Abstract Rammed earth materials have been applied to construct infrastructure for human society over thousands of years. The design of rammed earth is typically based on the results of unconfined compression (UC) tests, in which the differences in specimen dimensions and shapes can influence its physical properties greatly. However, there is a lack of strength prediction model that can consider the size effect. In this investigation, an experimental campaign is conducted with a testing matrix comprising two specimen shapes (cylindrical and cuboid), five height-to-width ratios (0.5, 1, 1.5, 2, and 2.5), three curing ages (7, 14, and 28 days), and three repetitive specimens. A total of 90 UC tests are performed. It is found that the behavior of rammed earth is governed by brittle failure. A shorter specimen fails in an hourglass pattern, whereas a longer specimen fails in a shear-dominated pattern. The strength of rammed earth reduces with the height-to-width ratio, and the difference in strength could be as large as 1.6–2 times for different curing ages. Two prediction models for cylindrical and cuboid specimens are proposed and calibrated using the measured data. Good agreement between the predictions and measurements is obtained, suggesting that the proposed models have a wide application range.
Luo et al. (Thu,) studied this question.