The simulation of a staged construction tailings dam

Question

Dear Plaxis, 
 I am simulating a staged construction of upstream tailings dam with the consideration of the actual construction sequence (step-by-step). 
 My first question is for this problem, which calculation type is appropriate to use, the consolidation or the fully coupled flow-deformation analysis . I thought the fully coupled flow-deformation analysis is more appropriate as we can see the water flow between the tailings and waste rock dike, which can not be simulated using consolidation if i understand well. 
 But I saw the issued report for the analysis of the tailings dams failure, the "consolidation" analysis were performed with the given duration time 
 And in the Plaxis2d consolidation analysis, the contour that corresponds to a power water pressure equal to zero is represented by a black line as shown in the below figures. It also agrees well with the contours of degree of saturation. 
 I presume that if the consolidation is used in the calculation, there should be no water flow. But we can see the water flow and the unsaturated tailings in the upper part of the dam close to the face (see the below two figures). Does it mean that we can still see the variation of saturation degree after using retention curves in tailings and dike when using consolidation calculation type? 
 
 My second question is how can we properly set the initial hydraulic condition, is it correct to set the global water table at the top surface of the initial geometry, and then choose "use pressure from the previous phase" for the consolidation calculation of the subsequent newly added tailings or we need to set a new water table for each stage after adding each new layer of tailings? 
 Thanks a lot for your help.

Dennis Waterman · Answer

Dear Jian Zheng, 
 Performing a fully-coupled analysis would indeed be the best representation of the construction of a tailings dam, but it's also the most difficult one for which accurate data on void ratio and permeability is required. Therefore we see that the process is often simplified using a consolidation analysis combined with steady-state groundwater flow. If the construction stages are relatively small and the permeability of the tailings is quite high using steady-state flow is not such a bad assumption. Of course you would have to update the steady-state flow field for every construction stage. 
 With kind regards, 
 Dennis Waterman

Dennis Waterman · Answer

Dear Jian Zheng, 
 1) I quote the last sentence of my previous answer: " Of course you would have to update the steady-state flow field for every construction stage." How you update it (by updating hydraulic boundary conditions or global level) is up to you. 
 2) Yes, you should. Because in this case during the construction process a steady state flow field will never be reached as construction goes faster than the change in flow field. Hence, using a series of steady-state solutions would not be correct. 
 3) In principle flow analysis works with hydraulic boundary conditions. The use of the global water level is just a quick way to assign those boundary conditions rather than have to put them in manually for every boundary line. Note that the global level itself is in fact not used in a flow analysis; only the resulting boundary conditions are used. However, if you do apply boundary conditions yourself, you have to make sure you apply them all. If nothing is applied, seepage is assumed. I'm not sure how the hydraulic heads you applied relate to the height of your model, but if h=2 is actually also the top of your dam h=1 is the top of the low land right from them I can imagine there is an issue with the horizontal lines as they're seepage, but also immediately on the side have a hydraulic head equal to the height. So the lines I coloured red in the figure below may be the problem if they're at h=2 and h=1 respectively. Maybe lowering the heads a bit to h=1.9 and h=0.9 solves it... But otherwise you would have to send your project to our support department for them to have a look at it. 
 With kind regards, 
 Dennis Waterman

Dennis Waterman · Answer

Dear Jian Zheng, 
 1) The criterion is flow .... if there is a transient flow field due to differences in phreatic levels you should use fully coupled analysis. If the phreatic flow field is steady-state, you can use consolidation + steady-state analysis. Excess pore pressures are not a criterion since both analyses will dissipate them. 
 2) There are a lot of mix-ups here. First of all, a fully-coupled analysis is not steady-state. In fact the whole purpose of a fully-coupled analysis is because it's not steady state: it's transient. Secondly, as I already explained in my previous answer under 3), a groundwater flow analysis (steady-state or fully coupled) doesn't use the global water level, it uses the hydraulic conditions on the boundaries. The global water level is just a means to apply the hydraulic conditions on the boundaries in a faster way than doing it manually line-by-line as it will apply a head on the boundary lines depending on the position of the global level. The calculation will use those hydraulic heads, not the global level itself. 
 3) Actually what you did now is not what I meant, but it's good to hear that it works. What I mean was that either 
 - you apply a slightly lower head on the vertical boundaries left and right - or apply a head over a short piece of the horizontal top close to the corner. 
 The "problem" is that the head on the top of the vertical boundary is exactly the same as the y-coordinate. So we're having a node that is a head but with pressure zero, which would indicate a seepage node. Hence, the node definition is ambiguous which may cause numerical issues. Better to explicitly define it. 
 With kind regards, 
 Dennis Waterman