Mid Sweden University

This thesis describes the exploration of a method to measurethe droplet size and the concentration of atmospheric liquid water. The purpose is to find a cost effective technique to detect the conditions for icing on structures.

Icing caused by freezing atmospheric water can be a signifi- cant problem for infrastructure such as power lines, roads and air traffic. About one third of the global installed wind power capacity is located in cold climates, where icing of rotor blades is one of the major challenges.

The icing process is complex and the result depends on a combination of the aerodynamic shape of the structure or airfoil, the velocity of the air and its contained water, the temperature, the mixing of snow and water, the concentration of liquid water and the Droplet Size Distribution (DSD).

The measurement method is based on a shadowgraph imag- ing system using light emitting diode (LED) light as background illumination and digital image processing. A prototype instru- ment has been constructed. The components were selected keeping the possibility of low-cost volume production in mind. The applications of a commercial instrument based on this tech- nique are e.g. real-time in-situ icing condition measurements and assimilation and verification of data in numerical weather models.

The work presented shows that measurements of the size and concentration of water droplets using shadowgraph images can be used for the comparison and validation of Numerical Weather Prediction (NWP) models and other instruments. The accuracy of the particle size measurement is high. The accuracy of the concentration measurement has the potential to become high due to the single-particle measurement range calibraiton. The precision of the instrument depends mainly on the number of images that is used to find each measurement value. The real-time performance of the instrument is limited by the image retrieval and processing speed and depends on the  precisionrequired.

Structural icing is a persistent challenge for the production of renewable energy from wind. It is mainly caused by supercooled atmospheric droplets of water, which are very common in cold climates. In the most exposed wind parks in Sweden, more than 10 per cent of annual energy production can be lost. Some properties of liquid water are included in current Numerical Weather Prediction (NWP) models and are used as input parameters for the estimation of icing, but they are rarely measured in-situ for verification or validation.

To address this problem, a new instrument was developed.This compilation thesis is a collection of five articles describing the development, testing and verification of this instrument. Finally, icing and ice loads are measured and compared with a standard model and a model using Artificial Intelligence (AI) and empirical data.

The new instrument, called Droplet Imaging Instrument(DII), is based on shadowgraph imaging using Light Emitting Diode (LED) light as background illumination and digital image processing. The components were selected with the possibility of low-cost volume production in mind. The applications of a commercial instrument based on this technique include, forexample, real-time in-situ icing condition measurements and assimilation and verification of data in NWP models. The instrument, alongside a reference instrument, was tested in two locations with different icing conditions. Shadowgraph imaging and its limitations as a measurement method for droplet size and concentration were investigated.

The work presented shows that measurements of the size and concentration of water droplets using shadowgraph images can be used for the comparison and validation of NWP models and other instruments. The Coefficient of Variation (CV) for a given value of the concentration is lower than 1.6 % for droplets 25 µm in diameter, based on uncertainty in the size measurement only. The accuracy of the sampling volume can be improved by measuring the background light intensity in the position of the measured droplet.

A fog chamber was used for initial tests. However, to evaluate models of ice accumulation, in-situ measurements are necessary. These measurements should use a temporal resolution of at least one sample per minute, preferably higher. With alimited amount of data, multivariate data analysis can be used to estimate the level of ice accretion. Together with a heuristic model of erosion/ablation, the resulting figures can be used to simulate the ice load.

All of the instruments, as well as many other components used during the described field measurements, did at some point break due to the difficult weather conditions. An instrument for measurement of icing conditions needs to be designed with high environmental protection and endurance. The results in the attached papers may help and motivate further technical development of instruments that can measure atmospheric liquid water in icing conditions.