Gradient based optimization has been used for the acoustic optimization of suction mufflers of household compressors with cooling capacities around 100 Watts. An optimum design has been sought, where the acoustic power transmitted through the muffler is simultaneously minimized around the cut-on frequencies of (i) the acoustic cavity resonances and (ii) the structural modes of the hermetic shell of the compressor. The transmitted acoustic power is calculated by use of the direct boundary element method. By use of the optimization procedure, it is attempted to reduce the transmitted sound power through the muffler into the cavity and hence to lower the ability of the source to acoustically excite the shell. Optimization is performed using a sequential quadratic programming method. Since the acoustic analysis is performed using a commercial acoustics program, the sensitivity analysis is performed using finite difference approximations. The design is parameterized using linear combinations of different perturbed model meshes. To this end the user has to specify allowable design changes within the mesh generator. For structured meshes, moderate design changes can be achieved by using this procedure without deteriorating the quality of the mesh. The design modifications suggested by the sequential quadratic programming method have been verified experimentally to give a significant reduction of the radiated power around the cavity and structural modes of the compressor.
展开▼