Abstract
Accurate estimation of the Hoek–Brown material constant (mi) is essential for reliable rock mass characterization, particularly in brittle, heterogeneous carbonate formations. This study introduces an optimized multi-stage triaxial compression test (ML-TCT) using a maximum secant modulus (E(sec) = max) criterion, designed to overcome common laboratory limitations like limited sample availability and the absence of automated control systems. Experimental tests were conducted on dolomitic limestone from the Shahbazan Formation using nine conventional single-stage tests (SL-TCT) and seven multi-stage tests under two protocols. The “loading–unloading” ML-TCT scheme proved superior, increasing data efficiency by generating 49 stress points from seven specimens compared to nine points from nine specimens in SL-TCT. Results show that ML-TCT yielded a higher and more representative mean mi value of 9.69, which is statistically significantly different from the SL-TCT results (p < 0.01). This higher mi is attributed to testing a single, persistent failure plane, which reduces inter-specimen variability. Conversely, SL-TCT produced unconfined compressive strength (σ(ci)) values that were 10–30 MPa higher, likely because cumulative micro-damage in the ML-TCT specimens weakens the rock’s cohesive component. Numerical validation using RS2 software demonstrated that the parameters derived from ML-TCT result in predictions of a larger plastic zone and greater crown displacement, indicating that the proposed method offers a more conservative approach for characterizing brittle rocks. Ultimately, it offers a practical, cost-effective alternative for estimating the mechanical properties of brittle rocks under constrained laboratory conditions.