KiSSAM is a multiphysics simulation software designed for high-performance and high-fidelity modeling of metal powder bed fusion (PBF) additive manufacturing processes at the mesoscale level.

The current version of KiSSAM supports simulations of both direct metal laser melting (DMLM) and electron beam melting (EBM) of PBF.

With high performance code on CUDA GPU, simulation of a laser track few millimeters long can be finished in hours on a desktop workstation with NVidia GPU.


The core of KiSSAM is the hydrodynamic solver based on Thermal Lattice Boltzmann Method with Volume of Fluid free surface tracking and phase transition between solid and liquid states.

Hydrodynamics includes:
  • surface tension and wetting,

  • Marangoni convection,

  • dumping drag force in mushy zone (-),

  • recoil pressure, mass losses,

  • energy deposition,

  • correct pressure in gas pores (important for keyholing).

Heat transfer is modeled in the melted metal, and in the surrounding area in bulk far from the melt pool.

Energy sources:
  • ray tracing with multiple reflections (for laser beam),

  • Monte Carlo simulation of electron trajectories with elastic and inelastic scattering in metal (for electron beam).

Energy sinks:
  • radiation cooling,

  • convection cooling,

  • evaporation cooling.

Additional physics:
  • models of evaporated metal and gas fluid flow in the gas chamber (in progress),

  • physically bases simulation of the sensors and detectors responses.

To be included in the future versions:
  • grain microstructure formation at the solidification front (the same as CAFE model) (in progress),

  • particle movement,

  • self-consistent liquid-gas modeling,

Process Map

In the simulation, any relevant physical parameters can be adjusted. This way, process maps can be multidimensional.

The parameters that can be configured in kissam, include
  • general parameters of powder stock:

  • material properties,

  • powder particle size distribution;

  • single tracks:

  • track length,

  • scan speed,

  • beam power,

  • beam size,

  • beam shape (to be included in future release),

  • beam type (laser, e-beam),

  • preheating temperature,

  • powder layer thickness;

  • single layer (zig-zag scan strategy), in addition to the parameters for single tracks:

  • hatch spacing,

  • delay between tracks,

  • layer sizes;

  • volume (multilayer) sample:

  • platform step,

  • platform rotation between layers,

  • delay between successive layers,

  • PBF process;

  • miscellaneous:

  • arbitrary scan strategies (e.g. wobbling);

  • additional volumetric samples themes (for future releases?) – important output includes roughness, porosity, shape of the sample):

  • thin bridges,

  • thin walls,

  • corners,

  • overhangs.

Simulation Output

Any aspect of simulation can be output for diagnostic. The key output data include

  • general:

  • effective absorbed energy as a function of time. It can be used as input in models for macroscopic physics;

  • spatial and temporal temperature distribution. It can be used for models of microstructure formation and macroscopic tensions;

  • melt depth map. Can be used for control over lack of fusion;

  • lack of fusion pores can be identified in the geometry output;

  • gas pores that tend to form in keyhole regime can be identified;

  • for single layer simulation:

  • scripts for geometry post-processing can be used to find surface properties, such as roughness;

  • in the volumetric sample simulations:

  • surface properties on the top, sides, downskin can be analyzed.

KiSSAM: DMLM and EBM simulation presets

The software can be used for simulation of any mesoscoping meltpool simulation. In the user interface, several presets are provided.

Single linear tracks:
  • arbitrary track length;

  • typical performance: 0.5 – 1 hr per 1 ms scanning time;

  • OUTPUT: Track morphology (including but not limited to melt depth, width, height); melt pool shape; temperature fields.

Single layers with a zigzag scanning strategy:
  • arbitrary layer area;

  • typical performance:

  • 6 hrs per 1 mm² scanning area (DMLM, beam size ~ 100 um, powder size > 10 um, spatial discretization 3 um);

  • 30 hrs per 10 mm² scanning area (EBM, beam size ~ 800 um, powder size > 40 um, spatial discretization 5 um);

  • OUTPUT: Surface morphology (including but not limited to melt depth, roughness); melt pool shape; temperature fields.

Multilayer samples (in progress):
  • arbitrary sample volume;

  • typical performance* TBD (projected for EBM square pillars with 5x5 mm2 base – 72 hrs per layer);

  • OUTPUT: Porosity; Surface morphology of layers; temperature fields.

Performance estimates are made for a single GeForce RTX 3090 24 GB GPGPU.

User Interface

  • Launch from bash-shell.

  • Standard input / output formats (vtk, vdb etc.):

  • images: png files with geometry cross-sections, 3D rendered simulation view,

  • video: wmv file with a 3D rendered simulation process,

  • volume data: vdb file contains volumetric data of the solid and fluid filling fraction at any time of the simulation process. Can be visualized with third party software,

  • temperature: vtk file of the rectilinear grid type for use in third party software or visualization in Paraview,

  • text files: Various diagnostics are output in text files that can be imported in third party diagnostics software, such as MS Excel, Matlab, etc.

  • library of material parameters that are often used in PBF is provided with the software. The materials are tested and verified in kissam simulations with the used of published experimental data,

  • both low-lever API and a library of presets for basic user-cases is provided,

  • multiple visualization methods,

  • runtime diagnostics for control over simulation flow.