Low-density volcanic rocks have specific geomechanical properties that require complex laboratory tests and characterization that are not usually available in common geotechnical studies. A pyroclastic rock behaves at sufficiently “low” stress levels as if it were a conventional rock under the action of an external load, but when subjected to higher stresses, the bonds between its particles can break, leading to a sudden decrease in its volume and the reorganization of its particles, thus forming a more compact structure than the initial one. This process is known as “mechanical collapse” and involves a drastic change in the properties, which is critical for engineering purposes. A compilation and analysis of many tests performed in our laboratory was conducted, considering several inputs such as origin rock block, lithotype class, dry density, particle density, porosity, failure stresses in Cambridge variables (p and q), deformation at failure (%), and isotropic collapse stress. The collapse stress is not available for all tests, then a reasonable prediction is made using the parabolic model of collapse. The objective of this research is to develop a machine learning procedure using this dataset to predict the isotropic collapse stress, which is challenging to measure in standard laboratory settings. This model will use a set of variables that can be easily obtained in the laboratory. Based on this evaluation, a technician can determine whether a complementary lab test is necessary or if failure is unavoidable within the given range of stresses. Two procedures were tested, XGBoost and artificial neural networks, both showing an outstanding prediction behavior.

Volcanic terrains exhibit a complex structure of pyroclastic deposits interspersed with sedimentary processes, resulting in irregular lithological sequences that lack lateral continuity and distinct stratigraphic patterns. This complexity poses significant challenges for slope stability analysis, requiring the development of specialized techniques to address these issues. This research presents a numerical methodology that incorporates spatial variability, nonlinear material characterization, and probabilistic analysis using a Monte Carlo framework to address this issue. The heterogeneous structure is represented by randomly assigning different lithotypes across the slope, while maintaining predefined global proportions. This contrasts with the more common approach of applying probabilistic variability to mechanical parameters within a homogeneous slope model. The material behavior is defined using complex nonlinear failure criteria, such as the Hoek–Brown model and a parabolic model with collapse, both implemented through linearization techniques. The Discontinuity Layout Optimization (DLO) method, a novel numerical approach based on limit analysis, is employed to efficiently incorporate these advances and compute the factor of safety of the slope. Within this framework, the Monte Carlo procedure is used to assess slope stability by conducting a large number of simulations, each with a different lithotype distribution. Based on the results, a hybrid method is proposed that combines probabilistic modeling with deterministic design principles for the slope stability assessment. As a case study, the methodology is applied to a 20-m-high vertical slope composed of three lithotypes (altered scoria, welded scoria, and basalt) randomly distributed in proportions of 15%, 60%, and 25%, respectively. The results show convergence of mean values after approximately 400 simulations and highlight the significant influence of spatial heterogeneity, with variations of the factor of safety between 5 and 12 in 85% of cases. They also reveal non-circular and mid-slope failure wedges not captured by traditional stability methods. Finally, an equivalent normal probability distribution is proposed as a reliable approximation of the factor of safety for use in risk analysis and engineering decision-making.