The procedure began by collating data from the Met office and Irish weather data. With this data an average rainfall map over the last 20 years was created. It should be noted that the turbine blades will inevitably experience varying rainfall rates throughout the year, which will, in turn, result in varying erosion rates. To map this phenomenon, the months of highest and lowest rainfall were chosen which were January and May, respectively. Showing the two extremes of rainfall months will provide more insight than a yearly average and allow for contrast to be observed between the 2 months. The result of this can be seen in Fig. 3.
Upon the completion of this map an experimental process in the Tribology laboratory at the University of Strathclyde was carried out to relate rainfall across the country to the degradation of wind turbine blades. There are many different ways to test for rain erosion [8]; however, the test method used for this case study is the whirling arm setup [9]. This consists of a large chamber with induced rainfall from hypodermic needles; in the centre of the chamber is a rotating arm which holds the material sample at the end. Adjusting the rotational speed of the spinning arm will adjust the impact velocity the sample encounters with the water droplet. The experimental setup can be seen in Fig. 1 and a schematic is shown in Fig. 2.
The material used in this experiment was G10 epoxy glass which is a similar glass fibre epoxy composite used within the wind turbine manufacturing industry [10]. The impact velocity was set to 60 ms−1 as this will simulate the leading edge of 2 MW turbines with a diameter of 100 m. One of the assumptions that was made to relate the experimental data to the weather data was that the wind turbines were always turning when it was raining. Although this might result in an overestimation in erosion, it is deemed a worst-case scenario that has a possibility of occurring.
All the variables which were kept constant including impact velocity, temperature and rainfall were all calibrated before the experimental campaign began. The impact velocity was calibrated using light transducer which was held close to the rotating shaft where a thin reflective strip was placed. A light source was focused onto the shaft and the transducer would give an electrical output when the light was reflected from the reflective strip each time it would rotate. The rain fall was calibrated by running the rain system for five hours and the water tank which feeds the rig was weighed periodically every 30 min to calculate the water consumed and hence the rainfall rate. The pump used was a peristaltic pump which proved to be extremely reliable and hence outputted 50 mm/h every 30 min. The temperature inside the rig was also measured and kept at 29° C, a temperature calibration test was carried out during the calibration of the rainfall where the temperature was measured using a probe inside the chamber and a reading was taken every 30 min for the five hours and the temperature only fluctuated ± 1° C.
The same procedure was carried out for all samples which included a 48-h drying period before measuring the mass and kept in the same container to ensure the conditions when the sample was drying were kept constant. After the 48 drying period the samples were weighed on a balance accurate to 0.00001 g and the mass of the sample was measured five times equally spaced out over one hour. From this a measurement error of ± 0.001% and standard deviation of 1.09E−05 was calculated for the neutral water and a measurement error of ± 0.019% and standard deviation of 1.16E−04 for the saltwater experiment.
The construction of the rainfall map (Fig. 3) allows the setup for the experiment to be finalised. The key to the map displays: Below 50 mm, 50 mm, 75 mm, 100 mm, 150 mm, 200 mm, 300 mm, 500 mm and above 500 mm. The rainfall rate of the whirling arm rig is 50 mm/h therefor the time each sample exposed to rain erosion can be determined, this is.
shown in Table 1.
Table 1 Exposure time in erosion rig to achieve required rainfall The chosen measurement for erosion is mass loss as a percentage of the original sample. The mass of the test material was measured before the experiment and after each exposure time. This would result in a direct numerical relation between the average monthly rainfall and the erosion as a mass loss.
This methodology was repeated with saltwater instead of rainwater to simulate offshore conditions. The saltwater used was a 3.5% saline solution which most accurately describes the seawater in the UK and Ireland [11]. The results and errors can be seen in Table 2.
Table 2 Mass loss results from erosion testing