GasTurbnLab: A Problem Solving Environment for Gas Turbine Simulation

Principal Investigators: Sanford Fleeter (ME), Elias Houstis, John Rice, Chenn Zhou (ME)

Research Associate: A. Catlin

Sponsor: DoE

This project applies the problem solving environment methodologies of PELLPACK, PYTHIA, and SciAgents to the simulation of the operability of gas turbines. Two principal physical phenomena are involved in this simulation:

Unsteady Interaction of Gas Turbine Engine Fluid-Thermal-Structure Components. The analysis of the unsteady operation of a gas turbine engine is an important aspect of propulsion system design. Namely, as efficiency requirements increase, stability margin's, i.e., rotating stall and surge margins, are necessarily reduced. To compensate, control systems are becoming more sophisticated including active performance seeking logic and neural networks. In addition, unsteady gas turbine engine operation often produces extreme loading for the turbomachinery blading, resulting in high cycle fatigue (HCF) failures. Thus, the accurate analysis of the peak blade row unsteady aerodynamic loading and vibratory responses is needed. The GasTurbnLab simulator will be developed to analyze the operability of a complete gas turbine engine.

Full Annular, Unsteady Flow in the Combustor. CFD gas turbine combustor modeling has generally been limited to isolated parts of the combustion system. Most models include only the reacting flow inside the combustor liner with assumed profiles and flow spits at the various liner inlets. Carefully executed models of this type can provide valuable insight into mixing performance, pattern factor, emissions and combustion efficiency. A CFD calculation for the unsteady flow through a complete annulus combustor — from the compressor diffuser exit to the turbine inlet — is needed. The combustor configuration should be representative of a lean, low emission design. The model should include an airblast fuel nozzle, dome and liner walls with dilution holes and cooling louvers.

1998
Annual Research Report

Department of
Computer Sciences