Node-level pore pressure reduction in coupled PorousFlow and SolidMechanics analysis

[vabufano]

Hello, I'm trying to perform a numerical analysis using the PorousFlow and SolidMechanics modules.
A soil cube is divided in half into two superimposed blocks (bottom block1 and top block2).
At time $t=0$, the cube is entirely submerged with the water table at ground level (at the top of block2) and the pore pressure (pp) distribution is hydrostatic.
What I would like to implement in my analysis is a reduction of the pp value for all nodes in block2 from their initial value (which will obviously be different depending on the node) to a pp value of 0 kPa.
I thought this might be possible with an auxvariable pp_aux that should simply be a copy of the pp variable, and with an auxkernel. Regarding the auxkernel, the only one that seems useful for my problem is PorousFlowPropertyAux, but this evaluates the variable at the element level and not at the node.
What could be an alternative way to do this? Also, regarding the time-dependent reduction of pp, do you think a boundary condition of the type MatchedValueBC could be useful?
Thank you!

P.S. I'm attaching an input file to illustrate the geometry and the problem. Note that this file only defines a hydrostatic pp distribution and does not yet include the time-dependent pp reduction in block2, as I am still looking for a solution for that.
column_erosion_no_decrease_w.txt

[GiudGiud]

Hello

What I would like to implement in my analysis is a reduction of the pp value for all nodes in block2 from their initial value (which will obviously be different depending on the node) to a pp value of 0 kPa.
I thought this might be possible with an auxvariable pp_aux that should simply be a copy of the pp variable, and with an auxkernel

PP is a solver variable for porous flow. So you cannot change it's value with an auxkernel. You need the solve to give you this lower value, possibly by changing the boundary conditions

[GiudGiud]

To perform what is described in point 2 I was thinking of using an AuxVariable and an AuxKernel.
Taking the above into account, is it possible to do this with an AuxVariable and an AuxKernel?

You can copy a nodal pp_var variable onto a nodal auxvariable using either the CopyValueAux or the ProjectionAux (same output)

[josebastiase]

okay.. maybe @GiudGiud can help you with this one.. I would do a restart and set ICs to zero at block 2, but im not sure...

[vabufano]

@GiudGiud Thanks! Any tips on how to set up a BC that gradually reduces the pp_values starting from their initial value at $t=0$? How can I recall the values copied with CopyValueAux or ProjectionAux in order to have something like $pp_{t=0}$*(1-t)?
@josebastiase Thanks! This is the last chance I'm going to take, because I really want to avoid having to restart the analysis if I can.

[GiudGiud]

How can I recall the values copied with CopyValueAux or ProjectionAux in order to have something li

you can chain auxkernels. So create a third variable using a ParsedAux.

Any tips on how to set up a BC that gradually reduces the pp_values starting from their initial value at

Not my field of expertise sorry!
I imagine you can use a function in a porous flow pressure/injection pressure BC and lower that

[navier-rgb]

@vabufano, did you solve this problem? I’m not entirely sure about your final objective. If you want the initial value of pp (at the start of the simulation) as a boundary condition for block 2, remember that BCs only accept values defined on the boundary, not values for all nodes in the entire block. Therefore, for all nodes in block 2, the most appropriate approach is to define IC for Block 2 instead of trying to impose this through a boundary condition.

What you need to do is to segment the ICs of the blocks and pass the initial pp values into a function (in the functions block). Then you can have a new IC for block 2.

[ICs]
  [pp_ic]
    type = FunctionIC
    variable = 'pp'
    function = ini_pp
    block = '1'
  []
  [pp_ic2]
    type = FunctionIC
    variable = 'pp'
    function = pp_ic2
    block = '2'
  []
[]
[Functions]
  [ini_pp]
    type = ParsedFunction
    expression = '(${y_top} - y) * 10.'
  []
  [pp_ic2]
    type = ParsedFunction
    vars = 'p'
    vals= 'ini_pp' # use postprocessor value pp_block2_avg if you want updated pp values
    expression = 'p + 100.'
  []
[]

If you want updated pp values at the start of each time step, you can use the values from the postprocessor:

[Postprocessors]
  [pp_block2_avg]
    type = ElementAverageValue
    variable = pp_aux
    block = '1'
  []
[]