The geometric and operational features of gas turbine engine combustors are receiving increased scrutiny due to a growing concern regarding environmental impact, performance, durability, and manufacturability. To minimize the risk associated with new projects, optimization of designs which are similar to those in current operation is attractive. To achieve this goal, a methodology that is efficient and can reveal interactions between parameters that affect performance is necessary. An approach which addresses these requirements is statistically designed experiments (e.g., multivariate experiments or “design of experiments”), which offers efficiency as well as the ability to identify interactions between variables. This approach was adopted and demonstrated in the present study utilizing a set of hardware specifically developed to allow multivariate experiments to be conducted. A radial mixer geometry consisting of four parameters (primary and secondary swirl vane angles, the presence of a venturi, and the co-/counter swirl sense) was examined. The design was developed to maintain constant effective area and overall dimensions. The responses selected for study were stability (i.e., reaction lean blow-out) and fuel distribution.
The results reveal that (1) the multivariate approach is effective in the present application, (2) the swirl sense between the primary and secondary swirlers play an influential role in determining the uniformity of the spray, (3) the size of the fuel spray area is affected by the mixer venturi and the swirl sense, (4) the symmetry of the fuel presentation is affected by the interaction of the swirler angles, (5) Lean Blow Out (LBO) is not affected by the parameters selected, and (6) the parameters affecting fuel distribution and combustion stability differ, indicating that the combustion performance is not described entirely by fuel distribution.