TRW’s inflatable restraint, seat belt and steering systems businesses have worldwide revenues of $3 billion. Last year, TRW Occupant Safety Systems produced more than 71 million seat belts, 24 million air bags and 6 million steering wheels. TRW is playing a leading role in the development and production of multistage air bags, which deploy the air bag at different levels based on the severity of the crash and the positions of the occupants. The firm’s multistage air bag production will double to 1.5 million units this year and grow to an estimated 9 million units in four years. TRW recently announced $1 billion in new annual safety systems contracts, giving the firm an industry-leading total of 24 safety systems integration contracts. The company also significantly expanded its safety systems capabilities with last summer’s opening of the $33 million Advanced Car Technology Systems Facility in Sailauf, Germany.
Development of HGIs
TRW has pioneered the development of heated gas inflators (HGIs), which provide equivalent performance to traditional pyrotechnic and augment inflators with hydrogen gas. The result is simple and environmentally friendly inflators, which are produced in Peterlee, England. Comprised of a sealed aluminum cylinder (bottle) filled with a stable, combustible gas mixture, which is ignited by a pyrotechnic squib, it contains and produces no toxic components. TRW has begun production of multistage HGIs for two European automakers, and North American auto makers have ordered both the dual-stage and single-stage HGIs.
The new inflators represent a significant design challenge. TRW engineers need to design many different versions, each with different output requirements, for the many automobile manufacturers that want to use the new inflators. Important design variables used to control HGI performance include the characteristics of the pyrotechnic squib ignitors, particularly the size and shape of the flame produced by the ignitors. These variables determine how the flame propagates in the bottle, which, in turn, controls the discharge rate and pressure from the bottle, which determines the inflation characteristics of the air bag.
Limitations of physical testing
The discharge rate and pressure of a particular inflator design are not difficult to determine. Engineers simply place the HGI in a tank and trigger the inflator while measuring the pressure in both the bottle and the tank as a function of time. This approach determines whether or not a particular design meets the requirements of the application but provides little or no information that can be used to improve the design in the usual situation where the initial design does not meet the requirements. This is because the test is in effect a black box – it neither provides the whole picture of the very complex process nor explains to you what unexpected effect has happened in the process. Visualizing and measuring the high-speed transient flow involving fast reaction and complex geometry is still very difficult. Also, the design experience gained through observations of non-reactive flows very often does not apply in reactive flow cases. Sometimes even the location of the choking point is confusing.
The normal approach to a problem like this is with CFD simulation, which makes it possible to visualize a flow field thoroughly and globally. CFD provides fluid velocity and pressure values throughout the solution domain for time-dependent problems with complex geometries and boundary conditions. As part of the analysis, a designer may change the geometry of the system or the boundary conditions such as inlet velocity, jet flow direction, etc. and view the effect on fluid flow patterns. CFD is an efficient and effective tool for generating detailed parametric studies, significantly reducing the amount of experimentation necessary to develop a device or to solve a design issue.
Challenges of HGI analysis
But an HGI is far more challenging than the typical fluid flow problem. Combustion of hydrogen in air is a complicated process that involves dozens of discrete steps. Another complicating factor is that combustion in an HGI involves both detonation and deflagration waves, where detonation is classified as sudden combustion in which the reaction front approaches the speed of sound. Finally, the strong interaction between chemical reaction and the transient flow nature 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. For these reasons, simulation of hydrogen air bag inflators had never been accomplished in the past.
TRW engineers, however, recognized that commercial CFD software had advanced to the point where it might be possible to simulate combustion accurately without the expense of developing a custom code. They felt that one particular product, FLUENT from Fluent Incorporated, Lebanon, New Hampshire, offers powerful capabilities in the areas of compressible flow modeling and simulation of complex chemical reactions. This code is based on a finite volume formation and offers the option of employing density-based schemes like Roe’s upwind scheme. Thus it has the advantage of maintaining computation stability in highly compressible, highly transient flow problems. Thanks to the advanced chemical reaction module in this commercial code, custom chemical equations can easily be implemented.
Simulating an HGI
TRW engineers implemented several different combustion mechanisms in order to compare their accuracy. They included 1-step, 2-step, 7-step, 8-step, 32-step and 38-step combustion mechanisms developed by various researchers. The reason for using so many different combustion mechanisms is that TRW engineers wanted to determine which was the simplest mechanism that would accurately predict HGI performance in order to conserve computational resources. The ability to implement the complicated and highly-accurate 38-step mechanism in simulations of detonation and deflagration wave propagation made it possible to tune the simpler mechanisms to provide accurate results under various flow conditions, such as different initial bottle conditions and fuel ratios.
Initially, TRW engineers modeled a simple detonation propagation problem in a long 1 cm diameter tube. An axi-symmetric model with proper initial conditions and uniform mesh density in the axial direction could be used. A lean hydrogen air mixture with 16% hydrogen was chosen, which was experimentally shown to generate a detonation with a speed of 1552 m/s. The CFD-computed detonation velocity was computed by tracking the middle point of the leading shock at different moments. All reaction models successfully reproduced the self-sustained detonation wave except the 7-step reaction. In all cases, the CFD-computed detonation velocity was very close to the measured value. For scientific study purpose, the detonation wave structure captured in the CFD solution is carefully examined and compared to the classical ZND (Zel’dovich, von Neumann and DÖring) theory to evaluate the validity of the ZND theory. An AIAA paper is published for this study. Similar simulations for the deflagration wave problems were also performed. These simulation efforts enable one to fine-tune a one-step reaction mechanism accurately for the large-scale computation of the inflator flow. Proper numerical requirements such as the mesh density in the deflagration flame region were also determined.
Impacting the design process
Making use of the experience gained in previous simplified simulations, TRW engineers began simulating more complicated and realistic inflators. Part of the results of a simulation of a HGI within a tank was demonstrated in Figure 1 below. These contours plots of water (combustion product) concentration at various moments properly show the flame growing process in the bottle. The pressure versus time histories also correlated well with the experimental measurements, as can be seen in Figures 2a and 2b. An additional check performed by TRW engineers was cutting open the bottle to look for heat markings on the bottom that would indicate that the flame reached the bottom of the bottle. As predicted by the simulation, the tests showed no heat marks.
Figure 1. Contours of H2O concentration at various moments in the simulation.
The key advantage of the large scale flow simulations is that, in addition to providing output results, they graphically depict the flow fields and particularly the movement of the front as combustion goes on. The simulations, which accurately model the geometry and chemical reactions in the premixed hydrogen-air mixture, revealed many significant aspects of the design that could not be determined from physical testing. For example the pulling force acting on the flame control device, which could not be determined from tests, was determined from CFD results.
Because the simulation is capable of capturing the interaction between the flow and the combustion, engineers were able to determine the sensitivity of the design to different design parameters in a global sense. In particular, the analysis results showed how each of these variables affects the growth of the flame in the hydrogen-air mixture. The flame characteristics are the main factor determining the performance of the inflator. Being able to determine their sensitivity to the main design variables was extremely important for effective performance control.
The ability to simulate accurately the reactive flow process of an HGI has substantially improved TRW’s design process. Rather than building prototypes as soon as they have developed a design concept, engineers can now achieve their design goal in far less time. This is mainly because they can visualize the whole process to understand exactly why the concept design is performing as it is, making it much easier to alter the design to achieve the desired characteristics. These methods have already been used in several custom designs for automotive OEMs and have proven their ability to streamline the design process.
Figure 2a. Comparison of bottle pressure history for a particular inflator design.
Figure 2b. Comparison of tank pressure history for a particular inflator design.