2. LS-DYNA load case taxonomy

A DFMEA is only as good as the load cases it reasons about. If the analysis team models a bracket as a static linear case while the field failure mode is a high-strain-rate impact, the failure mode will be scored against the wrong physics, the wrong severity, and the wrong detection control. This chapter builds the taxonomy that connects LS-DYNA solver choice and load-case construction to DFMEA line items, so that each failure mode is evaluated against a load case that actually reproduces its physics.

Solver formulation first: what implicit and explicit actually buy you

LS-DYNA offers two time-integration families, and the choice is a physics decision before it is a software decision. The explicit solver advances the solution with a central-difference scheme and a critical time step governed by the smallest element and the wave speed:

dt_crit ≈ L_e / c,   c = sqrt(E / ρ)   (1D bar approximation)
*CONTROL_TIMESTEP
$#  dtinit    tssfac    isdo    tslimn   dt2ms    lctm     erode   ms1st
       0.0       0.9       0       0.0    0.0        0        0        0

For a steel part with E = 210 GPa and ρ = 7.85e-9 t/mm³, c ≈ 5.17 mm/ms. A 2 mm characteristic element gives dt ≈ 0.39 µs, so a 50 ms event needs roughly 130,000 cycles — cheap per cycle, expensive in total, but unconditionally stable for contact and large deformation. The implicit solver solves a global stiffness system each increment and is unconditionally stable for linear problems, making it the right tool for static stiffness, modal, and long-duration quasi-static response where inertia is negligible.

The practical rule: explicit when the event is short, highly nonlinear, contact-dominated, or rate-sensitive; implicit when the response is smooth, inertia-free, and you need static equilibrium or eigenvalues. Mixed problems — a quasi-static crush that must be run explicitly — are handled by time scaling and energy-balance checks, not by switching solvers mid-run.

The load-case families and their DFMEA relevance

The taxonomy below is organized by the physical regime each family reproduces, because that regime determines which failure modes are even observable.

Two distinctions matter most for DFMEA scoring. First, impact and crash are not the same case: impact is a localized, often small-mass event (dropped tool, stone strike, foreign-object damage) where local strain rate and contact pressure dominate; crash is a system-level event where global collapse mode, intrusion, and energy partitioning dominate. A DFMEA that lists "impact" as a single line item will under-score the crash severity and over-score the impact occurrence. Second, durability is a different failure mechanism, not a longer dynamic run — it requires cycle counting and a fatigue model, and its detection controls are inspection-based rather than simulation-based.

Quasi-static cases: making explicit behave like implicit

When contact or large deformation forces an explicit run for a quasi-static problem, the standard technique is to scale the load duration so that kinetic energy stays a small fraction of internal energy. A reproducible example: a 200 mm steel bracket loaded to 15 kN over a physical 2 s. Running explicitly at real time would need ~5 million cycles; instead, compress the ramp to 20 ms and monitor energy:

*CONTROL_ENERGY
$#  hgen     rwen    slnten   rylen
       2        2        2        2
*DATABASE_GLSTAT
$#    dt
     0.0001
*LOAD_RIGID_BODY
$#  lcidd    lciddof    lciddir
         1         1         3

The acceptance criterion is the ratio of kinetic to internal energy in glstat: keep it below roughly 5% throughout the ramp, and confirm that the reaction force at the support matches the applied load within a few percent at the end of the ramp. If kinetic energy spikes, the ramp is too fast — lengthen it or add mass scaling with a documented target time step. This is the check that separates a defensible quasi-static result from a dynamic artifact, and it belongs in the DFMEA detection column as a simulation-verification control.

Impact and crash: rate effects and contact definition

For impact cases, the material card must carry rate sensitivity. A common error is to run a 10 m/s impact with a rate-insensitive *MAT_024 and report a strain that is 30–40% higher than a rate-sensitive model would give. Use *MAT_024 with the LCSS curve referencing a *DEFINE_CURVE of effective stress versus effective plastic strain rate, and verify that the strain rate in the critical element falls inside the calibrated range. Contact definition is equally decisive: *CONTACT_AUTOMATIC_SURFACE_TO_SURFACE with a sensible SOFT option and a friction coefficient consistent with the interface (0.1–0.3 for lubricated steel, 0.6–0.8 for dry). A crash case additionally needs an accelerometer output at the occupant reference point and an energy-absorption check — the internal energy of the crush zone should account for the majority of the initial kinetic energy, with hourglass energy below 5%.

Durability: from stress history to DFMEA occurrence

Durability load cases are built by extracting stress histories from the quasi-static or dynamic runs and feeding them to a fatigue model. In LS-DYNA this is typically a sequence of implicit static solves at load extremes, followed by *DATABASE_ELOUT extraction and an external fatigue tool. The DFMEA-relevant output is not a single stress but a damage ratio at the critical location, compared against the design target (for example, damage < 1.0 at the required cycle count with a safety factor of 2). Occurrence scoring should then reference the duty-cycle exposure — how many cycles the field population actually sees — rather than the simulation result alone.

Mapping load cases to DFMEA line items: a worked pattern

The taxonomy becomes useful when each DFMEA failure mode is explicitly bound to a load case and a verification criterion. Consider a steel mounting bracket with three identified failure modes: (1) plastic deformation under a 15 kN static proof load, (2) fracture under a 10 m/s stone impact, and (3) fatigue cracking at the weld toe over 10⁶ cycles. The mapping is:

• Mode 1 → quasi-static case, implicit or scaled explicit, acceptance: no element exceeds yield at 15 kN, and the KE/IE ratio stays below 5% if explicit.
• Mode 2 → explicit impact case with rate-sensitive *MAT_024, acceptance: no perforation and peak strain below the calibrated failure strain; verify strain rate in the critical element is within the test-calibrated range.
• Mode 3 → static load extremes plus fatigue post-processing, acceptance: damage ratio < 1.0 at 10⁶ cycles with a factor of 2.

Each acceptance criterion then becomes the detection control in the DFMEA, and its rigor determines the detection score. A mode verified only by a rate-insensitive impact run should not receive a strong detection rating, because the model does not reproduce the governing physics. This is the central discipline of the chapter: the load case must be physically faithful to the failure mode before its result can justify a DFMEA score.