Numerical studies for charge formation in combustion engines

Karri Keskinen

Research output: ThesisDoctoral ThesisCollection of Articles


Modern lean and low-temperature combustion (LTC) techniques form a pathway towards increased efficiency and reduced emissions in internal combustion engines. However, combustion abnormalities (associated with efficiency detriments and unburned hydrocarbon emissions) constitute a major engineering challenge augmented by the inherent cyclic variation in engines. Fuel-air charge combustibility is highly dependent on local in-cylinder metrics such as flow turbulence, mixture composition and temperature, for which experimental investigations are difficult. Thereby, it would be highly beneficial to develop fast and accurate computational tools capable of reliably describing charge formation phenomena such as turbulent mixing, wall heat transfer and thermal stratification. The primary aim of this dissertation is to enhance understanding and predictive accuracy of scale-resolving wall-bounded simulations pertinent to charge formation, hoping to facilitate improved comprehension and mitigation of combustion abnormalities in future studies. As a secondary objective, mixture formation trends are assessed in the context of gas direct injection, a modern fuel supply technique for lean charges.  The present computational fluid dynamics (CFD) studies include statistical (Reynolds-averaged simulation; RANS) and filtered (large eddy simulation; LES) turbulence modelling approaches. Model-centric near-wall approaches are emphasised due to their reduced computational load compared to direct numerical simulation (DNS) and wall-resolved LES. In particular, implementation and validation is carried out for a zonal hybrid LES/RANS method with a specific near-wall treatment (HLR-WT). Simplified engine setups and academic flow configurations are employed as test cases, with reference data including (1) measurements and DNS of academic and engine-like flows, in addition to (2) planar laser induced fluorescence (PLIF) imaging and high-resolution LES of gas jets.  Direct gas injection investigations highlight both independent and combined effects of injection pressure, timing and nozzle type. In scale-resolving simulations, hybrid LES/RANS methods provide varying improvements on coarse-grid LES, while methodology-specific characteristics should be acknowledged in practical utilisation. The novel methodological combination HLR-WT displays enhanced grid flexibility in academic configurations and promising results in engine-like flows. In particular, turbulent heat transfer characteristics, near-wall scaling and thermal stratification trends are relatively accurately reproduced in the highly demanding compression stroke with mild grid sensitivity. This is a positive indication of applicability in engineering-relevant, higher Reynolds number configurations.
Translated title of the contributionNumeerisia tutkimuksia polttomoottoreiden seoksenmuodostusta varten
Original languageEnglish
QualificationDoctor's degree
Awarding Institution
  • Aalto University
  • Larmi, Martti, Supervising Professor
  • Kaario, Ossi, Thesis Advisor
  • Vuorinen, Ville, Thesis Advisor
Print ISBNs978-952-60-8127-4
Electronic ISBNs978-952-60-8128-1
Publication statusPublished - 2018
MoE publication typeG5 Doctoral dissertation (article)


  • internal combustion engines
  • gas direct injection
  • hybrid LES/RANS
  • wall modelling
  • wall heat transfer


Dive into the research topics of 'Numerical studies for charge formation in combustion engines'. Together they form a unique fingerprint.

Cite this