A key problem overlooked during experiments to determine rock properties in the post-peak region

The constitutive relation and mechanical parameters should be consistent in the stability analysis of a given rock mass whether building a dam on it or excavating underground engineering in it. Rock mechanics experiments are usually used for confirming mechanics parameters, and the parameters should be relatively consistent under different experiment methods, such as uniaxial, tri-axial or plane stress compression experiments, which is a very important assumption but ignored sometimes because it is tenable in a general way; however the assumption is untenable for strain softening materials such as rock and concrete in the post-peak area, which is a question that has not fully been realized up to now. In this paper, the results of uniaxial and plane strain compression experiments are presented and discussed; it is found that the engineering or experimental style has obvious effects on the stress-strain relationship and especially on the mechanical parameters in the post-peak area. The size effect on softening materials is discussed in addition. 2004 Elsevier Ltd.