Poisson's ratio is a fundamental parameter in rock mechanics, traditionally often
treated as a constant. This study challenges that notion by developing a universal
empirical model that captures its significant stress-dependency from initial loading
to failure. The model was rigorously calibrated using 42 uniaxial compressive tests
on granite and successfully generalized to limestone, marl, sandstone, and saltstone,
representing a wide spectrum of mechanical behaviors from brittle to ductile. Its
predictive accuracy was quantified using root mean square error (RMSE), mean absolute
error (MAE), and mean absolute percentage error (MAPE). The model demonstrated high
fidelity, particularly at the critical crack initiation stage (sigma/sigma c = 0.5),
where errors were minimized. Across all five rock types and three calculation methods
(secant, average, tangent), the MAPE ranged from 1.4% to 3.7%, RMSE from 0.007 to
0.012, and MAE from 0.005 to 0.010. Performance was superior for ductile rocks like
marl (MAPE approximate to 1.9%) and saltstone (MAPE approximate to 1.4%) due to their
homogeneous deformation, while effectively capturing the more complex, crack-dominated
response of granite and sandstone. The research conclusively demonstrates that Poisson's
ratio is not a fixed material property but a dynamically evolving parameter. This
work provides a practical and robust framework that moves beyond the assumption of
a constant Poisson's ratio, offering a significant improvement for accurate deformation
analysis in numerical modeling, underground excavation design, and reservoir geomechanics.
