Numerical and Experimental Investigations of Deicing Performance for the Pneumatic Impulse Deicing Method

Abstract

Wind turbines operating in cold regions are more likely to sustain extreme icing, causing the degradation of aerodynamic performance and the loss of wind energy output. This paper proposed a pneumatic impulse deicing method based on the conventional pneumatic deicing boot. This method’s simplified numerical simulation model was built using the commercial software ABAQUS to explore whether this method can deice. After that, experimental investigations were carried out in an artificial climate chamber to verify the deicing performance of specimens using this proposed method. Numerical and experimental results show that the pneumatic impulse deicing method could eliminate the ice layer with a smaller displacement and shorter time than the conventional pneumatic deicing boot. The ice layer’s thickening and inflation pressure enhancement benefit the deicing performance. Experiment results proved that the pneumatic impulse deicing method performs better under inflation pressures of 1.5 MPa and 2.5 MPa with an ice thickness of 6 mm or 8 mm.

1. Introduction

Wind energy in cold climate areas is becoming popular throughout the world due to its high air density and wind potential [1,2]. However, wind turbines built here frequently expose extreme icing during the cold seasons. Even small amounts of ice accretion on the leading edge will seriously influence the aerodynamic performance of wind turbine blades [3,4,5]. Therefore, developing deicing technology is necessary to improve the stability and efficiency of wind turbines operating in cold regions.

Much inspiration for the IPS (ice protection system) used on wind turbines originates from aircraft IPSs because the aviation industry has been affected by icing events for decades [6]. The ice protection techniques for wind turbines can be classified into two types: passive techniques and active techniques [3,7]. The depressant fluid is one of the passive techniques, but the application is limited because of its non-biodegradable property. Another common passive technique is super-hydrophobic coating. In theory, because of their repulsive property, these coatings prevent the water drop from remaining on the protected surfaces [8,9]. However, it has been demonstrated that hydrophobicity is insufficient to prevent ice accumulation, and the coatings could only delay the freezing time under severe icing cases [10]. Compared with the passive techniques, the active techniques are more diverse and mainly include thermal and mechanical methods. Thermal methods prevent ice accumulation on wind turbine blades by maintaining the surface temperature above freezing [11,12,13,14,15]. Thermal protection systems that use hot air or heating resistance have been marketed, but they consume a lot of power at low temperatures and high wind speeds [3,7,10]. Mechanical methods refer to ice protection systems that apply a mechanical force to break the accumulated ice and accelerate its separation from the protected surface [16,17,18]. The ultrasonic deicing method is one of the mechanical methods. This method generates Lamb waves and horizontal shear waves in the structure. When the shear stress caused by these waves exceeds the ice adhesion strength, the ice layer will be detached from the plate [19]. Although this method has been successfully applied to aircraft wings, the application on wind turbine blades is still in the laboratory development stage due to the different materials of the structure. The pneumatic and electromagnetic impulse deicing methods are the two most typical mechanical methods, with characteristics of high deicing efficiency and low energy consumption [3,10]. The two mechanical methods have been implemented in aircraft wings successfully for decades, while the application on wind turbine blades is rarely reported.

The conventional pneumatic deicing method attaches a sheet of the rubber-made pneumatic deicing boot to the protected surface (shown in Figure 1) [20]. When there is a certain amount of ice on the protected surface, the pneumatic deicing boot expands under pressurized air, and the ice layer will separate from the protected surface under the combined influences of bending and shear stresses (shown in Figure 2) [21,22]. Displacement caused by the deformation of the pneumatic boot could reach 6~10 mm, and the duration of one operation period is about 1~2 s [22]. The research found that the aerodynamic performance degradation caused by the deformation is lower than due to ice accretion. However, the large deformation could lead to unwanted vibration and noise [23].

This work proposed a revised structure of the pneumatic deicing method, named the pneumatic impulse deicing method, to shrink the deformation and operation duration. A simplified numerical simulation model based on the revised structure was built using the commercial software ABAQUS to explore whether this method can deice. After that, experimental investigations were carried out in an artificial climate chamber to verify the deicing performance of specimens using this proposed method. Meanwhile, the deicing performances under different ice thicknesses and inflation pressures were recorded and discussed.

2. Structure of the Pneumatic Impulse Deicing Method

The revised structure buries the inflatable tube within the protected object, and the specimen with a pneumatic impulse deicing structure is shown in Figure 3. First, a groove is dug in the surface of the base board, and an expandable flat tube is pasted in the center of the groove. The modified epoxy resin is then used to pour the tube into the groove, and the epoxy resin is cured to form a deformable elastomer. The function of the elastomer is to help the structure’s surface return to the initial flat state after the pressure in the tube is released. Finally, a layer of metal skin is bonded on the surface of the elastomer to protect it, and both sides of the metal skin are fixed on the baseboard with rivets.

Under the inflated state, the pressurized air is applied rapidly into the inflation inlet. The flat tube deforms the metal skin outward through the elastomer, introducing bending stresses to break up the ice layer and shear stresses adequately at the ice/metal interface to promote the separation of the broken ice layer. Due to metal skin and the elastomer, the structure can withstand immense impulse pressure with minor deformation, usually less than 2 mm. Furthermore, the duration of the impulse pressure is less than 4 ms. After the pressurized air escapes from the air outlet, the metal skin and elastomer will return to their original form.

3. Numerical Simulation Methods

3.1. Dynamical Equation

The process of the pneumatic impulse deicing method involves structural dynamics. The dynamical equation of the finite element method is [16]:

$$[M]\{\ddot{\mathit{u}}(t)\}+[D]\{\dot{\mathit{u}}(t)\}+[K]\{\mathit{u}(t)\}=\{\mathit{F}(t)\}$$

(1)

where [M] is the mass matrix, [D] is the damping matrix, [K] is the stiffness matrix, $\{\ddot{\mathit{u}}(t)\}$ is the acceleration vector of the element node, $\{\dot{\mathit{u}}(t)\}$ is the velocity vector of the element node, $\{\mathit{u}(t)\}$ is the displacement vector of the element node, and $\{\mathit{F}(t)\}$ is the resultant force vector of the node.

The explicit integral method is used to solve the above equation, and the results with better convergence can be obtained quickly. So, the acceleration vector of the element node could be expressed as:

$$\mathit{a}(t_n)=M^{-1}[\mathit{P}(t_n)-{\mathit{F}}^{\mathrm{int}}(t_n)]$$

(2)

where F^int is the internal force vector of the element node and P is the external force of the element node.

Therefore, the velocity and displacement vectors of the corresponding node are, respectively:

$$\mathit{v}(t_{n+0.5})=\mathit{v}(t_{n-0.5})=0.5\mathit{a}(t_n)(\Delta t_{n-1}+\Delta t_n)$$

(3)

$$\mathit{u}(t_{n+1})=\mathit{u}(t_n)+\mathit{v}(t_{n+0.5})\Delta t_n$$

(4)

3.2. The Constitutive Equations

The equation of strain rate is:

$$\mathrm{d}\epsilon =\mathrm{d}{\epsilon }^{el}+\mathrm{d}{\epsilon }^{ck}$$

(5)

where dε^el is the elastic strain rate and dε^ck is the cracking strain rate.

The equation of stress rate is:

$$\mathrm{d}\sigma =[D^{el}](\mathrm{d}\epsilon -[T]\mathrm{d}{\epsilon }^{ck})$$

(6)

where [D^el] is the linear elastic stiffness matrix and [T] is the local crack coordinate transformation matrix.

3.3. The Brittle Cracking Model of the Ice Layer

In this work, the Rankine criterion was used for ice cracking detection, and the energy cracking criterion was used to define the crack zone’s subsequent failure characteristics. The equation of the Rankine criterion is:

$$C=C(t,{\sigma }^{\mathrm{I},\mathrm{II}})=0$$

(7)

where t is time, σ^I,II is stress (I and II represent the direct stress and shear stress, respectively), C = [C_nn C_tt C_ss C_nt C_ns C_ts]^T, C_nn, C_tt, and C_ss represent the direct stresses in three directions of the material and C_nt, C_ns, and C_ts represent the shear stresses in the three directions of the material, and C is the maximal principal stress. When C exceeds the material tensile strength, the crack shows up.

The equation of the energy cracking criterion is:

$$G_f^{\mathrm{I}}={\displaystyle \int {\sigma }_t^{\mathrm{I}}\mathrm{d}u}$$

(8)

where $G_f^{\mathrm{I}}$ is the material fracture energy and ${\sigma }_t^{\mathrm{I}}$ is the tensile stress of the element node.

3.4. Geometric Model and Parameters Settings

The dynamic/explicit module could achieve the abovementioned functions in the commercial software ABAQUS. The pneumatic impulse deicing structure is simplified to improve simulation convergence and efficiency as a three-ply bending model under impulse pressure. Because the pressure distribution in the length direction is the same, the length of the simulation model is only half of the actual length of the specimen to save calculation time (shown in Figure 4). The surfaces marked by red lines in Figure 4 are set as fixed. References [24,25] show that the shear stress at the ice/aluminum interface is less than 1.5 MPa. Therefore, the tangential behavior at ice/aluminum of the simulation chose 1.5 MPa. According to references [26,27,28], the fracture characteristic of the ice layer is given in

Table 1. The fracture characteristic of the ice layer.

Brittle Cracking		Brittle Shear		Failure Displacement
Cracking Stress	Cracking Strain	Retention Factor	Opening Strain
0.27	5 × 10−5	5 × 10−5	1	1 × 10−4

. Because the power loss of wind turbines can reach up to 36% under light icing conditions [29], the chosen ice thicknesses in this work include 2 mm, 4 mm, 6 mm, and 8 mm. Meanwhile, the applied inflation pressures include 0.5 MPa, 1.5 MPa, and 2.5 MPa with a duration of 4 ms. The other material parameters are given in

Table 2. The material’s parameters.

Parameters	Ice Layer	Metal Skin	Elastomer
Density/g/cm3	0.92	2.7	2.0
Elastic modulus/MPa	1500	70,000	400
Poisson ration	0.30	0.33	0.34
Length × width × thickness/mm	100 × 40 × 2 (4, 6, 8)	100 × 40 × 0.05	100 × 30 × 2

.

The mesh independence study was conducted first to explore the influence of mesh size on the simulation result. The mesh size includes 1 mm, 0.5 mm, and 0.25 mm. Three reference points on the metal skin are selected (shown in Figure 5), and the maximum tangential shear stress of the three points under different mesh sizes is given in Figure 6 (ice thickness is 2 mm, inflation pressure is 1.7 MPa).

According to Figure 6, the max error caused by the mesh size is 5.3%, which indicates that the influence of mesh size (using 1 mm, 0.5 mm, and 0.25 mm) on the simulation results could be ignored. To balance the prediction accuracy and computational time, after the mesh size convergence examination, the mesh size of all structures is set to 1 mm in both length and width directions. The mesh size in the ice thickness direction is 2 mm. The mesh type in this simulation is the eight-node linear hexahedron element. The model has 22,000~46,000 elements in total with different ice thicknesses.

4. Numerical Simulation Results

Based on the finite element method, the simplified pneumatic impulse deicing structure simulation model is built and computed via the commercial software ABAQUS. The deicing performances of different ice thicknesses under diverse inflation pressures are given in Figure 7. The deicing ratio (the ratio of deicing area to metal skin area) is introduced to compare the deicing performance more intuitively. Based on the structural dynamics, the initial acceleration of the ice layer can be used to reflect the impulse force working on itself, and the initial acceleration is positively correlated with the impulse force. The initial accelerations of the ice layer under different cases are given in Figure 8.

As shown in Figure 7, with the inflation pressure increasing, the deicing ratio of 2 mm ice shows an increasing trend. Enhancing the inflation pressure could improve the impulse force working on the ice layer (refer to Figure 8) and shear stress at the ice/aluminum interface (shown in Figure 9). Consequently, more ice layers will be peeled off. Because the shear stress at the ice/aluminum interface is less than 0.5 MP, part of the ice layer still sticks on the edges.

An interesting result is that the deicing ratio of the 4 mm ice layer under 0.5 MPa is much larger than under 1.5 MPa and 2.5 MPa. The explanation for the result is that the 4 mm ice layer is difficult to break off under 0.5 MPa due to the increased stiffness caused by the ice layer thickening, and although the shear stress on the edges could be ignored, the middle ice layer will drag off the ice layer on the edges by inertia (shown in Figure 10a). Conversely, the 4 mm ice layer cracks more easily under larger inflation pressure, and the ice layer on the edges cannot be dragged off by the middle ice layer (shown in Figure 10b).

When the ice thickness reaches 6 mm, the simulation shows better deicing performances under 0.5 MPa and 1.5 MPa, with deicing ratios of 99.9% and 100%, respectively. The stiffness of the ice layer increases further, which makes the ice layer harder to break up on the edges. Therefore, the 6 mm ice layer under 0.5 MPa or 1.5 MPa is peeled off as the 4 mm ice layer under 0.5 MPa inflation pressure. However, with the inflation pressure reaching 2.5 MPa, cracks show up on the edges of the ice layer, leading to a small part of the ice layer leaving on the aluminum surface.

For the 8 mm ice layer, only cracks appear when the inflation pressure is 0.5 MPa, and this is because the shear stresses at the ice/aluminum interface are less than 0.5 MPa (shown in Figure 11). However, the deicing ratios are both 100% when the inflation pressure is 1.5 MPa and 2.5 MPa. This result suggests that the thicker the ice, the easier it is to be peeled off as a whole.

The result of the maximum displacement of the aluminum skin is given in Figure 12. As we can see, with the inflation pressure increasing, the maximum displacement increases obviously. As a whole, the most significant displacement is less than 2 mm, which indicates that this revised pneumatic impulse deicing structure can shrink the deformation of the protected surface.

5. Icing and Deicing Experiments

5.1. Experiment Facility and Specimen

The experimental investigations were carried out in an artificial climate chamber (shown in Figure 13a). The chamber has a length of 4 m and an inner diameter of 2 m. The temperature in the chamber can be adjusted from −1.1 to −36 °C. Six icing nozzles operated by aerosolizing pressurized air and water are used to provide super-cooled water droplets. The bottom of the chamber is equipped with a high-power fan, which can realize multi-speed wind speed regulation. The icing parameters settings in the climate chamber are based on the environmental parameters recorded at Xuefeng Mountain Energy Equipment Safety National Observation and Research Station (shown in Figure 13b), which is located in Hunan Province, South Central China [30]. The climate belongs to the subtropical monsoon humid type, with an average annual rainfall of 1810 mm and an average annual temperature of 12.7 °C.

The facilities used in deicing experiments are shown in Figure 14. Firstly, a collecting cylinder is used to store pressurized air generated by an air compressor, and the pressurized air in the collecting cylinder is used to inflate the tube in the specimen. Then, a solenoid valve is used to control the inflation of the specimen. The solenoid valve is powered by a DC source. When energized, the valve opens, and the pressurized air rushes into the specimen. When powered off, the valve is closed, and the pressurized air escapes from the outlet of the specimen. A point switch is used to control the solenoid valve’s power-on and power-off, and the power-on duration is usually 3~4 ms. The specimens were manufactured according to Figure 3 and Table 2, with the length increasing to 400 mm.

5.2. Experiment Procedure

Based on the long-term monitoring of the icing phenomenon on Xuefeng Mountain, when the severe icing phenomenon (the ice type is mixed ice) occurs on wind turbine blades, the range of the main meteorological parameters is shown in

Table 3. Meteorological parameters of severe icing phenomenon on Xuefeng Mountain.

Icing Parameters	Temperature	Wind Velocity	Liquid Water Content
Value	−4~−8 °C	5.0~8.0 m/s	0.24~0.32 g/m3

[15]. Therefore, during the icing test, the climate chamber’s temperature, wind velocity, and liquid water content were set as −6 °C, 6.5 m/s and 0.28 g/m³, respectively. The test arrangement in the climate chamber is shown in Figure 15a. During the icing test, a vernier caliper was used to measure the ice thickness on the specimen every five minutes (Figure 15b). Under the icing condition, the ice type on the specimen’s surface was mixed ice. The thickness of the ice-covered specimen was measured at six random places. According to Formula (9), the average icing thickness is obtained. When the average ice thickness reaches the required value, stop spraying water and begin to carry out the deicing test.

$$h_{\mathrm{ave}}={\displaystyle {\sum }_i^6(h_i-h_b)}/6$$

(9)

where h_ave is the average ice thickness, h_i is the icing thickness measured at i time, and h_b is the thickness of the baseboard.

5.3. Deicing Test Results

The deicing experiment results are given in Figure 16. By drawing the residual ice layer’s shape, the deicing ratio could be calculated. The relative error of the deicing ratio between the simulation result and the experimental result is displayed in Figure 17.

As shown in Figure 16 and Figure 17, the experimental deicing areas are similar to the corresponding simulation results. The absolute relative error of the deicing ratio is less than 14.1%, which suggests that the numerical simulation method can accurately reflect the actual deicing performance.

The actual deicing performances using 0.5 MPa inflation pressure are worse than the corresponding simulation results. However, the deicing ratios of the experiments using 1.5 MPa or 2.5 MPa are usually larger than the simulation with a high probability. A possible explanation is that the actual inflation pressure, mainly referring to 0.5 MPa, dissipates faster than the simulation, which suggests that using higher inflation pressure, mainly referring to 1.5 MPa or 2.5 MPa, could bring better deicing performances.

When using 1.5 MPa or 2 MPa inflation pressure, the deicing ratios of 6 mm and 8 mm ice layer are larger than 2 mm and 4 mm, which suggests that the thickening of the ice layer could improve the deicing performance, and this proves the point that the thicker the ice, the easier it is to be peeled off as a whole.

The maximum displacement of the specimen under different inflation pressures by an oscilloscope and an infrared displacement meter is shown in Figure 18. Due to the ice layer having slight effects on the maximum displacement (refer to Figure 8), the results in Figure 18 are measured without icing conditions. As we can see, the maximum displacement under 0.5 MPa, 1.5 MPa, and 2.5 MPa is 1.06 mm, 1.58 mm, and 1.83 mm, respectively, and the relative errors with the corresponding simulation results are 14.1%, 13.1%, and 9.1% respectively. On the one hand, this indicates that the simulation results are consistent with the experiment results and proves that this revised pneumatic deicing structure can shrink the displacement of the protected surface. On the other hand, the displacement difference between 0.5 MPa and 1.5 MPa is larger than between 1.5 MPa and 2.5 MPa, which proves that the 0.5 MPa inflation pressure dissipates faster than 1.5 MPa and 2.5 MPa.

6. Conclusions

A simplified numerical simulation model based on the revised pneumatic deicing structure has been proposed in this paper to estimate the deicing performance of the ice layer. Corresponding experiments have been conducted on specimens to provide subsequent validation for the numerical model. According to the above investigations, deductions and suggestions of this article are as follows:

(1)

The experimental results are consistent with the numerical simulation results, and the maximum absolute relative error of the deicing ratio is 14.1%. Comparative analyses of the numerical and experimental results verify that the pneumatic impulse deicing structure can eliminate the ice layer with smaller displacement and shorter time than the conventional pneumatic deicing boot.

(2)

Simulation and experimental results indicate that the pneumatic deicing structure performs better under an ice thickness of 6 mm or 8 mm and an inflation pressure of 1.5 MPa or 2.5 MPa. The main reason for this phenomenon is that the ice layer is difficult to break off near the edges, and the middle ice layer will drag the ice layer on the edges off as a whole. In practical application, the deicing operation could be launched when the ice thickness reached a certain value, usually 6 mm, and the pneumatic impulse deicing method would perform better.

(3)

Although using 0.5 MPa inflation pressure could reduce the pressure charge time and save energy, according to the experiment results, the deicing performances are worse than the simulation results due to the fast dissipation. Therefore, a higher inflation pressure, such as 1.5 MPa or 2.5 MPa, is recommended in practical application.