Alex Meduvsky and Kwen Hsu
TRW Occupant Safety Systems
Edited by Sherri Koucky
Airbags have been available in vehicles for a number of years. So it might come as a surprise that the hydrogen inflators responsible for deploying bags have only recently been simulated in software.
The challenge in simulation is to accurately capture two qualities: the stiff hydrogen-air reaction and the complex flow geometry. Moreover, these factors must be modeled without consuming inordinate computational resources.
TRW engineers use commercial computational fluid-dynamics (CFD) software that allows for substantial complexity in chemical reactions. By creating the first accurate simulation of a hydrogen inflator, they can see how parts in the inflator influence combustion. This, in turn, makes it possible to fine-tune the critical combustion process for individual applications.
Development of HGIs
TRW pioneered the development of heated gas inflators (HGIs). These perform as well as traditional pyrotechnics and augment inflators with hydrogen gas. They are simple and environmentally friendly. An inflator is basically a sealed aluminum cylinder (bottle) filled with a stable, combustible gas mixture, ignited by a pyrotechnic squib. It contains and produces no toxic components. TRW has begun producing multistage HGIs for two European automakers, and North American automakers have ordered both the dual-stage and single-stage HGIs.
The new inflators pose a significant design challenge. Engineers must design many different versions, each with different output requirements.
Important design variables that control HGI performance include qualities of the pyrotechnic squib ignitors, particularly the size and shape of the flame from the ignitors. The flame characteristics are the main factor determining the performance of the inflator. Being able to gauge their sensitivity to the main design variables is extremely important for effective performance control. These variables determine how the flame propagates in the bottle. This, in turn, controls the discharge rate and pressure from the bottle, ultimately determining how the airbag inflates.
Limitations of physical testing
It’s not hard to predict the discharge rate and pressure of a particular inflator design. Engineers simply place the HGI in a tank and trigger the inflator while measuring the pressure in both the bottle and the tank over time. This approach shows whether or not a particular design can handle a specific application, but gives little that’s useful for improving the design. This is because the test is like a black box — it neither illuminates the whole complex process nor explains unexpected effects. And during tests, it’s quite difficult to visualize and measure the high-speed transient flow involving fast reaction and complex geometry.
Also, the design experience gained from observing nonreactive flows many times does not apply in cases of reactive flow. Reactive flow has a chemical reaction taking place along with the flow. Sometimes even the location of the choking point is confusing. Choking occurs when a hybrid mixture velocity becomes equal to the homogeneous equilibrium (HE) sound velocity.
The normal approach to a problem like this is CFD simulation, which makes it possible to visualize a flow field thoroughly. CFD can give fluid velocity and pressure values over time for geometries and boundary conditions that are both complicated. A designer may change the geometry or the boundary conditions such as inlet velocity and jet-flow direction to see the effect on fluid-flow patterns.
CFD is an efficient and effective tool for generating detailed parametric studies, significantly cutting the amount of experimentation needed during development. Simulations also reveal many significant design aspects that are impossible to see from physical testing. For example, physical tests won't reveal the pulling force acting on the flame control device. But CFD results can.
But an HGI is more challenging than the typical fluid-flow problem. A number of factors make simulation of hydrogen airbag inflators seemingly impossible. Combustion of hydrogen in air is a complicated process involving dozens of discrete steps. Both detonation and deflagration waves take place, detonation being sudden combustion in which the reaction front approaches the speed of sound. There is a strong interplay between chemical reaction and the transient flow of gas. This requires that the reaction front be modeled in a transient fashion, ruling out a simpler steady-state analysis. The flame expands and propagates significantly in both radial and axial directions.
Commercial CFD software called Fluent from Fluent Inc. of Lebanon, N.H., is advanced enough to handle compressible flow modeling and simulation of complex chemical reactions for HGI. This code is based on a finite-volume formation and can employ what are called density-based schemes, Roe's upwind being one. Roe's upwind scheme is a method of solving equations that predict fluid flow. It maintains computation stability in highly compressible, highly transient flow problems. The advanced chemical reaction module in this code lets it handle custom chemical equations.
Simulating an HGI
TRW engineers tested several different combustion mechanisms to compare accuracy. Some included combustion mechanisms with as many as 38 steps. The point was to determine the simplest mechanism able to accurately predict HGI performance. The most complicated of these let engineers tune the simpler mechanisms for accurate results under various flow conditions, such as different initial bottle conditions and fuel ratios.
Initially, developers modeled a simple detonation propagation problem in a long 1-cm-diameter tube. During an experiment, a mixture composed of 16% hydrogen and air generated a detonation with a speed of 1,552 m/sec. The task was then to simulate these results via CFD. CFD computes detonation velocity by tracking the middle point of the leading shock at different moments. Simulated detonation velocities nearly matched those measured. For scientific study purposes, engineers compared the detonation wave structure captured in the CFD solution to the classical ZND (Zeldovich, von Neumann, and Doring) theory. The ZND model describes the detonation wave as a shock wave, immediately followed by a reaction zone, or flame. Other simulations let engineers correctly fine-tune a one-step reaction mechanism for large-scale computation of inflator flow. These simulation runs also help determine requirements such as the mesh density in the deflagration flame region.
Experiences gained from simplified simulations guided engineers in simulating more complicated and realistic inflators.
A key advantage of the large-scale flow simulations is that they graphically depict the flow fields and how the front moves as combustion proceeds.
The ability to accurately simulate the reactive-flow process of an HGI has substantially improved TRW's design process. Rather than build prototypes of a design concept, engineers now get to their design goal more quickly. Visualizing the whole process lets them understand exactly why the concept design performs as it does, and simplifies the process of making changes.