12 20 Author: Joshua Tang Supervisor: A/Professor Evatt Hawkes Research Theme: Fundamental and Enabling research Non Premixed Combustion: Modelling Gas Turbine Flames using TPDF methods Research Theme: Fundamental and Enabling research Motivation and Background Non premixed combustion occurs in devices like gas turbines for power generation and aviation. These flames experience extinction if local mixing rates exceed combustion rates. Aviation devices demand smaller combustors which can increase mixing rates creating extinct regions. Power generation applications require lower Pre-processing involved extracting and smoothing of velocity data from the DNS using Matlab. The Composition-TPDF (C-TPDF) method requires velocity flow field information, a 2 equation model such as the k – ε model can be used, however, to reduce modelling errors DNS data was used for the simulation. Simulation Methodology: regions. Power generation applications require lower temperatures to reduce production of NO x , this results in lower combustion rates which can also lead to extinction. . The simulation was run using a Fortran 90 code which simulated the mixing and chemistry of the flame using the C-TPDF method and one of 3 mixing models, IEM, MC or EMST. The chemistry simulation method was identical to the DNS. Output from this program was processed using a specifically designed Matlab script. Results for each model were compared against the DNS In both devices extinction is a problem that can result in flame instability and increased pollutant formation. Extinguished regions can re-ignite, which can help to mitigate these model were compared against the DNS Results: Transported Probability Density Function (TPDF) methods can help to mitigate these effects. Modelling flame extinction and re-ignition is difficult due the random nature of turbulence and the complex interaction between the fluid flow and the underlying chemistry Figure 1: Visualisation of non premixed flames. Note the random nature of flame structures* Aim and Objectives • To develop computationally efficient methods to Figure 3: Temperature (Kelvin) colour maps for DNS and model simulations. Transported Probability Density Function (TPDF) methods offer compelling advantages, in particular, they treat the chemical reaction of the flow without approximation. However, a mixing model is needed to account for the effect of molecular diffusion. The effectiveness of TPDF methods relies heavily on the performance of the mixing model. Methodology: Simulation Scenario: simulate turbulent combustion • To assess the performance of Interaction with the mean (IEM), Modified Curl (MC) and Euclidean Minimum Spanning Tree (EMST) mixing models for use in TPDF methods. for DNS and model simulations. Figure 3 shows that the models can qualitatively predict the extinction and re-ignition events as there is little variance between the DNS and the model maps. A more quantitative measure of performance is given in Figure 3, the spatial temperature and mixture fraction distributions The simulation is based on a Direct numerical simulation (DNS) run by Hawkes, modelling a diffusion flame at Reynolds number (Re) = 2510, using Syn-gas (CO and H 2 ) as fuel. Figure 2 shows simulation configuration, there is one planar slab of fuel in counter flow with 2 slabs of oxidant on either side. Data from this simulation was provided as a validation data set. Figure 4: Temperature (left) and mixture fraction Conclusions: From Figure 4 it can be seen that all models under predict the peak temperature at the jet core (Y=0), with good agreement elsewhere in the simulation. The mixture fraction is over predicted at the jet core however shows almost perfect agreement in other regions for all models. Figure 4: Temperature (left) and mixture fraction (right) distributions at the final time-step Conclusions: The mixing models are all capable of qualitatively predicting the extinction and re-ignition as well as quantitatively predicting the spatial temperature and mixture fraction distributions for the given test case. ENGINEERING @ UNSW UNSW Figure 2: Simulation configuration.** •J. Jarosinski and B. Veyssiere. Combustion Phenomena: selected mechanisms of flame formation, propagation, and extinction. Taylor and Francis Group, 1st edition, 2009. ** D. Lignell, J. Chen, H . Schmultz (2010). Effects of Damköhler number on flame extinction and reignition in turbulent non-premixed flames using DNS. Combustion and Flame, Vol. 158, pp 949 – 963.
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
12
20
Author: Joshua Tang Supervisor: A/Professor Evatt Hawkes
Research Theme: Fundamental and Enabling research
Non Premixed Combustion: Modelling Gas
Turbine Flames using TPDF methods
Research Theme: Fundamental and Enabling research
Motivation and Background
Non premixed combustion occurs in devices like gasturbines for power generation and aviation. These flamesexperience extinction if local mixing rates exceedcombustion rates. Aviation devices demand smallercombustors which can increase mixing rates creating extinctregions. Power generation applications require lower
Pre-processing involved extracting and smoothing of velocity data from the DNS using Matlab. TheComposition-TPDF (C-TPDF) method requires velocity flow field information, a 2 equation model such as the k –ε model can be used, however, to reduce modelling errors DNS data was used for the simulation.
Simulation Methodology:
regions. Power generation applications require lowertemperatures to reduce production of NOx, this results inlower combustion rates which can also lead to extinction.
.
The simulation was run using a Fortran 90 code whichsimulated the mixing and chemistry of the flame usingthe C-TPDF method and one of 3 mixing models, IEM,MC or EMST. The chemistry simulation method wasidentical to the DNS.
Output from this program was processed using aspecifically designed Matlab script. Results for eachmodel were compared against the DNS
In both devices extinction isa problem that can result inflame instability andincreased pollutantformation. Extinguishedregions can re-ignite, whichcan help to mitigate these model were compared against the DNS
Results:
Transported Probability Density Function (TPDF) methods
can help to mitigate theseeffects. Modelling flameextinction and re-ignition isdifficult due the randomnature of turbulence andthe complex interactionbetween the fluid flow and
the underlying chemistry
Figure 1: Visualisation of non
premixed flames. Note the
random nature of flame
structures*
Aim and Objectives
• To develop computationally efficient methods to Figure 3: Temperature (Kelvin) colour maps
for DNS and model simulations.
Transported Probability Density Function (TPDF) methodsoffer compelling advantages, in particular, they treat thechemical reaction of the flow without approximation.However, a mixing model is needed to account for the effectof molecular diffusion. The effectiveness of TPDF methodsrelies heavily on the performance of the mixing model.
Methodology:Simulation Scenario:
simulate turbulent combustion
• To assess the performance of Interaction with themean (IEM), Modified Curl (MC) and EuclideanMinimum Spanning Tree (EMST) mixing models foruse in TPDF methods.
for DNS and model simulations.
Figure 3 shows that the models can qualitatively predictthe extinction and re-ignition events as there is littlevariance between the DNS and the model maps. A morequantitative measure of performance is given in Figure 3,the spatial temperature and mixture fraction distributions
The simulation is based on a Direct numerical simulation(DNS) run by Hawkes, modelling a diffusion flame atReynolds number (Re) = 2510, using Syn-gas (CO and H2)as fuel. Figure 2 shows simulation configuration, there isone planar slab of fuel in counter flow with 2 slabs of oxidanton either side. Data from this simulation was provided as avalidation data set.
Figure 4: Temperature (left) and mixture fraction
Conclusions:
From Figure 4 it can be seen that all models underpredict the peak temperature at the jet core (Y=0), withgood agreement elsewhere in the simulation. Themixture fraction is over predicted at the jet core howevershows almost perfect agreement in other regions for allmodels.
Figure 4: Temperature (left) and mixture fraction
(right) distributions at the final time-step
Conclusions:
The mixing models are all capable of qualitativelypredicting the extinction and re-ignition as well asquantitatively predicting the spatial temperature andmixture fraction distributions for the given test case.
ENGINEERING @ UNSW
UNSW
Figure 2: Simulation configuration.**
•J. Jarosinski and B. Veyssiere. Combustion Phenomena: selected mechanisms of flame
formation, propagation, and extinction. Taylor and Francis Group, 1st edition, 2009.** D. Lignell, J. Chen, H . Schmultz (2010). Effects of Damköhler number on flame extinction and reignition in turbulent non-premixed flames using DNS. Combustion and Flame, Vol.
158, pp 949 – 963.
Related Documents
