CB-Geo MPM

An MPM code for large-scale geomechanics simulations

Krishna Kumar, krishnak@utexas.edu
University of Texas at Austin

Kenichi Soga, Shyamini Kularathna, Jeffrey Salmond, Christopher Wilkes, Ezra Tjung, Tianchi Zhao, Thiago Araujo and Wentao Xu.

12th Annual MPM Workshop, Virginia Tech, USA
10th September 2019

CB-Geo Computational Geomechanics

CB-Geo MPM Open source code: https://github.com/cb-geo/mpm.
Documentation and validation https://cb-geo.github.io/mpm-doc/#/.
Benchmark problems: https://github.com/cb-geo/mpm-benchmarks.

View the CB-Geo website for more information and software tools

Large-scale geomechanics simulations

Large-scale multiphase multi-physics problems
Develop a modular system

Algorithms: Explicit, implicit, projection method
Mapping functions: linear, GIMP, CDPI

Complex geometries and boundary conditions
Distributed and shared-memory parallel

Kaikoura landslide simulation (Thiago Araujo., 2019)

HPC MPM code

Generic Templatised C++14
2D/3D MPM Code
Generalise Interpolation Material Point
Distributed MPI
Intel TBB parallelisation
Isoparametric elements
HDF5 data stores
Jupyter Notebooks integration
Material models:

Linear elastic
Mohr coulomb
Bingham fluid
Newtonian

Code overview

Factory design pattern

Factory implementation

              
//! A helper class to register a factory function
//! \tparam Targs variadic template arguments
template <typename Tbaseclass, typename Tderivedclass, typename... Targs>
class Register {
  public:
    //! Register with a given key
    explicit Register(const std::string& key) {
       // register the class factory function
       Factory<Tbaseclass, Targs...>::instance()
            ->template register_factory<Tderivedclass>(key);
    }
};

// Quadrilateral 4-noded element
static Register<mpm::Element<2>, mpm::QuadrilateralElement<2, 4>>
    quad4("ED2Q4");
// Quadrilateral 4-node-base GIMP element
static Register<mpm::Element<2>, mpm::QuadrilateralGIMPElement<2, 16>>
    quad_gimp16("ED2Q16G");

Iterate over particles / nodes/ cells functions

              
// Apply a given functor to all particles making use of tbb::parallel_for_each
// \tparam Toper          Type of the particle operation
// \param[in|out] oper    Specific operation applied to all particles
template <unsigned Tdim>
template <typename Toper>
void mpm::Mesh<Tdim>::iterate_over_particles(Toper oper) {
  tbb::parallel_for_each(particles_.cbegin(), particles_.cend(), oper);
}

// Iterate over particles to compute mass
// Compute mass
mesh_->iterate_over_particles(std::bind(
      &mpm::ParticleBase<Tdim>::compute_mass, std::placeholders::_1, phase));