Away from the recirculation zone, L MIN and L MODE : Decreased with T air, and, increased with Re air Were found not to be sensitive to U inj /U air Everything else being the same, L MIN increased relative to the pure (no bluff-body) co-flow experiments that were associated with lower u air ' When L MIN reaches the re-circulation zone, the highly intermittent ‘Spot-Wake Interactions’ behaviour replaces the ‘Random Spots’ Fig. 5 Experiments on the Autoignition of Ethylene Injected Concentrically into Confined Annular Jets of Hot Air Experiments on the Autoignition of Ethylene Injected Concentrically into Confined Annular Jets of Hot Air Interest in the effect of in-homogeneities and turbulence on autoignition is both fundamental and practical (critical in HCCI engines, important in diesel/CI engines and LPP gas turbines, to be avoided in SI engines, storage of flammables, et.c.) DNS of turbulent mixing layers: marginal propensity for earlier autoignition as the turbulence intensity (u air ') is increased Non-premixed counter-flow experiments: higher air temperature necessary for autoignition as u air ' is increased Non-premixed co-flow experiments: autoignition delayed as u air ' is increased Engine (e.g. HCCI) research: earlier autoignition as ‘mixing is enhanced’ So: What is the ‘effect of turbulence’ on the autoignition of ‘in-homogeneous flows’? Objectives Results Experimental Methods Motivation Non-premixed co-flow experiments in this apparatus with H 2 and C 2 H 2 injected into pure confined co-flows (as in Fig. 1, but w/out the bluff body), showed that as U air (and hence u air ') and/or U inj /U air were increased: The mean autoignition length increased non-linearly, so that, The mean residence time until autoignition was delayed Thus: Investigate a case for which u air ' increases independently of U air and U inj /U air In a practically relevant mixing configuration (akin to LPP premix ducts) Provide well-characterized data in a turbulent reacting flow in which the chemical and fluid-mechanical processes interact on the same scales that can serve as a challenging test-bed for the validation of advanced turbulence combustion models ConclusionsFurther Work Fig. 1. Apparatus Schematic. Mixing patterns for illustration. Fig. 2 (above). ‘Instantaneous’ (1ms exposure) OH * (310 ± 10nm) chemiluminescence of autoignition, from left-to-right: 1 st pair: T air = 1059K, T inj = 822K, U air = 17.8m/s, U inj /U air = nd pair: T air = 1051K, T inj = 848K, U air = 13.2m/s, U inj /U air = 3.1. Air was heated up to T air of 1100K and flowed upwards through a 3.0mm grid, around a bluff-body with U air up to 40m/s and into a well-insulated, fully transparent quartz tube The tube was open-ended, so experiments were done at atmospheric pressure Two tube/bluff-body sizes were used, but the blockage ratio, (D BL /D IN ) 2, was kept equal to 0.17 The grid ensured turbulent flow for all conditions; the macroscale Re air, based on the annular hydraulic diameter (D IN -D BL ) and U air was 1400 – 3600 The fuel was N 2 -diluted C 2 H 4, with mass fraction of C 2 H 4 in C 2 H 4 /N 2 equal to 0.74 Fuel was injected continuously and concentrically into the annular air jet behind the bluff-body with U inj = 10 – 80m/s, U inj /U air = 1.1 – 4.4 and T inj in the range 710 – 900K Autoignition occurred in the form of repeated ‘spotty’ flashes accompanied by a ‘popping’ sound Christos Nicolaos Markides * and Epaminondas Mastorakos Hopkinson Laboratory, Department of Engineering University of Cambridge Fig. 3 (left). Average, RMS and PDF post-processed images, calculated from 200 images taken during constant conditions: T air = 1066K, T inj = 745K, U air = 19.2m/s and U inj /U air = 3.2. Also, RMS and Abel transform of RMS for: T air = 1091K, T inj = 832K, U air = 38.2m/s and U inj /U air = 1.5. Fig. 4 (above). L MIN as a function of T air for various Re air and U inj /U air in ‘Random Spots’ and ‘Spot-Wake Interactions’ regimes. Showing re-circulation region extending from the injector to 1 – 3 D BL downstream. Fig. 5 (right). L MIN time series in ‘Random Spots’ and ‘Spot-Wake Interactions’. Note the high intermittency occurring approximately every 5 – 10 s, during which the autoignition location shifts abruptly to very short L MIN. Further evidence has been obtained, by comparison with homogeneous and more weakly turbulent flows, that turbulent mixing (through u air ') inhibits autoignition Turbulent mixing, even in this simple flow, can lead to extreme, possibly dangerous autoignition behaviour (here termed ‘Spot-Wake Interactions’) if the turbulence is strong enough (as it is in HCCI, Diesel/CI, LPP, SI) L IGN measured optically in the continuous - behaviour ‘Random Spots’ and ‘Spot-Wake Interactions’ regimes For each run, i.e. set of T air, T inj, U air and U inj conditions, 200 ‘instantaneous’ images, like those in Fig. 2, were used to compile three processed images: Average, RMS and PDF, as shown in Fig. 3 The minimum length from the RMS (L MIN ) and most probable from the PDF image (L MODE ), were correlated with the conditions, as shown in Fig.4 AverageRMSPDFRMS and Abel Transform Processed results from all instantaneous images of a single run Fig. 3 1 – 3 D BL 1 – 3 D BL Fig. 4 The discrepancy with the DNS can only be clarified if a link can be made between the variables: Re air, U air, U inj /U air and u air ', and, the mixture fraction ( ξ ) and scalar dissipation rate ( χ ) in the flows in which these experiments were performed ( Fig. 1 ) Preliminary results from acetone PLIF measurements of these variables suggest that the delaying effect can be explained in terms ξ and χ, but on a problem-specific basis C2H4C2H4 AIR Vacuum-Insulated Quartz Tube T air N2N2 Electrical Heaters Grid Autoignition Length ( L IGN ) T inj Insulation Autoignition Injector U inj U air Bluff Body Fig. 1 (*): & Fig. 2