 |
Optimization of Gas Turbine Combustors
|
|||||||||||||||||||||||||
| Pressure losses less that 6%, high temperature uniformity in the outlet flow:
Patter Factor reduced of 40% respect to the reference combustor. |
The present work describes a novel approach for the optimized
design of gas turbine combustors, that integrates a 0-D code,
CFD analyses and an advanced Multi Objective Game Theory
optimization algorithm.
Methodology
 |
3D CAD model of the combustor |
A proprietary 0-D code produces the baseline design of any generic combustor, given the required performances of the G.T., the machine characteristics, the fuel type and properties, the basic geometry (tubular or annular) and the combustion concept (i.e. lean premixed primary zone or diffusive processes). The following step is the optimization of the baseline design. To that purpose, a parametric CAD and mesh model should be prepared, as well as the needed set of macros to run the CFD analysis and post-process the results in batch mode.
 |
Optimization Workflow defined within modeFRONTIER |
Free parameters of the optimization process are position and size of the liner holes arrays, their total area and the shape of the exit duct. The three different objectives are the minimization of NOx emissions, pressure losses (Dp) and combustor exit Pattern Factor (Paf). This last index is a measure of the temperature uniformity of the produced gas flow. The lower it is, the better it is, specially in terms of thermal stress for the following turbine blades.
All the computational chain from the numerical definition of the variables' values to the results handling was automated by means of:
- a parametric CAD model of the device (CATIAv5 scripting language);
- a robust hexahedral multi-block meshing(ICEM CFD);
- a batch-mode use of the CFD solver, with automatic post-processing (ANSYS CFX).
The workflow automation and data handling, as well as the optimization engine, is provided by the modeFRONTIER optimization tool. The time and computational resources constraints allowed the authors to perform not more than one hundred different design evaluations: for this reason the modeFRONTIER's efficient and robust Multi Objective Game Theory algorithm was selected. The optimization was carried out in an automatic ad efficient way by modeFRONTIER, exploiting the distributed computing concept via queuing system. Parallelization was applied both to the single computation, both to the optimization itself.
 |
The computed designs in the space of the three objectives (each one to be minimized) |
 |
Results
- dry-low-emissions concept implemented;
- pressure losses less that 6%;
- high temperature uniformity in the outlet flow: Patter Factor reduced of 40% respect to the reference combustor.
The statistical analysis tools of modeFRONTIER were used to perform sensitivity studies of the considered parameters on the combustor performances, while a stochastic robustness analysis of the found solution is currently under progress.
 |
 |
Streamlines representing temperature and flow-field of the optimized combustor sector |
Correlations between some input parameters and objectives |
Article published in the Magazine: EnginSoft Newsletter English Special Issue 2006