No active order parameters

[ritamcj]

Dear all,
I am seeing the error "No active order parameters" in DeformedGrainMaterial in the phase field module. The relevant code snippet is below:

  const auto & op_to_grains = _grain_tracker.getVarToFeatureVector(_current_elem->id());

  // loop over active OPs
  bool one_active = false;
  for (MooseIndex(op_to_grains) op_index = 0; op_index < op_to_grains.size(); ++op_index)
  {
    if (op_to_grains[op_index] == FeatureFloodCount::invalid_id)
      continue;

    one_active = true;
    auto grain_id = op_to_grains[op_index];

    if (grain_id >= _deformed_grain_num)
    {
      rho_i = 0.0;
    }
    else
    {
       rho_i = _Disloc_Den_i[_qp];
    }
    rho0 += rho_i * (*_vals[op_index])[_qp] * (*_vals[op_index])[_qp];
  }

  if (!one_active && _t_step > 0)
    mooseError("No active order parameters");

My guess is that at a particular element, all grain order parameters become zero. I am unable to understand this. My simulation involves direct order parameter modification (through DiscreteNucleationForce) to simulate nucleation and growth. For new grains in reserved order parameters in GrainTracker, I have assigned a large nucleation force (nucleus_value = 50000). I wonder whether this causes the problem when too many nuclei are formed. I look forward to your suggestions.

Warm Regards,
Ritam Chatterjee

[GiudGiud]

@laagesen @SudiptaBiswas @jiangwen84 on phase field stuff please

[ritamcj]

Hi Sudipta,
Thank you for the reply. I have attached a pseudo input file, which should make it easier to understand my approach
Moose_github_21200_input_file.txt.

I had turned grain remapping off as I wanted my initial Ti-6Al-4V dual phase structure intact to evaluate the preferred location of nucleation at various interfaces such as alpha-alpha and alpha-beta. Can this cause new grains to not find any reserved order parameters? If that is the case, I will allow grain remapping and check again. A slower nucleation rate can work but I think it would eventually lead to the same issue due to unavailability of reserved order parameters for new grains.
Warm regards,
Ritam

[SudiptaBiswas]

If the remapping is turned off then how does the nuclei evolve over time? Without remapping, probably only option you have is increasing the number of order parameters.

[GiudGiud]

@ritamcj any update on this?

[ritamcj]

Hi Sudipta and Guillaume,
I apologize for the delay in responding as I was traveling. With grain remapping off, you correctly stated that the only option is to increase the number of order parameters. However, this significantly slows down the simulation. Hence, I switched on grain remapping, lowered the nucleation rate as well as the hold time of nuclei in the system. The simulation ran longer, but I am still trying to optimize to run up to large strain (~50%). I do accept your insights, and it can be marked as the solution. Many thanks for your explanations.
Warm regards,
Ritam

[YXY101676]

Hello, I want to know if the activation grains and the order parameters here are one-to-one correspondence, I may have misunderstood the code。