Search

Existing methods for predicting heat fluxes and temperatures in internal combustion engines, which take the form of correlations to estimate the heat transfer coefficient on the gas-side of the combustion chamber, are based on methodology developed over the past 50 years, often updated in view of more recent experimental data. The application of these methods to modern diesels engines is questionable because key technologies found in current engines did not exist or were not widely used when those methods were developed. Examples of such technologies include: high-pressure common rail and variable fuel injection strategies including retarded injection for nitrogen oxides emission control; exhaust gas re-circulation; high levels of intake boost pressure provided by a single- or double-stage turbocharger and inter-cooling; multiple valves per cylinder and lower swirl; and advanced engine management systems. This suggests a need for improved predicting tools of thermal conditions, specifically temperature and heat flux profiles in the engine block and cylinder head. In this paper a modified correlation to predict the gas-side heat transfer coefficient in diesel engines is presented. The equation proposed is a simple relationship between Nu and Re calibrated to predict the instantaneous spatially averaged heat transfer coefficient at several operating conditions using air as gas in the model. It was derived from the analysis of experimental data obtained in a modern diesel engine and is supported by a research methodology comprising the application of thermodynamic principles and fundamental equations of heat transfer. The results showed that the new correlation adequately predicted the instantaneous coefficient throughout the operating cycle of a high-speed diesel engine. It also estimated the corresponding cycle-averaged heat transfer coefficient within 10 per cent of the experimental value for the operating conditions considered in the analysis.

Experimental data from internal combustion (IC) engines suggests that the use of proprietary computational fluid dynamics (CFD) codes for the prediction of coolant-side heat transfer within IC engine coolant jackets often results in underprediction of the convective heat transfer coefficient. An experimental and computational study, based on a coolant gallery simulator rig designed specifically to reproduce realistic IC engine operating conditions, has been conducted to explore this issue. It is shown that the standard ‘wall function’ approach normally used in CFD models to model near-wall conditions does not adequately represent some features of the flow that are relevant in convective heat transfer. Alternative modelling approaches are explored to account for these shortcomings and an empirical approach is shown to be successful; however, the methodology is not easily transferable to other situations.

Over the last 40 years, several empirical correlations have been developed to estimate heat fluxes from the combustion chambers of internal combustion engines. Some of these expressions are based on correlations to compute the Nusselt number for forced convection in turbulent flow inside circular tubes. The fundamental suitability of this kind of empirical model in representing the highly complex processes of in-cylinder heat transfer is questionable, but in practice the models have steadily improved owing to contributions from numerous investigators. Other correlations have a less theoretical basis than those of the Nusselt number form. Formulae of this type have been obtained from the application of simple statistical techniques to large datasets, taking into account several engine operational parameters and engine types. The resulting correlations provide reasonable estimates but perform poorly when extrapolated or applied to novel concepts. In this paper, the most important correlations are reviewed against the features of a modern diesel engine, and research requirements for future modelling developments are identified and discussed.

A detailed programme of work has been undertaken to quantify the suitability of predictive methods for accurate determination of the levels of boiling heat transfer within an internal combustion (IC) engine cooling gallery simulator. An extensive array of experimental data has been obtained as the basis for the predictive validation. Working on the principle of superposition, the convective component of heat transfer has been represented by the established Dittus-Boelter correlation which has been extensively modified to account for developing boundary layers, surface roughness and nearwall viscous effects. The boiling component has been represented by the Chen model, modified for binary fluids and subcooling. For the IC engine cooling application it is concluded that the application of the Chen approach must be complemented by a convective heat transfer model that accurately represents the complex thermo-fluid situation being experienced within a developing flow.

Simple heat transfer correlations are known to underpredict the single-phase convective heat transfer coefficient when applied to internal combustion (IC) engine cooling passages. The reasons for such underprediction were investigated using a specially designed test rig which was operated under a wide variety of test conditions relevant to IC engine operation. Data from this rig study identified that undeveloped flow (fluid dynamically and thermally), surface roughness and fluid viscosity variation with temperature were the physical reasons responsible for the mismatch. Simple empirical heat transfer models have subsequently been extended to take account of these factors and are shown to give much improved correlation with rig data, and data from an engine study. The implications of this work for predicting engine heat transfer in a three-dimensional computational fluid dynamics environment are discussed.