Nilsson E.

The main noise sources of an industrial gas turbine installation is the air intake and exhaust. Big dissipative silencers are usually used to reduce the emitted noise. For gas turbine applications duct measurements before the silencer and chimney measurements after the silencer usually do not agree with the predicted sound insertion loss. This could be due to measuring problems or actual lack of attenuation of the silencer. The measurement environment in the duct is usually very difficult. For the air-intake the temperatures are normal but for the exhaust system the temperatures are 5-600oC with extremely turbulent flows. Usually only one or a few measurement positions are possible. To increase the knowledge of how to interpret the measurement results from the duct systems a study has been done on measurement technique in a scaled model duct with an air-flow in the duct. Flush mounted pressure transducer on a man-hatch is usually the only way to mount a transducer on the exhaust side. Scanning of the duct cross-section was done in the model measurements to reveal how well the flush measurement position represents the total pressure in a duct with flow. The measured levels are shown to depend on the choice of measuring position. For low frequencies the flush mounting is overestimating the average sound pressure level of the duct cross-section. For the mid-frequency range the levels are well estimated while the overestimation reoccurs at higher frequencies. The total sound pressure level is only exceeding the scanned value by a few dB. This is seen for real applications as well but the variations do not follow the scaling. The variations are seen for approximately the same frequency bands for both laboratory scale model and real scale measurements. The paper will present measurement results and discussions in order to verify the measurement results.