In turbulent inhomogeneous (or non-premixed) flows, fluctuations can exist in all scalars. In the context of the phenomenon of autoignition, the flow is also a reacting (or autoigniting) one and belongs to the general field of turbulent reacting flows. Scalars include both conserved/passive (mixture fraction) and reacting/active (species and temperature) quantities, with fluctuations present both in the (unconditional) scalars themselves and in the reactive scalars conditioned on a value of a conserved scalar; the latter being in contrast to non-reacting (or frozen) flows. In the context of such a turbulent inhomogeneous flow of hot oxidizer and colder fuel, and given the temperature, velocity and mixing fields of a particular configuration/geometry and the characteristics of various chemistries, we are specifically interested in predicting the location and timing of autoignition.
Current development of combustors for Diesel and Homogeneous Charge Compression Ignition (HCCI) engines depends mainly on an improved understanding and ability to predict the controlling phenomenon of autoignition. In these applications autoignition occurs in the presence of inhomogeneities and strong turbulence, which cannot be neglected and whose effect must be understood. The study of autoignition is also critical in applications where unwanted manifestations of this phenomenon are to be avoided, such as in the premix ducts of Lean Premixed Prevapourized (LPP) gas turbines, Spark Ignition (SI) engines and flammable material storage areas. In parallel, there is great academic interest in fundamental questions surrounding turbulent reactive flows where the chemical and fluid-mechanical timescales are of the same order, whence the mixing has a direct effect on the chemistry. With few exceptions, experimental autoignition investigations have concentrated on homogeneous mixtures, driven by the desire to understand the complex chemical kinetics of this phenomenon. Yet, our present understanding leads one to reason that autoignition in the turbulent inhomogeneous case cannot be understood at the fundamental level by extrapolation from homogeneous, or even laminar inhomogeneous studies.
Even though the role of turbulent mixing in determining the location of autoignition is becoming clearer, the indirect effect of turbulence on autoignition in such flows is far from understood, both in determining the eventual location and moment at which autoignition will occur and in the way it underpins the random nature of this location and timing. This lack of understanding is in part due to the extremely sensitive to initial conditions nature of autoignition that leads to a difficulty of achieving this phenomenon in a controlled experiment, and is further compounded by the general difficulty of obtaining well-characterized, accurate and detailed measurements in the harsh combustion environment. So far, the role of turbulent mixing has only been studied with Direct Numerical Simulations (DNS) and modelling. However, complications arise in these computational theoretical/numerical approaches as well that greatly inhibit a decisive uncovering of the important determining parameters and their effects. Due to the strong coupling of the underlying chemical and thermo/fluid-mechanical processes that control the phenomenon, the choice of chemical mechanisms, numerical limitations and simulation in non-representative flow length and time scales (Reynolds numbers), a certain insecurity is introduced in these findings. The convenient assumptions of scale-separation cannot be applied here and so it can be said that this problem truly lies on the boundaries where the fields of Thermodynamics, Fluid Mechanics and Chemistry meet.
Here, at the Hopkinson Laboratory, the Turbulent Autoignition Project has in parallel undertaken an experimental, theoretical, computational and modelling investigation that is concerned with the general problem of autoignition in turbulent inhomogeneous flows. This attempt is centered around the experimental study and consequent prediction of autoignition in turbulent, non-premixed (inhomogeneous) flows, with novel experiments in a co-flow, axisymmetric apparatus that has been set-up in the Hopkinson Laboratory. The Hopkinson Laboratory is situated in the Inglis Building of the Department of Engineering at the University of Cambridge and is part of the Energy Group, which belongs to Division A: Energy, Fluid Mechanics and Turbomachinery. If you are interested in an overview of all aspects of this work visit Autoignition in Turbulent Flows and the pages of Dr. Epaminondas Mastorakos. In the current pages we present and concentrate on the experimental approach to this problem; the design and realization of novel co-flow experiments capable of successfully demonstrating a variety of autoignition and related phenomena, the collection of well-characterized data and the analysis of subsequent results.
In the experiments performed in this work, gaseous fuels were injected, continuously and axisymmetrically (concentrically), into uniform high-temperature, turbulent co-flows of air, confined inside a well-insulated and fully optically accessible quartz tube. The flow and mixing fields in the tube were characterized with thermocouples, hot wires and Planar Laser-Induced Fluorescence (PLIF) of acetone. Original phenomena are reported concerning the emergence of autoignition in the form of 'spots', unsteady flame propagation and extinction or possible flashback. The observed phenomena are classified into operation regimes, the boundaries between which are explored, paying attention to flashback conditions. The frequency of appearance of the autoignition 'spots' was also measured, together with the sound and spectral signature of chemiluminescence produced during autoigniting conditions. Optical measurements of autoignition were made from which the location of autoignition was measured and used to estimate mean 'delay times' from injection. As would be expected by considerations of simple chemical kinetics and the mean scalar field, higher air temperatures and lower jet velocities were found to move the autoigniting regions in the tube closer to the injector. A more involved conclusion is that as the air velocity and hence turbulent fluctuations were increased, the scalar dissipation rate (gradients squared of the conserved scalar) at locations with 'most reactive' chemistry increased causing autoignition to be both shifted downstream and delayed. As autoignition moved downstream, the autoignition frequency and sound intensity decreased. Situations are presented that cannot be explained in terms of homogeneous delay time arguments and where the importance of the mixture fraction (normalized conserved scalar) and of the conditional (on the mixture fraction being most reactive) scalar dissipation rate are highlighted.
Back to Top