Simulation of atomization in design of biofuel/low temperature combustion

In the design chain of novel biofuels, there are two streams of information go into the simulation of atomization: one comes from biorefinery, which contains the physical properties of liquid biofuels, for example, density, viscosity and surface tension; the other comes from engine development, which contains the low temperature combustion operating conditions, such as ambient pressure, ambient temperature and injection pressure.

The multi-scale, multi-physics simulation of atomization of TMFB cluster is build on modern computational framework. It requires massive parallel computational infrastructure, which is supported by the Jülich Aachen Research Alliance (JARA), providing parallel computing, parallel I/O and parallel data management. Key components of the computational framework are the physical and mathematical models and numerical solvers for the high-fidelity simulations, which include 1) the high-performance unstructured finite-volume flow solver for the Navier-Stokes equations, 2) the level-set method based liquid-gas interface tracking and the surface tension model, 3) the Lagrangian method based droplet dispersion solver. The multicomponent evaporation model for fuel blending will be developed in the second phase the TMFB cluster.

Main results of the simulations are the droplet size and velocity distributions, as well as the 3 dimensional unsteady turbulent flow field. The feedback information also contains two streams of information. One goes towards biorefinary: which answers the questions – does the atomization of the novel biofuel similar as the traditional fuels? how critical are the influence of those properties on the fuel/combustion flexibility? As for the engine development, better understanding of the atomization process will assist designing advanced injection/mixture formation strategies for smokeless/NoX free combustion.



Atomization: Nondimensional Flow Parameters

Navier-Stokes equations with surface tension

The simulation will resolve the complete multiphase flow field near the injector nozzle. The governing equation of this process is the Navier-Stokes equations with surface tension force. To systematically analyze the fuel properties influence on the atomization process, the governing equations can be written non-dimensionally using dimensional analysis. Four independent flow parameters can be identified from them: the density ratio, the viscosity ratio, the Reynolds number and the Weber number. Both the fuel physical properties and the engine operation conditions are necessary information to derive the flow parameters.

At the same injection pressure 720Bar (72 MPa) and engine operation condition (air density is expected to be 22 kg/m3), using the liquid fluid properties of different reference fuels, the flow parameters control the atomization process are listed in the table. At those high Reynolds numbers, the flows are all turbulent. A fundamental characteristic of turbulent flows is that there exist a wide range of length, time and velocity scales in the flow field. To solve all these scales require high-performance computation. The computational mesh contains over 100 million cells (max 402 million); the smallest grid size is 1 micrometer. Typical simulation is running on hundreds of CPUs (maximum 4096 CPUs) for several days.


 

 

Simulation of atomization of iso-octane direct injection (spray removed)

 

Simulation of turbulent atomization of iso-octane direct injection


Compare droplet size distributions

The droplet size distribution, resulting from the atomization process, is important for the spray vaporization and combustion. Smaller droplets vaporize faster due to their larger surface to volume ratio compared to the larger droplets. Therefore, sprays contain mostly small droplets will have shorter evaporation time, leaving more time for the fuel-vapor/air mixing. A well-mixed fuel-air mixture once ignites, the combustion will behave more like a premixed flame; if the ignition happens during the mixing process, the combustion is still mixing controlled diffusion flame.