Thorsten Zirwes

Direct Numerical Simulation of Chemically Reacting Flows with the Public Domain Code OpenFOAM

A new solver for direct numerical simulation (DNS) of chemically reacting flow is introduced, which is developed within the framework of the open-source program OpenFOAM. The code is capable of solving numerically the compressible reactive flow equations employing unstructured grids. Therewith a detailed description of the chemistry, e.g. the reaction rates, and transport, e.g. the diffusion coefficients, has been accomplished by coupling the free chemical kinetics program Cantera. The solver implies a fully implicit scheme of second order for the time derivative and a fourth order interpolation scheme for the discretization of the convective term. An operator-split approach is used by the solver which allows solutions of the flow and chemistry with time scales that differ by orders of magnitude, leading to a significantly improved performance. In addition, the solver has proved to exhibit a good parallel scalability. The implementation of the code has first been validated by means of one-dimensional premixed flames, where the calculated flame profiles are compared with results from the commercially Chemkin code. To demonstrate the applicability of the code for three-dimensional problems, it has been applied to simulate the flame propagation in an explosion vessel of laboratory-scale. A computational grid with 144 million finite volumes has been used for this case. The simulation has been performed parallel on 8192 processors from the HERMIT cluster of HLRS. The calculated burning velocity agrees well with the experimental data.

Numerical Simulation of the Ignition of Fuel/Air Gas Mixtures Around Small Hot Particles

Abstract This study presents the simulation and detailed analysis of the ignition of initially quiescent fuel/air mixtures by small, stationary, laser-heated spherical particles. Our simulations cover a wide parameter space by varying the kind of fuel, stoichiometry, heating rate, radical surface destruction efficiencies as well as particle size. The results agree well with experimentally determined particle surface temperatures at the time of ignition over the whole range of parameters. The surface temperatures required for ignition strongly depends on the kind of fuel and increases in the order hydrogen, acetylene, ethylene, ethane, propane and methane. It also increases with decreasing particle size. By contrast, mixture stoichiometry and heating rate have a minor influence on the ignition temperatures. Comparisons with two-dimensional direct numerical simulations show that fast, but fully coupled one-dimensional simulations are sufficient to capture the details of the ignition event, permitting a systematic investigation for large number of conditions. At small particle radii (r≤2 mm) there exists a simple mapping of only two parameters, an apparent activation energy and a factor comprising thermo-physical properties of the gas phase that is able to estimate the particle surface temperature required for ignition. Such a map might be used for the safety assessment of ignition hazards by small hot particles as function of fuel, stoichiometry and particle size.

Ignition of combustible mixtures by hot particles at varying relative speeds

ABSTRACT Detailed numerical simulations have been performed to study the ignition behavior of hot spherical particles at atmospheric conditions. The particles move relative to a combustible gas with a velocity of 0–30 m s, which spans different flow regimes, from creeping flow to unsteady vortex shedding. The temperature of the particles’ surface increases linearly over time and is recorded at ignition for methane/air and hydrogen/air mixtures. For low relative velocities m s or Reynolds numbers , increases proportionally to or and the flow field is axisymmetric. For higher relative velocities, an unsteady vortex street forms behind the particle so that three-dimensional simulations are required. A correlation employing the van’t Hoff criterion yields linear correlations based on the Nusselt number and for both the low- and high-velocity ranges. For rich hydrogen flames at high velocities, the flame temporarily stabilizes near the hot particle in the recirculation zone downstream. As the surface temperature increases further, the flame suddenly starts to propagate downstream, leading to two distinct ignition events: local ignition at the particle’s surface and start of the propagation into the surrounding gas. The latter yields a much steeper increase of ignition temperature with incoming flow velocity.

A DNS Analysis of the Correlation of Heat Release Rate with Chemiluminescence Emissions in Turbulent Combustion

The essential correlation of heat release rate and chemiluminescence emission from turbulent combustion is quantitatively analyzed by means of direct numerical simulation (DNS) of premixed methane/air flames, employing a detailed reaction mechanism with 18 species and 69 elementary reactions, and the mixture-averaged transport method. One-dimensional freely propagating laminar flames have first been studied for different stoichiometries varying from fuel-lean to fuel-rich conditions. There, the local generation of the chemiluminescent OH* species correlates strongly with the heat released by the combustion reaction, especially in the fuel-lean range. Three-dimensional DNS have then been applied to calculate a synthetically propagating flame front subjected to different turbulent inflow conditions. Joint probability density functions of OH* concentration and heat release rate have been generated from the DNS results, showing a stronger scattering of the correlation curve compared to the corresponding laminar flame. As the chemiluminescence measurement gathers light only along one viewing direction, the line-of-sight integrated values of heat release and OH* concentration have been evaluated from the DNS, where the domain has been decomposed into a number of rays defined by a fixed viewing direction and a specific area. A quasi-linear relationship has been identified for these integral values, where the correlation becomes stronger for flames subjected to lower turbulence intensities or larger cross-section areas of the rays. A computational grid with 16 million finite volumes has been used for the DNS of the turbulent flames and the simulations have been performed in parallel with 3,600 processor cores from the Hazel Hen cluster of HLRS. Scale-up performance of the DNS code, which is based on the open-source program OpenFOAM, has been evaluated.

Numerical Simulation of Turbulent Combustion with a Multi-Regional Approach

The current work uses a multi-regional method for improving the computing performance of large-scale combustion simulations. In this manner, the solution of the isothermal flow within the burner is treated separately from the domain with combustion reaction. For the fresh gas flow within the nozzle only the Navier-Stokes equations for a non-reactive, fixed composition flow are solved, whereas the combustion model accounting for the chemical reactions is enabled in the ignition zone downstream of the burner. Because the chemistry solution takes a major part of the total computing time, the approach saves that part of execution time for the computing nodes located within the nozzle, where no chemical reaction occurs. In the present study, the potential of this methodology has been assessed by large eddy simulation (LES) of a model burner operated with a premixed methane/air flame. The multi-regional simulation showed consistent results with data obtained from the conventional single-regional computation. It however has been proven to be considerably faster than the comparable single-zonal LES, denoting an improved computing performance.

Automated Code Generation for Maximizing Performance of Detailed Chemistry Calculations in OpenFOAM

In direct numerical simulation of turbulent combustion, the majority of the total simulation time is often spent on evaluating chemical reaction rates from detailed reaction mechanisms. In this work, an optimization method is presented for speeding up the calculation of chemical reaction rates significantly, which has been implemented into the open-source CFD code OpenFOAM. A converter tool has been developed, which translates any input file containing chemical reaction mechanisms into C++ source code. The automatically generated code allows to restructure the reaction mechanisms for efficient computation and enables more compiler optimizations. Additional performance improvements are achieved by generating densely packed data and linear access patterns that can be vectorized in order to exploit the maximum performance on HPC systems. The generated source code compiles to an OpenFOAM library, which can directly be used in simulations through OpenFOAM’s runtime selection mechanism. The optimization concept has been applied to a realistic combustion case simulated on two peta-scale supercomputers, among them the fastest HPC cluster Hazel Hen (Cray XC40) in Germany. The optimized code leads to a decrease of total simulation time of up to 40% and this improvement increases with the complexity of the involved chemical reactions. Moreover, the optimized code yields good parallel performance on up to 28,800 CPU cores.