PorousFlow: how to visualise velocity on surfaces (paraview)

[Traiwit]

Hi guys,

Our mesh has surfaces that we apply BCs on them, I want to visualise the velocity normal to the surfaces, is it possible with MOOSE/Paraview?

[./topol_linear_TOPOLSRF_F005_UG1]
type = PorousFlowPiecewiseLinearSinkBEH
variable = porepressure
boundary = 'TOPOLSRF_F005_UG1'
pt_vals = '-1e9 1e9'
multipliers = '-1e9 1e9'
flux_function = 1
v = min_dist
PT_shift = -9810
fluid_phase = 0
[../]

I know I can visualise the surface via Sets on Paraview, but I only can show node properties (velocity is a mesh property)

Is there a way to work around this (like project element velocity value to the element surface [normal multiply?])?

Thank you very much!

Kind regards,
Traiwit

[GiudGiud]

Hello

You ll want to look at auxkernels for that.
Create a new auxiliary variable vec_times_n, const monomial

Then in the auxkernel, also boundary restricted, you can set the value with velocity * _normal

You ll want the DIffusionFluxAux but change from grad_u to just _u.

There might be one already somewhere, but cant seem to find it

Guillaume

[GiudGiud]

i'm still a bit confused, where does it take the velocity data from?

Do you not have velocity non linear variables?
And if not, cant porous flow compute them from the pressure field?

and how do I visualise this data?

You visualize it on the whole domain, then use the threhshold filter in paraview to only show the nonzero cells. They will all be near the boundary

[Traiwit]

Do you not have velocity non linear variables?
And if not, cant porous flow compute them from the pressure field?

hmm i'm using this to compute Velocity PorousFlowDarcyVelocityComponent

[./darcy_x]
order = CONSTANT
family = MONOMIAL
[../]
[./darcy_y]
order = CONSTANT
family = MONOMIAL
[../]
[./darcy_z]
order = CONSTANT
family = MONOMIAL
[../]
[]


[AuxKernels]
[./darcy_x]
type = PorousFlowDarcyVelocityComponent
variable = darcy_x
gravity = '0 0 -9.81'
component = x
[../]
[./darcy_y]
type = PorousFlowDarcyVelocityComponent
variable = darcy_y
gravity = '0 0 -9.81'
component = y
[../]
[./darcy_z]
type = PorousFlowDarcyVelocityComponent
variable = darcy_z
gravity = '0 0 -9.81'
component = z
[../]
[]

I'm still not sure, I will get back to you once there's an error.

You visualize it on the whole domain, then use the threhshold filter in paraview to only show the nonzero cells. They will all be near the boundary

This makes sense, thanks!

Kind regards,
Traiwit

[GiudGiud]

darcy_xyz could serve as the v_xyz parameters then I think.

btw if you feel like making a pull request this auxkernel could go in the moose/framework. We are keen on external contributions

[Traiwit]

@GiudGiud

I think it kinda almost works now, this is the updated code:

#include "AdvectiveFluxAuxBE.h"
#include "Assembly.h"

registerMooseObject("MOOSE_BEHApp", AdvectiveFluxAuxBE);

InputParameters
AdvectiveFluxAuxBE::validParams()
{
  InputParameters params = AuxKernel::validParams();
  MooseEnum component("x y z normal");
  params.addClassDescription("Compute components of flux vector for advection problems "
                             "$(\\vec{J} = \\vec{v} \\cdot \\vec{n})$.");
  params.addCoupledVar("v", "The x component of velocity");
  params.addParam<MooseEnum>("component", component, "The desired component of flux."); // Added this line
  return params;
}

AdvectiveFluxAuxBE::AdvectiveFluxAuxBE(const InputParameters & parameters)
  : AuxKernel(parameters),
    _component(getParam<MooseEnum>("component")),
    _v(coupledValue("v")),
    _normals(_assembly.normals()),
    _use_normal(getParam<MooseEnum>("component") == "normal")
{
  if (_use_normal && !isParamValid("boundary"))
    paramError("boundary", "A boundary must be provided if using the normal component!");
}

Real
AdvectiveFluxAuxBE::computeValue()
{
  RealVectorValue velocity(_v[_qp]);
  return velocity * _normals[_qp];
}

in this particular case, I only do it for flux_x

  [./flux_x]
    type = AdvectiveFluxAuxBE
    variable = flux_x
    v = darcy_x
    component = normal
    boundary = DRAINSRF_F008_UG4
    check_boundary_restricted = false
  [../]

just wondering why in the Darcy_x it shows negative value (which is correct based on the flow and drainage conditions) but when I project it to the surface (figure below), it becomes positive, is there anything that I missed?

Kind regards,
Traiwit

[GiudGiud]

How is the sideset oriented? The _normal attribute will depend on that

[Traiwit]

How is the sideset oriented? The _normal attribute will depend on that

I think it kinda makes sense now, it positive becase the velocity is flowing outward respect to the surface (hence positive) not to the xyz space. on the other hands if the flow is inward into the surface, it will be positive (i guess)

above is the surface where I apply BCs

[./drain_DRAINSRF_F008_UG4] 
type = PorousFlowPiecewiseLinearSinkBEH 
variable = porepressure 
boundary = 'DRAINSRF_F008_UG4'
pt_vals = '0 1e9' 
multipliers = '0 1e9' 
flux_function = 1
v = min
PT_shift = 0 
use_mobility = true
fluid_phase = 0
[../]

[GiudGiud]

I think it kinda makes sense now, it positive becase the velocity is flowing outward respect to the surface (hence positive) not to the xyz space. on the other hands if the flow is inward into the surface, it will be positive (i guess)

yes, as long as the sideset is consistently oriented

how is that surface/sideset defined? using a commercial mesher?
Does it consistently point the same direction?

We unfortunately do not currently have a way (coming this summer is the plan) to look at sideset directions from MOOSE

[Traiwit]

yes, as long as the sideset is consistently oriented

how is that surface/sideset defined? using a commercial mesher? Does it consistently point the same direction?

Yes i think so, from Abaqus.

We unfortunately do not currently have a way (coming this summer is the plan) to look at sideset directions from MOOSE

This will be awesome! looking forward to have this feature soon!

all good for now, thank you @GiudGiud!