Homec4science
Diffusion akantu (features/30-implementing-a-checkpoint-restart-mechanism) Browse

rAKA/src/model/diffusion_model/diffusion_model.cc
1db5ba830ccbfeatures/30-implementin…

Last Change
Raw File

diffusion_model.cc
View Options

/**
* Copyright (©) 2011-2023 EPFL (Ecole Polytechnique Fédérale de Lausanne)
* Laboratory (LSMS - Laboratoire de Simulation en Mécanique des Solides)
*
* This file is part of Akantu
*
* Akantu is free software: you can redistribute it and/or modify it under the
* terms of the GNU Lesser General Public License as published by the Free
* Software Foundation, either version 3 of the License, or (at your option) any
* later version.
*
* Akantu is distributed in the hope that it will be useful, but WITHOUT ANY
* WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR
* A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more
* details.
*
* You should have received a copy of the GNU Lesser General Public License
* along with Akantu. If not, see <http://www.gnu.org/licenses/>.
*/
/* -------------------------------------------------------------------------- */
#include "heat_transfer_model.hh"
// #include "dumpable_inline_impl.hh"
#include "element_synchronizer.hh"
#include "fe_engine_template.hh"
// #include "generalized_trapezoidal.hh"
#include "diffusion_law.hh"
#include "group_manager_inline_impl.hh"
#include "integrator_gauss.hh"
#include "mesh.hh"
#include "parser.hh"
#include "shape_lagrange.hh"
/* -------------------------------------------------------------------------- */
#include "dumper_element_partition.hh"
// #include "dumper_elemental_field.hh"
#include "dumper_internal_material_field.hh"
#include "dumper_iohelper_paraview.hh"
/* -------------------------------------------------------------------------- */
namespace akantu {
namespace diffusion {
namespace details {
class ComputeRhoFunctor {
public:
ComputeRhoFunctor(const DiffusionModel & model) : model(model){};
void operator()(Matrix<Real> & rho, const Element & element) {
auto law_id = model.getConstitutiveLawByElement()(element);
rho.array() = model.getConstitutiveLaw(law_id).getRho();
}
private:
const DiffusionModel & model;
};
} // namespace details
} // namespace diffusion
/* -------------------------------------------------------------------------- */
DiffusionModel::DiffusionModel(Mesh & mesh, Int dim, const ID & id,
const std::shared_ptr<DOFManager> & dof_manager,
const ID & dof_name, ModelType model_type)
: Parent(mesh, model_type, dim, id), dof_name(dof_name) {
AKANTU_DEBUG_IN();
this->initDOFManager(dof_manager);
if (this->mesh.isDistributed()) {
auto & synchronizer = this->mesh.getElementSynchronizer();
this->registerSynchronizer(synchronizer, SynchronizationTag::_diffusion);
this->registerSynchronizer(synchronizer,
SynchronizationTag::_diffusion_gradient);
}
registerFEEngineObject<FEEngineType>(id + ":fem", mesh, spatial_dimension);
this->mesh.registerDumper<DumperParaview>(id, id, true);
this->mesh.addDumpMesh(mesh, spatial_dimension, _not_ghost, _ek_regular);
AKANTU_DEBUG_OUT();
}
/* -------------------------------------------------------------------------- */
FEEngine & DiffusionModel::getFEEngineBoundary(const ID & name) {
return aka::as_type<FEEngine>(getFEEngineClassBoundary<FEEngineType>(name));
}
/* -------------------------------------------------------------------------- */
MatrixType DiffusionModel::getMatrixType(const ID & matrix_id) const {
if (matrix_id == "K" or matrix_id == "M") {
return _symmetric;
}
return _mt_not_defined;
}
/* -------------------------------------------------------------------------- */
void DiffusionModel::assembleMatrix(const ID & matrix_id) {
if (matrix_id == "K") {
this->assembleDiffisivityMatrix();
} else if (matrix_id == "M" and need_to_reassemble_capacity) {
this->assembleCapacityMatrix();
}
}
/* -------------------------------------------------------------------------- */
void DiffusionModel::assembleLumpedMatrix(const ID & matrix_id) {
if (matrix_id == "M" and need_to_reassemble_capacity) {
this->assembleCapacityMatrixLumped();
}
}
/* -------------------------------------------------------------------------- */
void DiffusionModel::assembleResidual() {
this->assembleInternalFlow();
this->getDOFManager().assembleToResidual(dof_name, *this->external_flow, 1);
this->getDOFManager().assembleToResidual(dof_name, *this->internal_flow, 1);
}
/* -------------------------------------------------------------------------- */
void DiffusionModel::predictor() { ++diffusion_release; }
/* -------------------------------------------------------------------------- */
void DiffusionModel::corrector() { ++diffusion_release; }
/* -------------------------------------------------------------------------- */
void DiffusionModel::initSolver(TimeStepSolverType time_step_solver_type,
NonLinearSolverType /*unused*/) {
DOFManager & dof_manager = this->getDOFManager();
this->allocNodalField(this->diffusion, 1, dof_name);
this->allocNodalField(this->external_flow, 1, "external_flow");
this->allocNodalField(this->internal_flow, 1, "internal_flow");
this->allocNodalField(this->blocked_dofs, 1, "blocked_dofs");
if (not dof_manager.hasDOFs(dof_name)) {
dof_manager.registerDOFs(dof_name, *this->diffusion, _dst_nodal);
dof_manager.registerBlockedDOFs(dof_name, *this->blocked_dofs);
}
if (time_step_solver_type == TimeStepSolverType::_dynamic ||
time_step_solver_type == TimeStepSolverType::_dynamic_lumped) {
this->allocNodalField(this->diffusion_rate, 1, dof_name + "_rate");
if (not dof_manager.hasDOFsDerivatives(dof_name, 1)) {
dof_manager.registerDOFsDerivative(dof_name, 1, *this->diffusion_rate);
}
}
}
/* -------------------------------------------------------------------------- */
std::tuple<ID, TimeStepSolverType>
DiffusionModel::getDefaultSolverID(const AnalysisMethod & method) {
switch (method) {
case _explicit_lumped_mass: {
return std::make_tuple("explicit_lumped",
TimeStepSolverType::_dynamic_lumped);
}
case _static: {
return std::make_tuple("static", TimeStepSolverType::_static);
}
case _implicit_dynamic: {
return std::make_tuple("implicit", TimeStepSolverType::_dynamic);
}
default:
return std::make_tuple("unknown", TimeStepSolverType::_not_defined);
}
}
/* -------------------------------------------------------------------------- */
ModelSolverOptions
DiffusionModel::getDefaultSolverOptions(const TimeStepSolverType & type) const {
ModelSolverOptions options;
switch (type) {
case TimeStepSolverType::_dynamic_lumped: {
options.non_linear_solver_type = NonLinearSolverType::_lumped;
options.integration_scheme_type[dof_name] =
IntegrationSchemeType::_forward_euler;
options.solution_type[dof_name] = IntegrationScheme::_temperature_rate;
break;
}
case TimeStepSolverType::_static: {
options.non_linear_solver_type = NonLinearSolverType::_newton_raphson;
options.integration_scheme_type[dof_name] =
IntegrationSchemeType::_pseudo_time;
options.solution_type[dof_name] = IntegrationScheme::_not_defined;
break;
}
case TimeStepSolverType::_dynamic: {
if (this->method == _explicit_consistent_mass) {
options.non_linear_solver_type = NonLinearSolverType::_newton_raphson;
options.integration_scheme_type[dof_name] =
IntegrationSchemeType::_forward_euler;
options.solution_type[dof_name] = IntegrationScheme::_temperature_rate;
} else {
options.non_linear_solver_type = NonLinearSolverType::_newton_raphson;
options.integration_scheme_type[dof_name] =
IntegrationSchemeType::_backward_euler;
options.solution_type[dof_name] = IntegrationScheme::_temperature;
}
break;
}
default:
AKANTU_EXCEPTION(type << " is not a valid time step solver type");
}
return options;
}
/* -------------------------------------------------------------------------- */
void DiffusionModel::assembleDiffisivityMatrix() {
AKANTU_DEBUG_ASSERT(this->getDOFManager().hasMatrix("K"),
"The K matrix has not been initialized yet.");
this->getDOFManager().zeroMatrix("K");
for_each_constitutive_law(
[](auto && diffusion_law) { diffusion_law.assembleDiffusivityMatrix(); });
}
/* -------------------------------------------------------------------------- */
void DiffusionModel::assembleInternalFlow() {
this->internal_flow->zero();
this->synchronize(SynchronizationTag::_diffusion);
for (auto ghost_type : ghost_types) {
for_each_constitutive_law([&](auto && diffusion_law) {
diffusion_law.computeDiffusivityGradU(ghost_type);
});
for_each_constitutive_law([&](auto && diffusion_law) {
diffusion_law.assembleInternalFlow(ghost_type);
});
}
}
/* -------------------------------------------------------------------------- */
auto DiffusionModel::getStableTimeStep() -> Real {
AKANTU_DEBUG_IN();
Real el_size{};
Real min_el_size = std::numeric_limits<Real>::max();
for (auto && type : mesh.elementTypes(spatial_dimension, _not_ghost)) {
auto nb_nodes_per_element = mesh.getNbNodesPerElement(type);
Array<Real> coord(0, nb_nodes_per_element * spatial_dimension);
FEEngine::extractNodalToElementField(mesh, mesh.getNodes(), coord, type,
_not_ghost);
for (auto && el_coord :
make_view(coord, spatial_dimension, nb_nodes_per_element)) {
el_size = FEEngine::getElementInradius(el_coord, type);
min_el_size = std::min(min_el_size, el_size);
}
}
Real min_dt = std::numeric_limits<Real>::max();
for_each_constitutive_law([&](auto && diffusion_law) {
min_dt = std::min(diffusion_law.getStableTimeStep(min_el_size), min_dt);
});
mesh.getCommunicator().allReduce(min_dt, SynchronizerOperation::_min);
AKANTU_DEBUG_OUT();
return min_dt;
}
/* -------------------------------------------------------------------------- */
void DiffusionModel::setTimeStep(Real time_step, const ID & solver_id) {
Model::setTimeStep(time_step, solver_id);
this->mesh.getDumper(id).setTimeStep(time_step);
}
/* -------------------------------------------------------------------------- */
void DiffusionModel::assembleCapacityMatrixLumped() {
AKANTU_DEBUG_IN();
if (!this->getDOFManager().hasLumpedMatrix("M")) {
this->getDOFManager().getNewLumpedMatrix("M");
}
this->getDOFManager().zeroLumpedMatrix("M");
for (auto && ghost_type : ghost_types) {
auto & fem = getFEEngineClass<FEEngineType>();
diffusion::details::ComputeRhoFunctor compute_rho(*this);
for (auto && type :
mesh.elementTypes(spatial_dimension, ghost_type, _ek_regular)) {
fem.assembleFieldLumped(compute_rho, "M", dof_name, this->getDOFManager(),
type, ghost_type);
}
}
need_to_reassemble_capacity_lumped = false;
AKANTU_DEBUG_OUT();
}
/* -------------------------------------------------------------------------- */
void DiffusionModel::assembleCapacityMatrix() {
AKANTU_DEBUG_IN();
auto ghost_type = _not_ghost;
this->getDOFManager().zeroMatrix("M");
auto & fem = getFEEngineClass<FEEngineType>();
diffusion::details::ComputeRhoFunctor rho_functor(*this);
for (auto && type :
mesh.elementTypes(spatial_dimension, ghost_type, _ek_regular)) {
fem.assembleFieldMatrix(rho_functor, "M", dof_name, this->getDOFManager(),
type, ghost_type);
}
need_to_reassemble_capacity = false;
AKANTU_DEBUG_OUT();
}
/* -------------------------------------------------------------------------- */
auto DiffusionModel::computeThermalEnergyByNode() -> Real {
AKANTU_DEBUG_IN();
auto time_step = this->getTimeStep();
Real ethermal = 0.;
for (auto && [n, flow] :
enumerate(make_view(*internal_flow, internal_flow->getNbComponent()))) {
Real E = 0.;
bool count_node = mesh.isLocalOrMasterNode(n);
if (count_node) {
E += (flow.array() * time_step).sum();
}
ethermal += E;
}
mesh.getCommunicator().allReduce(ethermal, SynchronizerOperation::_sum);
AKANTU_DEBUG_OUT();
return ethermal;
}
/* -------------------------------------------------------------------------- */
auto DiffusionModel::getEnergy(const ID & energy_id) -> Real {
AKANTU_DEBUG_IN();
Real energy = 0;
for_each_constitutive_law([&energy, &energy_id](auto && constitutive_law) {
energy += constitutive_law.getEnergy(energy_id);
});
// reduction sum over all processors
mesh.getCommunicator().allReduce(energy, SynchronizerOperation::_sum);
AKANTU_DEBUG_OUT();
return energy;
}
/* -------------------------------------------------------------------------- */
auto DiffusionModel::getEnergy(const ID & energy_id, const Element & element)
-> Real {
auto constitutive_law_element = element;
constitutive_law_element.element =
this->getConstitutiveLawLocalNumbering()(element);
return this->getConstitutiveLaw(element).getEnergy(energy_id,
constitutive_law_element);
}
/* -------------------------------------------------------------------------- */
std::shared_ptr<dumpers::Field>
DiffusionModel::createNodalFieldBool(const std::string & field_name,
const std::string & group_name,
bool /*padding_flag*/) {
std::map<std::string, Array<bool> *> uint_nodal_fields;
uint_nodal_fields["blocked_dofs"] = blocked_dofs.get();
auto field = mesh.createNodalField(uint_nodal_fields[field_name], group_name);
return field;
}
/* -------------------------------------------------------------------------- */
std::shared_ptr<dumpers::Field> DiffusionModel::createNodalFieldReal(
const std::string & field_name, const std::string & group_name,
__attribute__((unused)) bool padding_flag) {
if (field_name == "capacity_lumped") {
AKANTU_EXCEPTION(
"Capacity lumped is a nodal field now stored in the DOF manager."
"Therefore it cannot be used by a dumper anymore");
}
std::map<std::string, Array<Real> *> real_nodal_fields;
real_nodal_fields[dof_name] = diffusion.get();
real_nodal_fields[dof_name + "_rate"] = diffusion_rate.get();
real_nodal_fields["external_flow"] = external_flow.get();
real_nodal_fields["internal_flow"] = internal_flow.get();
real_nodal_fields["increment"] = increment.get();
if (auto it = real_nodal_fields.find(field_name);
it != real_nodal_fields.end()) {
return mesh.createNodalField(it->second, group_name);
}
return nullptr;
}
/* -------------------------------------------------------------------------- */
std::shared_ptr<dumpers::Field> DiffusionModel::createElementalField(
const std::string & field_name, const std::string & group_name,
bool /*padding_flag*/, Int /*spatial_dimension*/,
ElementKind element_kind) {
std::shared_ptr<dumpers::Field> field;
if (field_name == "partitions") {
field = mesh.createElementalField<Int, dumpers::ElementPartitionField>(
mesh.getConnectivities(), group_name, this->spatial_dimension,
element_kind);
}
bool is_internal = this->isInternal(field_name, element_kind);
if (is_internal) {
auto nb_data_per_elem =
this->getInternalDataPerElem(field_name, element_kind);
auto & internal_flat = this->flattenInternal(field_name, element_kind);
field = mesh.createElementalField<Real, dumpers::InternalMaterialField>(
internal_flat, group_name, spatial_dimension, element_kind,
nb_data_per_elem);
// homogenize the field
auto foo = dumpers::HomogenizerProxy::createHomogenizer(*field);
field =
dumpers::FieldComputeProxy::createFieldCompute(field, std::move(foo));
};
return field;
}
/* -------------------------------------------------------------------------- */
Int DiffusionModel::getNbData(const Array<Idx> & indexes,
const SynchronizationTag & tag) const {
AKANTU_DEBUG_IN();
Int size = 0;
auto nb_nodes = indexes.size();
switch (tag) {
case SynchronizationTag::_diffusion: {
size += nb_nodes * Int(sizeof(Real));
break;
}
default: {
AKANTU_ERROR("Unknown ghost synchronization tag : " << tag);
}
}
AKANTU_DEBUG_OUT();
return size;
}
/* -------------------------------------------------------------------------- */
void DiffusionModel::packData(CommunicationBuffer & buffer,
const Array<Idx> & indexes,
const SynchronizationTag & tag) const {
AKANTU_DEBUG_IN();
for (auto index : indexes) {
switch (tag) {
case SynchronizationTag::_diffusion: {
buffer << (*diffusion)(index);
break;
}
default: {
AKANTU_ERROR("Unknown ghost synchronization tag : " << tag);
}
}
}
AKANTU_DEBUG_OUT();
}
/* -------------------------------------------------------------------------- */
void DiffusionModel::unpackData(CommunicationBuffer & buffer,
const Array<Idx> & indexes,
const SynchronizationTag & tag) {
AKANTU_DEBUG_IN();
for (auto index : indexes) {
switch (tag) {
case SynchronizationTag::_diffusion: {
buffer >> (*diffusion)(index);
break;
}
default: {
AKANTU_ERROR("Unknown ghost synchronization tag : " << tag);
}
}
}
AKANTU_DEBUG_OUT();
}
/* -------------------------------------------------------------------------- */
Int DiffusionModel::getNbData(const Array<Element> & elements,
const SynchronizationTag & tag) const {
AKANTU_DEBUG_IN();
Int size = 0;
auto nb_nodes_per_element = 0;
for (auto && el : elements) {
nb_nodes_per_element += Mesh::getNbNodesPerElement(el.type);
}
if (tag == SynchronizationTag::_diffusion) {
size += nb_nodes_per_element * Int(sizeof(Real)); // temperature
}
size += Parent::getNbData(elements, tag);
AKANTU_DEBUG_OUT();
return size;
}
/* -------------------------------------------------------------------------- */
void DiffusionModel::packData(CommunicationBuffer & buffer,
const Array<Element> & elements,
const SynchronizationTag & tag) const {
if (tag == SynchronizationTag::_diffusion) {
packNodalDataHelper(*diffusion, buffer, elements, mesh);
}
Parent::packData(buffer, elements, tag);
}
/* -------------------------------------------------------------------------- */
void DiffusionModel::unpackData(CommunicationBuffer & buffer,
const Array<Element> & elements,
const SynchronizationTag & tag) {
if (tag == SynchronizationTag::_diffusion) {
unpackNodalDataHelper(*diffusion, buffer, elements, mesh);
}
Parent::unpackData(buffer, elements, tag);
}
/* -------------------------------------------------------------------------- */
} // namespace akantu
c4science · Help