Statistical Calibration of the HIFiRE Flight 2 Combustion Model

The objective of this work was to develop an approach to statistically calibrate the combustion model parameters of the HIFiRE Flight 2 combustor model. The flight experiment combustor geometry was identical to the HIFiRE Direct Connect Rig tested prior to the flight experiment to map engine operability. Pressure measurements from within the combustor were used to tune the computational fluid dynamics model of the isolator and combustor at tested dual-mode ramjet conditions. Two model parameters were considered: the turbulent Schmidt number and a parameter from the eddy dissipation concept heat release model. The calibration was performed using a Bayesian inference approach. The impact of including model discrepancy within the calibration framework was investigated. Additionally, the impact of calibrating the observation error alongside the model parameters was also evaluated. Results of the statistical calibration showed that the combustion model prior parameter distributions were significantly reduced as a result of the calibration. When including model discrepancy and calibrating the observation error alongside the model parameters, the turbulent Schmidt number posterior had a mean of 0.286 with a standard deviation of 0.015. The eddy dissipation concept parameter posterior had a mean of 0.684 and a standard deviation of 0.012. When not considering model discrepancy and calibrating the observation error, the calibrated model could not accurately capture the measurements. Mean parameter values were found to be similar to those from the full calibration, but the posterior parameter uncertainties were an order of magnitude smaller.