Variability of small scale model reinforced concrete and implications for geotechnical centrifuge testing

There are soil-structure interaction problems for which it is important to model both the relative soil-structure stiffness and strength. Examples from the earthquake engineering field include the design of resilient rocking-isolated foundations and the seismic stabilisation of slopes using piling. In both cases the aim is to ensure a preferred failure mode happens first in the soil instead of the structure i.e. controlled bearing failure of the foundation or soil yielding around piles. A recently developed model reinforced concrete for centrifuge testing can simulate stiffness and strength simultaneously, but suffers from variability in the material properties, as does the full-scale material. This paper presents a series of element tests on the variability of model reinforced concrete elements representative of large square monolithic bridge piers and slender square piles. Coefficients of variation for various material and element properties have been determined and shown to be similar to typical values for full-scale reinforced concrete elements obtained from the literature. It is also demonstrated that curing time beyond 28 days does not substantially affect strength and variability and that models of different absolute volume can be produced without inducing detrimental size effects. The results are used to discuss the selection of mean design strengths for model structural elements in centrifuge experiments using a quantitative statistical approach where there are competing structural and soil failure modes.