Banca de DEFESA: Daniely Amorim das Neves

Uma banca de DEFESA de DOUTORADO foi cadastrada pelo programa.
STUDENT : Daniely Amorim das Neves
DATE: 15/12/2023
TIME: 09:30
LOCAL: Microsoft TEAMS
TITLE:

TWO-PHASE FLOW PATTERN ANALYSIS FROM PRESSURE AND VIBRATION SIGNATURE

KEY WORDS:

two-phase flow, flow pattern identification, fluid-structure interaction.

PAGES: 211
BIG AREA: Engenharias
AREA: Engenharia Mecânica
SUMMARY:

The application of indirect approaches for characterising flow patterns and their parameters offers advantages by eliminating the subjectivity inherent in traditional inspection-based methodologies. In addition to being potentially cost-effective and easy to install, these approaches also enable integration with other supervisory systems and the Internet of Things. In this context, this work proposes methodologies for identifying flow patterns and characterising intermittent two-phase flow in horizontal pipes through non-invasive/non-intrusive pressure and acceleration sensors. For this purpose, signal processing techniques are employed in conjunction with physical models. Acceleration, dynamic pressure, and void fraction measurements are obtained from a horizontal line of liquid-gas two-phase flow. 23 experimental points are acquired in the stratified-smooth, stratified-wavy, slug, and dispersed bubble patterns. Fluid-structure coupling is investigated between pressure and structural waves, concentrating on the cut-off frequencies of the circumferential pipe wave modes. A methodology is proposed to demodulate acceleration signals around these cut-off frequencies, which can be estimated analytically. From the demodulated signal, it was possible to classify the flow pattern using values of Hurst exponent and Pearson and Spearman coefficients. Additionally, velocity and frequency parameters in Slug patterns were estimated with errors below 20\%. Furthermore, this work proposes an approach to identify the pattern of two-phase flow through non-invasive pressure measurements. Flow patterns are classified based on dimensionless thresholds from a nonlinear analysis of time series, namely the correlation dimension, Lyapunov, and Hurst exponents. The dynamics of the 23 experimental points are investigated, and the results indicate that the system response is chaotic for all flow patterns due to the positive sign of the Lyapunov exponent. From the Hurst exponent, it was possible to separate oscillatory flow patterns, such as slug patterns, from non-oscillatory flow patterns, such as stratified. The correlation dimension separated smooth and wavy stratified patterns from intermittent and dispersed bubble patterns. Subsequently, it is shown that these parameters form a multidimensional map that clusters the experimental points according to the dynamics of each flow pattern. In light of this, these results open new and interesting possibilities for applications in complex industrial systems.

COMMITTEE MEMBERS:
Externo à Instituição - OSCAR MAURICIO HERNANDEZ RODRIGUEZ - USP
Externo à Instituição - MICHAL KALKOWSKI - SU
Externo à Instituição - FABRICIO CESAR LOBATO DE ALMEIDA - UNESP
Presidente - 2126286 - ADRIANO TODOROVIC FABRO
Interno - 3375759 - TAYGOARA FELAMINGO DE OLIVEIRA