FEType Error with Adaptivity

[henryg9273]

Dear all,

I am trying to solve fluid-chemistry-thermal diffusion PDEs with several blocks in the Navier-Stokes module:

With the adaptivity block, an error message " ERROR: Bad FEType.family= 41 " always comes after the first converged step. Without the adaptivity block, the calculation can continue, but it is hard to converge. Below is the adaptivity block:

[Adaptivity]
marker = errorfrac #this line when commented, switches off adaptivity
max_h_level = 3
#steps = 2 #this line gets ignored in a transient run
[./Indicators]
[./error]
type = GradientJumpIndicator
variable = Cure
outputs = none
[../]
[./error2]
type = GradientJumpIndicator
variable = Temperature
outputs = none
[../]

[../]
[./Markers]
[./errorfrac]
type = ErrorFractionMarker
refine = 0.65
coarsen = 0.2
indicator = error
outputs = none
[../]

[../]
[]

I'm not sure about whether the types of marker and indicator are supported in the Navier-Stokes package or if it is the cause of the error.

Can I get some suggestions on this? The full script is attached below.

Thank you in advance!
Yuan

[GlobalParams]
supg = true
pspg = true
alpha = 1e-1
[]

[Mesh]
file = FluidNew.inp
uniform_refine = 0
construct_side_list_from_node_list = true
dim = 2
[]

[AuxVariables]
[vel_x]
order = FIRST
[]
[vel_y]
order = FIRST
[]
[]

[AuxKernels]
[vel_x]
type = VectorVariableComponentAux
variable = vel_x
vector_variable = velocity
component = 'x'
[]
[vel_y]
type = VectorVariableComponentAux
variable = vel_y
vector_variable = velocity
component = 'y'
[]
[]

[Variables]

[./velocity]
order = FIRST
family = LAGRANGE_VEC
[../]

[./Temperature]
order = FIRST
family = LAGRANGE
initial_condition = '20'
[../]

[./p]
[../]

[./Cure]
order = FIRST
family = LAGRANGE
initial_condition = '0.3'

[../]

[]

[Kernels]
[./mass]
type = INSADMass
variable = p
[../]

[./momentum_time]
type = INSADMomentumTimeDerivative
variable = velocity
[../]

[./momentum_convection]
type = INSADMomentumAdvection
variable = velocity
[../]

[./momentum_viscous]
type = INSADMomentumViscous
variable = velocity
[../]

[./momentum_pressure]
type = INSADMomentumPressure
variable = velocity
p = p
integrate_p_by_parts = true
[../]

[./temperature_advection]
type = INSADTemperatureAdvection
variable = Temperature
velocity = velocity
[../]

[./temperature_time]
type = ADHeatConductionTimeDerivative
variable = Temperature
specific_heat = 'cp'
density_name = 'rho'
[../]

[./temperature_conduction]
type = ADHeatConduction
variable = Temperature
thermal_conductivity = 'k'
[../]

[./coupledcurederv]
type = CoupledCureTimeDerivative
variable = Temperature
v = Cure

[../]

[./curederv]
type=TimeDerivative
variable = Cure
lumping=false

[../]

[./cureformula]
type=DCPDnonDgeneralPT
variable = Cure
v = Temperature #this is the coupled variable
Ttrig = '1'
Tintl = '0'
#_E = '82179.5'
#_n = '2.2275'
#_Kcat = '0'
_E= '110750'
_n= '1.7215'
_m= '0.77'
_cd= '14.4778'
_ad= '0.405'
[../]

[./cureadvection]
type=CureAdvection
variable = Cure
v = velocity
[../]

[]

[BCs]
[./no_slip]
type = VectorFunctionDirichletBC
variable = velocity
boundary = Top
[../]

[./no_slip2]
type = VectorFunctionDirichletBC
variable = velocity
boundary = Bottom
[../]

[./lid]
type = VectorFunctionDirichletBC
variable = velocity
boundary = Left
function_x = 'lid_function'
[../]

[./outlet_p]
type = DirichletBC
variable = p
boundary = Right
value = 0.0
[../]

[./Initiation]
type = DirichletBC
#type = NeumannBC
variable = Temperature
boundary = Left
value = '200'
[../]
[]

#length is 5*30mm, Rm is 5mm
[Materials]
[./DCPD]
block = 0
type = GenericConstantMaterial

    prop_names = 'specific_heat Hr density TConductivity A mu rho cp k'
prop_values = '1600000 350000000 9.8e-7 0.152 8.55e15 15000000 9.8e-7 1600000 0.152'  # polymer attributes

[../]
[ins_mat]
type = INSADMaterial
velocity = velocity
pressure = p
transient_term = true
integrate_p_by_parts = true
[]

[]

[Functions]
[./lid_function]
# We pick a function that is exactly represented in the velocity
# space so that the Dirichlet conditions are the same regardless
# of the mesh spacing.
type = ParsedFunction
value = '5e-5y(4-y)'
[../]
[]

[Preconditioning]
[./SMP]
type = SMP
full = true
solve_type = 'NEWTON'
[../]
[]

[Executioner]
type = Transient
num_steps = 10000

nl_rel_tol = 1e-5

end_time=200

nl_max_its=10
l_max_its=15

[./TimeStepper]
type = ConstantDT
dt = 0.01
# dt = 3.6e14
[../]

[./TimeIntegrator]
type = ImplicitEuler
[../]

#Preconditioned JFNK (default)
#solve_type = 'PJFNK'
petsc_options_iname = '-pc_type -pc_hypre_type'
petsc_options_value = 'hypre boomeramg'
#petsc_options_iname = '-pc_type'
#petsc_options_value = 'gamg'

[]

[Adaptivity]
marker = errorfrac #this line when commented, switches off adaptivity
max_h_level = 3
#steps = 2 #this line gets ignored in a transient run
[./Indicators]
[./error]
type = GradientJumpIndicator
variable = Cure
outputs = none
[../]
[./error2]
type = GradientJumpIndicator
variable = Temperature
outputs = none
[../]

[../]
[./Markers]
[./errorfrac]
type = ErrorFractionMarker
refine = 0.65
coarsen = 0.2
indicator = error
outputs = none
[../]

[../]
[]

[Controls]
[./bcs]
type = TimePeriod
disable_objects = 'BCs::Initiation'
start_time = '1.0' #'1718880' '1317808'
execute_on = 'initial timestep_begin'
[../]

[]

[Outputs]
execute_on = 'timestep_end' # Limit the output to timestep end (removes initial condition)
[./console]
type = Console
perf_log = true # enable performance logging
[../]
[./exodus]
type = Exodus
execute_on = 'initial timestep_end' # output the initial condition for the file
output_material_properties = true
show_material_properties = 'specific_heat thermal_conductivity'
file_base = ./Fluid/DCPD # set the file base (the extension is automatically applied)
interval = 10 # only output every 10 step
[../]
[]

[aeslaughter]

Can you provide the complete error message please? @lindsayad This is using Navier Stokes, not sure if it is related.

[henryg9273]

Thanks so much, Alex. I have invited you to the repository.

[lindsayad]

It doesn't appear that you included the mesh in your repository. I cannot run your input file without it

[henryg9273]

Sorry, I have just uploaded.

[lindsayad]

Ok you must be using a really old version of MOOSE because I get this error:

*** ERROR ***
The requested AD material property mu is declared as a regular material property. Either retrieve it as a regular material property with getMaterialProperty or declare it as an AD property wtih declareADProperty

This is an error application developers/analysts should have addressed almost a year ago. If I run into more of this, I won't be able to debug your issue at all

[lindsayad]

Ok after updating your input file, I do not get the error you reported. It appears that you were using quite an old version of MOOSE, and it's possible that the error you saw has since been fixed. Please update your MOOSE and then make sure you update libmesh as well.