Experimental Investigations of Different Microphone Installations for Active Noise Control in Ducts

A request on ventilation systems today is the feature of a low noise level. A common method to attenuate ventilation noise is to use passive silencers. However, such silencers are not suitable for the lowest frequencies and one solution is to use active noise control (ANC) to increase the noise attenuation in the low frequency range. Normally when using a feedforward ANC system to attenuate duct noise, both the reference microphone and the error microphone are exposed to airflow. As the airflow excites the diaphragm of the microphones, the microphone signals become contaminated by uncorrelated pressure fluctuations that are not part of the sound propagating in the duct. By reducing the flow velocity around the microphones, these uncorrelated pressure fluctuations can be reduced and the noise reduction improved. One way to reduce the flow velocity around the microphones is to place the microphones in outer microphone boxes connected to the duct via a small slit. In this paper a new practical design for the reduction of flow velocity around the microphones is presented; the microphone installation is based on a T-duct, and therefore it makes maintenance and especially construction easier, compared to the microphone box with a slit. Furthermore, comparative results concerning the performance of an ANC system for the two different microphone installations, the T-duct configurations and the microphone boxes with varying slit width, are presented. The results show that the active noise control performance is almost equal when using the suggested microphone installation as compared to when using a microphone box with a slit. M. Larsson, S. Johansson, L. Håkansson and I. Claesson INTRODUCTION A low noise level, which contributes to human well-being, is a factor of high importance in schools, factories, office buildings etc, as well as in our homes. In these environments ventilation systems constitute one well known noise source. The classical remedy to noise generated by such systems is passive silencers [1], i.e. dampers containing sound absorbing material. However, because low frequencies have long wavelengths, these passive silencers tend to be relatively large and bulky when used to attenuate noise in the low frequency range. A well known method to attenuate low frequency noise in various situations is active noise control (ANC) [2, 3]. While ANC is best suited for low frequencies, passive silencers are best suited for higher frequencies and therefore a combination of the two often is an attractive solution. A single-channel feedforward adaptive control system used to attenuate ventilation noise generally consists of two microphones, one loudspeaker and a control unit. One microphone –a reference microphone– is placed upstream relative the loudspeaker. The reference microphone detects the noise propagating in the duct and generates a reference signal which is fed to the control unit that steers the loudspeaker. Downstream from the loudspeaker, the other microphone –an error microphone– is placed. The error microphone senses the residual noise after control and generates an error signal which is also fed to the controller. The referenceand error signals allow the controller to adjust itself to continuously minimize the acoustic noise sensed by the error microphone. It does this by creating an output via the loudspeaker that is based on the reference signal and out of phase with the sound propagating in the duct by the time it reaches the placement of the error microphone. A continuous problem when applying ANC to duct noise is the airflow present in the ducts that the microphones are exposed to. Placing the microphones in airflow will result in contamination of the microphone signals, since they will contain a contribution of turbulence pressure fluctuations arising when the airflow excites the diaphragm of the microphones. A high level of turbulence compared to the level of noise propagating in duct will lead to less correlation between the referenceand error signals. This in turn results in a decreased performance of the ANC system [2, 3]. Therefore it is essential to reduce the amount of uncorrelated turbulence fluctuations which not are a part of the propagating sound, to optimize the noise attenuation potential of the active control system. A common way to do so is by placing the microphones in outer turbulence boxes connected to the duct via a small slit [3, 4]. As shown in [4] the performance of an ANC system applied to duct noise can be significantly improved by placing the referenceand error microphones in outer microphone boxes. However, such microphone boxes implies a new construction of the duct pieces in which the microphones are placed. In this paper a microphone installation based on a standard T-duct is presented. Since the microphone installation is based on a duct piece already manufactured, eliminating the need for the development of new ICSV13, July 2-6, 2006, Vienna, Austria duct pieces, this of course makes it an attractive solution to manufacturers of ventilation systems. Furthermore, comparative results concerning the performance of an ANC system with different microphone installations; T-duct configurations and microphone boxes with varying slit width are presented. The results show that the active noise control performance is almost equal or better when using the suggested T-duct based microphone installation as compared to when using a microphone box with a slit. THE EXPERIMENTAL SETUP The measurements were carried out in a laboratory at Lindab AB in Farum, Denmark and on a duct system built in a laboratory at Blekinge Institute of Technology (BTH), Sweden. The duct system used at BTH was circular with a length of approximately 21 meters and a diameter of 315 mm. The system was equipped with a standard axial fan (Lindab CK315), a passive silencer (Lindab SLU 100) and a draught valve close to the fan to regulate the airflow. In these measurements two different airflows were used; 3,2 m/s with the draught valve closed and 6,7 m/s with the draught valve completely open. The passive silencer, the loudspeaker and the error microphone were located near the duct outlet. The attenuation was evaluated in the error microphone. Figure 1 is a diagrammatic illustration of the experimental setup at BTH which henceforth will be referred to as setup1. The laboratory at Lindab AB had the possibility Loudspeaker Error mic Microphone box Microphone box