Incorrect initial stress distribution within soil if inserting liner

Hi, I am trying to model a large pipe pile in soil, where the pile is modelled with liner elements. As the first step, I need to establish a proper initial geostatic conditions, which is easy to do without the liner (pile). However, once I include the liner (even zero density of the liner elements and small stiffness), I cannot have reasonable stress field - positive or near-zero horizontal stress inside the pile. Do you know the reason and solution? I can send you files if required. In the following, I am providing some figures:
Fig 1: Density and boundary conditions.
Fig 2: Stress conditions with ‘zone initialize-stresses ratio = 1.0’, and then ‘model cycle 0’. Everything look good.
Fig 3: Final vertical stress conditions with ‘model solve convergence = 1.0’. Things become a lot worse.
Fig 4: Final horizontal stress conditions with ‘model solve convergence = 1.0’. Does not look good.

As a general rule, it is impossible to diagnose a FLAC3D issue with any certainty without a complete data file so we can see exactly what you are doing.

However - from the images I would speculate that the liner-zone coupling springs are too soft, so the zones in contact with the liner are de-stressed after the liner is installed.

for the rule of thumb on recommended coupling stiffness.
Also see the STRUCT LINER INITIALIZE COUPLING command (below), which will initialize coupling spring forces to approximately balance the state of stress in the contacting zones.

Thanks, David!
I have tried a range of stiffness of the liner-zone coupling springs, which does not help. Also, on “STRUCT LINER INITIALIZE COUPLING”, I have done quite a few analyses for different problems, and never found it works properly. For the current problem, this command makes the stress distribution even worse.
Anyway, I am sending through my files, hoping you guys can find the reasons.

The problem was that at this stage you input a very low cohesion to the links. yielding/fails occurred to these links. It is recommended a very high link cohesion during this (initial stress) stage and input/modify a realistic value after this stage.

Thanks, Cheng! I have tried by “coupling-cohesion-shear=1000.0”. However, results did not improve much - still odd stress distribution. I have checked the shear stress along the liner, which is only around 5 kPa. Have you looked at my files? Maybe send your version back to me for me to try.

I also input an large coupling-yield-normal values:

coupling-yield-normal=1e10 coupling-yield-normal-2=1e10
coupling-cohesion-shear=1e10 coupling-cohesion-shear-2=1e10

The key is to avoid link fails during this stage.

BTW, a better way would setup liners after the in-situ stress has been setup and run to equilibrium.

Hi Cheng, I have implemented exactly as you advised. The results are improved (see Figure 1 below), but still not fully correct yet. I have checked the link state - no fail.
Could you pleaes elaborate “set up liners after the in situ stress has been set up…”? Actually I have tried this, but failed. Apparently the zones cannot be separated anymore for inserting liners, if the initial stress state has been set up.
Theoretically, it does not make sense to me why inserting of a weightless liner will affect the initial stress state which was in balance, though numerically it may happen due to the way the liner is interacting with the soil. In other words, I do not really understand why FLAC3D still needs to run many steps and return a wrong stress distribution while the initial stress distribution established with “zone initialize-stresses ratio 1.0” should already be in equilibrium automatically. I guess it must be due to inserting of liner. However, unfortunately for some reason “stru liner initialize coupling” is not working in this case. Is it because of curved shape of the liner?
In addtion, when I was checking, I noticed something odd for “Liner Target of Link Side 1” (see Figure 1 below). There is some green shade above the soil indicating presence of link between the node and zone. However, there is no zone above the top of the soil. For Side 2 (outside), it seems to be more reasonable - The top of the green shade is flush with the top of the soil (See Figure 2 below).

I got quite reasonable results. See the Figure below. I also sent you my modified datafile.

1 Like

Thank you, Cheng, very much! The reason why I cannot have your results previously is because I have also give high cohesion to the interface of the liner elements (links between node and zone) which are located above the top of the soil. I thought this is not important since there is no node-zone link above the top of the soil (since no zone). But for some reason, on slide 1, there is node-zone links above the tope of the soil, as I mentioned in my previous msg, and also the red part of the shade on the figure below. I do not know how this was generated when I created the liners, but I have to find a way to delete them as they are the trouble makers.

Unfortunately, the intermediate step of setting up initial stress state with unrealistically high cohesion and tension does not really help at the end. After this, once you change cohension and tension back to near zero (keep the same friction angle), which are the realistic values for the liner-zone interface, everything is messed up again.
It looks to me that all these issues are due to the incorrect coupling-shear and coupling-normal along the edge of the liner elements, see figures below. For the 2nd figure, the unreasonable normal stress is only at the tip of the pipe pile.