Numerical simulations of gas turbine combustion

Ke Su, Purdue University

Abstract

A numerical investigation of gas turbine combustion is conducted using the KIVA-3V code. Off-design conditions and spray parameters have significant effects on flow patterns, kinetic reactions and spray dynamics in the combustor, which consequently influences combustion performance and NOx emissions. This study provides insight into physical and chemical processes in combustion, and evaluates variations of combustion performance and NOx emissions in off-design conditions. Combustor flows with non-uniform inflows are simulated. It is shown that the circumferential non-uniform profile has stronger influence on the overall temperature distribution (OTDF) than the radial profile. Meanwhile, the aerodynamic interaction of transient compressor and combustor flows and its effects on combustion are investigated. Transient inflows are assumed as sinusoidal functions of time. Three flow patterns, namely the quasi-steady, transition and steady patterns corresponding to frequencies of n ≤ 80, 80 ≤ n ≤ 320, and n ≥ 320 Hz, are classified. For flows in quasi-steady pattern, OTDF and NOx emissions vary with time in sinusoidal trends. As the frequency increases, oscillations of OTDF and NOx emissions become milder. When in steady pattern, flow oscillations are so weak that as if the combustion in steady state. Further, the operability of combustor is analyzed, which provides an effective method for analysis of combustion performance. In addition, resonance happens to combustor flows at the frequency of 240 Hz, which intensifies fluctuations of flows and leads to more deviations of combustion performance. A group spray model is developed for efficient and realistic simulation of spray dynamics. It uses droplet groups to represent a spray, employs the Lagrangian approach to trace each group, and considers turbulent dispersion of droplets within the group. A group evaporation correction factor, dependent on droplet number density, size and Sherwood number, and the group size, is introduced to consider the effects of spray parameters on droplet evaporation. It is found that the group model needs only 1/50 of droplet parcels required by the traditional SSF model and thereby saves 40∼80% of computation time. It is used in the simulation of gas turbine combustion at low power conditions, and promising results are achieved.

Degree

Ph.D.

Advisors

Zhou, Purdue University.

Subject Area

Mechanical engineering

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS