2. Key standards and guidelines

A crash simulation credibility plan does not exist in a vacuum. Every acceptance threshold you write must be traceable to a recognized standard, a regulatory protocol, or a documented internal procedure. The four reference families most frequently cited in automotive crash CAE are ASME V&V 10, ASME V&V 20, ISO 26262, and the NHTSA/IIHS test protocols. They answer different questions, and confusing their scopes is one of the most common sources of weak acceptance criteria.

The essential distinction is this: ASME V&V 10 governs how you build and defend a computational model; ASME V&V 20 governs how you quantify and judge the numerical uncertainty of a simulation result against experimental data; ISO 26262 governs the functional-safety argument that determines how much confidence a simulation must provide; and the NHTSA/IIHS protocols define the physical load cases, boundary conditions, and pass/fail metrics that your model must reproduce. A credible plan links all four: the protocol defines the scenario, V&V 10 defines the model development process, V&V 20 defines the quantitative acceptance test, and ISO 26262 defines the required confidence level and documentation rigor.

ASME V&V 10: the model credibility framework

ASME V&V 10 (Standard for Verification and Validation in Computational Solid Mechanics) is a process standard. It does not give you a number; it tells you what evidence you must assemble before any number is meaningful. The standard separates verification — is the code solving the equations correctly? — from validation — does the model represent the physical reality for the intended use? For crash work, verification includes mesh convergence studies, energy balance checks (hourglass energy, added mass, internal energy versus initial kinetic energy), and unit/coordinate-system audits. Validation is the comparison against physical crash tests.

V&V 10 also introduces the concept of validation hierarchy: unit-level tests (a single material coupon or a spot weld), subsystem-level tests (a bumper beam or a B-pillar sub-assembly), and full-system tests (the complete vehicle barrier impact). A credible plan does not jump straight to full-vehicle correlation. It builds evidence bottom-up, so that when a full-vehicle result deviates, you can localize whether the error came from a material card, a joint model, or a system-level boundary condition. In practice, a plan that cites V&V 10 should explicitly list which phenomena are validated at which level, and which are only assessed qualitatively.

ASME V&V 20: quantifying the acceptance test

ASME V&V 20 (Standard for Verification and Validation in Computational Fluid Dynamics and Heat Transfer) is often borrowed for crash work because it provides the quantitative machinery that V&V 10 deliberately leaves open. Its central idea is the validation comparison error:

E = S − D

where:
  S = simulated value of a validation variable (e.g., peak B-pillar intrusion)
  D = experimental value of the same variable
  E = comparison error

Acceptance criterion:  |E| ≤ U_val
  U_val = combined validation uncertainty
        = sqrt(U_num² + U_input² + U_D²)

  U_num   = numerical uncertainty (discretization, convergence)
  U_input = uncertainty from input parameters (material, friction, strain rate)
  U_D     = experimental measurement uncertainty

The practical consequence is important: a simulation is not "wrong" merely because it differs from the test. It is unacceptable only when the difference exceeds the combined uncertainty of both the simulation and the experiment. This forces the plan to document experimental uncertainty (sensor accuracy, test-to-test scatter) alongside numerical uncertainty. A common failure mode is to set a threshold such as "peak force within 10%" without ever establishing whether 10% is larger or smaller than U_val. If U_val is 15%, a 10% threshold is unjustifiably strict; if U_val is 4%, it is too loose.

ISO 26262: how much confidence is required

ISO 26262 is a functional-safety standard for road vehicles, not a simulation standard. Its relevance to a credibility plan is indirect but decisive: it determines the rigor that the simulation evidence must carry. The Automotive Safety Integrity Level (ASIL A–D) assigned to a safety goal sets the required confidence in the argument that a system meets its safety requirements. When simulation is used as evidence in that argument — for example, to demonstrate occupant protection performance or structural integrity of a battery enclosure — the plan must show that the model is fit for that specific purpose and that its limitations are documented.

In practice, this means the credibility plan should record the ASIL context of each simulation, the intended use of the result, and the consequence of the model being wrong. A high-ASIL use case justifies a broader validation matrix, more physical tests, and tighter U_val; a low-ASIL or concept-stage study may accept a coarser correlation. ISO 26262 also drives traceability: every acceptance threshold should be linked to the requirement it supports, so an auditor can follow the chain from safety goal to load case to model to validation evidence.

NHTSA and IIHS protocols: the physical reference

The regulatory and consumer-test protocols define what the simulation must reproduce. NHTSA's FMVSS standards specify the mandatory load cases — for example FMVSS 208 (occupant crash protection, frontal and side), FMVSS 214 (side impact), and FMVSS 301 (fuel system integrity, rear impact) — including barrier type, impact speed, impact angle, and the injury or integrity criteria that constitute pass/fail. IIHS protocols add consumer-facing tests such as the moderate overlap frontal, small overlap frontal, and side impact, with their own barrier geometry, speed, and rating metrics.

For acceptance purposes, the protocol supplies the validation variables: the specific quantities you compare. Typical choices include barrier force–displacement curves, B-pillar or door intrusion at defined heights, occupant compartment acceleration pulses, HIC and chest deflection from the dummy, and post-test structural deformation patterns. The plan should name each variable, its measurement location, its experimental uncertainty, and the acceptance threshold derived from U_val. It should also state which protocol revision is being used, because barrier face geometry and speed tolerances change between revisions and a model validated against one revision is not automatically valid for another.

Bringing the four together in a plan

A workable structure is to treat the protocol as the scenario definition, V&V 10 as the evidence-building process, V&V 20 as the quantitative acceptance test, and ISO 26262 as the confidence and traceability requirement. The table below shows how a single acceptance entry can be assembled from these sources.

The example values above are illustrative and assumed for teaching purposes, not measured results. The point is the logic: the decision is not "9 mm is close enough" but "9 mm exceeds the combined uncertainty, so the model is not yet validated for this variable and the discrepancy must be investigated." That investigation would typically examine the dominant contributors to U_input — in a side impact, often the door trim material model, the barrier honeycomb crush behavior, and the friction at the barrier–door interface — and may require an additional subsystem test to isolate the cause.

The practical outcome of this chapter is a mapping discipline: for every acceptance threshold in the credibility plan, record the protocol that defines the scenario, the V&V 10 validation level and evidence, the V&V 20 uncertainty budget and comparison criterion, and the ISO 26262 confidence context. When those four are present and consistent, the acceptance decision becomes defensible rather than a matter of engineering opinion.