The Challenge

The north wall of the open pit had experienced significant deformation over an area spanning approximately eight benches above the nearest ramp. Ground-based radar monitoring recorded displacement velocities of up to 54 cm/day in August 2019. Rainfall between November 2019 and April 2020 resulted in cumulative displacements exceeding 1 m.

These observations raised questions of the potential extent of failure caused by further rainfall. To continue safe operations, modeling needed to go beyond conventional slope stability analysis to simulate the post-failure movement and deposition of the material.

This type of analysis is challenging in large-strain FLAC3D models, where severe deformation can produce highly distorted zones that compromise numerical stability. MPM avoids this mesh distortion because material points move through a background computational grid that is reset each timestep, making the method particularly suitable for extreme deformation and runout problems. However, applying MPM to a large 3D model can be computationally intensive.

The Modeling Approach

A FLAC3D model was constructed and calibrated under small-strain conditions before being evaluated in large-strain mode. The rock mass was represented using a Hoek-Brown model with perfectly plastic behavior, while weaker materials associated with fault damage zones were represented using a Mohr-Coulomb criterion. (Note: In the original model shown in the following figure, faults were represented as zone joints, which are not yet implemented in MPoint, so those zones were were simulated as weaker damaged rock using a Mohr-Coulomb criterion.) The interpreted failure mechanism was controlled by wedges formed by intersecting major structures, along with elevated pore pressures near the slope face.

The calibrated small-strain model reproduced the location and general extent of the observed instability, establishing the basis for investigating post-failure behavior. Three approaches were then evaluated:

1. Fully meshed FLAC3D model: The model was continued in large-strain mode.
2. Fully MPM MPoint model: The full volume of material was represented by material points.
3. Hybrid FLAC3D-MPoint model: Only the region expected to undergo large deformation used material points, while the remainder of the model used FLAC3D zones.

In the coupled model, FLAC3D zones are converted to material points by transferring material properties, stresses, and displacements from one to the other to maintain continuity in the mechanical response.

Comparing the Three Approaches

The fully meshed FLAC3D model captured the developing instability, but numerical stability deteriorated rapidly once displacement reached approximately three times the element size, or about 20 m. At that stage, severe mesh distortion prevented the calculation from reliably continuing long enough to predict the full runout.

Full MPM MPoint Model

Converting the entire model to material points allowed the calculation to continue through acceleration, large deformation, flow, and deposition. Unlike the fully meshed calculation, MPoint could represent the substantial displacement associated with the failure without the calculation being halted by mesh distortion.

The simulation indicated that most of the failed material would be contained on the ramp below the failure rather than reaching the pit bottom. However, the fully MPM model also highlighted a modeling consideration: material-point interactions produced an apparent separation near the pit bottom because material points are considered part of the same continuum when they come close together, preventing interpenetration when it is likely that a sliding contact exists between the failing material and underlying rock. This unrealistic behavior is not present in the hybrid modeling method.

Hybrid FLAC3D-MPoint Model

In the hybrid simulation, only the region expected to undergo large deformation was represented by material points. Adjacent FLAC3D zones interacted with these points using hybrid points that transferred information between the two numerical formulations. A frictional wall was also applied along zone surfaces below the failure region. This prevented unrealistic penetration or merging while providing a surface along which the failed material could slide.

The hybrid model produced a similar practical conclusion to the fully MPM simulation: the failure was largely contained by the ramp below rather than propagating to the pit bottom. At the same time, the hybrid approach eliminated the unrealistic material interaction observed in the fully MPM calculation and was approximately 10 times faster.

Results

The comparison demonstrated the complementary strengths of FLAC3D and MPoint.

The fully meshed approach with FLAC3D alone was effective for representing pre-failure behavior and progressive instability but became limited once deformation became sufficiently large to severely distort the computational mesh. The fully MPM with MPoint alone approach extended the simulation into the post-failure regime, capturing large displacement, velocity evolution, runout, and deposition.

The hybrid approach provided the key advantages of both methods: FLAC3D zones could continue representing relatively stable portions of the rock mass, while material points were concentrated in the area undergoing extreme deformation. For this case, the approach also provided a computational advantage with significantly fast runtimes over the fully MPM model.

Most importantly for the engineering question being investigated, the MPoint-based simulations indicated that most of the failed material would be contained on the ramp below the instability, although a limited volume could potentially reach the pit bottom if failure progressed under more intense rainfall conditions.

Benefits of the Hybrid Approach

The study demonstrates how combining FLAC3D and MPoint can provide a practical workflow for slope analyses requiring both stability and consequence assessment. The hybrid approach can:

• Extend analysis beyond failure initiation to evaluate post-failure displacement, runout, and deposition.
• Reduce mesh-distortion limitations by using material points where extreme deformation develops.
• Maintain efficient continuum modeling in portions of the domain where strains remain relatively small.
• Connect progressive failure and runout within a consistent framework, reducing the need to transfer results between separate stability and runout models.
• Improve computational efficiency compared with a fully MPM model, with the hybrid implementation in this study running approximately 10 times faster.

The hybrid FLAC3D-MPoint approach is particularly relevant for high-consequence slopes where understanding whether failed material could reach infrastructure, working areas, or other exposed assets is essential to informed decision-making to reduce risk.

Looking Forward

The hybrid FLAC3D-MPoint framework offers an efficient way to connect conventional slope stability analysis with post-failure consequence assessment. Rather than stopping the analysis at the onset of instability as in traditional methods, the hybrid approach can continue following the material through large deformation, runout, and deposition while reserving the more computationally intensive MPM formulation for the portions of the model that require it.

The study also emphasizes that reliable quantitative predictions of runout distance, failed volume, and deposition geometry depend on appropriate selection of the constitutive model and calibration of post-peak material behavior against observed failures where data are available. Further development and validation are needed before the methodology can be applied routinely for quantitative post-failure prediction.

Ultimately, the approach provides a path toward evaluating both whether a slope may fail and what could happen after it does within a single mechanically consistent modeling framework.

Reference

Cancino, C., Hazzard J., & Coetzee, C. (2026). A hybrid FLAC3D-MPM approach for post-failure slope behaviour. In Proceedings, EUROCK 2026: Risk Management in Rock Engineering, (Skopje, North Macedonia, September 2026).