I did an analysis of pipe burst-induced transient flow (flow only) in unsaturated soils. I did the analysis as below:
(1) Assign pressure head at boundaries and evaluate the pore water pressure with suction using steady state flow method;
(2) Assign pipe burst pressure near the top surface (closed flow BC), and evalute the wetting extent using transient flow method.
However, during my analysis, very high suction points were observed in the front of the wetting extent (as shown in the suction plot below, where black line means the phreatic level).
This is confusing to me, why such high suction can be found near the wetting extent? In particular, it seems that when I use a finer mesh, the suction points become less, but still exist.
Hope someone can tell me how to address this issue. Thanks!
(a) Initial suction distribution after Step 1 (steady state flow) is achieved.
(b) Distribution of suction during transient flow after pipe burst starts
(c) Comparison of suction in coarser and finer mesh
The area with suctions is the area in between the original phreatic level and the wetting area. This area hasn’t been wetted yet by the pipe burst and has therefore still the original suction as from before the pipe burst.
Suction is caused by the capillary flow against gravity, so the suction in a point depends on how deep is the saturated flow below that point. The fact that there is fully saturated soil just above has no influence on the suction.
For transient analysis, element size is dependent on time step used. Larger time step requires smaller element size for accurate results and vice versa.
Dear Faseel,
Thanks for your prompt reply.
The problem is, suction values in front of the wetting edge during transient flow is higher than these initial values achieved in the previous steady state flow analysis.
For example, at the location of x = 2.6 and y = 5.6, the initial suction was about 20 - 30 kPa (Figure 1 below). However, it increased to more than 50 kPa in the subsequent transient flow analysis (Figure 2 below). Moreover, if I built a higher slope (say 30 m height), this suction increament reached more than 200 kPa due to the transient flow. This seems out of expectation.
Figure 1 Initial suction at steady state flow
Figure 2 Updated suction in transient flow analysis
Dear Haifeng,
The area with the original suctions is only a few elements large and the peak value seems to occur in one single node. So it seems that indeed it's a matter of accuracy as my colleague Faseel already hinted, and it needs some serious refinement of the mesh. Generally wetting problems need very fine meshes.
With kind regards,
Dennis Waterman
Dear Dennis,
Seems that finer mesh provides a possible solution to my problem. I would try it. Thanks for your information!
Regards,
Haifeng