LS-DYNA vs Abaqus/Explicit: Explicit Dynamics Solver Comparison for Crash and Impact

LS-DYNA vs Abaqus/Explicit: Explicit Dynamics Solver Comparison for Crash and Impact

I've used both LS-DYNA and Abaqus/Explicit enough to have real opinions about them. LS-DYNA was my first explicit solver, so I'm probably a bit biased, but I've come to appreciate Abaqus/Explicit for different reasons. If you're trying to decide which one to use, let me break down the differences I've actually experienced.

History and Market

| | LS-DYNA | Abaqus/Explicit | |---|---|---| | Developer | LSTC (Livermore Software, now ANSYS) | Dassault Systèmes SIMULIA | | Origin | Lawrence Livermore National Lab (1980s) | Hibbitt, Karlsson & Sorensen (1980s) | | Primary industry | Automotive crash, defense, manufacturing | Aerospace, defense, general nonlinear | | Market share (crash) | ~70% (automotive industry standard) | ~20% | | Market share (forming) | ~60% | ~25% |

Solver Architecture

| Feature | LS-DYNA | Abaqus/Explicit | |---------|---------|-----------------| | Time integration | Central difference | Central difference | | Mass matrix | Lumped (diagonal) | Lumped (diagonal) | | Element library | Extensive (200+ element types) | Comprehensive | | Material models | 250+ material models | 50+ built-in + UMAT/VUMAT | | Contact | 30+ contact types | General contact (auto-detect) | | Parallel | SMP + MPP (domain decomposition) | MPI + OpenMP | | GPU | Yes (CUDA) | Yes (CUDA) | | Keyword/Input | Text keyword file (.k) | Input file (.inp) or CAE |

Element Formulation

| Element | LS-DYNA | Abaqus/Explicit | |---------|---------|-----------------| | Shell (default) | Belytschko-Tsay (ELFORM 2) | S4R (similar to B-T) | | Shell (enhanced) | ELFORM 16 (fully integrated) | S4 (enhanced strain) | | Solid (default) | ELFORM 1 (1-point hex) | C3D8R (reduced integration) | | Solid (improved) | ELFORM 2 (S/R full) | C3D8 (full integration) | | Tetrahedron | ELFORM 10, 13 | C3D10M (modified, good) | | Beam | ELFORM 1 (Hughes-Liu) | B31 (similar) | | Truss | ELFORM 3 | T3D2 | | Hourglass | Multiple types (IHQ 1-6) | Enhanced (default) |

Key difference: LS-DYNA offers more element formulation choices (ELFORM parameter), allowing fine-tuning for specific problems. Abaqus has fewer but well-validated options.

Material Models

| Material | LS-DYNA | Abaqus/Explicit | |----------|---------|-----------------| | Elastic | MAT_001 | Built-in | | Plastic kinematic | MAT_003 | Built-in | | Johnson-Cook | MAT_015 | Built-in | | Piecewise linear | MAT_024 | Built-in (tabular) | | Concrete damage | MAT_159 (RHT) | Concrete damaged plasticity | | Composite (Hashin) | MAT_022, MAT_054/055 | Built-in (Hashin) | | Rubber (hyperelastic) | MAT_027, MAT_077 | Ogden, Mooney-Rivlin | | Foam | MAT_057, MAT_083 | Built-in (crushable foam) | | Strain rate | Cowper-Symonds, Johnson-Cook | Johnson-Cook, rate-dependent | | Failure/erosion | *MAT_ADD_EROSION | Damage evolution + deletion | | User material | User-defined (Fortran) | VUMAT (Fortran) | | Total material models | 250+ | 50+ built-in |

Key difference: LS-DYNA has significantly more built-in material models (250+ vs 50+). For special materials (foams, honeycomb, soil, concrete), LS-DYNA likely has a pre-built model. Abaqus relies more on VUMAT user subroutines.

Contact

| Feature | LS-DYNA | Abaqus/Explicit | |---------|---------|-----------------| | General contact | *CONTACT_AUTOMATIC_SINGLE_SURFACE | General Contact (auto-detect) | | Surface-to-surface | *CONTACT_AUTOMATIC_SURFACE_TO_SURFACE | Surface-to-surface | | Node-to-surface | *CONTACT_AUTOMATIC_NODES_TO_SURFACE | Node-to-surface | | Eroding contact | *CONTACT_ERODING_SURFACE_TO_SURFACE | Built-in (with element deletion) | | Forming contact | *CONTACT_FORMING_SURFACE_TO_SURFACE | Surface-to-surface (forming) | | Self-contact | Automatic (in single surface) | Automatic (in general contact) | | Friction | Static + dynamic (FS, FD) | Coulomb (μ) | | Contact damping | DC parameter | Contact damping | | Penalty scaling | SFS, SFM | Scale factor |

Key difference: LS-DYNA offers more contact types (30+), including specialized types for forming, erosion, and tied interfaces. Abaqus general contact is simpler to set up (auto-detect) but less customizable.

Parallel Performance

| Feature | LS-DYNA | Abaqus/Explicit | |---------|---------|-----------------| | SMP | Up to 16 cores | Up to 8-16 cores | | MPP | Up to 1000+ cores | Up to 1000+ cores (MPI) | | Domain decomposition | Automatic (recursive) | Automatic | | GPU acceleration | Yes (single GPU, multi-GPU) | Yes (single GPU) | | Scaling efficiency | Excellent (MPP) | Very good (MPI) | | Large models | 10M+ elements (MPP) | 10M+ elements (MPI) |

Key difference: LS-DYNA has slightly better MPP scaling for very large models (10M+ elements), but both are comparable for most applications.

Automotive Crash

| Feature | LS-DYNA | Abaqus/Explicit | |---------|---------|-----------------| | Industry adoption | ~70% of automotive OEMs | ~20% of automotive OEMs | | Dummy models | LSTC dummy models (free) | SIMULIA dummy models | | Barrier models | Standard (NCAC, free) | Standard | | Airbag | *AIRBAG_PARTICLE (advanced) | Built-in (fluid cavity) | | Spot weld | *CONSTRAINED_SPOTWELD | Fastener (connector) | | Seatbelt | *MAT_SEATBELT | Built-in (seatbelt element) | | Pedestrian impact | Standard | Standard | | Regulatory compliance | FMVSS, EuroNCAP (validated) | FMVSS, EuroNCAP |

Key difference: LS-DYNA is the automotive crash industry standard. Most OEMs (Toyota, Honda, Ford, GM, BMW, VW) use LS-DYNA for crash. Dummy and barrier models are widely available and validated.

Metal Forming

| Feature | LS-DYNA | Abaqus/Explicit | |---------|---------|-----------------| | Adaptive meshing | *CONTROL_ADAPTIVE (refinement) | ALE adaptive meshing | | Forming contact | *CONTACT_FORMING (specialized) | Surface-to-surface | | Springback (implicit) | *CONTROL_IMPLICIT (switch) | Abaqus/Standard import | | FLD | Built-in FLD evaluation | Custom (field output) | | Industry adoption | ~60% | ~25% | | Tool compensation | *CONTROL_SPRINGBACK_COMPENSATION | Custom workflow |

Key difference: LS-DYNA has better adaptive meshing (element splitting) and built-in FLD evaluation. Abaqus uses ALE for adaptive meshing (different approach). LS-DYNA's springback transition to implicit is smoother (same solver).

User Experience

| Feature | LS-DYNA | Abaqus/Explicit | |---------|---------|-----------------| | Pre-processor | LS-PrePost (free), HyperMesh, ANSA | Abaqus/CAE, HyperMesh, ANSA | | Post-processor | LS-PrePost (free) | Abaqus/CAE, LS-PrePost | | Input format | Keyword file (.k, text) | Input file (.inp, text) | | GUI workflow | Limited (LS-PrePost basic) | Full GUI (Abaqus/CAE) | | Documentation | Manual + dynasupport.com | Manual (comprehensive) | | Community | Large (dynasupport, forums) | Large (SIMULIA community) | | Training | LSTC training, YouTube | SIMULIA training, courses |

Key difference: Abaqus/CAE provides a better integrated GUI workflow (model setup, mesh, submit, post-process in one tool). LS-DYNA workflow is more text-file oriented, with external pre/post-processors.

Pricing

| | LS-DYNA | Abaqus/Explicit | |---|---|---| | License type | Annual lease | Annual lease (in Abaqus suite) | | Single solver | ~$15,000-25,000/year | ~$20,000-30,000/year (with Standard) | | MPP add-on | Additional cost | Included | | Dummy models | Free (LSTC) | Additional cost (SIMULIA) | | Total (with dummies) | ~$20,000-30,000/year | ~$30,000-40,000/year |

Key difference: LS-DYNA is generally cheaper, especially with free dummy models. Abaqus includes both Standard and Explicit in one license (value if both are needed).

When to Choose LS-DYNA

• You work in automotive crash (industry standard)
• You need many specialized material models (250+)
• You need specialized contact types (forming, eroding)
• You work with metal forming (adaptive meshing, FLD)
• You need blast and ballistic simulation
• Your company/partners use LS-DYNA (file compatibility)
• You need free dummy and barrier models
• You prefer keyword file control (full parameter access)
• You need MPP for very large models

When to Choose Abaqus/Explicit

• You need both implicit and explicit in one workflow (smooth transition)
• You need general contact (auto-detect, simpler setup)
• You need VUMAT user subroutines (larger community)
• You need Python scripting for automation
• You work in aerospace or defense (SIMULIA adoption)
• You prefer GUI workflow (Abaqus/CAE)
• Your company already uses SIMULIA products
• You need submodeling (global-local analysis)
• You need CEL (coupled Eulerian-Lagrangian) for fluid-structure

Can You Use Both?

Some companies use both:

• LS-DYNA for crash and forming (industry standard)
• Abaqus/Explicit for nonlinear drop test and impact (when Abaqus/Standard is also used)

The keyword/input files are not compatible. Models must be rebuilt in the other tool.

My Take

Here's how I decide: if I'm doing automotive crash or metal forming, LS-DYNA is the obvious choice — it's what the industry uses, the dummy models are free, and the material library is unmatched. If I'm already using Abaqus/Standard for a project and need to run an explicit drop test or impact analysis, Abaqus/Explicit makes more sense because I can import the model directly. Switching between the two isn't practical — the file formats are completely different. So pick the one that fits your industry and stick with it.