3. Inflator modeling and pyrotechnic gas generation

The inflator is the energy source of the entire airbag system, and its model determines whether downstream bag pressure, mass flow, and deployment timing are physically credible. In a coupled gas-dynamics simulation, the inflator is not a simple pressure boundary; it is a transient chemical reactor whose output depends on propellant burn rate, chamber thermodynamics, and nozzle choking. This section develops the standard modeling chain: burn-rate law, mass generation, mass flow rate through the exit, temperature-dependent gas properties, and tank-test validation.

Burn Rate Laws

Solid pyrotechnic propellant (typically sodium azide-based or non-azide formulations) burns according to a pressure-dependent law. The most common form is Vieille's law, also called the Saint-Robert burn law:

r_b = a * P_c^n

where r_b is the linear burn rate (m/s), P_c is the combustion chamber pressure (Pa), a is the burn-rate coefficient, and n is the burn-rate exponent. For typical airbag propellants, n ranges from 0.3 to 0.7, and a is calibrated from strand-burner or closed-bomb tests. The mass generation rate follows from the burning surface area A_b and propellant density rho_p:

dm_gen/dt = rho_p * A_b * r_b = rho_p * A_b * a * P_c^n

The burning area A_b is not constant. For a cylindrical grain, it may be the outer surface, an internal perforation, or both, and it evolves as the grain regresses. In many production inflators, the propellant is a disc or wafer stack, so A_b is approximated as constant until the grain is consumed, then drops to zero. This simplification is acceptable for the first 20–40 ms, which covers the critical deployment window.

Mass Flow Rate Through the Exit

Combustion products leave the inflator through nozzles or orifices into the bag. The flow is almost always choked at the throat because the chamber-to-bag pressure ratio far exceeds the critical value (about 1.9 for diatomic gases). The choked mass flow rate is:

dm_out/dt = C_d * A_t * P_c * sqrt(gamma / (R * T_c)) * (2 / (gamma + 1))^((gamma + 1) / (2 * (gamma - 1)))

Here C_d is the discharge coefficient (0.8–0.95 for drilled orifices), A_t is the total throat area, gamma is the ratio of specific heats, R is the specific gas constant, and T_c is the chamber temperature. The chamber pressure itself evolves from the balance between generation and outflow, coupled with energy release:

dP_c/dt = (gamma - 1) / V_c * (Q_gen - Q_out) - gamma * P_c / V_c * dV_c/dt

where Q_gen is the chemical energy release rate, Q_out is the energy carried out by the exiting gas, and V_c is the chamber volume (which may change if the inflator has a moving component). In a fixed-volume inflator, the last term vanishes.

Temperature-Dependent Gas Properties

Combustion gas temperature in an airbag inflator typically peaks between 1200 K and 2200 K, then decays as gas expands and heat transfers to the housing and bag fabric. Treating gamma and R as constants is a common source of error. A better approach is to tabulate c_p(T) for the gas mixture and compute:

gamma(T) = c_p(T) / (c_p(T) - R)
R = R_universal / M_mix

For a typical nitrogen-rich mixture with M_mix around 24–28 g/mol, R is approximately 290–340 J/(kg·K). At 1500 K, gamma may be 1.28; at 300 K, it rises to about 1.40. This variation changes the choked flow rate by 5–10%, which is significant when predicting bag pressure at 30 ms. A practical implementation uses a lookup table indexed by temperature, with linear interpolation between points.

Tank Test Validation

The standard validation method is the closed-tank test. The inflator discharges into a rigid tank of known volume V_tank (typically 60 L or 100 L), and the tank pressure P_tank(t) is recorded. The simulation must reproduce this curve. The tank pressure is governed by:

dP_tank/dt = (gamma_tank - 1) / V_tank * (dm_out/dt * c_p * T_c - P_tank * dV_tank/dt)

Since the tank is rigid, dV_tank/dt = 0. The validation procedure is:

1. Run the tank test at three temperatures: −35 °C, 23 °C, and 85 °C.
2. Record P_tank(t) at 1 kHz or higher.
3. Run the simulation with the same initial conditions and compare peak pressure, time-to-peak, and the decay tail.
4. Adjust C_d, a, and the gas property table until the RMS error is below 5% of peak pressure.

A typical acceptance criterion is that the simulated peak pressure is within ±5% and the time-to-peak is within ±2 ms of the measured value. If the simulated pressure rises too slowly, the burn rate coefficient a is likely too low. If the peak is too high but the decay is correct, the throat area A_t or discharge coefficient C_d is too small.

Practical Calibration Workflow

A reproducible calibration sequence for a new inflator model is:

1. Fix geometry: measure A_t and V_c from the CAD model. These are not tuning parameters.
2. Estimate a and n: use closed-bomb data if available. If not, start with n = 0.5 and adjust a to match the initial pressure rise rate.
3. Tune C_d: match the peak tank pressure. A 10% change in C_d typically shifts peak pressure by 8–12%.
4. Refine gas properties: if the decay tail is too fast, increase c_p at high temperature; if too slow, decrease it.
5. Validate at three temperatures: the same parameter set must work at −35 °C, 23 °C, and 85 °C. If it does not, the burn rate law is missing a temperature-dependent term, often an Arrhenius factor exp(-E_a / (R T)).

A common failure symptom is a simulated pressure that rises correctly but never decays. This usually means the propellant mass is exhausted in the model but the tank pressure equation is missing the heat-loss term to the tank wall. Adding a convective loss term h A_wall (T_tank - T_ambient) with h around 20–50 W/(m²·K) corrects the tail without affecting the peak.