The Mohr-Coulomb Failure Criterion That Dictates Rock Fracture in Hydraulic Fracturing

Hydraulic fracturing succeeds only when the fluid pressure pumped into a wellbore exceeds the rock's inherent shear strength and the confining tectonic stress, a threshold defined by the Mohr-Coulomb failure criterion. While the classic linear equation accurately predicts fracture networks in shallow, conventional reservoirs, modern deep shale operations require generalized non-linear models to prevent costly under-pressurization.[7]

The foundation of this geomechanical threshold was formalized in the early 20th century, building on Coulomb's 1773 friction work and Mohr's 1900 state of stress theory. The resulting linear equation states that shear strength equals the rock's inherent cohesion plus the product of the normal stress and the tangent of the internal friction angle. In practical terms, this means a rock's resistance to breaking increases proportionally as the surrounding earth squeezes it tighter.[5][6]

Petroleum engineers rely on this mathematical relationship to design stimulation programs. By plotting the maximum and minimum principal stresses as a semi-circle—known as a Mohr circle—against the rock's failure envelope, operators determine exactly when a formation will yield. Typical intact reservoir rocks exhibit an internal friction angle between 30 and 45 degrees, requiring surface pumps to deliver fluid at 8,000 to 10,000 pounds per square inch to initiate a fracture network.[4][6]

However, the transition to unconventional reservoirs has exposed the physical limitations of the linear model. A 2026 analysis published in SciOpen evaluated the dynamic Mohr-Coulomb stability of natural fractures in deep shale under hydraulic fracturing-induced stress superposition. The researchers demonstrated that as fluid permeates the complex, pre-existing fracture networks of shale, the localized stress fields alter dynamically, rendering a static, linear failure envelope insufficient.[1]

The primary mechanical blind spot of the classic criterion is its dimensionality. As geologist John Handin noted in his foundational 1969 Journal of Geophysical Research paper, the standard Mohr-Coulomb model "assumes that the intermediate principal stress has no influence on failure." In a highly anisotropic environment like a deep shale basin, where horizontal stresses vary wildly in different directions, ignoring that middle vector leads to inaccurate pressure forecasts.[2]

To bridge this gap, computational modelers have developed the Generalized Mohr-Coulomb failure criterion. Detailed in a 2023 MDPI engineering review, this non-linear approach incorporates all three principal stresses. By rounding the sharp corners of the classic hexagonal yield surface in three-dimensional stress space, the generalized model provides a continuous, mathematically smooth envelope that better matches true triaxial laboratory tests.[3]

Re-evaluating intact rock strength through this generalized lens changes the operational math. A 2016 study in Solid Earth examined friction and failure under varying confining pressures, confirming that the linear model often underestimates the rock's true strength at high depths. When operators use the linear model in deep shale, they risk under-pressurizing the well by 15% to 20%, resulting in shorter fracture half-lengths and stranded hydrocarbons.[4][7]

The trade-off between the two models centers on data acquisition costs. The linear Mohr-Coulomb criterion requires only two parameters—cohesion and the friction angle—which can be derived from standard, relatively inexpensive triaxial compression tests on core samples. The generalized model demands true triaxial testing equipment, which applies independent pressures along three axes, a specialized procedure available in far fewer commercial laboratories.[3][5]

Consequently, the industry is splitting its approach based on reservoir complexity. Software developers are now embedding both criteria into real-time fracture simulators. The deciding factor for a completion engineer is no longer which equation is universally correct, but whether the specific geological target justifies the $50,000 to $100,000 premium for the true triaxial core data required to populate the generalized model.[7]