CFD predictions of wake-stabilised jet flames in a cross-flow

Abstract

This study describes an investigation into predicting the major flow properties in wake-stabilised jet flames in a cross flow of air using first-and second-order turbulence models, applied within a Reynolds-averaged Navier-Stokes (RANS) modelling framework. Standard and re-normalisation group (RNG) versions of the k-turbulence model were employed at the first-order level and the results compared with a secondmoment closure, or Reynolds stress model (RSM). The combustion process was modelled using the laminar flamelet approach together with a thermal radiation model using the discrete ordinate method. The ability of the various turbulence models to reproduce experimentally established flame appearance, profiles of velocity and turbulence intensity, as well as the combustion efficiency of such flames is reported. The results show that all the turbulence models predict similar velocity profiles over the majority of the flow domain considered, except in the wake region, where the predictions of the RSM and RNG k-models are in closer agreement with experimental data. In contrast, the standard k-model over-predicts the peak turbulence intensity. Also, it is found that the RSM provides superior predictions of the planar recirculation and flame zones attached to the release pipe in the wake region. 1.IntroductionA large quantity of oil field associated gas, and waste hydrocarbons from the process industry, is flared globally each year. With the predicted increase in oil production in the next decades, a significant increase in flaring is expected [1,2]. However, flare operations can result in the formation of pollutants such as unburned hydrocarbons, CO, NO x and noise and the effect of a cross-flow of wind can be significant. In order to minimise these effects, a detailed understanding of the flow features and structures of these non-premixed jet flames in a cross-flow (JFICF) is necessary. Depending on the value of the jetto-cross-flow momentum flux ratio, R= j u j 2 / cf u cf 2 , where j , cf and u j , u cf are the density and velocity of the fuel jet and the cross-flow, respectively, the flow features that arise in the JFICF have an important influence on the flame's combustion characteristics.For flares operating at low values of R, typically R ≤ 1.0, the flame is either seated on the flare stack or stabilised in its wake [3]. The latter case is also known as a wake-stabilised flare which belongs to a flow regime where the flame is severely bent-over by the cross-flow and has high turbulence intensities in the near-field [4]. This category of flare is commonly found in offshore crude oil production facilities.The three main flow features in the wake-stabilised flare include: a counter-rotating vortex pair (CVP)[5], coherent structures around the upper surface of the jet [3] and the planar recirculation zone in the wake of the stack [6]. The latter zone is part of the low-pressure region where a secondary flame is attached to the release pipe, as shown in Fig. 1, and the vortex dynamics and mixing in this ...

Reduced combustion mechanism for C1–C4 hydrocarbons and its application in computational fluid dynamics flare modeling

“…Modeling efforts in this area have been carried out using different approaches (Busupally andDe, 2016, Kronenburg et al, 2000;Thomas et al, 2008). Further, numerical simulations of different flames in a cross-flow air has been conducted using various combustion and turbulence models to study the effect of fuel and cross-flow velocities on pollutant properties (Lawal et al, 2010(Lawal et al, , 2013Aboje et al, 2015). The present work discusses two different models in CFD to predict flare efficiency and soot emission using the developed combustion mechanism LU 3.0.1.…”

Quantification of Gas Flaring from Satellite Imagery: A Comparison of Two Methods for SLSTR and BIROS Imagery

“…Flaring activity based on remote sensing estimates is uncertain [ 76 , 77 , 78 ], as are emissions estimates, due to uncertain emission factors [ 79 , 80 , 81 , 82 , 83 ], on the one hand, and to limitations from both the ground-based data used for the fitting (e.g., [ 32 , 77 , 84 ]), in the case of the calibration-based approach, and the sensors (e.g., [ 85 ]), in the cases of the calibration-based and the physico-chemical-based approaches, on the other hand. The former include geographic and temporal biases: the reporting processes, the gas characteristics and the combustion conditions, for example, may have a strong local dimension and limited applicability, become outdated, be inconsistent or highly uncertain.…”

Large eddy simulations of wake-stabilised flares

“…6(b). Therefore, it is clear that RANS is only able to resolve the secondary flame region in wake-stabilised flares at very low values of R [10], whilst the LES is capable of resolving this complex flow feature at both low and high values of R. It is very important to predict this secondary flame region correctly because it is where low combustion efficiency occurs producing significant quantity of emissions of unburned hydrocarbons (UHCs), as will be discussed further later.…”

“…In light of the substantial computational effort involved in LES-based calculations, the objective of the present investigation is to examine the potential benefits and limitations of the LES approach, over RANS simulations [10] employing a Reynolds stress turbulence model, in improving predictions of the complex flame structure that influences the flare combustion efficiency, temperature and emissions. The complex flow features of interest include the secondary flame region on the lee side of the flare stack, fragmentation of the flame by the cross-flow, the counter-rotating vortex pair and its influence on global mixing within the flow field.…”