When I started my career in geotechnical engineering, slope stability analysis was already a central part of the tailings dam design process. People in the generation before me were running limit equilibrium models by hand, plotting circles on graph paper, and breathing a sigh of relief when the factor of safety came out above 1.5. Fast forward 40 years, and the tools have become far more sophisticated—finite element / finite difference methods, 3D modeling, coupled hydro-mechanical analyses, and probabilistic frameworks.
And yet, after all this time, one truth has remained constant: slope stability analyses are never the whole story. They are tools, not answers. Here are the five key limitations I’ve seen—lessons learned from both successes and failures—that every engineer should keep in mind.
1) The Factor of Safety Is Not a Guarantee
We often present a neat number: FS = 1.5 for static conditions, FS = 1.1 for seismic. It looks precise, scientific, and comforting. But that single number hides a world of assumptions—soil strength parameters, pore pressures, failure surface shapes, boundary conditions, sensitivity to parameter selection. A factor of safety isn’t a prediction of performance; it’s a measure of confidence in a model. I’ve known facilities to fail with calculated FS values well above 1.0 because the inputs didn’t reflect reality. Which brings me to the next point.
2) Input Data Is Almost Always the Weak Link
Stability models are only as good as the strength and pore pressure parameters fed into them. Garbage in, garbage out. Laboratory triaxial tests on tailings samples may not capture field conditions. Piezometer data provides snapshots, not complete pictures. Variability across a foundation or an embankment is often simplified into a single “representative” value. The greatest source of error isn’t the math—it’s the data. That’s why I place such emphasis on field monitoring and updating models as the facility evolves.
3) Failure Surfaces Are Rarely as Simple as Our Models
Most slope stability methods—whether limit equilibrium or finite element / finite difference—assume a relatively simple failure mechanism: circular or planar slip surfaces. But real-world failures may be messy, especially in complex multi-component embankments or foundation conditions. They can also involve progressive failure, strain-softening, static liquefaction, or a combination of mechanisms. The 2014 Mount Polley failure and other well-documented cases show that complex failure modes often defy our textbook models. If you only look for circles, you may miss the real risks.
4) Probability and Uncertainty Are Underappreciated
For years, we’ve relied almost exclusively on deterministic analyses: one set of inputs, one output. But soils don’t behave deterministically. There’s scatter, uncertainty, and spatial variability in every parameter we measure. Probabilistic approaches (e.g., reliability-based design) offer a better sense of the likelihood of failure, but they remain underused in practice. Even then, probability is only part of the picture—we still need engineering judgment.
5) Analyses Don’t Replace Observational Data
Perhaps the biggest lesson of my career is that stability analyses must be married with real-time observation. Piezometers, inclinometers, settlement markers, satellite InSAR, and good old-fashioned site inspections provide the feedback loop that no model can. I’ve seen facilities where analyses looked fine, but instrumentation told a different story—and it was (most often) the instruments that turned out to be right.
Final Reflections
Slope stability analyses are invaluable. They guide design, support permitting, and give us a structured way to think about risk. But they are not crystal balls. The real danger comes when we treat them as absolute truth instead of informed estimates. After 40 years, I view slope stability analyses as one piece of a larger puzzle—complemented by rigorous site investigations, conservative water management practices, thoughtful construction approaches, and continuous monitoring.
If there’s one takeaway, it’s this: don’t fall in love with the neat numbers that models give you. Stay humble, stay observant, stay curious, and remember that the ground doesn’t read your reports. Mother Nature has no contract with your assumptions.