Flamelet-generated manifolds for the simulation of non-premixed laminar co-flow flames doped with polycyclic aromatic hydrocarbons

The removal of tar from biomass derived fuel gases is considered to be a bottleneck in the large-scale application of biomass gasification. Tar is made up of Polycyclic Aromatic Hydrocarbons (PAHs) consisting of fused aromatic rings. In the currently most promising technologies, a significant role of oxygen radicals can be observed. By introducing a limited amount of oxidizer ( 0.2) to hot biogas, it has been proven possible to convert naphthalene (C10H8) into smaller hydrocarbons [1] [2]. In this process diffusion flames will appear. A better understanding of the processes taking place is necessary to tackle scaling issues and to optimize the current design of the reactor. Modeling of combustion applications very often requires the use of detailed chemistry models in two or even three dimensional simulations. If, for instance, accurate predictions of NO or PAH/soot formation are required , complex reaction mechanisms involving many chemical species and reactions have to be used. It is well-known that the use of such detailed reaction mechanisms results in high computational costs and that efficient methods are needed to decrease the computational burden. For instance, in the flamelet-generated manifold (FGM) approach [3], a low-dimensional manifold is created by using solutions of one-dimensional flamelet equations. Applying FGM leads to a reduction of the computational cost up to several orders. The goal of this paper is to investigate the ability of the flamelet-generated manifold (FGM) approach for the numerical simulation of multidimensional laminar diffusion flames doped with PAHs. FGMs constructed from premixed and non-premixed (counterflow) one-dimensional flamelets have initially been applied to an undoped laminar methane/nitrogen flame in a co-flow of air. These results are compared to the full chemistry (FC) model. Inclusion of phenomena such as differential diffusion with constant Lewis numbers for each species and NO formation are also studied.When unity-Lewis numbers are considered, an FGM which consists of counterflow diffusion flamelets is able to predict temperature and species concentrations in good agreement with the FC solution. Manifolds constructed from premixed flamelets give less good results. Differential diffusion effects are hard to capture to a satisfactory level. Including non-unity Lewis numbers in solving the one-dimensional flamelets leads to great improvement in the results, but using a simplified model in the application of the FGM, results in significant deviations. The formation of thermal NO is included by solving an additional transport equation and retrieving its source term from the manifold. In comparison to direct look-up, large improvements are observed. It is demonstrated, that in respect to the traditional steady Z- approach, large improvements are made in predicting NO mass fractions.The fuel flow is doped with several different PAH species (ranging from 1- to 4-ring) to simulate the tar content present in genuine producer gas originated from an indirect gasifier. The flamelet approach is based on the assumption that the time scales of chemical processes are much smaller than the flow time scales. Considering PAHs, this might not be the case. So an additional transport equation (as is done for NO in the undoped situation) is introduced to improve the prediction of PAH chemistry. Experimental results contribute in validating the 2D numerical models. The experimental setup consists of a co-flow burner, a heated fuel line and an injection device which allows accurate and controlable injection of liquid PAHs. Temperature measurements concerning the injection of toluene (400 ppm) have been performed. For qualatively measuring PAHs in a laminar doped diffusion flame LIF (Laser Induced Fluorescence) is applied. Fluorescence appears to be a promising non-intrusive technique with regard to identification of the species and spatial resolution [4-6]. The dopants considered are 1-to 4 ring PAHs (benzene, toluene, naphthalene, phenanthrene and pyrene). The PAHs which are solid at room temperature are dissolved in 2,-heptanone before injection. Concentrations range from 1000 to 2000 ppm of PAHs in the fuel flow. These concentrations are of the same order as the tar concentrations found in genuine producer gas. Wavelenghts considered with a YAG laser are 532 and 266 nm. LII (Laser Induced Incandescence) is also applied to visualize the soot in the top of the flame. Qualitative measurements of OH are executed by means of chemiluminescence. Experimental results will be compared with the 2D numerical results. After validation the numerical tools will be applied to model the current reactor and improve its design.