Joint experimental and numerical study of the influence of flame holder temperature on the stabilization of a laminar methane flame on a cylinder

Abstract

The mechanisms controlling laminar flame anchoring on a cylindrical bluff-body are investigated using DNS and experiments. Two configurations are examined: water-cooled and uncooled steel cylinders. Comparisons between experimental measurements and DNS show good agreement for the flame root locations in the two configurations. In the cooled case, the flame holder is maintained at about 300 K and the flame is stabilized in the wake of the cylinder, in the recirculation zone formed by the products of combustion. In the uncooled case, the bluff-body reaches a steady temperature of about 700 K in both experiment and DNS and the flame is stabilized closer to it. The fully coupled DNS of the flame and the temperature field in the bluff-body also shows that capturing the correct radiative heat transfer from the bluff-body is a key ingredient to reproduce experimental results.

Introduction

The burnt gas temperatures reached in combustion chambers usually exceeds the maximum temperatures which can be sustained by most materials, especially metals used in engines. Therefore, cooling these walls as well as all chamber elements in contact with the flame is mandatory for combustion chamber designers. While cooling is obviously needed to preserve walls, its effects on the flames themselves has received less attention and is usually neglected in many CFD approaches. Flame/wall interaction, for example, is a field of combustion which has not been investigated yet with sufficient care [1], [2], [3], [4], [5], [6]. In most cases, authors measure or compute the maximum wall heat fluxes induced by the flame but do not investigate the effects of the wall on the flame itself.

In the field of simulation, most models [7], [8], [9], [10], [11] assume adiabatic flows. For premixed flames, the famous BML (Bray Moss Libby) approach, for example, which is the workhorse of many theories for turbulent premixed flames [12], [13] assumes that a single variable (the progress variable c) is sufficient to describe the flow: this is true only when the flow is adiabatic. In the same way, many usual methods for chemistry tabulation such as FPV [14], FPI [15] or FGM [16] assume that chemistry can be described using only two variables, the mixture fraction z and the progress variable c, which implies that the flames must be adiabatic.¹ Considering that wall heat fluxes in most chambers correspond to approximately 5–40% of the chamber total power, assuming adiabaticity is clearly not compatible with the high-precision methods which are sought today. Note that computing the interaction between the flame and the wall requires to compute both the flow and the temperature within the walls simultaneously: the LES code must be coupled with a heat transfer code within the combustor walls. This task is not simple [19], [20] because time scales are usually very different (a few milliseconds in the flow and a few minutes in the walls).

Among all walls present in a chamber, flame holders play a special role because they control the most sensitive zone of the chamber: the place where the flames are anchored. Any temperature change of the flame holder will induce a change of position for the flame roots and therefore a change in stability and efficiency. The coupling mechanisms between heat transfer within flameholder and flame stabilization have not been analyzed in detail yet. In a series of recent papers [21], [22], [23], the MIT group has numerically studied the stabilization of premixed flames on square flame holders and shown that the location of the flame roots but also the blow-off limits were strongly affected by the temperature of the flame holder.

The present study focuses on a similar question: which differences in flame anchoring are observed when the temperature of the flame holder varies from a low (typically 300 K) to a high value (700 K). To obtain such a large variation in temperature, a premixed laminar methane/air flame is stabilized on a cylindrical flame holder. Two flame holders are used, with exactly the same external shape. The first one has an internal water cooling system, leading to a surface temperature close to 300 K. The second one is a full, solid cylinder which is uncooled, leading to a surface temperature close to 700 K.

Both experiments and DNS are used to analyze the differences in flame structure near the flame holder. Simulations are performed in dual mode: the flow is computed with DNS using a 13 species kinetic scheme for CH₄/air flames [24] while the temperature in the solid is computed with a heat transfer solver, coupled to the flow solver. The simulations, performed for cooled and uncooled flame holders, reveal drastic differences in flame root location and flow topologies. They also show that radiative heat transfer must be taken into account to predict the flame topology for the uncooled case.

Section 2 presents the experimental setup. The tools used for the coupled flow/solid simulation are described in Section 3. Results for the cooled flame holder are discussed in Section 4 before presenting the uncooled case in Section 5. Finally Section 6 discusses the influence of radiative heat fluxes on the flame stabilization when the flame holder is uncooled.