Noise identification of confined orifice flow from sparse experimental data using a pressure decomposition framework

The present study proposes a pressure decomposition framework designed to decouple hydrodynamic and acoustic components from sparse acoustic measurement data, effectively identifying the flow-induced noise of a confined orifice in lithography. The framework involves three primary steps: peak detection, mode decomposition, and component identification. By employing spectral analysis and spectral proper orthogonal decomposition, the framework extracts key information on discrete tonal frequencies, amplitudes, and waveforms, reconstructing coupled hydrodynamic and acoustic pressures into new modal representations. Component decomposition is further achieved through wavenumber–frequency spectrum analysis, revealing the characteristic phase velocity of the reconstructed modes. An acoustic experiment was conducted using a microphone array to evaluate the noise identification performance. The findings indicate four characteristic zones within the fluid dynamic and acoustic pressure pulsations, with acoustic components prevailing in the low and mid-frequency ranges, particularly associated with large-scale vortex structures. Finally, the production mechanisms of the identified hydrodynamic and acoustic pressure pulsations were further revealed by solving the eigenvalue problem of the compressible linearized Navier–Stokes equations in the frequency domain. The results support that the decomposed sound pressure features a low attenuation factor, allowing for long-distance propagation with minimal loss.