LES for Internal Combustion Engine Flows [LES4ICE]

IFPEN / Rueil-Malmaison - 11-12 December 2018



Further improving the environmental performances of internal combustion engines (ICE) increasingly requires moving beyond traditional design based on a cycle averaged approach (RANS), and to reliably predict and control individual engine cycles under realistic operating conditions. Large-Eddy Simulation (LES) offers this unique potential and opens up new avenues for extending the scope of application of CFD for ICEs,

Since its 1st edition in 2008, the biannual LES4ICE conference provides a forum for exchange concerning research and development of LES and related experimental techniques for their application to ICE flows.

It brings together researchers and engineers working in the field of piston engine combustion to debate the state of the art in LES applied to ICEs and examine advanced experimental techniques capable of supporting and validating its development.

Published research in recent years has demonstrated the ability of LES to yield an unprecedented detailed insight into non- cyclic characteristics of flow and combustion in ICE. In particular, its application to spark-ignition research engines was shown to reproduce experimental findings on cyclic variability of intake flow, its interaction with direct fuel injection, and on the resulting cyclic combustion variability (CCV). First attempts also concerned the exploitation of such advanced simulations in order to identify the sources of flow and combustion variability in an effort to limit them by design in the future. Recent work also showed the potential of LES to provide an unprecedented insight into the link between CCV and knock. LES not only allowed a quantitative prediction of knock intensity and limits, but also yielded a detailed insight into the phenomena at stake and in particular on destructive knocking modes related to a coupling between auto-ignition and acoustic waves inside the cylinder.

Despite less prominent, the development and application of LES to study Diesel spray combustion has also received increasing attention, in relation to the important collaborative research effort undertaken in the frame of the Engine Combustion Network (ECN).

Despite the present and foreseeable progress in terms of supercomputer performance, LES meshes compatible with a practical usage will still be far from resolving all the relevant space and time scales of ICE flows. Reliable LES predictions in a realistic time frame thus still require the availability of sub-grid scale models able to accurately reproduce the effects of unresolved scales. Although it can be considered that existing models already allow addressing many of the phenomena at stake, further research is required to increase the reliability and domain of application of LES methods.

Published research indicates that sub-grid scale models for turbulence alleviating the need for an a priori choice of model constants proved efficient to predict the complex internal aerodynamics during the full engine cycle. However, the accurate modelling of unresolved near-wall flow still requires dedicated research work aimed at ensuring an accurate prediction of wall friction, heat losses and turbulence generation, at a cost compatible with a practical usage.

Widely used Discrete Particle Methods are reported to yield satisfactory predictions of fuel sprays under engine condition, as even coarse meshes allow resolving a part of the generated flow entrainment and resulting convective mixing in LES. Eulerian/Eulerian approaches are essentially used in LES of the flow inside injectors or near the nozzle exit. Such detailed in-nozzle flow LES could in particular allow an accurate imposition of unsteady injector outflow conditions necessary to yield accurate LES of the fuel spray. On-going research concerns all aspects of high pressure fuel injection, the coupling between different approaches used in specific flow regions, the modelling of super-critical thermodynamic conditions, or the formation of liquid wall films and related pool fires.

In terms of engine combustion, published models for spark ignition, premixed turbulent flame propagation based on flamelet approaches, and inexpensive pre-tabulated chemistry approaches to fresh gases’ auto-ignition were shown to allow addressing key phenomena. On-going research topics concern further improvements in the modelling of the early phases of spark ignition of importance for predicting minimum ignition energy or misfires, turbulent combustion models valid outside the flamelet regime in order to address highly diluted and leaned-out combustion and increased turbulence levels, or the formulation of advanced turbulence-chemistry interaction models that combine accurate turbulent combustion models with detailed auto-ignition or pollutant chemistry.

LES attracts increasing interest from the automotive industry in relation to its potential of increased predictivity and of extending the domain of application of CFD to non-cyclic flow and combustion phenomena not yet addressed in early design phases. /In this context research work must address methodological aspects aimed at ensuring reliable and accurate LES results without a priori experimental knowledge. Another key aspect is the development of  numerical methods allowing to reduce related pre-processing and return times in order to make them compatible with an industrial usage.

There also is a need for methods allowing extracting meaningful information from the important amount of data generated in LES, but also in engine experiments. This is essential to be able to efficiently exploit such complex databases in order to gain a better understanding of phenomena at stake in non-cyclic engine combustion and to support the formulation of reduced order models to be used in industrial design processes.

Finally, a successful LES research is strongly dependent on the availability of dedicated high-resolution quantitative experimental techniques, and on their application to detailed studies of engine flow and combustion under realistic operating conditions. Such experimental research is not only necessary to yield validation data, published recent research also exemplified how a combined usage of LES and advanced diagnostics could truly yield an unprecedented detailed insight into presently poorly understood and mastered engine phenomena.

LES4ICE aims at proposing its participants the unique opportunity to keep up with the relevant worldwide research in these fields.


LES applications to ICE

  • LES for predicting & understanding non-cyclic engine phenomena: cyclic combustion variability, fast transients, extreme cycles, rare events, …
  • Predicting and characterising abnormal combustion (knock, superknock) with LES
  • Detailed LES studies of interactions between intake aerodynamics, fuel injection and combustion
  • LES of in-nozzle injector flows and its link to fuel sprays
  • High-fidelity LES of mixture preparation and combustion phenomena


Experiments for LES

  • Combined usage of advanced diagnostics and LES for yielding a better understand and mastering of  ICE flows
  • Methods allowing to extract meaningful information from large experimental and LES databases
  • Experimental techniques with a potential for supporting the development of LES
  • Experimental databases for validating LES (simplified geometries and engines)


LES methodology

  • Numerical methods adapted for LES
  • Quality criteria for LES of ICE flows
  • Mesh convergence studies of LES for ICE
  • Comparing & validating LES with experimental evidence
  • UQ adapted for LES of ICE


LES models for ICE flows

  • Sub-grid scale turbulence
  • Accounting for turbulent wall boundary layers
  • Fuel injection & fuel spray modelling
  • Turbulence-chemistry interaction
  • Accounting for detailed chemistry in LES