The paper presents a new approach to modelling of the heating and evaporation of gasoline fuel droplets with a specific application to conditions representative of internal combustion engines. A number of the components of gasoline with identical chemical formulae and close thermodynamic and transport properties are replaced with characteristic components leading to reducing the original composition of gasoline fuel (83 components) to 20 components only. Furthermore, the approximation to the composition of gasoline with these components is replaced with a smaller number of hypothetical quasi-components/components as previously suggested in the multi-dimensional quasi-discrete (MDQD) model. The transient diffusion of quasi-components and single components in the liquid phase as well as the temperature gradient and recirculation inside the droplets, due to the relative velocities between the droplets and the ambient air, are accounted for in the model. In the original MDQD model, n-alkanes and iso-alkanes are considered as one group of alkanes. In this new approach, the contributions of these two groups are taken into account separately. The values for the initial model parameters were selected from experimental data measured in a research engine prior to combustion. The results are compared with the predictions of the single-component model in which the transport and thermodynamic properties of components are averaged, diffusion of species is ignored and liquid thermal conductivity is assumed to be infinitely large, or approximated by those of iso-octane. It is shown that the application of the latter models leads to an under-prediction of the droplet evaporation time by approximately 67% (averaged) and 47% (iso-octane), respectively, compared to those obtained using the discrete component model, taking into account the contributions of 20 components. It is shown that the approximation of the actual composition of gasoline fuel by 6 quasi-components/components, using the MDQD model, leads to an under-prediction of the estimated droplet surface temperatures and evaporation times by approximately 0.9% and 6.6% respectively, for the same engine conditions. The application of the latter model has resulted in an approximately 70% reduction in CPU processor time compared to the model taking into account all 20 components of gasoline fuel.

No abstract

This paper reports the application of optical tomography and Chemical Species Tomography (CST) to the characterisation of the in-cylinder

A conventional laminar vortex ring model is generalized by assuming that the time dependence of the vortex ring thickness is given by the relation = a t b , where a is a positive number and 1/4 6 b 6 1/2. In the case in which a = √ 2ν, where ν is the laminar kinematic viscosity, and b = 1/2, the predictions of the generalized model are identical with the predictions of the conventional laminar model. In the case of b = 1/4 some of its predictions are similar to the turbulent vortex ring models, assuming that the time-dependent effective turbulent viscosity ν * is equal to . This generalization is performed both in the case of a fixed vortex ring radius R 0 and increasing vortex ring radius. In the latter case, the so-called second Saffman's formula is modified. In the case of fixed R 0 , the predicted vorticity distribution for short times shows a close agreement with a Gaussian form for all b and compares favourably with available experimental data. The time evolution of the location of the region of maximal vorticity and the region in which the velocity of the fluid in the frame of reference moving with the vortex ring centroid is equal to zero is analysed. It is noted that the locations of both regions depend upon b, the latter region being always further away from the vortex axis than the first one. It is shown that the axial velocities of the fluid in the first region are always greater than the axial velocities in the second region. Both velocities depend strongly upon b. Although the radial component of velocity in both of these regions is equal to zero, the location of both of these regions changes with time. This leads to the introduction of an effective radial velocity component; the latter case depends upon b. The predictions of the model are compared with the results of experimental measurements of vortex ring parameters reported in the literature.

No abstract

The fully Lagrangian approach (FLA) to the calculation of the number density of inertial particles in dilute gas-particle flows is implemented into the CFD code ANSYS Fluent. The new version of ANSYS Fluent is applied to modelling dilute gas-particle flow around a cylinder and liquid droplets in a gasoline fuel spray. In a steady-state case, the predictions of the FLA for the flow around a cylinder and those based on the equilibrium Eulerian method (EE) are almost identical for small Stokes number, Stk, and small Reynolds number, Re, (Re = 1, Stk = 0.05). For the larger values of these numbers (Re = 10, 100; Stk = 0.1, 0.2) the FLA predicts higher values of the gradients of particle number densities in front of the cylinder compared with the ones predicted by the EE. For transient flows (Re = 200), both methods predict high values of the number densities between the regions of high vorticity and very low values in the vortex cores. For Stk ≥ 0.1 the maximal values predicted by FLA are shown to be several orders of magnitude higher than those predicted by the EE. An application of FLA to a direct injection gasoline fuel spray has focused on the calculation of the number densities of droplets. Results show good qualitative agreement between the numerical simulation and experimental observations. It is shown that small droplets with diameters dp = 2 µm tend to accumulate in the regions of trajectory intersections more readily, when compared with larger droplets (dp = 10 µm, dp = 20 µm). This leads to the prediction of the regions of high number densities of small droplets.KEY WORDS: Gas-particle flow Particle number densities Eulerian approach Fully Lagrangian approach Gasoline fuel sprays.

A phenomenological study of vortex ring-like structures in gasoline fuel sprays is presented for two types of production fuel injectors: a low-pressure, port fuel injector (PFI) and a high-pressure atomizer that injects fuel directly into an engine combustion chamber (G-DI). High-speed photography and phase Doppler anemometry (PDA) were used to study the fuel sprays. In general, each spray was seen to comprise three distinct periods: an initial, unsteady phase; a quasi-steady injection phase; and an exponential trailing phase. For both injectors, vortex ring-like structures could be clearly traced in the tail of the sprays. The location of the region of maximal vorticity of the droplet and gas mixture was used to calculate the temporal evolution of the radial and axial components of the translational velocity of the vortex ring-like structures. The radial components of this velocity remained close to zero in both cases. The experimental results were used to evaluate the robustness of previously developed models of laminar and turbulent vortex rings. The normalized time, , and normalized axial velocity, , were introduced, where tinit is the time of initial observation of vortex ring-like structures. The time dependence of on was approximated as and for the PFI and G-DI sprays respectively. The G-DI spray compared favourably with the analytical vortex ring model, predicting , in the limit of long times, where α = 3/2 in the laminar case and α = 3/4 when the effects of turbulence are taken into account. The results for the PFI spray do not seem to be compatible with the predictions of the available theoretical models.

A meshless method for modelling of 2D transient, non-isothermal, gas-droplet flows with phase transitions, based on a combination of the viscous-vortex and thermal-blob methods for the carrier phase with the Lagrangian approach for the dispersed phase, is developed. The one-way coupled, two-fluid approach is used in the analysis. The method makes it possible to avoid the 'remeshing' procedure (recalculation of flow parameters from Eulerian to Lagrangian grids) and reduces the problem to the solution of three systems of ordinary differential equations, describing the motion of viscous-vortex blobs, thermal blobs, and evaporating droplets. The gas velocity field is restored using the Biot-Savart integral. The numerical algorithm is verified against the analytical solution for a non-isothermal Lamb vortex and some asymptotic results known in the literature. The method is applied to modelling of an impulse two-phase cold jet injected into a quiescent hot gas, taking into account droplet evaporation. Various flow patterns are obtained in the calculations, depending on the initial droplet size: (i) low-inertia droplets, evaporating at a higher rate, form ring-like structures and are accumulated only behind the vortex pair; (ii) large droplets move closer to the jet axis, with their sizes remaining almost unchanged; and (iii) intermediate-size droplets are accumulated in a curved band whose ends trail in the periphery behind the head * Corresponding author Email address: O.Rybdylova@brighton.ac.uk (O. Rybdylova)Preprint submitted to International Journal of Heat and Fluid Flow October 28, 2015 of the cloud, with larger droplets being collected at the front of the two-phase region.