4. Gas dynamics modeling in MADYMO

Curtain airbag inflation is fundamentally a gas dynamics problem. The restraint performance of a side curtain depends not only on the fabric geometry and the folded package, but also on how the inflator mass flow is distributed inside the cushion over the first 20–50 ms. MADYMO offers three distinct gas modeling families for this task: the uniform pressure model, the hybrid gas particle (HGP) model, and the CFD-based approach using a finite-volume gas domain coupled to the structural mesh. Selecting the right one is a trade-off between fidelity, runtime, and the physical questions you need to answer.

Uniform Pressure Model

The uniform pressure model treats the entire cushion volume as a single thermodynamic control volume. At each time step, the solver computes the internal pressure from the ideal gas law using the inflator mass flow rate, the gas temperature, and the instantaneous chamber volume:

P = (m · R · T) / V

where:
  m = instantaneous gas mass in the chamber [kg]
  R = specific gas constant [J/(kg·K)]
  T = gas temperature [K]
  V = current chamber volume from the structural mesh [m³]

The inflator is defined by a tank pressure curve or a mass flow rate table, typically derived from a tank test. The pressure is applied uniformly as a normal load on all fabric elements. This approach is computationally cheap and numerically robust, which makes it suitable for early design iterations, parameter sweeps, and full-vehicle side impact runs where the curtain is one of many restraints.

The key limitation is that it cannot capture pressure gradients. In a real curtain, the gas enters at the inflator port and must travel along the length of the cushion. During the first 10–15 ms, the pressure near the inlet is significantly higher than at the far end. The uniform pressure model smears this out, which can lead to inaccurate predictions of deployment timing, cushion shape, and contact force against the occupant. It also cannot represent gas leakage through seams or vent holes in a spatially resolved way.

Hybrid Gas Particle Model

The hybrid gas particle model divides the cushion volume into a set of discrete gas particles or chambers connected by flow paths. Each particle carries its own mass, internal energy, and temperature. Gas flows between particles through orifices or porous connections, driven by the pressure difference. The structural mesh provides the volume of each chamber, and the gas state is updated using a one-dimensional flow relation at each connection.

In MADYMO, the HGP model is typically configured with a set of GAS_PARTICLE definitions and FLOW_PATH elements. A practical curtain model might use 5–15 particles distributed along the cushion length, with flow paths representing the main gas channel and the connection to the upper and lower chambers. The inflator feeds one or more inlet particles. Vent holes and seam leakage are modeled as additional flow paths to ambient.

The HGP approach captures the essential physics of finite inflation speed: the pressure wave travels along the cushion, the far end inflates later, and the local pressure history depends on the flow resistance of the internal passages. It is significantly faster than CFD and is the standard choice for production curtain airbag validation in MADYMO. The main modeling challenge is calibrating the flow path coefficients. These are often derived from component tests or from CFD simulations of the gas volume alone.

CFD-Based Approach

The CFD-based approach solves the Navier-Stokes equations on a gas domain that is coupled to the structural fabric mesh. In MADYMO, this is typically done through a co-simulation with a CFD solver or through the built-in finite-volume gas module. The gas domain is discretized into cells, and the inflator is modeled as a mass flow inlet with a specified temperature and gas composition. The fabric motion provides the moving boundary condition.

This approach resolves the detailed flow field: jet formation at the inflator outlet, recirculation zones, pressure gradients along the cushion, and the effect of internal tethers and baffles on gas distribution. It is the most accurate method but also the most expensive. A typical curtain CFD model may require 10–100 times the runtime of an HGP model. It is therefore reserved for detailed design studies, root-cause analysis of deployment anomalies, or calibration of simpler models.

The table below summarizes the three approaches for a typical 2.5 m curtain airbag with a hybrid inflator:

Practical Selection and Calibration Workflow

A robust curtain airbag program typically uses all three models in a staged workflow. Start with the uniform pressure model to establish the basic inflation timing and cushion shape. Once the folded geometry and material properties are stable, transition to the HGP model and calibrate the flow paths against available test data. Use the CFD model selectively to investigate specific issues such as delayed inflation at the rear end or excessive pressure near the B-pillar.

When calibrating the HGP model, the most sensitive parameters are the flow path discharge coefficient and the effective flow area. A common starting point is a discharge coefficient of 0.6–0.8 for sharp-edged orifices and 0.3–0.5 for porous fabric passages. The inflator mass flow curve should be taken from a tank test at the same temperature as the intended operating condition. If the predicted cushion pressure at the far end is too low, reduce the flow resistance; if the inlet pressure spikes unrealistically, increase the flow area or reduce the discharge coefficient.

The SVG below illustrates the three gas modeling approaches applied to the same curtain airbag geometry, showing how the pressure field representation differs between them.

In practice, the HGP model is the workhorse for curtain airbag validation. It captures the essential finite inflation speed and pressure distribution while remaining compatible with full-vehicle simulation timelines. The uniform pressure model remains useful for early-stage concept studies, and the CFD approach provides the detailed insight needed when the HGP model cannot resolve a specific physical phenomenon. The key is to match the model fidelity to the engineering question at hand, and to validate each level against the one above it before moving to the next.