Professional Documents
Culture Documents
W GONG Finite Element Analysis of A Wind Turbine Tower
W GONG Finite Element Analysis of A Wind Turbine Tower
W GONG Finite Element Analysis of A Wind Turbine Tower
Content
1. Prototype model introduction .......................................................................................... 3 2. Finite element model ....................................................................................................... 5 2.1 DIANA FEA program ............................................................................................... 5 2.2 Model setup ............................................................................................................... 5 3. Model verification ........................................................................................................... 8 3.1 Model geometry verification ..................................................................................... 8 3.2 Eigen modes analysis in DIANA .............................................................................. 8 3.3 Verification of eigen modes .................................................................................... 12 4. Static Analysis ............................................................................................................... 12 4.1 Static wind only....................................................................................................... 13 4.2 Static wave only ...................................................................................................... 18 5. Dynamic Analysis ......................................................................................................... 19 5.1 Dynamic wind only ................................................................................................. 19 5.2 Dynamic wave only................................................................................................. 31 6. Simulated earthquake spectral response analysis .......................................................... 31 7. Fatigue analysis ............................................................................................................. 35 8. Coupled wind-wave load analysis ................................................................................. 36 8.1 Static coupled wind-wave load analysis.................................................................. 37 8.2 Dynamic coupled wind-wave load analysis ............................................................ 40
Geometry Geometry of the finite element model is modeled in accordance with the geometrical shape of the prototype. The tower is composed of different sections of conical and cylindrical parts. Thickness of the tower is 200 mm for the lower part height of 89m and 150mm for the above sections. The turbine nacelle and rotors are excluded in the model; however, the weight of these components is modeled as vertical force applied at the top of the tower. Material Tower is made of steel with material parameters as following: Youngs Modulus 2.0E+5 kN/mm2 Poison Ratio 0.3 Density 7.9E-9 t/mm3 Yield Model Von Mises Hardening Parameters 0.275 kN/mm2, 0; 0.330 kN/mm2, 0.025
Finite Element Type Two types of finite element are used for the model. The tower body is modeled using curved shell element and the bottom foundation and the top of the tower is modeled using flat shell element. For the curved shell element, Q20SH (quadrilateral curved shell, 4 nodes, linear) is applied and for the flat shell element, Q20SF (quadrilateral flat shell, 4 nodes, linear, mindlin) is applied. Boundary Condition As stated above, simplification converts the original prototype from a floating offshore structure into a bottom founded structure. Therefore, in the finite element model, boundary condition is simply modeled as pinned bottom nodes. The nodes of the tower bottom are restricted to have translational motions in the X, Y and Z directions.
1. Self weight of the tower 2. Turbine tower weight modeled as vertical pressure at the top of the tower 3. Wind loads (static & dynamic) 4. Wave loads (static & dynamic) 5. Simulated earthquake spectrum
3. Model verification
3.1 Model geometry verification Based on the descriptions stated above, the geometry parameters of the total tower model are: Total Area = 3.44153e+009 [mm] Total Volume = 6.34156e+011 [mm] Total Mass = 5009.83 [T] The total mass of the model is in consistence with the mass of the prototype structure and thus the models geometry is verified. 3.2 Eigen modes analysis in DIANA From DIANAs analysis, the first 12 eigen modes of the structure and the respective eigen frequencies are computed. This eigen mode analysis is helpful for verification of the model on the first order and also useful for providing reference for the further dynamic analysis.
Fig 5(a) 1st eigen mode perspective view and top view
Fig 5(b) 2nd eigen mode perspective view and top view
In summary, the eigen frequencies of the first 12 modes are shown in Fig 6. Eigen analysis shows the structures twisting behaviour only starts from the 9th mode with eigen frequency of 7.58Hz. For the first 8th modes, structural behaviour is only flexural. Due to the axially symmetrical shape, each eigen mode pair has same characteristics for the X and Y directions.
3.3 Verification of eigen modes According to Laurens B. Savenije (December 2009), frequency parameters of fixed wind turbine tower structure are given in the following table 1. For fixed type, the first 3 eigen frequencies are about 0.3Hz, 0.3Hz and >0.9Hz according to the table. This is consistent with the eigen analysis computed as above and thus it can be approximately believed that the finite element model of the tower is sound, provided many simplifications are applied here.
Table 1 Natural periods of a floating and fixed turbine
4. Static Analysis
After verification of the model, analysis of the structural behaviour subject to external loadings is performed. External loadings are mainly from two sources: wind and wave or atmospheric and hydraulic forces. Each force is composed of static (average) component and dynamic (fluctuating) component. Characteristics of the wind load and the wave load are conceptually similar but yet the sections of the tower that are subject to the loadings are different. Wind force is loaded on the turbine rotors and the tower body also and wave loading is majorly on the region of the tower body that is near the water surface. Here, firstly, the influence of the wind load and the wave load are studied separately.
Coupled impacts from the two will be analyzed in a later stage, after a better understanding of the separate contributions. For the wind impact, static wind force and dynamic wind force are analyzed: 4.1 Static wind only Load parameters: Wind speed: V=13m/s (Beaufort scale 7 wind speed) Impact area: surface area formed by the rotors (round surface with diameter D=82.4m) Drag coefficient: Cd=2.0 Resulted force magnitude: F=1163.79kN Force direction: X direction Loaded area: tower top Analysis results Structures deformation and the von Mises stress are computed according to this static wind force condition:
From the analysis results, under the static wind force of wind speed 13m/s, the maximum displacement of the tower at the top is about 288 mm and the maximum von Mises stress at the tower bottom is around 0.0288 kN/mm2, which is far less than the yield stress of the material. In order to study the nonlinear behaviour of the tower, firstly the structural response of a larger force load is analyzed here linearly, which is going to provide a reference for the nonlinear result in the later stage: Load parameters: Wind speed: V=50m/s (Beaufort scale 12 wind speed) Impact area: surface area formed by the rotors (round surface with diameter D=82.4m) Drag coefficient: Cd=2.0 Resulted force magnitude: F=17215.78kN Force direction: X direction Loaded area: tower top Analysis results
The above results show that under the static wind force of wind speed 50m/s, the maximum displacement is about 4.05 m and the maximum von Mises stress is 0.297 kN/mm2, which exceeds the materials initial yielding stress. For the same conditions of wind speed 50m/s, nonlinear analysis is performed and the results are shown below:
Fig 9(a) nonlinear analysis X direction displacement under static wind at 50m/s
Fig 9(b) Nonlinear analysis Von Mises stress under static wind at 50m/s
The above shows that results of the nonlinear analysis differ from results of the linear analysis, even if the loading conditions are exactly the same, Nonlinear analysis gives a slightly larger deformation response of the structure while the stress condition is minor, with maximum von Mises stress of 0.279 kN/mm2, which is about 6% less than the result of the linear analysis. 4.2 Static wave only As stated above, for static analysis, the characteristics of the wind load and the wave load are similar and the different is the loading area on the structure. The static wind load is applied at the top of tower, shown in the previous section. However, for the static wave load, it should impact on the region around the water line, depending on the scale of the wave height. For reference, wave force on cylindrical structure can be computed as: F = Fdrag + Finertia Fdrag = 0.5CdAu2 Finertia = CmVa
With sufficient wave load parameters (u, Cd, etc) resulted static wave force can be calculated according to the above formulas and static analysis of the wave load on the tower can be performed the same way as that of the wind load in DIANA. Here the static wave load analysis will thus not be redundantly performed again.
5. Dynamic Analysis
5.1 Dynamic wind only In addition to the static analysis in the former section, dynamic analysis of the tower is performed in this section. The dynamic wind load condition is modified via a small scale wind gust data (Fig 10). In order to produce a more continuous result, this small scale gust is copied several times to simulate a more variable condition. The simulated result is thus shown in Fig 11.
Fig 10 Small scale wind gust data by Laurens B. Savenije (December 2009)
13 12.5 12 11.5
Wind Speed(m/s)
10
20
30
Time(s)
Based on the above loading condition, response of the tower is computed and the results are shown below. For transient analysis in DIANA, 5 different time integration schemes are available. In order to give a better understanding of the different characteristics of these 5 different schemes, each of the schemes is applied based on the same loading conditions above. The size and number of time step is an important issue for performing transient analysis in DIANA. Initially, the input time step size is set as 0.1s. This means for the total duration of the 30s of the external loading, there would be 300 time steps. Also, as the first eigen mode period of the tower is about 2.16s. This means one period could contain about 20 results points which are regarded as sufficient for the analysis result. Results of the different time integration schemes are thus shown below respectively:
300
250
Displacement(mm)
200
150
100
50
0 0 5 10 15 20 25 30
Time(s)
400
350
300
250
Force(kN)
200
150
100
50
-50 0 5 10 15 20 25 30
Time(s)
350 300
Displacement(mm)
250 200
150 100 50
0 -50 0 5 10 15 20 25 30
Time(s)
600
500
400
300
Force(kN)
200
100
-100
-200 0 5 10 15 20 25 30
Time(s)
350 300
Displacement(mm)
250 200
150 100 50
0 -50 0 5 10 15 20 25 30
Time(s)
600
500
400
300
Force(kN)
200
100
-100
-200 0 5 10 15 20 25 30
Time(s)
350 300
Displacement(mm)
250 200
150 100 50
0 -50 0 5 10 15 20 25 30
Time(s)
600
500
400
300
Force(kN)
200
100
-100
-200 0 5 10 15 20 25 30
Time(s)
350 300
Displacement(mm)
250 200
150 100 50
0 -50 0 5 10 15 20 25 30
Time(s)
700
600
500
400
Force(kN)
300
200
100
-100
-200 0 5 10 15 20 25 30
Time(s)
Conclusion: Based on the results from the above 5 different time integration schemes, the New Mark, Wilson Theta, Hilbert Hughes Taylor and the Implicit Runge Kutta schemes show very similar results with only the Euler backward scheme being significantly different from the other four. For the results from the first four schemes, it can be seen that the response of the tower has a period that is in accordance with the first eigen mode of the structure (frequency 0.462Hz, period 2.164s). The amplitude range of the first four schemes also share common characteristics. However, for the Euler implicit scheme, the results only reflect the characteristics of the external loading (compare Fig 11 and Fig 12) and the eigen mode characteristics of the structure itself are completely damped out in the computed results. Therefore, it can be concluded that using the first four schemes seem to be much safer when performing dynamic analysis and thus the following analysis will be using the New Mark scheme only for simplicity reason. Before performing nonlinear dynamic analysis, there is a need to answer the above problem: why the result from the Euler backward scheme is so different from the others. Theoretically, the Euler backward is of first order accuracy while the other four schemes have higher order accuracy. Therefore, a trial with a smaller time step size is made to study the result of the Euler backward scheme again. For this time, time step size is chosen as 0.01s resulting 3000 time steps for the duration of the external loading. This time step size is 1/10 of the original one used. The result of a smaller time step size using Euler backward scheme is shown below:
450 400 350 300 Displacement(mm) 250 200 150 100 50 0 -50 0 5 10 15 Time(s) 20 25 30
Compare Fig 12 and 17, it is easily shown that reducing the time step size generates a better result that is more in consistence with the other schemes for the Euler backward one. The sacrifice is of course the apparently increased amount of time for the computation. Also, compare Fig 17 and Fig 13(a), 14(a), 15(a) and 16(a), it can be seen that the amplitude of the Euler backward result is generally smaller than that of the other four schemes. The reason is probably that Euler backward scheme has a larger numerical damping on the results. After answering the characteristics of different time integration schemes, nonlinear dynamic analysis will be performed in the following. As stated above, for simplicity, only New Mark scheme will be used. 6) Nonlinear analysis without damping using New Mark scheme:
450 400
350 300
Displacement(mm)
250 200
150 100 50
0 -50 0 5 10 15 20 25 30
Time(s)
600
500
400
300
Force(kN)
200
100
-100
-200 0 5 10 15 20 25 30
Time(s)
Through Fig 18, it can be seen that even nonlinear analysis function is applied (including physical and geometrical nonlinear considerations), the result is almost the same as the result from linear analysis (Fig 13). Therefore, in order to give a better understanding of the nonlinear characteristics, damping is applied for the following section. 7) Nonlinear analysis with damping using New Mark scheme: Damping parameters: In the transient dynamic analysis, it is necessary to apply Rayleigh damping. Rayleigh damping is characterized by the two constants a and b, whereas the damping coefficient of the steel is about 0.025. So now the question is how to determine the coefficients a and b for given . If a modal analysis would be used, the damping coefficients 1 and 2 for the two lowest frequencies 1 and 2 will equal if a = 212 b = 2 where
1 (1 ) = 2 2 1
For this wind turbine tower, based on eigen mode analysis stated previously, the lowest two frequencies 1 and 2 are 2.902 rad/s and 10.864 rad/s respectively. Therefore, the Rayleigh damping constants a and b are computed as: a = 0.115 b = 0.00363
450 400
350 300
Displacement(mm)
250 200
150 100 50
0 -50 0 5 10 15 20 25 30
Time(s)
600
500
400
300
Force(kN)
200
100
-100
-200 0 5 10 15 20 25 30
Time(s)
From the above results, it is easily seen that the damping influences the response of the structure to the loading to a large degree. Amplitudes of the structural response are reduced and the inner force of the structure is also decreased. And as time goes by, the damping has more positive effects on the structure. In addition, comparing Fig 17 and 19(a), it can be seen the nonlinear analysis with damping result is somehow similar to the linear analysis result of Euler backward scheme. This again shows that Euler backward scheme introduces much more numerical damping than other time integration schemes. 5.2 Dynamic wave only As for dynamic wave load, a similar transient analysis could also be performed in DIANA. Similar to the static load analysis, the loaded area of the dynamic wave load is also different from that of the dynamic wind load. The time load shape depends on the local wave environment. An example of the dynamic wave load can be:
Again, for simplicity reason, dynamic wave analysis will not be performed again here.
Resulting structural response of the above earthquake loading is shown below. Results of two different rules are both presented here. ABS rule
SRSS rule
Conclusion: Based on the above results, it can be seen that ABS rule gives a much larger structural response than the SRSS rule. Under the ABS rule, the maximum structural deformation is in the order of 620 mm and the maximum stress is about 0.0820 kN/mm2, given the condition of the earthquake spectrum of Fig 21. The structural response is in the linear stage which means if the simulated earthquake spectrum is the same as the in situ external force, then the structural would be safe from failure.
7. Fatigue analysis
Fatigue analysis of the structure is performed in this part. For fatigue analysis, a so-called Whler diagram is required. According to the steel material of the structure, the following material data will be used: SWOEHL 0.4 kN/mm2 NWEOHL 10000000
Based on the above material properties and under the load of self weight only, the result is shown as following:
From the above result, the number of load cycles to fatigue failure range from 6.54106 to 9.98106. The weakest part is at the toe of the structure and the strongest part is the upper tower of the structure.
8.1 Static coupled wind-wave load analysis Wind load parameters Wind speed: V=13m/s (Beaufort scale 7 wind speed) Impact area: surface area formed by the rotors (round surface with diameter D=82.4m) Drag coefficient: Cd=2.0 Resulted force magnitude: F=1163.79kN Force direction: X direction Loaded area: tower top Wave load parameters For wave load, a sample of the loading condition is shown below. For static load, the magnitude of the force will be approximated via equalizing the amplitude of the wave force of this fluctuating loading situation, that is to say, the static wave force will be set as about 2000kN which is the amplitude of the dynamic wave load in Fig 25.
Impact area: section of tower submerged in water Resulted force magnitude: P=2kPa Loaded area: tower surface submerged in water
Results:
Fig 26(b) Total absolute structural deformation under static wind-wave load
The above shows that structural deformation in the Y direction due to the static wave force is less than the structural deformation in the X direction due to wind force. However, the combined wind-wave force would give the maximum deformation at the top of the tower a displacement around 280 mm. The maximum von Mises stress in this situation is about 0.0318 kN/mm2 at the bottom of the tower. The value is about 10% higher than the situation of the wind load alone (Fig 7(b)). 8.2 Dynamic coupled wind-wave load analysis For the dynamic coupled wind-wave load analysis, the following load conditions are applied simultaneously on the structure. Rayleigh damping is also included for this analysis. Wind load condition Wind load is simulated as periodical force acting on the top of the tower in the X direction. The phase and amplitude is determined through the wind gust condition in the following figure. This results in a force with amplitude of about 1076 kN.
13 12.5 12 11.5
Wind Speed(m/s)
10
20
30
Time(s)
Wave load condition For wave load, the force is in the same phase as the wave elevation variation in the figure below, which means the wave load is approximated as a sine wave with period of 10 s. The amplitude of the load is 2 kPa on the section of the tower submerged in water. The load is acting in Y direction.
For the two different dynamic loads above, they have different amplitudes, periods and they also act in different parts of the tower in perpendicular directions. The results of the structural response to this complex loading condition are shown as below:
450 400
350 300
Displacement(mm)
100 50
0 -50 0 5 10 15 20 25 30
Time(s)
Fig 29 Top point X direction displacement under coupled dynamic wind-wave load
150
100
Displacement(mm)
50
-50
-100
-150 0 5 10 15 20 25 30
Time(s)
Fig 30 Top point Y direction displacement under coupled dynamic wind-wave load
450 400
350 300
Displacement(mm)
100 50
0 -50 0 5 10 15 20 25 30
Time(s)
Fig 31 Top point absolute displacement under coupled dynamic wind-wave load
Fig 32 Von Mises stress under coupled dynamic wind-wave load at T=5s
From the above results, it is shown that the eigen modes in the structural response to the fluctuating wave load in Y direction are more damped out than the situation of the wind load. Due to the mild wave load applied here, the structural response in Y direction is quite minor compared with that in X direction and thus the total absolute deformation situation at the tower top is more or less similar to the result of that in X direction only. The maximum deformation at the tower top is about 370 mm at T=5s (Fig 31). At the time of maximum structural response, the von Mises stress condition of the structure is given in Fig 32. The situation is not far from that under the dynamic wind load condition only (Fig 19(c)). Yet, the resulting stress is about 10% larger in this case, due to the presence of the wave load. The structural response to the coupled wind-wave load is more or less the combination of the response to each load condition, with some nonlinear coupling effects present. Analysis of different structural response characteristics under dynamic wind and wave loads computed in DIANA According to the above analysis, there are some interesting characteristics about the results computed in DIANA. Compare both the dynamic wind and wave load conditions (Fig 27 & 28), although there are many differences, there are also some common points. Both loads have a period of 10 s, which is simplified in the input data in this analysis. The wind gust load is purely positive because the wind gust direction does not change during the loading period. For the dynamic wave load, it is simplified as a sine wave in this analysis with changing forcing directions. However, the period of the wave load is the same as the period of the wind gust load. Compare structural responses to the wind and wave loadings in Fig 29 & 30, interestingly, even the loading conditions share the common characteristics as stated above, the results are somehow different. Both results are using the New Mark time integration scheme with the same scheme parameters and time step size, but the response to the wind load has an apparently larger influence of the structural eigen period while the response to the wave load has only some slight hints of the structural eigen period. For both cases, the structural response influence from the eigen modes is damped more and more with the time goes by. For the duration between 25s and 30s, there is almost no hint of the eigen mode which is explained by the Rayleigh damping applied in this analysis. In order to provide a more consolidated reference for the above issue, another transient analysis with a much smaller time step size is performed. The time step size is now 0.001s for a period of 10s. New Mark scheme is still used. The structural responses to the wind and wave loads in two different directions are shown below. From Fig 33 & 34, it can be seen that even the time step size is reduced to a much more conservative level, the result is still the same case as what is got in the above situation. Still, the response to the wind gust load is subject to a larger influence from the structural eigen mode and the response to the idealized sine wave load has a much smaller influence from the structural eigen mode.
Displacement(mm)
250
10
Time(s)
Fig 33 Top point X direction displacement under coupled dynamic wind-wave load (t=0.01s)
100 80 60 40
Displacement(mm)
20
10
Time(s)
Fig 34 Top point Y direction displacement under coupled dynamic wind-wave load (t=0.01s)
For the problem stated above, the theoretical answer is that the structural response to the wind load has a larger influence of the structural eigen mode because the dynamic wind gust load still has some minor components of lower periods which are closer to the
structural eigen period even if it looks like to have a period of 10s. The wave load is idealized as a sine wave with a period of 10s and thats why it doesnt show much hint from the structural eigen mode. On the other side, for the situation of forced motion of damped system, the result is composed of two parts: damped free vibration and damped forced vibration. As time goes by, the damped free vibration will be faded away while the damped forced vibration will dominant the result and this is why the influence of the structural eigen mode will be damped out as manifested in the above analysis results.