Ultrasonic Two-Phase Flow Measurements Based on Pattern Recognition Techniques

The state-of art of ultrasonic two-phase flow measurements is characterised by a number of different approaches commonly based on the identification and characterisation of individual voids (bubbles, plugs etc.) applying the techniques of ultrasonic testing. The recorded individual events are integrated to extract parameters as void fraction or volume flow rates. The main limitation of these methods arises from the complicated structure of two-phase flow at higher void fractions which leads to multiple diffractions of the sound beam. The measurement is therefore limited to low void fractions or a simple flow structure.
The main idea of the present work was to overcome these limitations by means of pattern recognition. An ultrasonic beam crossing the two-phase flow is modulated by the changing structure of the voids passing by and therefore the through-transmission signal must contain information about the parameters of the two-phase flow even if information about individual flow effects cannot be derived. Therefore it was supposed that a pattern recognition algorithm trained with signals obtained at known conditions is able to identify the set of the flow parameters (flow rates, void fraction etc.) in an unknown situation.