how to set up boundary conditions in tcl for Brick8N_u_p_U?

[ouyang]

Hi everyone,
I am using Brick8N_u_p_U element to analyze a underground mined panel within saturated porous medium. Here are my questions:

(1). For zero pore pressure boundary condition, for example at a node 6747, I can do this:

fix 6747 0 0 0 1 1 1 1

The first "1" above means zero pore pressure is given at the node 6747 based on the OpenSees menu. But if I want to add a non-zero pore pressure, say 9999, at the node 6747, I can not find any instructions in the menu. Is that correct if I set it up in the tcl file as follows:

...
fix 6747 0 0 0 1 1 1 1
...
pattern Plain 1 Linear {
load 6747 0 0 0 9999 0 0 0
}

(2) Is there any way I can want to specify water flux/per unit time (10 ft^3/second) at a node 123? is that correct if I do it like this:
...
fix 123 0 0 0 0 1 1 1
---
pattern Plain 1 Linear {
load 123 0 0 0 0 10 10 10
}


Thanks for any help.

Ouyang

[silvia]

you might want to contact the authors of that material.

[Boris]

Hello There,

for u-p-U formulation, each of the nodes has 7DOFs (3 u, 1 p and 3 U) and they behave (almost) the same way as regular DOFs...


It is very important to realize that both p and U describe BC for the same pore fliud so they are not independent.
So if you specify pore pressure as constrained (you set it to 0.0 by putting 1 for BC (in SAP parlance)) than you have to be carefull with also preventing movement of that fluid (by putting 1 1 1 for U). You can prevent movement of water in some directions (where it is physical to do so) but if you completely fix U (1 1 1) you essentially make it undrained and p cannot be constrained...



for specifying initial values for pore pressure (p) or flux (through U) you have to look at the actuall definition of RHS for this u-p-U system. I would suggest that you look at section 8.2.4. (pages 188 and 189) of my lecture notes...

Boris

[ouyang]

Hi Boris,
Thanks for your help.
Based on the actual definition (eqs. 8.61 and 8.63) of RHS on page 189 of your lecture note, these virtual "loads" are the summation of tractions, body forces and pore pressure at BC. Based on my understanding, if I want to specify the pore pressure at some nodes on BC, I have to "manually" figure out what are the nodal forces at those nodes on BC contributed by body forces and tractions. Then these virtual "loads" can be obtained and specified in tcl like the format

pattern Plain 1 Linear {
load 123 fsx fsy fsz ff ffx ffy ffz
}

Here are my questions:
1. Since the node forces contributed by the body forces are element-dependent, it is not easy to estimate them manually. Practically, if I don't know those node forces, I can not calculate the virtual "loads" like "fsx fsy fsz ff ffx ffy ffz ". As a result, BC can not specified. In the meantime, if BC is unknown, the nodal forces at those nodes on BC contributed by body forces can not be analyzed by Opensees with Brick8N_u_p_U.
2. As indicated the above "pattern Plain ....", must "ff" be always zero in the tcl file based on the definitions in equation 8.61 of your lecture note?
3. It seems that the flux at BC can not be specified in RHS of equations 8.61 and 8.63. Your lecture note (page 183, above equation 8.26) says "on \Gamma_w, the normal outflow \dot{w}_n is specified". This is related to that new variable U. Any suggestions?

ouyang

[Boris]

since pore pressure is a DOF you do specifiy it as such (initial displacements...).

there is no forcing term for p on the RHS...


that equation (8.26) gives conditions for strong form (before discretization), while if you look at equations 8,62 and 8.63, they give you the appropriate forms after deiscretization...

Boris Jeremic