Assessment of elliptic flame front propagation characteristics of iso-octane, gasoline, M85 and E85 in an optical engine

Abstract

Premixed fuel–air flame propagation is investigated in a single-cylinder, spark-ignited, four-stroke optical test engine using high-speed imaging. Circles and ellipses are fitted onto image projections of visible light emitted by the flames. The images are subsequently analysed to statistically evaluate: flame area; flame speed; centroid; perimeter; and various flame-shape descriptors. Results are presented for gasoline, isooctane, E85 and M85. The experiments were conducted at stoichiometric conditions for each fuel, at two engine speeds of 1200 rpm (rpm) and 1500 rpm, which are at 40% and 50% of rated engine speed. Furthermore, different fuel and speed sets were investigated under two compression ratios (CR: 5.00 and 8.14). Statistical tools were used to analyse the large number of data obtained, and it was found that flame speed distribution showed agreement with the normal distribution. Comparison of results assuming spherical and non-isotropic propagation of flames indicate non-isotropic flame propagation should be considered for the description of in-cylinder processes with higher accuracy. The high temporal resolution of the sequence of images allowed observation of the spark-ignition delay process. The results indicate that gasoline and isooctane have somewhat similar flame propagation behaviour. Additional differences between these fuels and E85 and M85 were also recorded and identified.

Introduction

The current issues with our hydrocarbon based economy and its effects on climate change and human life are well documented (for instance [1]). These environmental and socio-political issues are among the most motivating research drivers, providing impetus for research in renewable energy and design-to-specification fuels [2], [3], [4], [5]. Nevertheless, developed as well as developing countries still rely to a great extent on conventional fuels powering conventional engines. There is still a lot of room for considerable improvement in understanding the chemical reaction and flame-propagation processes, and reducing the emissions of these engine-fuel combinations. One of the most important ways to analyse combustion processes in engines is to employ 3D-CFD codes, with incorporation of various well refined fuel oxidation and flame propagation mechanisms [6], [7]. The models and codes need validation with experimental work accurately describing the exact nature of these in-cylinder processes.

Although, flame is defined as the luminous part of the burning gases caused by highly exothermic, rapid oxidation [8]. For simplicity in this study, the earliest and relatively short plasma state of the glowing charge was also considered as a flame. For both moving and standing flames, the flame front is the indicator of where gases heat up and start emitting light [9], [10]. This front is considered to consist of two regions: preheat and reaction zones. For instance, Fig. 1 illustrates the top view of the reaction and preheat zones in the chamber of the optical-access engine used in this paper.

The combustion process in SI engines can be divided into four main stages: spark and flame initiation; initial flame kernel development; turbulent flame propagation; and flame termination [11]. The first two stages are of high importance in terms of in-cylinder pressure development [12], [13], [14], [15], [16]. These four stages are influenced by: spark energy and duration [17]; spark plug design and orientation [18]; in-cylinder flow field [19]; cyclic cylinder charging [20]; in-cylinder composition [21]; and other related factors. A detailed literature survey on the effects of these parameters on the four stages of combustion appeared in [12].

The flame speed $S_n$ (which can be measured from images of the spatial–temporal development of the flame) is given by [9], [22]:

$$S_n=v_g+u_n$$

where $v_g$ is the gas expansion velocity immediately adjacent to the flame front and $u_n$ is the stretched laminar burning velocity of combusting air fuel mixture [23]. The turbulent burning velocity equals the laminar velocity with the added effect of the flow field, geometry; wrinkling of the flame front; pressure effects on flame thickness; history of the flame [24]. The effect of the turbulent flow field is crucial for the first and second stage of combustion. It has been shown that the smallest flame kernels are distorted shortly after ignition [25]. The laminar velocity is an intrinsic property of a combustible fuel, air and burned gas mixture. That is defined as the velocity, relative to and normal to the flame front, with which unburned gas moves into the front and is transformed to products [26].

Turbulent burning velocity plays a prime role and directly effects the in-cylinder pressure development, i.e., engine performance. Turbulent burning velocity and laminar burning velocity are important physical properties of fuel air mixtures. It is essential that both of these velocities are derived experimentally from flame speed and in-cylinder pressure measurements [9], [14], [11], [22]. The work produced by an engine is related to the flame speed as can be inferred from the following. The burned mass of charge is given by

$$m_b(t)=(\overline{S_x}\overline{S_y}\overline{S_z})(t){\rho }_b(t)\mathit{Sf}(t)\text{,}$$

where $\overline{S_x}\text{,}\overline{S_y}\text{,}\overline{S_z}$ are the average flame speeds in the $x\text{,}y\text{,}z$ directions. These can be determined by dividing the flame radius along an axis by the elapsed time from ignition. Sf is a shape specific function. The burning of fuel releases energy to the working fluid in the cylinder, given by [26], [4]:

$$m_b\mathit{LHV}-({\mathit{mc}}_V\mathrm{d}T)-\mathrm{\Sigma }{h}_i{\mathit{dm}}_i-\mathrm{d}{Q}_{\mathit{ht}}=p\mathrm{d}V$$

The rate of burning of the air–fuel mixture affects the chemical energy change of the fluid, and this directly affects the indicated work and power output. In Eq. (3) the work done on the piston $p\mathrm{d}V$ equals the energy released from the burning fuel $m_b\mathit{LHV}$, minus the energy required to heat up the charge ${\mathit{mc}}_V\mathrm{d}T$, minus the heat transfer to walls $\mathrm{d}{Q}_{\mathit{ht}}$, and adjusted by the masses leaving or entering the chamber $\mathrm{\Sigma }{h}_i\mathrm{d}{m}_i$. Note: term $\mathrm{\Sigma }{h}_i\mathrm{d}{m}_i$ can be positive (during fuel injection) or negative (flow to crevice volumes or blow by). Therefore engine performance is highly dependent on flame propagation characteristics within the cylinder.

In previous engine research images of flames in cylinders showed a significant enflamed volume, but the pressure measurements were not accurate or sensitive enough to indicate the evolving flame kernels [21], [15]. Therefore, optical investigation of combustion is preferred to pressure tracing at the early combustion stages. The practical realisation of visual access to a combustion chamber of a working piston engine is not easy, with any of the visible, ultra violet spectra or laser radiation approaches [43], [44], [45], [46]. The fluctuating pressure at high temperature, the limited strength of transparent materials and the geometrical constrains kept investigators from studying optical engines at real working conditions. In most cases the engine speed and CR were kept low in order to observe the propagating flames. In previous investigations the effect of changing engine speeds and equivalence ratios were studied. However, because of the tight cylinder geometries, there has been no optical data recorded in the same engine at different compression ratios. Another major difficulty is the time scale of rapid oxidation. The average of temporal resolution that can be found in the literature is about 0.2–0.4 ms. Only one paper included data at higher temporal resolution, which could potentially provide insight to the earliest and faintest flames [32]. It has also been reported that fouling of the optical ports limits the length of operation time [18]. The experimental conditions and a general summary of the most relevant work on flame speed measurements and other investigations in optical engines can be found in Table 1. A comprehensive review of experimental investigation techniques in reciprocating-piston engines is in [47].

It has been shown that the shape of the evolving flame kernel has a major effect on the in-cylinder combustion processes [14]. Generally, previous studies have assumed that the propagation of the spark-initiated oxidation is isotropic, i.e. spherical flame propagation [15], [16], [18], [21], [25], [28], [29], [39], [38], [30], [31], [32], [33], [34], [35], [36]. Only a few studies mentioned different flame-front geometries [14], [15], [30], [39] and looked into implications arising from the assumption that the flame front surface had a spherical geometry. However, these shapes were not described mathematically and detailed analyses were not carried out.

Even though the in-cylinder flame front is a three-dimensional flame, in most studies flame-speed measurements are measured from two-dimensional projections of the images. Applying the isotropic propagation assumption, the two-dimensional projected contours of spherical flames can be digitized and their various geometrical properties determined. Actual flame speeds and flame shapes were measured in a small number of studies, where the flame radii were calculated using the “equivalent radius” (EQR) method [39], [25], [15], [38], [16], [32], [33], [40], [41], which determines the radius from the measured area:

$$r=\sqrt{\frac{A}{\pi }}$$

where r is the flame radius and A is the area of the projected region. There has been no attempt to refine this assumption.

Many of the early investigators (that established the fundamentals of optical engine work) due to limitations of available tools used hand tracing methods to delineate the boundaries and/or had low number of samples (3–6 measurements averaged) [19]. Later papers do have a larger number of measurements, but statistical distribution of their findings was not documented [37]. Cyclic variability in engines is a widely studied phenomenon [48], [27], [18], [12]: the nature of the processes prior to ignition, the ignition itself and combustion instabilities cause fairly high standard deviation of in-cylinder measurements. Therefore statistical tools and high numbers of samples are needed in order to keep errors in the results low.

In previous studies with optical-access engines the main choice of fuels were pure hydrocarbons (HC), such as propane and isooctane. Less attention was paid to practical fuels such as gasoline and alcohol blends (Table 1). It is a usual practice to use isooctane as a surrogate of gasoline in engine related research purposes as these two fuels have similar physical properties. Moreover, gasoline is a mixture of hydrocarbons with a composition that is not guaranteed, whereas isooctane is an easily available pure chemical. Previous flame-propagation studies in optical engines did not compare flame propagation characteristics of gasoline and isooctane to verify the two fuels behaved in a similar fashion. Alcohols and blends with gasoline (or isooctane) have been used in piston engines since the engine itself was invented. At present, bio-alcohols are proposed among the candidates for future fuels. Many studies have investigated their emission and performance qualities [49], [50], [51], [52], [40], but the literature is lacking the relevant optical-engine data. Usually each published study concentrates on one engine geometry (e.g. one compression ratio) and one fuel. There are very few optical data available on comparison of different fuels in the same engine operating conditions. Table 2 lists some of properties of the fuels tested in these engines (from [26], [53]).

The main contribution of this paper is statistical characterisation of non-spherical and non-isotropic aspects of flame propagation. A specifically-designed multi-fuel optical engine was used to compare flame-propagation characteristics of isooctane and gasoline. E85 and M85 were also investigated as practical alternative spark-ignition engine fuels and to fill in the gaps in the flame-propagation data base. E85 and M85 were “research grade”: they were mixed in house using pure alcohols and isooctane. CR of 8.14:1 and 5:1 were chosen to test the fuels: the higher to simulate real engine conditions; and the lower one to provide sufficient contrast from the higher one. Utilising the capability of an extremely sensitive and fast camera, high temporal resolution was achieved, allowing investigation of phenomena like ignition delay and early flame kernel formation. The large number of samples allowed mathematical statistics to be used to find the typical distribution of the measured data. It was concluded that elliptical flame structures describe flame propagation more accurately than spherical flame structures in many cases. Therefore a new and more detailed set of combustion data with these fuels has been obtained, and it can be used for validation of CFD and emissions studies.