Advanced methods for validating combustion instability predictions using pressure measurements
The utility of high fidelity (HF) pressure measurements from Continuous Variable Resonance Combustor (CVRC) is explored in trying to find better ways to validate in-house developed simulation models with experiment results; as well as developing predictive method using pressure time series for instability onset. ^ Phase averaged wave travelling speed calculation from concurrent simulation and experiment case show comparable results, however to use an averaged quantity in turbulence driven combustion instability environment, cycle to cycle variations are averaged out and many details of combustion instability are lost. ^ A Cross-correlation method is applied to CVRC HF experimental pressure data. Analysis of the data indicated very high pressure signal correlation between different chamber location during combustion instability, confirming the concept of a travelling wave inside of an acoustically closed combustion chamber. Phase angle shifts during CVRC translating oxidizer post length (LOP) tests show interesting combustion behavior change (shifting of combustion heat release location), indicating the potential of HF pressure measurements in studying detailed behavior during combustion instability. Cross-correlation analysis is also applied to selected pressure cycles in fixed post cases to find the speed of sound and mean flow velocity of the chamber to compare simulation and experiment. ^ Another attempt at high-fidelity comparison is to convert the CVRC HF pressure time series into a visibility graph using visibility algorithm, then plotting the log- log degree distribution (DD) to see how well simulation pressure data compare to experiment results. The analysis result for fixed post 13.97 cm length shows the pressure signal behavior change that could be used for predicting instability onset. Simulation vs. experiment DD plot also shows the potential for validation. By plotting simulation and experiment DD together, locations in the chamber where the simulation pressure signal detail matches better with the measured pressure signal are identified.^
William E. Anderson, Purdue University.
Off-Campus Purdue Users:
To access this dissertation, please log in to our