An automated, multi-disciplinary optimization procedure for sub-sonic gas turbine compressor blades is presented. Evolutionary optimization algorithms are coupled with existing tools for geometry generation, mechanical integrity analysis and Q3D flow analysis for design and off-design conditions. Aerodynamic and mechanical objectives and constraints are formulated based on the standard design criteria. The feasibility of the approach is tested by automatically designing different rotor blades for the subsonic compressor region. First results are promising. All rotor blades show similar profile shapes, which underlines the robustness of the optimization procedure. The blades are characterized by a pronounced front loading which leads to a large (predicted) operating range. A special focus in this paper is on a 3D-blade parameterization, which by default leads to smooth blades, and on the assessment of the off-design behavior. The considered optimization algorithm shows a fast and robust convergence even from randomly initialized blades.
Flame stabilization in a swirl-stabilized combustor occurs in an aerodynamically generated recirculation region which is a result of vortex breakdown. The characteristics of the recirculating flow are dependent on the swirl number and on axial pressure gradients. Coupling to downstream pressure pulsations is also possible. Flame stability and emission formation depend on flow and mixing properties. The mixing properties of the investigated burner can be influenced by the position and the amount of fuel injection into the burner. The fuel injection is controlled by two different setups using (a) 8 proportional valves to adjust the mass flow for each fuel injector individually or using (b) 16 digital valves to include or exclude fuel injectors along the distribution holes. The objectives are the minimization of NOx emissions and the reduction of pressure pulsations of the flame. These two objectives are conflicting, affecting the environment and the lifetime of the combustion chamber, respectively. A multi-objective evolutionary algorithm is applied to optimize the combustion process. Each optimization run results in an approximation of the Pareto front by a set of solutions of equal quality, each representing a different compromise between the conflicting objectives. One compromise solution is selected with NOx emissions reduced by 30%, while mainaining the pulsation level of the given standard burner design. Chemiluminescence pictures of this solution showed that a more uniform distribution of heat release in the recirculation zone was achieved. The results were confirmed in high pressure single burner tests. The suggested fuel injection pattern has been successfully implemented in engines with sufficient stability margins and good operational flexibility. This paper shows the careful development process from lab scale tests to full scale pressurized tests.
This paper analyzes the influence of uncertainties on large-scale industrial radial compressors. The goal is to verify the robustness of new compressor designs towards operational and geometrical uncertainties. Key uncertainties for compressors are identified. Detailed measurements are conducted and modeled by probability distributions. These models then serve as input for the uncertainty analysis. For the uncertainty analysis different techniques are compared. The focus is on computationally efficient techniques, i.e., the required number of CFD simulations for the uncertainty analysis should be minimal. Promising methods are the Probabilistic Collocation Method and techniques that use response surface models in combination with the i-optimal design approach for generating sample points. Pure Monte Carlo simulations are not practical as they are too time-consuming. The key uncertainty considered here is the surface roughness. The effect is modeled with a commercial CFD RANS solver. The uncertainty analyzes show a performance decrease and variation for roughness values above the limit of hydraulically smooth surfaces.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.