Independent Articles

‹ Back

Are Your Consultants Doing Your Stability Analyses Correctly?

 

Slope stability analysis is one of the most common services provided by geotechnical and tailings consultants. It is also one of the easiest analyses to present in a way that looks sophisticated, while the underlying model may not have been properly understood.

 

Software such as GeoStudio SLOPE/W and Rocscience Slide has made it possible to model increasingly complex slopes, embankments, pore pressure conditions, loading cases and slip surface geometries. This is useful. However, increased software capability does not remove the need for engineering judgement. In fact, it makes engineering judgement more important.

 

One issue I have seen is the direct extraction of cross-sections from 3D surfaces or survey models, followed by the import of near-identical, highly irregular geometry into a 2D stability model. The consultant may claim that this is done “for accuracy”. But is it really accuracy, or is it simply the unfiltered transfer of survey noise into a limit equilibrium model?

 

A 2D slope stability analysis is not a drawing exercise. It is a numerical idealisation of a physical problem. The purpose of the cross-section is to represent the critical geometry in a way that is technically meaningful for the method being used. If the embankment crest, slope, beach, foundation contact or tailings surface contains many small irregularities, the question should be: do these features materially affect the potential failure mechanism, or are they simply adding numerical clutter?

 

Limit equilibrium programmes work by dividing the potential sliding mass into slices or columns and solving force and/or moment equilibrium. The factor of safety is then calculated based on the available shear strength relative to the mobilised shear stress. The method is powerful, but it remains an approximation. The modeller is not recreating reality; the modeller is creating a defensible engineering representation of reality.

 

This distinction matters.

 

Adding every small survey kink, point, step and surface irregularity does not automatically make the model more accurate. In some cases, it may do the opposite. It can create localised geometric artefacts, unrealistic slice shapes, convergence issues, or changes in the calculated factor of safety that are more related to model construction than physical behaviour. A highly detailed section may look impressive in a report, but impressive is not the same as correct.

 

The better question is not: “How many data points did the consultant import?”

 

The better question is: “Has the consultant idealised the geometry in a way that is appropriate for the analysis method, material behaviour, pore pressure condition and expected failure mechanism?”

 

There is also a broader problem in how slope stability results are reported. I recently saw a discussion from a consultancy explaining why factors of safety should be reported to three decimal places. I disagree with this approach.

 

A factor of safety of 1.347 may look more precise than 1.3 or 1.35, but the additional digits can create a false sense of certainty. Stability models are built on assumptions: geometry, material zoning, shear strength parameters, pore pressure interpretation, loading conditions, construction history, strain compatibility, and the selected method of analysis. Laboratory test results themselves contain variability. Field conditions contain even more variability.

 

Yes, the calculation may output three decimal places. That does not mean the engineering conclusion is accurate to three decimal places.

 

In many cases, the difference between 1.347 and 1.352 is not practically meaningful. It may be less significant than the uncertainty in the pore pressure surface, the selection of undrained strength ratio, the interpretation of foundation stratigraphy, or the assumed phreatic condition. Reporting excessive decimal places can therefore become a form of technical theatre: it gives the appearance of precision without necessarily improving understanding.

 

This is not an argument against slope stability software. It is an argument for better use of slope stability software.

 

A good stability assessment should demonstrate that the consultant understands the method being used, the limitations of the method, the sensitivity of the result, and the physical behaviour of the facility. The report should explain why the selected cross-section is critical, why the geometry has been idealised in a particular way, why the selected parameters are appropriate, how pore pressures were represented, and what the result actually means for the facility.

 

Clients should be cautious when they see models that appear overly detailed but are poorly explained. They should also be cautious when small differences in factors of safety are presented as if they are highly significant.

 

A stability model should not be judged by how complicated it looks.

 

It should be judged by whether it is technically defensible, physically meaningful, appropriately conservative, and clearly explained.

 

So the next time a consultant presents a stability analysis, ask a few simple questions:

• Why was this cross-section selected?
• Was the geometry idealised, or simply imported?
• Were unnecessary surface irregularities removed?
• What checks were done to confirm that the result is not controlled by model artefacts?
• How sensitive is the result to pore pressure, material strength and geometry?
• And, most importantly, does the reported factor of safety reflect engineering judgement, or just software output?

 

Because a stability analysis is not correct simply because the software produced a number.

-Yoshlin Govender (CEO & Tailings Expert)