Problem description

For a seepage boundary the kernel determines in each step and for each node the exact boundary condition during a steady state groundwater flow analysis. Depending on the situation the boundary condition may change between a closed flow boundary (no flow) or a prescribed head boundary (inflow or outflow of water depending on the calculation).

In order to determine if the node is a closed or prescribed head boundary we analyse the discharges through the node. For very small discharges the boundary is assumed to be closed, otherwise there is a prescribed head value. To prevent numerical oscillations we use certain tolerances on these discharge values.

Due to an issue with these tolerances it may happen that (part of) the seepage boundary is incorrectly set to the wrong boundary conditions. As a result the calculation could converge to the wrong solution. In the example below part of the seepage boundary was not closed but remained a prescribed head boundary causing an unrealistic inflow of water. Consequently the calculation converged to the wrong solution.

Figure 1: incorrect steady state solution

Figure 2: correct steady state solution

The issue mentioned here occurs for models with relatively low discharges in the nodes along the seepage boundary. This is in general caused by low permeable soils. Unfortunately “low permeable” can not be indicated with a more specific value here since this is dependent on the exact situation at hand.

Realise that some models are more susceptible of this issue than others. A typical situation where this may occur is a sloped embankment. A typical situation where this most likely will not occur is the bottom of an excavation pit where outflow occurs along the full seepage boundary.

Solution

This issue has been solved in PLAXIS 2D AE and PLAXIS 2D Classic.

In older versions: for steady-state groundwater flow situations, a possible work around may be to upscale all permeabilities in the model with the same factor (typical factors are 10, 100, etc.). Note that the steady-state solution is not dependent on the absolute values of the permeability but only on the relative differences in permeability between the soil layers.

When you are uncertain about your results please send your packed project to support (contact page) so we may further investigate your model.