Explicit vs. Implicit Solvers: Accuracy vs. Efficiency in Large‑Deformation Crash Simulations

Direct answer

For a short-duration drop/crash event dominated by high-frequency contact and large deformation, the explicit solver is the right choice for peak contact force prediction. Implicit integration is not more accurate here; in crash-like regimes it frequently fails to converge or smears contact peaks, so its larger steps do not translate into useful accuracy. Use implicit only for the quasi-static phases around the impact.

1. Peak contact forces in a 50 mm drop test

Explicit central-difference integration resolves the contact event at its natural timescale: the stability-limited step (element length divided by wave speed) is orders of magnitude smaller than the contact duration, so integration error is negligible and force peaks are captured faithfully, provided the mesh, hourglass control, and contact definitions are sound. Implicit Newmark-type schemes allow larger steps, but each step requires a global solve with Newton iterations that often stall or diverge under severe contact, buckling, or material failure — and even when they converge, a step that spans many contact oscillations numerically damps the peak. A completed implicit run is not evidence of a correct peak.

2. Time-step criteria and damping

For explicit, the step must stay below the Courant limit of the smallest element; accuracy work then focuses on mesh quality and energy bookkeeping rather than step size. Mass scaling can raise the stable step, but verify its cost: keep added mass to a small fraction of critical parts and confirm hourglass energy is a small fraction of internal energy with a smooth total-energy history. If you still run implicit for comparison, limit the step so that many steps fall within the contact duration (dozens, not single digits) and prefer an undamped or minimally damped integrator — algorithmic damping parameters directly attenuate the peaks you are trying to measure. Exact parameter names and defaults differ between codes and versions (LS-DYNA, Abaqus, Radioss), so confirm against your vendor's current theory manual rather than copying settings across solvers.

3. When hybrid explicit–implicit partitioning pays off

A hybrid workflow becomes advantageous when the model mixes timescales: use implicit for gravity preload, bolt pretension, or slow positioning of rigid tooling, then switch to explicit for the impact. This avoids burning millions of explicit steps on quasi-static phases while keeping the violent transient in the robust integrator. For mixed rigid–deformable models, it is most attractive when the deformable region is locally refined (driving the explicit Courant limit down) while the rest of the structure is stiff or rigid — partitioning lets the fine region dictate cost only where it matters. If the whole model is uniformly fine and the entire event is transient, pure explicit is simpler and usually cheaper.

Verification before committing

Run a reduced representative model both ways (and explicit with/without mass scaling) and compare deformed shape, peak force, and energy histories. Agreement within a few percent on peak force, with clean energy balance, is the practical gate — not solver completion.