High Pressure Optical Measurements of Temperature at Turbine Rotor Inlet Conditions

The measurement of combustion product gas temperature is valuable for the development and control of many combustion systems. In gas turbines engines, measurement of the rotor inlet temperature remains particularly challenging because of harsh operating conditions and limited access. The integrated spectral band ratio (ISBR) method is a non-intrusive optical emission temperature measurement technique suitable for this application. As with two-color pyrometry, the ISBR method utilizes the ratio of spectral band emission, but instead of using emission from a gray broadband surface, emission from water vapor bands in the near infrared region are used. Instead of using the Planck equation to correlate temperature to the emission ratio, a correlation between water spectral band emission and temperature is derived from a model of water vapor emission. The model shows the correlation is relatively insensitive to water concentration and measurement pathlength. In this paper, the ISBR method was extended from an atmospheric to a pressurized system as part of a larger effort to measure rotor inlet temperature in a functioning gas turbine engine.

Optical fibers made of sapphire were used to transmit the radiative signal from the post combustion zone to a Fourier Transform Infrared Spectrometer (FTIR) without the need for cooling. Gas emission spectral bands, nominally 100 cm^-1 wide between 4600 and 6200 cm^-1 were investigated. Optical line-of-sight and thermocouple point measurements were obtained during two temperature sweeps; one at high load and one at low load (pressures of 1.2 and 0.7 MPa, respectively). The three thermocouples placed in the flow at an axial distance 76 mm downstream of the optical measurement but at various radial distances from the centerline showed a stratified, non-uniform temperature with differences between thermocouples on the order of 180 K. The average of these thermocouples was consistently approximately 200 K lower than the temperature obtained from the optical measurement. The difference in temperatures can be attributed to radiative cooling bias of the thermocouple, the random uncertainty of averaging both measurements, and the potential for inlet air cooling between the optical and thermocouple locations. The change in optical temperature during the sweep was nearly the same but slightly higher (3% over an 87 K sweep) than the change in average thermocouple temperature. Repeatability of the optical measurement at a given operating condition was on the order of ± 15 K and the absolute uncertainty of a single optical measurement was estimated to be ± 70 K. A linear correlation with an R-squared value of 0.97 was also found between raw optical signal and thermocouple measurements suggesting that once a calibrated measurement is obtained, changes in temperature can be determined using a correlation of the raw signal to produce the temperature.