (CPF). Link to CPF website for this Project here.
Steel column bases in seismically braced frames and other similar structures must be designed for high uplift or tensile forces. A common detail for this connection involves anchors embedded in the footing with a plate at their lower end, also embedded in the footing. This detail is increasingly prevalent in construction practice, since it is exempt from the strength calculations of ACI 318 Appendix D. However, no experimental data or validated design guidelines are available to support the design of this detail. As a consequence, approaches from other similar situations (such as punching shear of slabs) are adapted for this purpose. To address this practical need, this report presents tension tests on two full-scale specimens featuring this anchorage detail. The main variable examined in the experiments is the embedment depth, such that two depths – 12 inches and 18 inches, are tested. The test specimens exhibit a classic concrete failure cone extending upwards from the edges of the embedded base plate. The experimental data provides evidence that the anchorage detail provides an effective means to carry high tensile loads. The data is evaluated against three strength models, including the ACI-318 Appendix D method, the ACI 318 punching shear equation, and the Concrete Capacity Design (CCD) method. It is determined that the Appendix D method is significantly conservative (average test-predicted ratio 1.34), because it does not consider the beneficial effects of the embedded plate. On the other hand, the punching shear method is unconservative (average test-predicted ratio 0.62) because it does not explicitly incorporate the size effect in concrete. The CCD method shows most promise, with an average test-predicted ratio of 0.99. Limitations of the study include the small size of the test set, and the absence of reinforcement in the specimens.