Metodologia de projeto e otimização acústica de mufflers reativos aplicados a grupos geradores utilizando algoritmo genético com método dos elementos finitos

Abstract

This work proposes the development of an optimized reactive acoustic silencer for exhaust systems of generator sets, based on a case study that adopts as reference a commercial model with an average attenuation of 35 dB(A), employed in an industrial environment. The main objective is to reduce the noise emitted and ensure compliance with current standards. Although essential in several applications, generator sets constitute significant sources of noise pollution, especially during the exhaust phase of internal combustion engines. The research employs numerical simulations based on the Finite Element Method (FEM), integrated with Genetic Algorithms (GA), to explore nonconventional silencer geometries, such as configurations with lateral inlet and asymmetric baffles. The objective function was defined to maximize the Transmission Loss (TL) within a frequency range of 1500–2500 Hz, strategically selected based on the primary noise source of the system. Partial experimental validation was conducted using the twoload method, in accordance with ASTM E2611-09. In addition, field tests based on the ABNT NBR ISO 8528-10 standard were proposed. The results demonstrate the robustness of the Genetic Algorithm and the potential of the methodology to generate simpler and lower-cost reactive silencers while maintaining acoustic performance comparable to the combined silencer (reactive and resistive) adopted in the commercial practical case. This finding is corroborated by simulations and initial tests, reinforcing the effectiveness of the proposed approach.