Search

Industrial gas turbines are complex systems, and their proper analysis requires specific knowledge of the different components interacting with each other. One of the challenges is to accurately predict the overall performance of the system. To do so, a performance analysis tool can be used but it needs to rely on representative characteristics from the different elements. One of the key items required for the performance derivation is the understanding of the total cooling and leakage air (TCLA) and its distribution within the machine. A secondary air flow (SAF) tool is used to evaluate TCLA, but it needs to be “connected” with the overall performance model. The source locations of the SAF system are relatively straightforward to handle, but the sink (or dump) locations occurring in the turbine section require a detailed understanding of the pressure distribution within the flowpath. As a matter of fact, a change in SAF distribution to the turbine yields a change in turbine work output and load distribution that needs to be captured in the overall engine performance assessment. Furthermore, a change in turbine inlet boundary conditions also yields a change in pressure values for the SAF system. For these reasons, an accurate performance modeling of a gas turbine requires a so-called Synergy Loop to converge the overall boundary conditions of all its interacting modules and sub-models. This paper will describe the method used to integrate different commercial and in-house software to complete the Synergy Loop. The authors will also describe the different iteration steps possible between the specific components and the recommended iteration loop sequence for fast convergence.

Turbine secondary flow system calculations are usually performed by utilizing the thermal-fluid network approach. The network consists of branches and nodes. Fluid branches usually describe the flow resistance of the correspondent fluid path section, while fluid nodes are used to connect fluid branches between each other. Fluid flow in branches and nodes is described by the set of conservation equations such as mass and momentum equations. For fluid nodes, usually total or static pressure is used as a variable in these equations. However, such an approach may yield a system of equations that cannot be solved (for theoretical cases where zero resistance branches are present) or may produce significant differences in results compared to the more precise CFD solution (for real cases with resistances). This paper provides an approach on how to create a model of a fluid lossless junction in order to solve the mentioned problems. Such a junction enables the connection of any number of inflow and outflow fluid branches without bringing any flow resistance, which allows fluid branches to handle flow resistance influence by themselves. The proposed model is based on the idea, that each node should contain not just one pressure (static or total) as a variable, but two variables — both static and total pressures. Such a node can be used to model junctions of different types with flow mixing, as well as separation and sudden cross-sectional area changes. With this approach it is also possible to model chambers by modifying just one equation. Also, this new method can be applied for both compressible and incompressible calculation types and can handle chocked flows as well.