How Ice Accretion Affects Wind Turbines
For the wind energy industry a 'cold climate' site refers to a location that might experience significant periods of time or frequency of icing events or low temperatures outside the operational limits of standard wind turbines. In recent years many countries in Europe, Asia and North America have had to develop wind farms in cold climate regions. The main reasons for this choice are good wind energy resources in high altitudes (e.g. Switzerland with sites at 800 metres above sea level), and the higher installation and O&M costs for offshore when compared with land-based wind farms. This new development has brought some new challenges for designers, manufacturers and operators. One of these challenges is icing of the wind turbine blades and its effects on the aerodynamics and responses of the wind turbine. This article addresses ice accretion on the blade and its effects on the aerodynamic properties of the rotor.
By Mahmoud Etemaddar, PhD candidate, Institute of Marine Technology, NTNU, Norway
The parameters that affect wind turbine icing can be divided in to two main groups: environmental parameters and system parameters. The environmental conditions for atmospheric icing are characterised by four parameters: liquid water content of the air (LWC), mean water droplet diameter (MVD), ambient temperature (T) and relative humidity (RH). LWC is the measure of mass of water in the air (this is typically measured in a standard volume of air such as grams per cubic metre), MVD is the mean diameter of water particles in the air in micrometres and RH is the ratio of the partial pressure of the water vapour in the air–water mixture to the standard water vapour pressure of a flat sheet of pure water in those conditions. In general the rate of ice accretion increases with LWC and MVD and reduces with increasing the temperature. The RH for icing simulation is usually equal to 100%. The average range of LWC and MVD for wind turbine atmospheric icing can be 0.01 to 0.15g/m3 and 5 to 40µm respectively.
Water particles in the air can remain in the liquid phase even below freezing temperature, and these are called supercooled water particles. This is the main cause of icing in land-based wind farms during the winter time when the temperature goes below 0°C. These water particles in the air are frozen as soon as they crash into solid obstacles such as wind turbine blades. This phenomenon is called atmospheric icing.
For offshore wind farms that are located in cold climate regions such as the north of the North Sea, icing may originate either from seawater (spray icing) or from supercooled freshwater particles in the air. The sea-spray icing is mainly important for the lower level part of a wind turbine tower. For example for offshore multi-megawatt wind turbines observations show that above 4 metres only very slight icing is caused by spray directly from waves at wind speed less than 25m/s, and therefore only atmospheric icing is usually observed on the rotor.
Atmospheric Icing on Wind Turbine Blades
The ice starts growing from the crashing point, which is always the leading edge of the blade, and then extends to both sides of the blade. This means the ice is mainly accumulated on the first half of the chord length from the leading edge of the blade. Sometimes, due to the 'round back' effect of unfrozen water, the ice can also grow even on the trailing edge. Although atmospheric conditions, as well as incoming wind speed, are almost the same for the entire length of the rotor, ice growth is not constant. This is because some system parameters, in addition to the environmental parameters, affect the icing of the blade. Relative wind speed and object thickness are the two main system parameters for ice accumulation. The relative wind speed (VR) at each point along the blade is the superposition of incoming wind speed (Vw) and shaft speed times the distance from the rotor centre (?.r). The relative wind speed increases linearly from the root to the tip, while the incoming wind speed is almost constant. The other parameter is the thickness of the icing object, which for a wind turbine blade is the thickness of the blade. Due to structural requirements the blade thickness reduces from the root to the tip and the ice distribution thus varies in both the chord-wise and span-wise directions of the blade. This is mainly because the relative wind speed increases and blade thickness reduces along the blade from the root to the tip. The rate of ice accretion increases with relative wind speed and reduces with object thickness. Balancing both these factors means that ice thickness increases from the root to the tip of the blade.
The outer appearance and internal structure of ice is different according to the atmospheric condition. The type of ice encountered on wind turbines is generally divided into three groups, mainly based on the density. Glaze ice is hard, almost bubble free, clear, homogeneous ice with a density close to that of pure ice, 920kg/m3. Hard rime ice is rather hard, granular, and with a density of 600–900kg/m3. Finally, soft rime ice is white or opaque ice with a loosely bonded structure and a density of less than 600kg/m3. There are two main effects which result from icing on the wind turbine blade. These are change in aerofoil geometry due to macro-scale effects and change in surface roughness due to micro-scale effects. The first can change the flow distribution around the blade (and can be simulated with sufficient accuracy using potential flow theory) while the second affects the boundary layer and flow separation near the aerofoil surface.
Effects of Icing on Aerodynamics
As mentioned above, the primary effect of icing is change to the leading edge geometry and blade surface roughness, which alter the aerodynamic properties of the blade depending on the type and size of the ice build-up. Mass distribution along the blade, and thereby blade natural frequencies, may also be changed by icing. This effect is of secondary importance compared to the aerodynamic effects because of the small ratio of the accumulated ice mass to the blade mass at the first stage of ice accumulation.
Rotor aerodynamics are characterised by two generalised aerodynamic coefficients, which are the power (Cp) and thrust (CT) coefficients. These parameters are a function of the blade pitch angle ?, the tip speed ratio TSR (which is the ratio of the rotor tip speed to the incoming wind speed), as well as lift (CL) and drag (CD) coefficients for each of the aerofoil sections along the blade. The lift and drag coefficients are directly affected by icing due to the change in geometry and surface roughness of the blade. The shaft speed is indirectly affected by the lift and drag coefficients of the blade when moving at above the rated wind speed through the actions of the controller. Taking into account the non-linearity of the wind turbine system one can imagine how complicated the effect of icing on the performance of the wind turbine can be. If the wind turbine is considered as a system with wind speed as an input, it is obvious that the wind speed is not affected by icing. But the controller performance could be altered by icing. Some controller parameters, such as pitch actuator gains, depend on the slope of the power coefficient curves, but the effect of icing on controller performance is a secondary effect and one can focus only on the change in lift and drag coefficients of the blade to see the primary effects of icing. In the following section the effect of icing on the lift and drag coefficients of the blade, and consequently their effects on power and thrust coefficients, will be described.
As an example, the ice accumulation on the blade after 4 hours working under atmospheric icing conditions (T = –10°C, LWC = 15g/m3, MVD = 15µm and RH = 100%) were simulated with LEWICE on a NREL 5MW virtual wind turbine. The results show the negligible icing on the inner two-thirds of the blade and linear ice growth on the outer one-third of the blade. This part of the blade is constructed using the NACA 64618 profile. Figure 1, in the middle, shows the 2D ice profiles on this part of the blade at five different longitudinal sections.
The lift and drag coefficients of the aerofoil 2D sections can be determined by wind tunnel test experiments or by computational methods such as CFD; each method has its own benefits and difficulties. The rational approach is to set up a numerical model for one or a few sections, validate it against some experimental results and then use it for the rest of the blade. The turbulent method based on the Reynolds Average Navier Stokes (RANS) is one of the approved methods in the aviation and wind energy industries to estimate the aerofoil aerodynamic properties. The lift and drag coefficients of a clean NACA profile, compared to five 'iced' NACA profiles, are illustrated on the left side and right side of Figure 1 respectively. It is obvious that at this level of icing and with a low angle of attack (which is roughly between –7 and 7 degrees) the drag coefficients are much more affected by icing when compared to lift coefficients. This means that the effect of icing on aerodynamic coefficients is mainly due to a change in drag coefficients. Because of the non-linearity of the system, it is difficult to specify directly the effect of the change in lift and drag coefficients on the response. This is why it is wise to investigate the effect of icing on power and thrust coefficients which are representative for the whole rotor. Figure 2 shows the power and thrust coefficients as a function of tip speed ratio for four different pitch angles from 0 to 15 degrees before and after icing. The solid lines and dashed lines are the results for clean and iced rotors respectively.
Results and Conclusion
These results show that the icing reduces power outputs in the whole operating region of the wind turbine, while the thrust is increased in some regions and reduced in others. As a result of this work it is now easier to investigate the effect of icing on the response of the wind turbine. In variable speed, pitch regulated wind turbines the whole operating region can be divided into two sections. The first (region (1)) starts when cut-in wind speed is achieved (i.e. the minimum wind speed at which power production is practicable) and finishes when rated wind speed is reached (where the output power reaches the maximum power of the wind turbine), while region (2) is the region where the turbine is working at its rated wind speed until cut-out wind speed is reached (where the wind turbine has to be shut down due to structural limit states and safety). In region (1) tip speed ratio is constant and pitch angle is set to a small value near zero. When the turbine goes to region (2), as the wind speed increases the shaft speed remains constant while the tip speed ratio starts to reduce as the pitch angle is increased by the pitch controller in order to alleviate the loads on the structure. As illustrated by Figure 2, in below rated wind speed, both the power and thrust will be reduced due to icing; thereby loads and responses due to aerodynamic loads should be reduced. Above rated wind speed, where TSR starts reducing and pitch angle starts increasing, we should expect less deviation from nominal power and higher thrust compare to the clean rotor.
These results shows that in below rated wind speed one can shift the cut-in wind speed to a higher value to get the beneficial minimum power without having to worry about additional structural loads, but one should also shift the cut-out wind speed to a lower value (to prevent the wind turbine from failure in above rated wind speed); however it seems likely that the rated power will only be reached at higher wind speeds. Region (2) therefore occurs over a narrower band of wind speeds in the iced condition when compared to non-iced normal conditions. However, because these days all wind turbines that are designed for cold climate regions are equipped with de-icing or anti-icing systems the wind turbine behaviour under icing conditions described in this study should rarely be encountered 'in the field'. Nevertheless, the study should prove useful for ice-detection, safety and system optimisation purposes.
Biography of the Author
Mahmoud Etemaddar is a PhD candidate at the Institute of Marine Technology at NTNU (2009–2013). His PhD project is ‘Offshore Wind Turbine Analysis under Fault Conditions’. He has an MSc in Marine Engineering (2000–2005) from Amirkabir University of Technology, Tehran, Iran, and worked as a Marine and Offshore Platform Engineer between 2005 and 2009.{/access}
For the wind energy industry a 'cold climate' site refers to a location that might experience significant periods of time or frequency of icing events or low temperatures outside the operational limits of standard wind turbines. In recent years many countries in Europe, Asia and North America have had to develop wind farms in cold climate regions. The main reasons for this choice are good wind energy resources in high altitudes (e.g. Switzerland with sites at 800 metres above sea level), and the higher installation and O&M costs for offshore when compared with land-based wind farms. This new development has brought some new challenges for designers, manufacturers and operators. One of these challenges is icing of the wind turbine blades and its effects on the aerodynamics and responses of the wind turbine. This article addresses ice accretion on the blade and its effects on the aerodynamic properties of the rotor.
By Mahmoud Etemaddar, PhD candidate, Institute of Marine Technology, NTNU, Norway
{access view=!registered}Only logged in users can view the full text of the article.{/access}{access view=registered}
Atmospheric Icing and Sea-Spray IcingThe parameters that affect wind turbine icing can be divided in to two main groups: environmental parameters and system parameters. The environmental conditions for atmospheric icing are characterised by four parameters: liquid water content of the air (LWC), mean water droplet diameter (MVD), ambient temperature (T) and relative humidity (RH). LWC is the measure of mass of water in the air (this is typically measured in a standard volume of air such as grams per cubic metre), MVD is the mean diameter of water particles in the air in micrometres and RH is the ratio of the partial pressure of the water vapour in the air–water mixture to the standard water vapour pressure of a flat sheet of pure water in those conditions. In general the rate of ice accretion increases with LWC and MVD and reduces with increasing the temperature. The RH for icing simulation is usually equal to 100%. The average range of LWC and MVD for wind turbine atmospheric icing can be 0.01 to 0.15g/m3 and 5 to 40µm respectively.
Water particles in the air can remain in the liquid phase even below freezing temperature, and these are called supercooled water particles. This is the main cause of icing in land-based wind farms during the winter time when the temperature goes below 0°C. These water particles in the air are frozen as soon as they crash into solid obstacles such as wind turbine blades. This phenomenon is called atmospheric icing.
For offshore wind farms that are located in cold climate regions such as the north of the North Sea, icing may originate either from seawater (spray icing) or from supercooled freshwater particles in the air. The sea-spray icing is mainly important for the lower level part of a wind turbine tower. For example for offshore multi-megawatt wind turbines observations show that above 4 metres only very slight icing is caused by spray directly from waves at wind speed less than 25m/s, and therefore only atmospheric icing is usually observed on the rotor.
Atmospheric Icing on Wind Turbine Blades
The ice starts growing from the crashing point, which is always the leading edge of the blade, and then extends to both sides of the blade. This means the ice is mainly accumulated on the first half of the chord length from the leading edge of the blade. Sometimes, due to the 'round back' effect of unfrozen water, the ice can also grow even on the trailing edge. Although atmospheric conditions, as well as incoming wind speed, are almost the same for the entire length of the rotor, ice growth is not constant. This is because some system parameters, in addition to the environmental parameters, affect the icing of the blade. Relative wind speed and object thickness are the two main system parameters for ice accumulation. The relative wind speed (VR) at each point along the blade is the superposition of incoming wind speed (Vw) and shaft speed times the distance from the rotor centre (?.r). The relative wind speed increases linearly from the root to the tip, while the incoming wind speed is almost constant. The other parameter is the thickness of the icing object, which for a wind turbine blade is the thickness of the blade. Due to structural requirements the blade thickness reduces from the root to the tip and the ice distribution thus varies in both the chord-wise and span-wise directions of the blade. This is mainly because the relative wind speed increases and blade thickness reduces along the blade from the root to the tip. The rate of ice accretion increases with relative wind speed and reduces with object thickness. Balancing both these factors means that ice thickness increases from the root to the tip of the blade.
The outer appearance and internal structure of ice is different according to the atmospheric condition. The type of ice encountered on wind turbines is generally divided into three groups, mainly based on the density. Glaze ice is hard, almost bubble free, clear, homogeneous ice with a density close to that of pure ice, 920kg/m3. Hard rime ice is rather hard, granular, and with a density of 600–900kg/m3. Finally, soft rime ice is white or opaque ice with a loosely bonded structure and a density of less than 600kg/m3. There are two main effects which result from icing on the wind turbine blade. These are change in aerofoil geometry due to macro-scale effects and change in surface roughness due to micro-scale effects. The first can change the flow distribution around the blade (and can be simulated with sufficient accuracy using potential flow theory) while the second affects the boundary layer and flow separation near the aerofoil surface.
Effects of Icing on Aerodynamics
As mentioned above, the primary effect of icing is change to the leading edge geometry and blade surface roughness, which alter the aerodynamic properties of the blade depending on the type and size of the ice build-up. Mass distribution along the blade, and thereby blade natural frequencies, may also be changed by icing. This effect is of secondary importance compared to the aerodynamic effects because of the small ratio of the accumulated ice mass to the blade mass at the first stage of ice accumulation.
Rotor aerodynamics are characterised by two generalised aerodynamic coefficients, which are the power (Cp) and thrust (CT) coefficients. These parameters are a function of the blade pitch angle ?, the tip speed ratio TSR (which is the ratio of the rotor tip speed to the incoming wind speed), as well as lift (CL) and drag (CD) coefficients for each of the aerofoil sections along the blade. The lift and drag coefficients are directly affected by icing due to the change in geometry and surface roughness of the blade. The shaft speed is indirectly affected by the lift and drag coefficients of the blade when moving at above the rated wind speed through the actions of the controller. Taking into account the non-linearity of the wind turbine system one can imagine how complicated the effect of icing on the performance of the wind turbine can be. If the wind turbine is considered as a system with wind speed as an input, it is obvious that the wind speed is not affected by icing. But the controller performance could be altered by icing. Some controller parameters, such as pitch actuator gains, depend on the slope of the power coefficient curves, but the effect of icing on controller performance is a secondary effect and one can focus only on the change in lift and drag coefficients of the blade to see the primary effects of icing. In the following section the effect of icing on the lift and drag coefficients of the blade, and consequently their effects on power and thrust coefficients, will be described.
As an example, the ice accumulation on the blade after 4 hours working under atmospheric icing conditions (T = –10°C, LWC = 15g/m3, MVD = 15µm and RH = 100%) were simulated with LEWICE on a NREL 5MW virtual wind turbine. The results show the negligible icing on the inner two-thirds of the blade and linear ice growth on the outer one-third of the blade. This part of the blade is constructed using the NACA 64618 profile. Figure 1, in the middle, shows the 2D ice profiles on this part of the blade at five different longitudinal sections.
The lift and drag coefficients of the aerofoil 2D sections can be determined by wind tunnel test experiments or by computational methods such as CFD; each method has its own benefits and difficulties. The rational approach is to set up a numerical model for one or a few sections, validate it against some experimental results and then use it for the rest of the blade. The turbulent method based on the Reynolds Average Navier Stokes (RANS) is one of the approved methods in the aviation and wind energy industries to estimate the aerofoil aerodynamic properties. The lift and drag coefficients of a clean NACA profile, compared to five 'iced' NACA profiles, are illustrated on the left side and right side of Figure 1 respectively. It is obvious that at this level of icing and with a low angle of attack (which is roughly between –7 and 7 degrees) the drag coefficients are much more affected by icing when compared to lift coefficients. This means that the effect of icing on aerodynamic coefficients is mainly due to a change in drag coefficients. Because of the non-linearity of the system, it is difficult to specify directly the effect of the change in lift and drag coefficients on the response. This is why it is wise to investigate the effect of icing on power and thrust coefficients which are representative for the whole rotor. Figure 2 shows the power and thrust coefficients as a function of tip speed ratio for four different pitch angles from 0 to 15 degrees before and after icing. The solid lines and dashed lines are the results for clean and iced rotors respectively.
Results and Conclusion
These results show that the icing reduces power outputs in the whole operating region of the wind turbine, while the thrust is increased in some regions and reduced in others. As a result of this work it is now easier to investigate the effect of icing on the response of the wind turbine. In variable speed, pitch regulated wind turbines the whole operating region can be divided into two sections. The first (region (1)) starts when cut-in wind speed is achieved (i.e. the minimum wind speed at which power production is practicable) and finishes when rated wind speed is reached (where the output power reaches the maximum power of the wind turbine), while region (2) is the region where the turbine is working at its rated wind speed until cut-out wind speed is reached (where the wind turbine has to be shut down due to structural limit states and safety). In region (1) tip speed ratio is constant and pitch angle is set to a small value near zero. When the turbine goes to region (2), as the wind speed increases the shaft speed remains constant while the tip speed ratio starts to reduce as the pitch angle is increased by the pitch controller in order to alleviate the loads on the structure. As illustrated by Figure 2, in below rated wind speed, both the power and thrust will be reduced due to icing; thereby loads and responses due to aerodynamic loads should be reduced. Above rated wind speed, where TSR starts reducing and pitch angle starts increasing, we should expect less deviation from nominal power and higher thrust compare to the clean rotor.
These results shows that in below rated wind speed one can shift the cut-in wind speed to a higher value to get the beneficial minimum power without having to worry about additional structural loads, but one should also shift the cut-out wind speed to a lower value (to prevent the wind turbine from failure in above rated wind speed); however it seems likely that the rated power will only be reached at higher wind speeds. Region (2) therefore occurs over a narrower band of wind speeds in the iced condition when compared to non-iced normal conditions. However, because these days all wind turbines that are designed for cold climate regions are equipped with de-icing or anti-icing systems the wind turbine behaviour under icing conditions described in this study should rarely be encountered 'in the field'. Nevertheless, the study should prove useful for ice-detection, safety and system optimisation purposes.
Biography of the Author
Mahmoud Etemaddar is a PhD candidate at the Institute of Marine Technology at NTNU (2009–2013). His PhD project is ‘Offshore Wind Turbine Analysis under Fault Conditions’. He has an MSc in Marine Engineering (2000–2005) from Amirkabir University of Technology, Tehran, Iran, and worked as a Marine and Offshore Platform Engineer between 2005 and 2009.{/access}
