3. Airbag modeling approaches in LS-DYNA

In LS-DYNA, a side airbag is not modeled as a solid membrane that simply pushes the occupant. Instead, the gas inside the bag is represented by one of several dedicated airbag algorithms, and the pressure they generate is applied to the fabric as a distributed load. Choosing the wrong algorithm is one of the most common reasons a side airbag model either runs for hours or produces an unrealistically stiff cushion. This section compares the three principal approaches — Uniform Pressure (UP), Control Volume (CV), and Corpuscular Particle Method (CPM) — and explains when each is appropriate.

Why the gas model matters

A deploying side airbag is a fluid–structure interaction problem. The inflator discharges hot gas at high mass flow rate; the fabric stretches, unfolds, and vents gas through seams and vent holes. The pressure acting on the fabric depends on how much gas is inside, the bag volume, the gas temperature, and the venting behavior. LS-DYNA offers three levels of fidelity for this physics, and they differ mainly in how the gas state is computed and how the pressure is distributed over the fabric.

Uniform Pressure (UP) method

The Uniform Pressure method is the simplest and fastest. The bag is defined as a closed surface, and LS-DYNA computes a single scalar pressure from a user-supplied pressure-versus-time curve or from a simple analytical relation. That pressure is applied uniformly to every element of the bag surface. There is no gas mass, no temperature, and no volume feedback.

Typical keyword usage is minimal:

*AIRBAG_SIMPLE_PRESSURE_VOLUME
$#   sid    sidtyp      rbid      vsca      psca      vini
         1         0         0       1.0       1.0       0.0
$#   lcpr      lcvol
         100         0

Here lcpr points to a load curve of pressure versus time. Because the pressure is prescribed, the bag will not "feel" the occupant: if the occupant compresses the bag, the pressure does not rise. This makes UP suitable only for early-stage packaging studies, for airbags that are already fully inflated before contact, or for cases where the inflator output is well characterized and occupant interaction is not the focus. It is also useful as a debugging tool because it is numerically robust and cheap.

Control Volume (CV) method

The Control Volume method treats the bag as a thermodynamic control volume. LS-DYNA tracks the bag volume from the deformed mesh, the gas mass from the inflator mass flow rate, and the gas temperature from an energy balance. Pressure is then computed from the ideal gas law (or a user-supplied equation of state) and applied uniformly to the bag surface. Venting is handled through *AIRBAG_REFERENCE_GEOMETRY and porous fabric definitions.

A representative setup looks like this:

*AIRBAG_WANG_NEFSKE
$#   sid    sidtyp      rbid      vsca      psca      vini
         1         0         0       1.0       1.0       0.0
$#   lcmt      lcpr      lcvol      lct      lca
         200         0         0       201         0
$#   c23       c24       c25       c26
       0.0       0.0       0.0       0.0

The key inputs are the mass flow rate curve lcmt and the temperature curve lct. Because the pressure now depends on the instantaneous bag volume, the bag becomes load-path sensitive: when the occupant compresses it, volume drops and pressure rises. This is the workhorse method for side airbag occupant protection studies. It is accurate enough for most regulatory and consumer-test load cases, and it is significantly cheaper than CPM.

The main limitation is that CV assumes the gas is well mixed and that pressure equilibrates instantly throughout the bag. For a long, narrow side curtain airbag, or during the very early milliseconds when the gas jet has not yet filled the bag, this assumption is questionable. The pressure is also uniform, so local effects near the inflator nozzle are not captured.

Corpuscular Particle Method (CPM)

CPM represents the gas as a large number of discrete particles. Each particle carries mass, momentum, and energy, and moves through the bag volume, colliding with the fabric and with other particles. Pressure on the fabric is computed from the momentum transfer of particle–wall collisions. This captures gas dynamics that CV cannot: jetting from the inflator, non-uniform pressure distribution, and the finite time required for gas to propagate along a long bag.

CPM is activated through *AIRBAG_PARTICLE:

*AIRBAG_PARTICLE
$#   sid    sidtyp      rbid      vsca      psca      vini
         1         0         0       1.0       1.0       0.0
$#   lcmt      lcpr      lcvol      lct      lca
         200         0         0       201         0
$#   npart     npartm     ...
      200000         0

The particle count npart is the critical numerical parameter. Too few particles produce a noisy, non-physical pressure signal; too many make the run prohibitively expensive. A practical starting point for a side airbag is 100,000–300,000 particles, with a sensitivity check at half and double that count. CPM is the method of choice when the deployment sequence itself is under study — for example, when the bag must unfold through a tight gap between the seat and the door trim, or when the timing of the bag reaching the occupant is critical.

Choosing between the three methods

The decision is driven by the question being asked, not by a universal ranking. The table below summarizes the practical trade-offs.

A common engineering workflow is to start with UP to validate the mesh and contact definitions, switch to CV for the main occupant protection matrix, and use CPM only for the specific load cases where deployment timing or gas jetting is the subject of investigation. This staged approach keeps the overall project schedule realistic.

Verification and common failure symptoms

Regardless of method, verify the gas model before trusting the occupant response. A useful check is to inflate the bag in free space (no occupant, no contact) and plot bag pressure and volume versus time. For CV, the pressure should rise smoothly and plateau as the inflator mass flow ends. For CPM, the pressure should converge as particle count increases; if doubling the particle count changes peak pressure by more than about 5%, the model is under-resolved.

Typical failure symptoms and their causes:

• Pressure spikes at t = 0: the initial bag volume is too small or the mass flow rate curve starts with a non-zero value. Check the reference geometry and the first point of lcmt.
• Bag never inflates: the sid in the airbag keyword does not match the segment set ID, or the segment set normal direction is inverted.
• Noisy pressure in CPM: particle count too low, or the time step is too large relative to particle transit time. Reduce the time step or increase npart.
• Unrealistic stiffness in CV: vent area or porosity is missing, so gas cannot escape. Confirm the porous fabric definition and vent hole segments.

The SVG below summarizes the three methods as a decision map, showing the key inputs, the gas-state representation, and the resulting pressure behavior on the fabric.

In practice, the choice is rarely permanent. A side airbag program typically begins with UP to confirm that the folded bag unfolds without self-penetration, moves to CV for the occupant injury criteria matrix, and reserves CPM for the few cases where the deployment trajectory or the local pressure near the inflator is the subject of the study. Understanding what each method does and does not represent is what allows that staged workflow to be planned deliberately rather than discovered by trial and error.