AUTO-REGRESSIVE MODELING FOR FLOW PATTERN CHARACTERISATION

Erik J.W.M. Legius, Robert F. Mudde, Harry E.A.Van Den Akker

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

The two-phase flow of air/water has been investigated in two 'complex' geometries: a vertical pipe with contraction and a line/riser system with a bend. Pressure recordings and optical fiber measurements have been analysed using Auto-Regressive modeling. This type of modeling describes the data as a linear combination of previous data. It is discussed how the transition from one flow regime to an other is influenced by the flow obstructions. The Auto-Regressive modeling provides a quantitative measure of the similarity between two flow realisations. It is found that identification of flow regimes based on this measure coincides with visual observation. With the Auto-Regressive technique the width of the transition could also be quantified. Furthermore it is shown, that the upward bend stabilises the slug flow in the vertical part downstream of the bend: the transition towards churn flow is postponed.

Original languageEnglish
Title of host publicationHeat Transfer
Subtitle of host publicationVolume 5 � Numerical and Experimental Methods in Heat Transfer
PublisherAmerican Society of Mechanical Engineers (ASME)
Pages323-333
Number of pages11
ISBN (Electronic)9780791826744
DOIs
Publication statusPublished - 1998
Externally publishedYes
EventASME 1998 International Mechanical Engineering Congress and Exposition, IMECE 1998 - Anaheim, United States
Duration: 15 Nov 199820 Nov 1998

Publication series

NameASME International Mechanical Engineering Congress and Exposition, Proceedings (IMECE)
Volume1998-Q

Conference

ConferenceASME 1998 International Mechanical Engineering Congress and Exposition, IMECE 1998
Country/TerritoryUnited States
CityAnaheim
Period15/11/9820/11/98

Fingerprint

Dive into the research topics of 'AUTO-REGRESSIVE MODELING FOR FLOW PATTERN CHARACTERISATION'. Together they form a unique fingerprint.

Cite this