wilkandy/ba_hun

Repository for creating a new MOOSE based application

ba_hun's Issues

make BAPiecewiseLinearSinkFlux postprocessor

need lumped mass postprocessor

Just like RichardsMass, but uses lumped version.

update error macros with streaming to use new moose functions

PR coming...

remove RealTensorValueAux in favour of MaterialRealTensorValueAux

BAHalfCubicSink incomplete type 'class Function'

Update of moose requires different Function includes, maybe?

Add BAMaterial

This has AuxVariable inputs that define zones that define the permeability and porosity. This is so that i don't need lots of different blocks in the input mesh: just one block with different zones

BAHalfCubicSinkAux needed

This is so i can easily create a map of actual ET

BAHalfCubicSink and BAHalfCubicSinkFlux cutoff enhancement

At present it is a Real. I should change to a Function, as then extinction depth can more easily be defined as varying over the model domain.

BAMaterial - porosity needs to be treated like permeability

BAPolyLineSink needs to record porepressure at the point

This is because PointValue Postprocessor appears to segfault on the BA meshes. However, a DiracKernel placed at the same point doesn't segfault. So i can put a BAPolyLineSink with zero flux at the point, and record the porepressure

Remove kernels/simple_diffusion.test

BASumPostprocessor divide by a postprocessor instead of a Real

BASumPostprocessor

make a PolyLineSink where P0 is a function

Sink strength = C(P - P0)

i want P0 to be a function of t

BAPolyLineSink should cache elements

This is temporary, until DiracKernel does the multi-processor caching, or something else is fixed so this isn't necessary

Include modflow_shift in VG curve specification

Modflow doesn't properly model unsaturated flow. I can partially replicate this by artificially shifting the porepressure in the suction curve to give full saturation for modflow_shift <= P < 0. Possibly this won't effect results of the BA models, but the point is they also might run faster, being more linear. I'm just adding this issue so i can experiment.

put max and min permeability and porosity in BAMaterial

conform to moose standards

Use new input parameter construction

Replace the legacy input parameter construction with the new input parameter construction.

Do not turn on all modules

BaHun needs RICHARDS only.

make a BAHalfCubicSink and BAHalfCubicSinkFlux

Just like RichardsGaussian, but with Cubic. Then the extinction depth can be specified exactly without the "trailing Gaussian tail".

Fix unit testing

Currently it uses CPPUNIT. Change to GTEST.

BAMaterial should be able to determine the transverse direction

For swept meshes

BAPolyLineSink has start and end times

Remove unused params

Bug Description

Due to new changes in the way moose handles default behavior for unused parameters, multiple tests are now hitting errors for their unused parameters. There are only a handful but they must be addressed before the PR for MOOSE can succeed.

Steps to Reproduce

The specific input files are viewable in the attached MR, and can be seen on the civet job here: https://civet.inl.gov/job/1077111/

Impact

The Moose update will not be mergeable until this change is made, so it is of a high importance. Removing the tests should have little impact but should be verified by this repo's developers.

general-head BC with auxvar

probably derive from RichardsPiecewiseLinearSink, and have an auxvar that multiplies the flux. RichardsPiecewiseLinearSink only has a function that multiplies the flux.

Jacobian test for PETSc-3.9.4

PETSc-3.9.4 has a little different way to compute FD Jacobian. If users are intending to test the Jacobian at the frist time step, we have to use num_steps = 1

BASumPostprocessor should add zero if w=0

make a max permeability change in BAMaterial

Stop using deprecated interface in AppFactory

Migrate to smart pointers in main.C

Make BAMaterial have constant perm and por, indep of qp