An experimental study of the detailed flame transport in a SI engine using simultaneous dual-plane OH-LIF and stereoscopic PIV

Abstract

Understanding the detailed flame transport in IC engines is important to accurately predict ignition and combustion phasing, rate of heat release and assess engine performance. This is particularly important for RANS and LES engine simulations, which often struggle to accurately predict flame propagation and heat release without first adjusting model parameters. Detailed measurements of flame transport in technical systems are required to guide model development and validation.

This work introduces an experimental dataset designed to study the detailed flame transport and flame/flow dynamics for spark-ignition engines. Simultaneous dual-plane OH-LIF and stereoscopic PIV are used to acquire 3D measurements of unburnt gas velocity (${\stackrel{\rightharpoonup }{U}}_{gas}$), flame displacement speed (S_D) and overall flame front velocity (${\stackrel{\rightharpoonup }{U}}_{Flame}$) during the early flame development. Experiments are performed in an optical engine operating at 800 and 1500 RPM with premixed, stoichiometric isooctane-air mixtures. Analysis reveals several distinctive flame/flow configurations that yield a positive or negative flame displacement for which the flame progresses towards the reactants or products, respectively. For the operating conditions utilized, S_D exhibits and inverse relationship with flame curvature and a strong correlation between negative S_D and convex flame contours is observed. Trends are consistent with thermo-diffusive flames, but have not been quantified in context of IC engines. Flame wrinkling is more severe at the higher RPM, which results in a broader S_D distribution towards higher positive and negative velocities. Spatially-resolved distributions of ${\stackrel{\rightharpoonup }{U}}_{gas}$ and S_D are presented to describe in-cylinder locations where either convection or thermal diffusion is the dominating mechanism contributing to flame transport. Findings are discussed in relation to common engine flow features, including flame transport near solid surfaces. Findings provide a first insight into the detailed flame transport within a technically relevant environment and are designed to support the development and validation of engine simulations.

Introduction

In many combustion systems, such as spark-ignition (SI) engines, combustion efficiency is largely determined by the speed at which a flame transverses through a fuel-air mixture. Accordingly, one of the overarching goals in combustion science is to understand the detailed mechanisms of flame transport. In this context, we are concerned with the speed at which a wrinkled flame front traverses a certain distance. In turbulent combustion, the velocity of the flame front is denoted as the vector sum between local convection (${\stackrel{\rightharpoonup }{U}}_{gas}$) and thermal diffusion (S_D) in the flame normal direction ($\stackrel{\rightharpoonup }{n}$) [1]:

$${\stackrel{\rightharpoonup }{U}}_{Flame}={\stackrel{\rightharpoonup }{U}}_{gas}+\stackrel{\rightharpoonup }{n}·S_D$$

S_D is commonly referred to as the flame displacement speed and is a central quantity in the understanding and modelling of turbulent premixed combustion [2]. Numerous numerical and experimental studies have been performed to understand the intrinsic behavior of S_D with respect to flame structure [e.g., [3], [4], [5], [6], [7]], flame stretch (including strain and curvature) [e.g., [6], [7], [8], [9]], turbulence intensity [e.g., 10,11], and pressure [e.g. 12]. These fundamental investigations are often performed for a well-defined flame geometry (e.g., Bunsen, spherical or flat flame) within a well-characterized flow environment. Findings from such laboratory-scale flames have revealed the intrinsic behavior of S_D with respect to various physicochemical and turbulent flow properties, which has greatly expanded our knowledge of flamelet theory [13–14].

While detailed investigations in laboratory-scale flames have provided a comprehensive understanding of flame dynamics in well-defined flame/flow geometries, flame behavior in technical systems (e.g., SI engines) is far less understood. In SI engines, we are not only interested in the intrinsic behavior of S_D at high pressures, but also the capacity of ${\stackrel{\rightharpoonup }{U}}_{gas}$ to provide a fast, efficient flame transport. Additionally, it is important to resolve the stochastic nature of ${\stackrel{\rightharpoonup }{U}}_{gas}$ because previous studies have demonstrated that local flow variations can lead to favorable or unfavorable flame development, including misfires [15], [16], [17], [18].

A comprehensive understanding of flame transport in SI engines requires detailed experimental measurements resolving the coupling between chemical reaction and the turbulent flow. Laser diagnostics such as flame imaging coupled with velocimetry measurements are well-suited to achieve this goal. Several investigations have utilized simultaneous flame-flow imaging methodologies in internal combustion (IC) engines [15], [16], [17], [18], [19], [20], but only a few investigations have studied flame transport in detail [21,22]. Mounaïm-Rousselle et al. utilized high-speed laser tomography and particle image velocimetry (PIV) to resolve local 2D values of S_D in a boosted optical SI engine [21]. Using a short laser pulse separation, Mie scattering images resolved a planar view of flame displacement, while 2D ${\stackrel{\rightharpoonup }{U}}_{gas}$ velocities were directly measured from PIV. Measurements evaluated spatially-averaged S_D values as the flame progressed in time for different dilution levels.

Recognizing limitations in 2D measurements, Peterson et al. utilized a multi-planar approach to resolve local 3D flame displacement speeds in an optical engine [22]. The approach, originally proposed by Trunk et al. [23], utilized dual-plane laser induced fluorescence (LIF) imaging of the hydroxyl radical (OH) to resolve flame displacement on parallel planes. A 3D flame surface was constructed between each plane, providing the flame normal direction in 3D space. Stereoscopic PIV (SPIV) was simultaneously applied with OH-LIF to measure the convection of the reconstructed flame surface. Measurements successfully resolved distributions of ${\stackrel{\rightharpoonup }{U}}_{gas}$ and S_D during early flame propagation. Findings revealed a broad distribution of S_D and the importance of measuring flame/flow quantities in 3D, which testified to the complex nature of flame development during the early stages of combustion.

The multi-dimensional, multi-parameter measurements presented in [22] provide a unique opportunity to investigate detailed flame transport and flame/flow dynamics in IC engines. This is particularly important for the development of predictive combustion models utilized in CFD engine simulations. While many sophisticated flame propagation models exist, many models struggle to accurately describe heat release without first adjusting model parameters. Overall, there is a lack of detailed measurements describing flame transport, which limits the development of more predictive flame models. In turn, this limits the use of CFD to be an effective design tool.

The work presented in this study expands on the measurements in [22] to introduce an experimental dataset designed to study detailed flame transport and flame/flow dynamics in IC engines. Measurements of S_D, ${\stackrel{\rightharpoonup }{U}}_{gas}$, and ${\stackrel{\rightharpoonup }{U}}_{Flame}$ are presented for premixed, stoichiometric isooctane-air mixtures at two different engine speeds of 800 and 1500 RPM. Measurements are acquired at a single crank-angle degree (°CA) during the early flame development phase when less than 5% of the mixture is consumed. Findings reveal distinctive flame/flow configurations that yield a positive or negative flame displacement for which the flame progresses towards the reactants or products, respectively. Locally resolved S_D is evaluated with respect to 2D flame curvature along flame surfaces to describe the flame behaviour that is potentially responsible for yielding negative S_D values. Finally, spatially resolved distributions of S_D and ${\stackrel{\rightharpoonup }{U}}_{gas}$ are presented for each engine speed to describe in-cylinder locations where either thermal diffusion or convection is the more dominating mechanism contributing to flame transport.