Volume 19, Issue 3 (September 1996)
Model Size Effects in Centrifuge Models of Granular Slope Instability
Two assumptions behind geotechnical laboratory testing are: first, that soil with grains small enough to behave as a continuum in a full-scale field situation will also behave as a continuum in a small laboratory test; and, second, that side boundary effects in a laboratory test are either negligible or, if not negligible, do not dominate behavior but are quantifiable so that a correction can be made for their effects. Data are presented here from 61 centrifuge models of slopes brought to failure to assess side boundary and grain-size effects on model behavior. Three granular soils of different grain sizes were used to build models of different scales and widths. Model slopes were brought to failure by increasing self-weight, which simulates increasing full-scale height while maintaining slope geometry. To simulate the behavior of soil with cohesion, a small negative pore pressure was applied to the soils using a vacuum.
Models failed with wide, shallow slope failures. There were two effects on model behavior arising from grain size. First, it appears that the model failure “surface” may have to be more than 30 grain diameters below the soil surface before grain-size effects are negligible. Secondly, the repeatability of model results was also affected by grain size: coefficients of variation in failure data for a single group of identical models were 10% or less when slope height was 158 grain diameters or greater; coefficients of variation increased markedly for smaller model slopes with heights 56 grain diameters or less. When effects of model width were examined, it appears that model width may have to exceed model slope height by a factor of 5 for shallow wide slope failure to develop unimpeded by side boundaries.