What is explicit dynamics?

Explicit dynamics is structural simulation for events that happen fast and violently: crash, impact, drop test, blast, ballistics, forming and stamping. The physics is still solid mechanics, but the numerical method is fundamentally different from the FEA used for static and slow-dynamic problems.

The distinguishing feature is time integration. An implicit solver takes a large time step and solves a system of equations to find a state that satisfies equilibrium at the end of it. An explicit solver computes acceleration directly from the forces it already has, steps forward by a tiny increment, and repeats — with no equation system to solve and no iteration at all.

Explicit integration is only stable if the time step is smaller than the time a stress wave takes to cross the smallest element in the model. That is the Courant condition, and it is not a tuning parameter — exceed it and the solution does not degrade, it explodes.

For a steel part meshed at a few millimetres, that step is on the order of a microsecond. Simulating 100 milliseconds of a drop event therefore takes on the order of a hundred thousand steps. Each step is very cheap, because there is no matrix to factorise, but there are an enormous number of them.

The critical consequence for practice: one small element sets the cost of the entire model. A single sliver element somewhere unimportant can halve your time step and double your run time. Mesh quality in explicit work is not about accuracy so much as about affordability, and checking the smallest element before launching a run is a habit worth forming.

• Use explicit when the event is measured in milliseconds, when contact is complex and changing constantly, when material is failing and elements are being deleted, or when an implicit solver simply will not converge because the problem is too nonlinear.
• Use implicit for static loads, slow processes, and any problem where the answer is an equilibrium state. Simulating a one-second event explicitly is usually a mistake — a million time steps to answer a question a single implicit solve would have settled.

A related trap is mass scaling, which artificially increases density in the smallest elements to raise the stable time step. It is standard practice and it works, but it adds inertia that is not real. Used lightly it saves hours; used heavily it quietly changes the physics, and the added kinetic energy is something to watch rather than assume away.

In an impact simulation, the solver spends most of its effort deciding what is touching what. Surfaces come into contact, slide, separate and self-contact as a structure folds onto itself. Contact definition, more than mesh density, is what most often decides whether a crash model is credible.

Materials are equally decisive, and they are strain-rate dependent: most metals and nearly all polymers are stronger and behave differently when deformed quickly. A material model calibrated from a slow tensile test will misrepresent an impact. Failure criteria — when an element has taken enough damage to be deleted — determine whether your model tears where the real part tears, and they are notoriously mesh-sensitive.

This is why explicit work leans harder on validation than almost any other kind of simulation. The standard sanity check is energy balance: total energy should be conserved, hourglass energy from under-integrated elements should stay small relative to internal energy, and added mass from scaling should stay modest. If the energy plot is wrong, nothing else in the result is worth reading.

OpenRadioss is the significant development here: Altair open-sourced the Radioss explicit solver under AGPL-3.0 in 2022, making a genuine, industrially validated crash code openly available for the first time. It is used for automotive crash, drop test and impact, and it is a different class of tool from a general-purpose FEA code with an explicit option bolted on.

Several general-purpose open-source FEA codes also carry an explicit capability. Those are adequate for drop test and simple impact, and honestly none of them is a crash code — worth knowing before choosing one for the job.