Numerical investigation of uncertainties in experiments for flame structure analysis and particle synthesis
The objective of the presented work is the numerical investigation of experiments for the study of nanoparticle producing laminar flames. Popular, state of the art numerical tools for the investigation of the detailed structure of laminar flames are based on strong assumptions, leading to one-dimensional, steady state approximations of the transport and conservation equations. Unfortunately, these assumptions are often violated. The main focus of this thesis is on the quantification of uncertainties and model errors introduced by thermodynamic and gas dynamic effects, geometric constraints of the reactor and invasive measurement technique in real experiments by means of detailed multi-dimensional numerical simulations of the particle forming, reacting flow. Five different experimental setups were investigated in close collaboration with the research groups conducting these experiments. First, the impact of a molecular beam sampling nozzle on the structure of an atmospheric, premixed hydrogen/oxygen flame. Second, the probing from a low-pressure flat flame of methane and oxygen, where additionally the impact of buoyancy at large heights above the burner was quantified. Third, iron pentacarbonyl doped, particle forming, low-pressure flames of hydrogen/oxygen and methane/oxygen, with and without probing nozzle. Fourth, an atmospheric, sooting, ethylene/oxygen stagnation flow flame with a probing orifice in the stagnation plane. And finally a silica producing low-pressure flame of hydrogen/oxygen doped with hexamethyldisiloxane, used for laser induced fluorescence measurements of SiO. All these investigations required two- and three-dimensional models of the burner and housing geometry in order to capture all deviations from the one-dimensional assumptions. The results presented in this thesis demonstrated the need for such detailed investigations and also contributed to the improvement of experiments and of the kinetic models derived from the measurements. It could be shown that the one-dimensional assumptions are valid only for a very limited set of conditions and that empirical correction formulas are not of general validity. However, the deviations from the idealizing assumptions can be quantified by complementary fluid mechanical simulations of the individual experiment. Based on these finding, the multi-dimensional flow simulation became a permanent element of the experimental workflow at the Institute for Combustion and Gas Dynamics.