Search Results

A detailed analysis is performed upon the results of CFD combustion simulations of several diesel fuel spray flame experiments. Simulations are validated against measurements from a constant volume combustion chamber testcase [9]. Particular emphasis is made in the analyses to identify mechanisms associated with the ‘lift-off’ phenomena characteristic of contemporary high injection pressure diesel engine combustion. A recently developed industry state of the art RANS hybrid combustion model (Extended Coherent Flame Model - 3 Zones) [41] is used which takes account of both a propagating (premixed) flame combustion mode as well as the conventionally assumed diffusion flame mode used in most diesel combustion models. The location of and development of a propagating reaction front, obtained from analysis of the progress variable within the model, is studied in relation to the lift-off behaviour.

This paper uses CFD methodology to simulate a prototype Diesel engine operating at high peak pressures (HPP). Under these conditions the accurate estimation of the level of thermomechanical stress on metal components is crucial for the design process. CFD simulations have been performed of flow, combustion and heat transfer to provide detailed insight into the in-cylinder behaviour of the engine. Particular emphasis was put on improving wall heat transfer predictions which have been compared with detailed local time-resolved surface heat transfer measurements. It is demonstrated that heat transfer strongly depends on flame spread via flow field and spray-related processes. Hence local heat transfer measurements also provide a stringent testing ground for spray and combustion model performance. Additionally it is shown that widely-used empirical heat transfer correlations are incapable of estimating the critical level and nature of thermal loading.

The objective of this paper is to present a numerical simulation method for the calculation of an unsteady, one-dimensional flow and heat transfer in the branched intake manifolds of multi-cylinder engines. The method operates on the one-dimensional differential conservation equations for a variable-area duct network with friction and heat transfer at the walls. The latter processes are represented by appropriate drag and heat transfer coefficient correlations, as also are the losses which occur at junctions and other geometrical irregularities. The equations are solved by a time-marching finite-volume method, on a computational mesh in which the velocities are located between the pressures which drive them.

This paper describes a spray impingement model which is formulated on the basis of literature findings and mass, momentum and energy conservation constraints. The model involves the analysis of the relevant impingement regimes and the associated post-impingement characteristics. In order to reflect, to some extent, the stochastic nature of the impingement process, a random procedure is introduced to determine some of the quantities representing the post-impingement status. This allows secondary droplets resulting from splash to have a distribution of sizes and velocities. Numerical simulations for several test cases are presented, showing satisfactory agreement with experimental data.

A mathematical model of formation and transport of liquid films, incorporating a droplet-wall impaction model and exchange mechanisms with the gas-phase, has been developed and incorporated into the STAR-CD computational fluid dynamics code. It has been applied to a test case representation of the multi-point fuel injection in four stroke SI engines. The results indicate that the major features of droplet impaction and film development are reproduced by the model. The qualitative agreement with data in the region of spray impaction is good.

This paper presents a mathematical model for the prediction of the dynamic characteristics of wall films formed by impinging sprays. The model takes into account the impingement pressure due to bombardment of impinging droplets, tangential momentum transfer resulting from oblique droplet impingement on the film surface, and the gas shear force at the film surface. The general transport equations of mass, momentum and energy for wall film flows are established in the boundary-layer framework. It is shown that this set of equations can be substantially simplified if local equilibrium occurs and a dimensional analysis is performed to identify the conditions for the applicability of the local equilibrium model. Solution of the full film equations is obtained by an efficient hybrid integral/numerical method, which allows numerical calculations to be performed in a two-dimensional framework. An implicit finite volume scheme is employed for this purpose.

In this study three-dimensional CFD calculations of the gas motion and spray characteristics of a small (1.9l), high-speed direct-injection Diesel engine are presented and evaluated. The calculations were performed using the SPEED code, developed within the European IDEA-EFFECT project: it uses fully implicit finite volume methodology in conjunction with an unstructured mesh to represent the full complexities of the engine geometry and solve the equations governing the gas motion, fuel spray evolution and subsequent fuel/air mixing. Submodels for particular aspects of these processes developed by various partners in the project are incorporated. The accuracy of the predictions is assessed through comparisons with detailed LDA measurements of the velocity field during the induction and compression strokes up to the time of ignition, as well as with quantitative measurements of the spray penetration and local droplet velocities. Moderately good agreement is obtained.

The extension is described of an existing multidimensional method of calculating in-cylinder air motion to the representation of the injection of a liquid fuel spray. Sample calculations are presented of the droplet and gas motion and fuel-air mixing in an axisymmetric representation of an open-chamber direct-injection engine, in the absence of combustion, and are believed to be the first in which a realistic representation of the gas-phase turbulence behaviour is employed. One of the more important findings is that the spray induces velocities and turbulence levels in the gas which are comparible to, and sometimes greater than, those produced by other mechanisms such as swirl and squish. It is concluded however that considerable further work is required to make such models truly predictive and detailed experimental data is urgently required to assist this task.

A method is described of calculating the flow, temperature and turbulence fields in cylinder configurations typical of a direct-injection diesel engine. The method operates by solving numerically the Navier Stokes equations that govern the flow, together with additional equations representing the effects of turbulence. A general curvilinear-orthogonal grid that translates with the piston motion is used for the calculations in the complex-shaped piston bowl, whilst an expanding/contracting grid is used elsewhere. Predictions are presented showing the evolution of the velocity and turbulence fields during the compression and expansion phases of a motored engine cycle, for various shapes of axisymmetric piston bowl and various initial swirl levels. These results illustrate the strong influence of these factors on the TDC flow structure.

Progress is described on the development of a multi-dimensional method for the prediction of the detailed in-cylinder events in a firing D.I. Diesel engine. An existing method incorporating fluid dynamics and spray representations is extended to include a combustion model, of a kind which allows in an approximate way for both chemical-kinetic and turbulence effects on the burning rate. An example calculation is presented which demonstrates that, with appropriate adjustments to the empirical coefficients of the combustion model, the method produces qualitatively realistic predictions of the major phases of the combustion process, including ignition, premixed burning and diffusion burning. The results also serve to illustrate the usefulness of multidimensional methods in revealing the causes of inadequate performance.

Multidimensional model predictions and laser Doppler anemometer measurements are presented of the flow in a motored model engine equipped with a central shrouded valve. Although the accuracy of prediction, as assessed against the data, is at best moderate, the simulation is sufficiently close to provide valuable insight into the flow behaviour. an important finding in this regard is that the shrouded valve generates a long-lived tumbling vortex which is sustained and amplified by the compression process and in turn causes amplification of the turbulence, the TDC levels of which are more than twice those observed in similar studies with non-shrouded valves. It is concluded that inlet arrangements which produce such tumbling motions are likely to lead to enhanced flame propagation rates.

Multidimensional calculations axe presented of the behaviour of sprays injected into the combustion chambers of motored reciprocating engines, in circumstances giving rise to three-dimensional spatial variations in the droplet and gas flow fields. The calculations were performed using the implicit EPISO algorithm, extended to include a Lagrangian description of the spray. The gas-phase turbulence is represented by the κ-ε model and its effect on the droplets is modelled stochastically. Two applications of the method are reported: one involves the simulation of, and comparison with data from, published experiments on a laboratory engine fitted with a single-hole injector in the cylinder wall. The second case is a demonstration calculation for a direct-injection Diesel with a cylindrical piston bowl and a four-hole injector.

Multidimensional calculations and laser Doppler anemometry measurements are presented of the air flow in a research diesel engine motored at 900 rpm with a compression ratio of ∼8.5. The engine comprised the cylinder head of a Ford 2.5L high speed direct-injection diesel mounted on a single cylinder Fetter engine modified to provide optical access for LDA measurements in a toroidal piston-bowl. The accuracy of the predictions is assessed against ensemble-averaged velocity data and found to be sufficient to allow better understanding of the flow in production engine geometries under realistic operating conditions.

In this paper a multidimensional method is described and evaluated for the prediction of Diesel sprays. The method, which shares many features with similar approaches developed elsewhere, embodies an Eulerian description of the gas flow and a stochastic Lagrangian treatment of the spray droplets. Gas phase turbulence effects are represented by the k-ε model and their influence on the droplets is modelled stochastically, as are the processes of collision and coalescence. Comparisons are made with the spray penetration measurements in a quiescent bomb of Yule et al [1] covering a range of pressures and temperatures. Satisfactory agreement is obtained at combinations of low pressure and temperature and high pressure and temperature, but not for high-pressure, low-temperature cases. The sensitivity of the predicted penetration rates to the assumed initial droplet size distribution is relatively weak, but the calculated vaporization rates are strongly sensitive.

The results are reported of a combined experimental and computational study of steady flow through an axisymmetric valve/port assembly, the main objective of which was to assess the accuracy of the multidimensional model predictions of this flow. Measurements of the discharge coefficient, mean velocity and the turbulent Reynolds stress fields were obtained by hot-wire anemometry at various valve lifts. These were supplemented by flow visualisation studies. Predictions were made using a finite-volume method employing a body-fitted computational mesh and the k-ε turbulence model. Good agreement was found at low lifts, but at higher values this deteriorated due to the inability of the turbulence model to provoke the flow separations which occurred in the experiments. The conclusion is that for both idealised and practical ports multidimensional predictions will be of limited accuracy until better turbulence models become available.

Multi-dimensional modelling of the flow and combustion promises to become a useful optimisation tool for IC engine design. Currently, the total simulation time for an engine cycle is measured in weeks to months, thus preventing the routine use of CFD in the design process. Here, we shall describe three tools aimed at reducing the simulation time to less than a week. The rapid template-based mesher produces the computational mesh within 1-2 days. The parallel flow solver STAR-CD performs the flow simulation on a similar time-scale. The package is completed with COVISEMP, a parallel post-processor which allows real-time interaction with the data.