| OCR Text |
Show Comparison of Local and Global Combustion Efficiencies for the TCEQ 2010 Flare Study Air-Assisted Flare using High Performance Computing Michal Hradisky, Philip J. Smith Institute for Clean and Secure Energy, University of Utah ABSTRACT There is an ever-increasing role of simulations in understanding complex systems such as the combustion occurring during operation of industrial flares. Extensive research, both experiments and simulations, has been conducted for simplified flares in well-controlled environments. However, experimental, as well as computational, studies of full-scale flares present great challenges when characterizing the effect of operational and environmental parameters on the combustion efficiency of full-scale flares. Often, for the full-scale flare studies, the experiments themselves, along with the instrumentation, can be very complex and expensive, and can be very difficult to characterize in terms of accuracy. For simulations of full-scale flare operations, computational resources can impose significant restrictions on the models, and their associated accuracy, that can be used to represent the system. In this paper we focus on the air-assisted flare experiments from the TCEQ 2010 Flare Study and compare the local combustion efficiency as measured by the extractive probe to the overall, global, combustion efficiency for a given experiment using both Reynolds-Averaged Navier-Stokes (RANS) as well as Large Eddy Simulation (LES) models. As such, we illustrate the role of simulation in determining combustion efficiencies of flares to help with safe and responsible development of combustion systems in the 21st century. INTRODUCTION Flaring is used to burn waste gasses from industrial operations, such as disposing of unrecoverable gasses during oil well drilling process, or to burn wasted products from refineries. Flares are also indispensable in emergency conditions, such as pressure relief and emergency depressurization in events such as plant-wide power failure or unexpected weather conditions. Currently, flares cannot be subjected to standard emission testing techniques, and to date, there is no experimental instrument that would directly measure combustion efficiency. Therefore, the Environmental Protection Agency (EPA) provides a list of operational guidelines in the Code of Federal Regulations (CFR) 40 Section 60.18, which regulate maximum flare tip velocity as a function of the net heating value of gas being combusted, as well as place restrictions on visible emissions and time for which these visible emissions can occur. However, these guidelines rely on proper operation of these flares, and within the specified operational parameters. It is unlikely that these regulations capture all possible operational conditions that occur for an actual full scale flare. For instance, the same fuel and assist media flow rates, which adhere to the guidelines, can produce two very different combustion efficiencies if subjected to different ambient conditions, such as cross wind velocities. Ideally, a real-time combustion efficiency monitoring technique could be applied that would aid in adjusting operational parameters during the flaring event to maximize the combustion efficiency, decrease the emissions, and still adhere to the required guidelines. As mentioned, currently, there is not an experimental instrument that would directly measure combustion efficiency. However, there are several measurement techniques that relate emissions or elemental composition to the combustion efficiency. Such techniques include Infrared Hyper-Spectral Imaging Technology, Passive Fourier Transform Infrared Spectroscopy (PFTIR) and Active Fourier Transform Infrared Spectroscopy (AFTIR), Tunable Diode Laser Absorption Spectroscopy (TDLAS), and Differential Absorption LIDAR (DIAL). These techniques are also referred to as remote sensing technologies. Generally, these techniques rely on line-of-sight measurement (or small cone) and do not take into account the highly-unsteady and highly-variable mixing of turbulent buoyant plumes and the effects of unsteady eddies on the resulting combustion efficiency. They only collect information regarding to the concentration of gasses for the line-of-sight measurement, based on which they infer the combustion efficiency. Further, these techniques produce differing results based on the line-of-sight chosen and therefore cannot determine the overall combustion efficiency of any given flare. There are several emerging technologies which also take into account the rapidly changing dynamics of the plume along with the emission composition. However, these techniques are yet to be applied to full scale flaring systems. In 2009 and 2010 the Texas Commission on Environmental Quality (TCEQ) contracted with the University of Texas at Austin to conduct a full-scale flare experiments to examine impacts of the air and steam assist media on combustion efficiency, determine how operating conditions relate to requirements in 40 CFR Section 60.18, and also compare currently available experimental techniques for measuring emissions from these flare experiments. This study was also supported by the Air Quality Research Program. In conjunction with several remote sensing technologies, extraction probe was also used to evaluate composition of emissions. In this paper we illustrate the role of simulations to determine combustion efficiency for full scale flares. While simulations have their own challenges, coupled with experiments, they can provide additional information regarding the highly unsteady nature of the plume in terms of both composition as well as velocities. Depending on the turbulence and combustion models used, we can expect varying computational costs along with varying accuracy of results. Because experimental flaring systems are generally large in size, to fully resolve all pertinent geometric features of the flare and still capture the plume far downstream, large domains with fine meshes must be used. This places constraints on the computational requirements needed to complete the simulation. For our simulations, we have used High Performance Computing (HPC) systems. These systems allow us to capture fine geometric details of the flare as well capture the flames and plume downstream. SIMULATION DESCRIPTION For this paper, we have ran simulations of the air assisted flare used in the TCEQ 2010 study. This air assisted flare is designated as EEF-LH-24/60 and was provided by John Zink. We have used the technical drawing in the TCEQ final report to construct our CAD representation of the flare. The technical drawing attached in the TCEQ report does not provide full geometric details needed to construct the CAD model that would represent the actual flare. Therefore, we have made several assumptions that would best represent the geometry as it appears in the drawing and in several pictures available in the report. The key details we estimated included the dimensions of the fuel tips as well as the angles and dimensions for the air plenum at the flare tip; these details could potentially influence simulation results. We have used commercially available code Star-CCM+ developed by CD-a d a p c o t o c r e a t e t h e CAD representation of the air-assist flare, including the extraction probe used to determine experimental combustion efficiency, as well as to run our simulations. The geometry used for our computations is shown in Figures 1 and 2. Figure 1 shows the overall domain used for our computations, whereas Figure 2 shows close up of the flare tip. Figures 3 and 4 show Figure 1. Overall simulation domain. detailed views of the mesh used for the simulation inside the flare tip as well as for the entire domain, respectively. Cell size is very fine near the tip of the flare, with increasing size farther away from the flare tip. This technique is used to decrease the computational load and concentrate the cell resolution on the areas of interest. For all simulations we have employed the Progress Variable Model. We are using 3D unsteady formulation with Large Eddy Simulation (LES) model. Our fuel and air boundary conditions are based on test point A2.4 Run 3, as shown in Table E-1 of the final TCEQ 2010 report. Therefore, all results assume the fuel stream flow rate to be 354 lb/hr and air stream flow rate to be 148,533 lb/hr. We are using pure propene as the fuel. To resolve such wide range of scales present in this problem, each simulation is run on a high performance computer on 600 cores. Figure 2. Close up of the John Zink TCEQ air assisted flare tip. Figure 3. Meshing detail of the TCEQ air assisted flare tip. Figure 4. Overall domain computational mesh containing about 25 million computational cells. SIMULATION ANALYSIS The LES simulation results for the TCEQ 2010 air assisted flare are shown in Figures 5 and 6. Figure 5 shows temperature contours near the flare tip, while Figure 6 shows the velocity distribution near the flare tip and the probe. The computed global as well as local combustion efficiencies are shown in the presentation. For this run, the experimental extraction probe measurements were converted to 91.3% c o m b u s t i o n e f fi c i e n c y . However, it is important to note that the combustion efficiency of 91.3% was computed based on the plume composition from the extraction probe, and therefore, is highly dependent on the location of the probe in the plume. Figure 5. Temperature contours for the LES simulation of experiment A2.4 Run 3. Figure 6. Velocity contours for the LES simulation, which includes both the flare as well as the extraction probe, of experiment A2.4 Run 3. REFERENCES 1) Texas Commission on Environmental Quality PGA No. 582-8-862-45-FY09-04 Tracking No. 2008-81, TCEQ 2010 Flare Study Final Report. 2) A Technology for Measuring Combustion Efficiency of Industrial and Field Flares (integrating measurements and simulations), Philip J. Smith, AFRC 2010 3) Environmental Protection Agency 40 CFR 60 and 63, Standards of Performance for New Stationary Sources: General Provisions 4) Star-CCM+ User Guide, v 8.02 |
