Phase Discrimination in Marine Icing Using a Coplanar Capacitive Array

Abstract

This paper describes the development of an array of coplanar capacitive sensors applied to marine icing. Current atmospheric icing monitoring systems consider single phase conditions in their operation. Marine icing conditions present a unique environment where the liquid water phase effects cannot be neglected and require a novel approach. We have conducted an initial proof of concept and propose a new icing monitoring system which can distinguish between the individual phases. A numerical model confirmed our initial hypothesis of the system's ability to discriminate the multiphase domains based on the array of geometrically dissimilar capacitive sensors. In addition, we also developed a novel experimental technique based on a comparative study under constant conditions to eliminate the need for an independent ice accretion monitoring system normally required in sensor development. The new approach promises a better characteristic in marine icing monitoring systems or in similar applications where multiphase dielectric is present.

Marine Icing Sensor with Phase Discrimination

“…The dotted line corresponds to 0.35 mm dielectric coating layer. The graph demonstrates a linear independence in a similar way to the gap variation analysis in [ 12 ] though not as profoundly. The FEM parameters are listed in Table 1 .…”

“…This phenomena was utilized in discriminating each individual phase, ice and water. Our team developed a multi-dimensional least-squares-method [ 12 ] to map the measured data to the measurands. The least-squares method was later substituted by a more robust and more accurate technique based on artificial neural networks method [ 13 ].…”

“…We have conducted a series of ice layer deposition experiments in a similar way described in [ 12 ] using a water saturated roller. The system was placed in a deep freezer at −20 C and was initially tempered for 1 h. Then, over a period of two hours, we have periodically rolled a water layer onto the sensor surface which is illustrated in Figure 4 .…”

“…We have developed a sensor array that allows a flexibility to be applied on curved surfaces such as wind turbine wings. Our earlier work described in [ 12 ] utilized a planar array of electrostatic sensors of different spacing that have been instrumented by a custom circuitry based on LC oscillator. Here we report on using a constant-spacing electrostatic sensor array interfaced and conditioned by a commercial four-channel capacitive pickup board by Texas Instruments.…”

“…Previously we processed the array signals by a multidimensional least squares method [ 12 ] and in another approach by artificial neural networks [ 13 ]. In this work, we applied a machine learning strategy centered around a decision tree method.…”