Elsevier

Combustion and Flame

Volume 172, October 2016, Pages 326-341
Combustion and Flame

Characterisation of two-stage ignition in diesel engine-relevant thermochemical conditions using direct numerical simulation

https://doi.org/10.1016/j.combustflame.2016.06.010Get rights and content

Abstract

With the goal of providing a more detailed fundamental understanding of ignition processes in diesel engines, this study reports analysis of a direct numerical simulation (DNS) database. In the DNS, a pseudo turbulent mixing layer of dimethyl ether (DME) at 400 K and air at 900 K is simulated at a pressure of 40 atmospheres. At these conditions, DME exhibits a two-stage ignition and resides within the negative temperature coefficient (NTC) regime of ignition delay times, similar to diesel fuel. The analysis reveals a complex ignition process with several novel features. Autoignition occurs as a distributed, two-stage event. The high-temperature stage of ignition establishes edge flames that have a hybrid premixed/autoignition flame structure similar to that previously observed for lifted laminar flames at similar thermochemical conditions. A combustion mode analysis based on key radical species illustrates the multi-stage and multi-mode nature of the ignition process and highlights the substantial modelling challenge presented by diesel combustion.

Introduction

Future diesel engines will be required to attain cost-effective reductions in pollutant emissions while improving fuel economy. These improvements will arguably be easier to attain with improved fundamental understanding of diesel engine combustion, which can be built into practically useful computational models to be used in engine design. With the goal of providing this understanding, experimental and numerical studies have been carried out.

Experiments conducted in optically accessible diesel engines and chambers have visualised ignition and flame stabilisation using high speed cameras and laser diagnostics [1], [2], [3]. The results suggest that diesel combustion is a multi-stage and multi-mode process. Observations of natural chemiluminescence, soot luminosity, and planar laser-induced fluorescence (PLIF) have been able to infer the existence of low- and high-temperature autoignition, premixed combustion, and nonpremixed combustion (diffusion flames). Such visual observations form the basis for conceptual models of diesel combustion that have been proposed for conventional [1], [2], [3] and low-temperature [3] diesel engines .

In conventional diesel engines, the fuel is injected near the end of the compression stroke and the air charge is not substantially diluted with exhaust gasses. This results in highly stratified combustion with locally very rich mixtures (having high values of mixture fraction, ξ) and high temperatures. Following ignition, a quasi-stable lifted flame develops. The lifted flame is thought to exhibit a nonpremixed (diffusion flame) region, centred on the stoichiometric mixture fraction (ξST), and a rich premixed region within the fuel jet, and, in typical ambient conditions, a region of first stage ignition (or a “cool flame”) due to low-temperature chemistry (LTC) that exists ahead of and/or inside the high-temperature chemistry (HTC) region [2], [3], [4]. The lifted flame is anchored at the most upstream location of nonpremixed combustion (the lift-off length, LOL). High levels of NOx formation are thought to occur in the hot, near-stoichiometric diffusion flame, while high levels of soot formation are thought to occur in the hot and rich inner region. Strategies to reduce pollutant formation are therefore focused on lowering peak temperatures and increasing the degree of premixing (entrainment) prior to combustion, which is closely related to parameters such as the homogeneous ignition delay time, τ, and the LOL. The reduction of pollutants therefore requires an understanding of the physical processes that govern ignition and flame stabilisation.

Autoignition in stratified mixtures at atmospheric conditions has been extensively studied (e.g., see the review by Mastorakos [5]). A chief conclusion is that autoignition is sensitive to both ξ and the scalar dissipation rate, χ, such that autoignition first occurs at locations where the local mixture is near the homogeneous most reactive ξ value, ξMR [6], and the χ is low [6], [7], [8], [9], [10], [11], [12], [13]. Experiments of autoigniting, turbulent, stratified flows at atmospheric pressure have been performed by Markides et al. [14], [15], [16]. In those studies, the configuration of an axis-symmetric fuel jet, co-flowing with hot air was considered. A statistically stationary state was achieved, whereby the rapid succession of independent autoignition events (kernels) formed at “random spots” (in physical space) and were convected downstream without establishing a connected flame surface. A direct numerical simulation (DNS) targeting similar conditions was also conducted [17], which confirmed that the kernels formed where the local mixture was near ξMR and experiencing low χ rates. Another recent experiment of a turbulent, autoigniting jet simultaneously measured ξ and temperature [18]. The results provided further experimental evidence that ignition kernels form preferably at ξMR locations where the χ is low. For diesel engine-relevant conditions, temporally and spatially resolved experimental observations of autoignition do not currently exist.

The stabilisation mode of turbulent lifted flames is an open topic of research. At atmospheric pressure, a range of stabilisation mechanisms have been proposed based on theories of, e.g., premixed flame propagation [19], nonpremixed flamelet extinction [20], edge flame propagation [21], [22], [23], large scale turbulence [24], and autoignition [25], [26], [27], [28], [29], without a clear consensus having yet been arrived at [30], [31]. For engine-relevant conditions, the stabilisation method is even less clear due to difficulties in taking well-resolved measurements. The available conceptual models [1], [2], [3] do not explicitly state which stabilisation mode or modes are responsible for diesel flame stabilisation and there remains ambiguity as to the flame structure in the vicinity of the flame base. For example, diagrams of quasi-stable diesel flames [2], [3] are consistent with both flame stabilisation by two-stage autoignition and edge flame propagation into a region of first-stage autoignition (or a cool flame). A review study of stabilisation modes in diesel jets by Venugopal and Abraham [32] concluded that multiple stabilisation modes (with contradictory assumptions) could each partially explain the stabilisation behaviour observed in experiments. This indicates that the underlying physical behaviour of diesel combustion is not well understood.

A major impediment to modelling diesel combustion is the limited fidelity of experimental observation. Existing techniques are limited to qualitative descriptions of the flame structure and global observables such as LOL and τ, but additional information is required when designing and evaluating combustion models, particularly when comparing incongruous models. Additional well-resolved observations at diesel engine-relevant conditions are therefore required.

An alternative approach is to conduct DNS which serve as “numerical experiments” that fully resolve all spatial and temporal scales. The main limitation to DNS is the extreme computational cost, which is highly sensitive to the range of scales in the target simulation and the complexity of the underlying physics. Diesel combustion exhibits: multi-phase behaviour, thousands of chemical reactions, and an extremely large range of scales, which makes an engine-level application of DNS infeasible. However, by making appropriate simplifying assumptions, idealised DNS may be conducted that complement physical experiments and provide additional insight into diesel combustion. The growth of high performance computing resources has enabled several DNS investigations of diesel engine-relevant combustion at idealised conditions [33], [34], [35], [36], [37], [38], [39], [40], [41], the main findings of which are reviewed here.

Sreedhara and Lakshmisha [33] conducted a three-dimensional (3D) DNS study of n-heptane ignition in decaying isotropic turbulence. Ensemble-averaged results showed that high-temperature ignition occurred over a range of rich ξ values (approximately corresponding to ξMR) experiencing low χ. That result was consistent with prior ignition studies conducted at non-engine-relevant conditions [5].

DNS studies of the ignition of laminar [34] and turbulent [35], [36] two-dimensional (2D) mixing layers were conducted by Mukhopadhyay and Abraham. The effect of mixing layer thickness (σ) on τ was studied in the laminar cases and it was observed that at low values of σ (highly stratified, corresponding to high χ values), σ and τ were negatively correlated. The turbulent ignition cases observed that both the first- and second-stages of autoignition were delayed by high values of χ [36] and the second-stage of ignition influenced the χ field due to enhanced diffusion at the leading edge of the ignition (which promoted χ) and reduced compositional gradients behind the ignition front due to heat release (which lowered χ) [35].

A DNS study of the autoignition of a sparse spray of n-heptane at a pressure of 24 bar was performed by Borghesi et al. [37]. The visualisation of the temperature field showed that the autoignition occurred in a “spotty” pattern. Statistics conditioned upon ξ and χ also revealed a two-staged ignition where the second-stage of ignition was negatively correlated with χ. The high-temperature ignition was also observed to occur first in mixtures slightly richer than ξMR.

Several DNS studies targeting diesel condition and using dimethyl ether (DME) fuel have recently been performed [38], [39], [40], [41]. DME is an attractive diesel surrogate fuel as it exhibits a similar two-stage ignition and negative temperature coefficient (NTC) regime of ignition delay times also observed for diesel fuels [42]. DME is also a low-sooting, potentially renewable fuel with a high cetane number, and a potential alternative to diesel fuel in real engines [43]. DME can also be modelled with compact chemical mechanisms [44], which is attractive for DNS.

A study of laminar lifted flame stabilisation at diesel engine-relevant conditions using a 30 species reduced chemical mechanism for DME was conducted by the present authors [38]. That study considered the parametric variation of oxidiser temperature over a range which spanned the NTC regime. The study identified that hybrid edge flame/autoignition structures and stabilisation modes may exist at diesel-relevant thermochemical conditions and that a gradual transition from primarily edge flame propagation to primarily autoignition stabilisation was observed with increasing oxidiser temperature. For the case of 900 K oxidiser temperature, a main tribrachial (triple) flame was observed with an additional, fourth branch upstream of the stabilisation location due to LTC. The four-branched edge flame was termed a tetrabrachial flame and it broadly resembled the flame structure presented in conceptual models of conventional diesel combustion. This result suggested that a combination of both autoignition and edge flame propagation could contribute to diesel flame stabilisation.

Subsequent DNS at similar conditions by Deng et al. [39], [40] also observed the hybrid flame structures and a stabilisation mode transition from edge flame propagation to autoignition with increasing oxidiser temperature for a fixed inlet velocity [39] or increasing inlet velocity for a fixed oxidiser temperature [40]. Chemical explosive mode analysis (CEMA) was used which provided further evidence in support of the findings from the previous study [38].

A 2D turbulent mixing layer DNS of ignition [41] was recently conducted by the present authors at identical thermochemical conditions to the lifted laminar flame case with an oxidiser temperature of 900 K [38]. The focus of the mixing layer study was on the behaviour of LTC and its interactions with the high-temperature ignition. It was found that the LTC develops initially as a first-stage autoignition in lean mixtures (low ξ values), approximately equal to the first stage, most reactive mixture fraction (ξMR, 1) conditioned upon low χ values. The LTC then transitioned to a diffusively supported “cool flame” which propagated into richer ξ. The cool flame propagation behaved as a low-temperature deflagration, not an ignition front due to gradients in the first stage ignition delay time (τ1). The cool flame shortened the onset of both LTC and HTC reactions preferentially at richer mixtures, causing high-temperature ignition kernels to develop in mixtures much richer than the most reactive mixture fraction value (ξMR) as calculated in a homogeneous reactor. The results also identified additional features of interest that were not analysed in Ref. [41], such as a transition from discrete autoignition “kernels” to edge flame formation and propagation. The presence of LTC and HTC reactions, and autoignition and edge flame propagation, produced a complex overall ignition.

A major finding from the recent DNS studies using detailed DME chemistry [38], [39], [40], [41] is that diesel-relevant combustion can involve a large number of combustion modes. For LTC, autoignition and deflagrations (cool flames) are observed; for HTC, autoignition, rich premixed, lean premixed, and nonpremixed (diffusion flames) are observed. This potentially leads to very complex ignition and stabilisation processes. As preliminarily identified [41], all of these combustion modes may occur in a single ignition case, and these modes may overlap in time. The details of this process may be important for understanding and modelling diesel combustion, and it therefore merits further consideration.

The main objective of this study is to expand upon the analysis presented in Ref. [41], which was primarily concerned with the LTC behaviour, using the same dataset (To reviewers: Ref. [41] is included in the supplementary materials to this submission). In the present work, the overall ignition process is considered, including the HTC features of ignition kernel and edge flame evolution. The following specific questions will be addressed:

  • What is the qualitative behaviour of the LTC and HTC modes during ignition?

  • How do the statistics of LTC and HTC evolve in ξ space and what is the effect of conditional fluctuations of χ?

  • How do the autoignition kernels transition to propagating edge flames?

  • Are the edge flames similar to those observed in lifted, laminar studies [38], [39], [40]?

  • How does the edge flame propagation speed respond to fluctuations in χ and the upstream chemical state?

  • What combustion modes are present and how do they evolve in time?

The results are presented in six parts. Firstly, the overall ignition process is characterised in terms of the qualitative and statistical behaviour in Sections 3.1 and 3.2, respectively, expanding upon the cursory overview provided in Ref. [41]. For orientation, a brief description of the LTC and its effect on the high-temperature ignition follows in Section 3.3, which restates the chief findings presented in Ref. [41]. The growth of the autoignition kernels and the formation of edge flames is detailed in Section 3.4. In Section 3.5, the propagation speeds of edge flames are measured in terms of their displacement speeds, and correlations are presented with respect to local mixing rates and chemical progress ahead of the edge flames. Finally, in Section 3.6, a method of combustion mode analysis based upon key species mass fraction thresholds is proposed, which discriminates between various modes of combustion. The combustion mode analysis unifies many observations from the prior results and comparisons are made with conceptual models of diesel combustion.

Section snippets

Configuration

The numerical method and the configuration of the computational domain are identical to that presented in Ref. [41], so only a brief overview is provided here for orientation.

The objective in setting up the DNS was to represent the ignition of a fuel-stratified ignition in diesel-relevant thermochemical conditions. As already mentioned, DME was selected for the fuel due to the availability of a short chemical mechanism and its reproduction of the essential features of diesel autoignition,

Qualitative description

A qualitative description of the overall ignition process is now provided. To this end, Figs. 1 and 2 show instantaneous images for a fixed window of the domain (1.2 mm × 1.2 mm in size). The domain sample shown is representative of the entire domain. As already mentioned, the DNS was introduced in Ref. [41], where an overview was presented in terms of a time-series of heat release rates. In this section, the description is expanded to several additional variables, in order to gain a more

Conclusions

A direct numerical simulation (DNS) of diesel engine-relevant ignition was conducted under idealised conditions. The computational domain was a two-dimensional mixing layer of dimethyl ether (DME) and air with a spectrum of isotropic pseudo turbulence imposed as an initial condition in order to match the Damkhöler number estimated at the flame base of a diesel jet. The thermochemical conditions were selected in order to approximate a diesel-relevant environment.

The ignition occurred as a

Acknowledgements

This work was supported by the Australian Research Council. The work at Sandia National Laboratories was supported by the Combustion Energy Frontier Research Center, an Energy Frontier Research Center funded by the US Department of Energy (DOE), Office of Science, Office of Basic Energy Sciences under Award no. DE-SC0001198. Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy under contract DE-AC04-94AL85000.

References (68)

  • C.N. Markides et al.

    An experimental study of hydrogen autoignition in a turbulent co-flow of heated air

    Proc. Combust. Inst.

    (2005)
  • C.N. Markides et al.

    Measurements and simulations of mixing and autoignition of an n-heptane plume in a turbulent flow of heated air

    Exp. Thermal Fluid Sci.

    (2007)
  • L. Vanquickenborne et al.

    The stabilization mechanism of lifted diffusion flames

    Combust. Flame

    (1966)
  • C. Muller et al.

    Partially premixed turbulent flame propagation in jet flames

    Symp. (Int.) Combust.

    (1994)
  • L. Muñiz et al.

    Instantaneous flame-stabilization velocities in lifted-jet diffusion flames

    Combust. Flame

    (1997)
  • J. Buckmaster

    Edge-flames

    Prog. Energy Combust. Sci.

    (2002)
  • J. Broadwell et al.

    Blowout of turbulent diffusion flames

    Symp. (Int.) Combust.

    (1985)
  • YooC. et al.

    A DNS study on the stabilization mechanism of a turbulent lifted ethylene jet flame in highly-heated coflow

    Proc. Combust. Inst.

    (2011)
  • LuoZ. et al.

    Chemical explosive mode analysis for a turbulent lifted ethylene jet flame in highly-heated coflow

    Combust. Flame

    (2012)
  • C. Lawn

    Lifted flames on fuel jets in co-flowing air

    Prog. Energy Combust. Sci.

    (2009)
  • S. Sreedhara et al.

    Autoignition in a non-premixed medium: DNS studies on the effects of three-dimensional turbulence

    Proc. Combust. Inst.

    (2002)
  • S. Mukhopadhyay et al.

    Influence of compositional stratification on autoignition in n-heptane/air mixtures

    Combust. Flame

    (2011)
  • S. Mukhopadhyay et al.

    Influence of heat release and turbulence on scalar dissipation rate in autoigniting n-heptane/air mixtures

    Combust. Flame

    (2012)
  • G. Borghesi et al.

    Complex chemistry DNS of n-heptane spray autoignition at high pressure and intermediate temperature conditions

    Combust. Flame

    (2013)
  • A. Krisman et al.

    Polybrachial structures in dimethyl ether edge-flames at negative temperature coefficient conditions

    Proc. Combust. Inst.

    (2015)
  • DengS. et al.

    Autoignition-affected stabilization of laminar nonpremixed DME/air coflow flames

    Combust. Flame

    (2015)
  • DengS. et al.

    Stabilization of laminar nonpremixed dme/air coflow flames at elevated temperatures and pressures

    Combust. Flame

    (2015)
  • R. Sankaran et al.

    Direct numerical simulations of turbulent lean premixed combustion

    J. Phys. Conf. Ser.

    (2006)
  • WangY. et al.

    Direct numerical simulation of ignition in turbulent n-heptane liquid-fuel spray jets

    Combust. Flame

    (2007)
  • ChenJ. et al.

    Terascale direct numerical simulations of turbulent combustion using S3D

    Comp. Sci. Discov.

    (2009)
  • E.R. Hawkes et al.

    Estimates of the three-dimensional flame surface density and every term in its transport equation from two-dimensional measurements

    Proc. Combust. Inst.

    (2011)
  • O. Chatakonda et al.

    On the fractal characteristics of low Damkohler number flames

    Combust. Flame

    (2013)
  • M. Talei et al.

    Ignition in compositionally and thermally stratified n-heptane/air mixtures: a direct numerical simulation study

    Proc. Combust. Inst.

    (2015)
  • P. Domingo et al.

    Triple flames and partially premixed combustion in autoignition of non-premixed turbulent mixtures

    Symp. (Int.) Combust.

    (1996)
  • Cited by (0)

    View full text