Factor of safety drop below 1.0 when running a C-phi reduction

Dear James,

Rarely Safety calculations could give the user the SumMsf value lower than 1.0. 

It means most likely that the previous phase is not stable, probably there is no equilibrium on this phase.

First of all I will advice you to refine the mesh and increase Max steps on the Safety phase (500–1000). We need to know that for defined number of steps we receive the stable value of SumMsf.

If it does not help, please add Nil-phase after the previous phase (Safety phase − 1): just add a phase with Plastic type of calculation and calculate it. There is no changes on the phase. The Soil body seems to be collapse likely appears.

Maybe the reason is in material parameters, geometry, loads, mesh quality and coarseness, phase settings (including iterative procedure settings) etc.

Might be interesting:

— Article Safety analysis and displacements
— Webinar Using PLAXIS to Determine Factors of Safety

Hope it helps.

Thx Rushan, your suggestion are very helpful

Dear James

Does your problem solve?

Dear James,

If the the strength reduction factor SMsf firs reaches a peak value and then after a while suddenly drops below 1 (often all the way down to zero) then the previous step is usually correct and the problem is because of very high strain levels. During safety analysis very high deformations can occur leading to a quite high absolute inaccuracy while the relative accuracy still fulfills the 1% requirement. This high absolute inaccuracy may give large local peak stresses and cause the calculation to diverge, dropping the SMsf below 1. Quite often it comes with the error message "Accuracy not reached in last step", but not always.

So this drop can be ignored when:
  - First a stable value of SMsf is reached before it suddenly drops.
  - The sudden drop occurs at very high levels of deformation (sometimes not even meters but kilometers)
  - Intermediate results (hence, before the drop) show a clear and sensible failure mechanism

Often the problem is that users choose too many additional steps, and so deformations increase to very high levels causing the abovementioned problem. When a peak value of SMsf is reached after let's say 50 steps, there is no need to calculate 500 steps - it will actually most likely only lead to numerical problems. If the peak value is stable while displacement are rapidly increasing for another 20-30 steps it's usually clear failure was reached.


With kind regards,

Dennis Waterman