With the increase of wind power capacity in cold climate, an improved understanding of the effects of ice accretion on wind turbine blades would prove to be useful to better estimate the power losses due to icing. Currently, computational fluid dynamics (CFD) is used in conjunction with blade element momentum (BEM) theory to provide power losses estimations. However, most studies so far do not line up with the experimental observations by generally underestimating icing losses. An accurate ice accretion model is a crucial part of the icing estimation process. This paper proposes a method to improve the modelling of ice accretion on a wind turbine nacelle by combining variable high-resolution atmospheric data with the LEWICE software. To validate the methodology, 8 icing events were selected over the winter of 2018–2019, in a Canadian wind farm. Atmospheric data were obtained by a Meteorological Conditions Monitoring Station (MCMS) and images were provided by a camera, both installed on a wind turbine nacelle. The models were compared with experimental data provided by camera images of a reference cylinder. This paper shows that the usage of the LEWICE model improves the concordance with the camera observations compared to the Makkonen model. It was found that the wind orientation on the nacelle could vary significantly during an event and caused conditions outside of the validation range of LEWICE. It was observed that the choice of the meteorological conditions averaging time has a significant impact on the final ice shape. This paper provides valuable insights on the use of high-resolution data and on the LEWICE ice accretion model for ice modelling on the nacelle structures. It becomes a first step before transposing ice accretion to the blades to provide better performance predictions.

随着寒冷气候下风力发电容量的增加，更好地了解积冰对风力涡轮机叶片的影响将被证明有助于更好地估计结冰引起的功率损失。目前，计算流体动力学 (CFD) 与叶片元素动量 (BEM) 理论结合使用以提供功率损失估计。然而，到目前为止，大多数研究并没有通过普遍低估结冰损失来与实验观察结果一致。准确的积冰模型是结冰估算过程的关键部分。本文提出了一种通过将可变高分辨率大气数据与 LEWICE 软件相结合来改进风力涡轮机机舱积冰建模的方法。为了验证该方法，在 2018-2019 年冬季，在加拿大的一个风电场中选择了 8 次结冰事件。大气数据由气象条件监测站 (MCMS) 获得，图像由安装在风力涡轮机机舱上的摄像机提供。这些模型与参考圆柱体的相机图像提供的实验数据进行了比较。本文表明，与 Makkonen 模型相比，LEWICE 模型的使用提高了与相机观察结果的一致性。结果发现，机舱上的风向在事件期间可能会发生显着变化，并导致超出 LEWICE 验证范围的条件。据观察，气象条件平均时间的选择对最终冰的形状有显着影响。本文提供了有关使用高分辨率数据和 LEWICE 积冰模型对机舱结构进行冰建模的宝贵见解。