The detailed picture: interaction of chemical and physical processes

A laminar, stationary, one-dimensional flame can be used as a simple model of a combustion process that takes into account both chemical kinetics and transport processes.


Such stationary flames are shown schematically in Fig. 2. In the case of a premixed flame, the starting materials flow into a region where high-temperature burned waste gas is present. The chemical reactions that lead to ignition and combustion begin on contact with this region, as described above. Relative to the oncoming gas mixture, then, the flame represents a reaction front that travels into the unburned region due to transport processes. With non-premixed flames, fuel and oxidant flow separately into the burned region; in this case, the reaction occurs simultaneously with the mixing of fuel and oxidant. In both cases, the reaction takes place not ‘autonomously’ but under the substantial influence of physical processes such as the continuous flow of fresh gas into the reaction zone, and through heat conduction and diffusion of chemical species between the various zones.

These physical processes introduce an explicit spatial factor into the mathematical description of combustion. The combustion’s fundamental physical-chemical equations, therefore, no longer constitute a system of ordinary differential equations, but a very large and difficult to solve system of partial differential equations. The solution of these equations yields the complete behaviour of a combustion system for specified conditions, including pollutant formation, including for technologically important processes.



Schematic diagram of flame-like combustion. Premixed case: fuel and oxidant are completely mixed before the reaction begins.

Non-premixed case: fuel and oxidant are mixed in the reaction zone. Mixing and reaction take place simultaneously.

This type of calculation, then, should be the sought goal, and in actual fact such so-called direct numerical simulations (DNS) of combustion processes are carried out. Fig. 3 shows on the left the result of the DNS of a turbulent hydrogen/air flame, based on the spatial distribution of the OH radical (see Elementary chemical kinetics). To the left of the flame there is the unburned hydrogen/air mixture, to the right the burned waste gas; the reaction zone is identified by the increase in the OH radical. Since these computations solve the physical and chemical equations in all details and with full spatial and temporal resolution, the simulation provides a very realistic picture of combustion. For example, it shows the typical wrinkling of the flame due to the turbulent flow field and the presence of the OH radical in the burned zone (waste gas) to the right of the flame front. The interaction between physical and chemical processes, especially complex in the case of turbulent combustion, is taken into account fully in DNS, which in principle leads to quantitatively accurate results.


Such detailed simulations are suitable for studying fundamental combustion phenomena and for validating models for a simplified description. They are used often where it is difficult or impossible to obtain the necessary data from experiments. Of particular interest here is the technically very important combustion in multiphase systems, where liquid fuel in the form of droplets evaporates in hot air and then ignites and burns. Detailed simulation studies of such configurations, as shown in Fig. 3 on the right for a fuel droplet surrounded by a laminar flow of hot air, are important aids in the development of better combustion methods, for example in diesel engines and turbines.

Unfortunately, performing such DNS as aids in the development of real technical systems is subject to very tight practical and methodical limitations. The problem, on the one hand, is the massive computing time needed. Even with the rapid increase in computing velocity and capacity, complete computation of technically relevant systems (such as e.g. reciprocating engines, gas turbines or furnaces in power plants) will remain quite impractical for many years. For example, the two-dimensional computation of a turbulent flame (across an area of 1 cm2) shown in Fig. 3 on the left requires several days’ computing time; for technically relevant systems, DNS would take many months or even years. Even if the problem of computing time could be solved, DNS would still not be suitable for the modelling of technical combustion systems since the resulting huge amount of detailed information is only very difficult to manage. It is, therefore, vital to search for a simplified — but nevertheless sufficiently accurate — description of combustion.




Result of a direct numerical simulation (DNS) of the change over time in a turbulent hydrogen/air flame. The concentration field of the species OH, shown in false colours, has dimensions of 1 cm × 1 cm. The simulation demonstrates realistically how the turbulence wrinkles the flame.


Time sequence (simulated) of the temperature field at the ignition of a methanol droplet in a hot air stream travelling from left to right. The image on the left shows the temperature field immediately after ignition, which takes place downstream from the droplet. Thereafter, the flame front migrates towards the droplet (image on the right, 0.12 ms later).



Start | Previous | Page: 1 - 2 - 3 - 4 - 5 - 6 | Next | End