Document Type Master's Dissertation Author Gouws, Johannes Jacobus URN etd-04212008-074344 Document Title Combining a one-dimensional empirical and network solver with computational fluid dynamics to investigate possible modifications to a commercial gas turbine combustor Degree MEng (Mechanical) Department Mechanical and Aeronautical Engineering Supervisor
Advisor Name Title Mr R M Morris Committee Chair Mr J A Visser Committee Co-Chair Keywords
- turbine combustor
- fluid dynamics
Date 2007-09-05 Availability unrestricted AbstractGas turbine combustion chambers were traditionally designed through trial and error which was unfortunately a time-consuming and expensive process. The development of computers, however, contributed a great deal to the development of combustion chambers, enabling one to model such systems more accurately in less time. Traditionally, preliminary combustor designs were conducted with the use of one-dimensional codes to assist in the prediction of flow distributions and pressure losses across the combustion chamber mainly due to their rapid execution times and ease of use. The results are generally used as boundary conditions in three- dimensional models to predict the internal flow field of the combustor. More recent studies solve the entire flow field from prediffuser to combustor exit. This approach is, however, a computationally expensive procedure and can only be used if adequate computer resources are available.
The purpose of this study is two-fold; (1) to develop a one-dimensional incompressible code, incorporating an empirical-based combustion model, to assist a one-dimensional network solver in predicting flow- and temperature distributions, as well as pressure losses. This is done due to the lack of a combustion model in the network solver that was used. An incompressible solution of flow splits, pressure losses, and temperature distributions is also obtained and compared with the compressible solution obtained by the network solver; (2) to utilise the data, obtained from the network solver, as boundary conditions to a three-dimensional numerical model to investigate possible modifications to the dome wall of a standard T56 combustion chamber. A numerical base case model is validated against experimental exit temperature data, and based upon that comparison, the remaining numerical models are compared with the numerical base case. The effect of the modification on the dome wall temperature is therefore apparent when the modified numerical model is compared with the numerical base case.
A second empirical code was developed to design the geometry of axial straight vane swirlers with different swirl angles. To maintain overall engine efficiency, the pressure loss that was determined from the network analysis, of the base case model, is used during the design of the different swirlers. The pressure loss across the modified combustion chamber will therefore remain similar to that of the original design. Hence, to maintain a constant pressure loss across the modified combustion chambers, the network solver is used to determine how many existing hole features should be closed for the pressure loss to remain similar. The hole features are closed, virtually, in such a manner as not to influence the equivalence ratio in each zone significantly, therefore maintaining combustion performance similar to that of the original design. Although the equivalence ratios in each combustion zone will be more or less unaffected, the addition of a swirler will influence the emission levels obtained from the system due to enhanced air-fuel mixing.
A purely numerical parametric analysis was conducted to investigate the influence of different swirler geometries on the dome wall temperature while maintaining an acceptable exit temperature distribution. The data is compared against the data obtained from an experimentally validated base case model. The investigation concerns the replacement of the existing splash-cooling devices on the dome wall with that of a single swirler. A number of swirler parameters such as blade angle, mass flow rate, and number of blades were varied during the study, investigating its influence on the dome wall temperature distribution.
Results showed that the swirlers with approximately the same mass flow as the existing splash-cooling devices had almost no impact on the dome wall temperatures but maintained the exit temperature profile. An investigation of swirlers with an increased mass flow rate was also done and results showed that these swirlers had a better impact on the dome wall temperatures. However, due to the increased mass flow rate, stable combustion is not guaranteed since the air/fuel ratio in the primary combustion zone was altered.
The conclusion that was drawn from the study, was that by simply adding an axial air swirler might reduce high-temperature gradients on the dome but will not guarantee stable combustion during off-design operating conditions. Therefore, a complete new hole layout design might be necessary to ensure good combustion performance across a wide operating range.
© University of Pretoria
Filename Size Approximate Download Time (Hours:Minutes:Seconds)
28.8 Modem 56K Modem ISDN (64 Kb) ISDN (128 Kb) Higher-speed Access dissertation.pdf 6.71 Mb 00:31:02 00:15:58 00:13:58 00:06:59 00:00:35