Skip to main navigation Skip to search Skip to main content

Study on the internal pressure via the combination of the proper orthogonal decomposition and dynamic mode decomposition methods

  • School of Civil Engineering, Harbin Institute of Technology

Research output: Contribution to journalArticlepeer-review

Abstract

The internal pressure induced by a dominant wall opening threatens the safety of claddings in the building envelope. This paper investigates the influence factors of internal pressure by conducting the wind tunnel test. The combination of proper orthogonal decomposition (POD) and dynamic mode decomposition (DMD) methods is applied to analyze the mechanism of internal pressure attenuation and amplification. POD–DMD methods can inhibit unreal modes with high frequencies effectively and extract the physical modes, corresponding to the mean pressure and Helmholtz resonance. The most dangerous internal pressure is induced by the windward and sideward openings. The high-order DMD mode at the Helmholtz frequency reveals the dynamic of air flow through the opening. The mode energy in the internal space is strengthened by a smaller internal volume, a larger opening area, a thick opening, and a lower wind velocity. Current wind loading codes may be unconservative to evaluate the internal pressure with a dominant wall opening in the testing flat-roofed building when the wall opening ratio (the opening area/the total wall area) is in the range of 1%–3.1%. The non-dimensional empirical fitting formulas based on the covariance integration principle are proposed to estimate the internal pressure of the flat-roofed and cylindrical-roofed buildings. The estimating errors are within 10% of testing values, which is much lower than that estimated by the previous approach.

Original languageEnglish
Article number078108
JournalPhysics of Fluids
Volume38
Issue number7
DOIs
StatePublished - 1 Jul 2026

Fingerprint

Dive into the research topics of 'Study on the internal pressure via the combination of the proper orthogonal decomposition and dynamic mode decomposition methods'. Together they form a unique fingerprint.

Cite this