One-dimensional variations: Blades Report on task 5 of seminar

Dutch Offshore Wind Energy Converter project

DOWEC 10070 rev 2 Name: J.F. Kooij (LMGH)

Written by:

Released by: version Date 2

30-09-03

Signature:

Date: 30-09-03

R. v/d Berg (LMGH) No of pages 19 New document in PDF

DOWEC 10070 rev 2

30-09-2003 Page: 1 of 19

Summary:

In this document several different blade alternatives were investigated: • Blades with structural pitch, • Carbon fibre blade and • Blade with an aerodynamic root.

Contents 1 Introduction............................................................................................................................... 3 1.1 Overview 3 1.2 Blade 3 2 Structural pitch ......................................................................................................................... 4 2.1 Blade variations 4 2.2 Analyses 6 2.3 Results 6 2.3.1 Tip deflection ............................................................................................................................... 7 2.3.2 Maximum bending moment......................................................................................................... 8 2.3.3 Fatigue loading ........................................................................................................................... 9

2.4 Conclusions and recommendations 10 3 Carbon blades............................................................................................................................ 11 3.1 Introduction 11 3.2 Comparison of model properties 11 3.2.1 3.2.2 3.2.3 3.2.4 3.2.5

Mass ............................................................................................................................................. 11 Stiffness ........................................................................................................................................ 12 Tip deflection ............................................................................................................................... 12 Eigenfrequencies ......................................................................................................................... 13 Static moment and inertia ........................................................................................................... 14

3.3 Costs 14 3.4 Conclusions 15 4 Aerodynamic blade root .......................................................................................................... 16 4.1 Possible modifications 16 4.2 Input 16 4.3 Results 18 4.4 Conclusions and recommendations 19 References ....................................................................................................................................... 19

DOWEC 10070 rev 2

30-09-2003 Page: 2 of 19

1 Introduction 1.1

Overview

On the DOWEC seminar held on 15/16 January 2002 several alternative designs were thought of (see ref. [2]). A few were labelled as interesting to research. In task 5 alternative blade designs were described. The following alternatives (with their objective) were listed: • A blade with structural pitch Improve elastic behaviour; reduce (fatigue) loading through designing the directions of the natural bending modes. • A carbon fibre blade Determine influence of CFRP blade on LPC. • A blade with an aerodynamic root Reduce aerodynamic root losses. LM Glasfiber Holland BV investigated all three alternatives with some corporation of ECN and TU Delft. ECN was responsible for the calculation load sets (if needed) and TU Delft gave advice on aerodynamic behaviour of profiles (used for aerodynamic root). The structural design was down by LM Glasfiber. 1.2

Blade

The DOWEC 6 MW blade has been used for all calculations in this report. The input data for this blade can be found in ref [1] and [3]. The first document (ref. [1]) describes the chord, twist and thickness distribution of a few different concepts, from which one is chosen to be the concept baseline. From this baseline a structural design has been made, which resulted in the baseline blade design, which is described in ref. [3]. The baseline blade is designed for a rotor diameter of 129 meter, so the radius is 64.5 meter. For the blade a hub of 1.8 meter is used, this makes a blade length of 62.7 meter. The chord and twist distribution is shown in figure 1. 14

5

Chord (m)

Twist (deg)

12

4 10

3

8 6

2

4

1 2 radius (m)

0 0

10

20

30

40

50

radius (m)

0 60

0

10

20

30

40

50

60

Figure 1: Chord and twist distribution

DOWEC 10070 rev 2

30-09-2003 Page: 3 of 19

2 Structural pitch For bigger blades it is assumed that aero-elastic behaviour becomes more important. Therefore some structural variations are performed. This would have as objective to improve the aeroelastic behaviour and to reduce (fatigue) loading through modifying the directions of the natural bending modes of the blades. 2.1

Blade variations

The most influence on the bending modes would be the location of the main spar together with the location of the webs. All variations were modelled using Focus with which all properties were calculated. To keep the structural variations as simple as possible (and thus a better overview) only the direction of the main spar is altered. The main spar can be rotated around a point and in these variations these points are the root and the tip. The rotation angle of the main spar is 0.5° for all variations (+5 or –5 in the names of the variations) as can be seen in figure 2.

Figure 2

If the shift is made for the suction and pressure side independently, eight different variations can be obtained (see table 1). Name Fixation point Movement on suction side Movement on pressure side

tt+5 Tip TE TE

tt-5 Tip LE LE

tt+5-5 Tip TE LE

tt-5+5 Tip LE TE

tr+5 Root TE TE

tr-5 Root LE LE

tr+5-5 Root TE LE

tr-5+5 Root LE TE

Table 1: Blade variations (in grey the final selected ones)

DOWEC 10070 rev 2

30-09-2003 Page: 4 of 19

For all variations the principle stiffness, the location of the centre of gravity in chordwise direction and the natural frequencies can be obtained easily. These properties are shown in figures 3 to 5 and form a basis for the selection of variations to do further research on. 10

[deg]

5 0 -5 -10 baseline

-15

tt-5+5 tt-5

-20

tt+5-5 tt+5

-25

tr-5 -30

tr+5-5 tr+5

-35 -40 0

10

20

30

40

50

60

70

Figure 3: Principle stiffness 0,6 [m]

0,5

baseline tt-5+5

0,4

tt-5 tt+5-1

0,3

tt+5 tr-5

0,2

tr+5-5 tr+5

0,1 0,0 -0,1 -0,2 -0,3 0

10

20

30

40

50

60

70

Figure 4: Centre of gravities (distance to centre line)

DOWEC 10070 rev 2

30-09-2003 Page: 5 of 19

110%

flap-1 edge-1

105%

100%

95%

90% original

tt-5+5'

tt-5'

tt+5-5'

tt+5'

tt-5'

tr+5-5'

tr+5'

Figure 5: Natural frequencies

As not all proposed blade variations seem to have an effect on the aerodynamic behaviour, the following variations are selected in consultation with ECN (the reason why is also noted): • tt-5+5: rotation of principal stiffness axis towards feathering enlarges contribution of flapwise displacement to second (leadwise) blade bending mode, hence increasing its aero-elastic damping; • tr+5-5: idem; • tt+5: no rotation of principal stiffness axis, similar effect on the two lowest blade natural frequencies, hence enabling investigation of this effect alone; • tt+5-5: rotation of principal stiffness axis towards stall, similar effect on blade natural frequencies. 2.2

Analyses

To get an indication of the influence of the variations some representative load cases are selected. Most representative would be normal production at and above rated wind speed and almost at cut-out wind speed. Additional to this, normal production below rated wind speed seems advisable and some cases with a yaw misalignment. This results in the following load cases: • 008: Nominal production 8 m/s • 012: Nominal production 12 m/s • 018: Nominal production 18 m/s • 024: Nominal production 24 m/s • 231: Nominal production 12 m/s, -20º yaw misalignment • 234: Nominal production 25 m/s, -20º yaw misalignment Stentec has performed all analyses using Phatas (see ref. [4]). 2.3

Results

As every load case gives a lot of results the most interesting features will be presented. These are the following: tip deflection, maximum bending moment and fatigue loading.

DOWEC 10070 rev 2

30-09-2003 Page: 6 of 19

2.3.1

Tip deflection

The tip deflection can be presented in two different ways, i.e. as the maximum tip deflection or as the minimum tower clearance. The explanation speaks for itself, but blade manufacturers favour to talk about the tip deflection and the turbine manufacturers would rather stick to minimum tower clearance. First of all the minimum tower clearances are presented for all load cases in figure 7. 140%

baseline

-tr+5-5

018

024

-tt+5

-tt+5-5

-tt-5+5

130% 120% 110% 100% 90% 80% 008

012

231

234

Figure 6: Minimum tower clearance for all load cases

As clearly can be seen the variation tt+5 has for all load cases a higher value for the tower clearance, which means that the tip deflection for this variation is presumed to be lowest. When taking for all load cases the minimum value of the tower clearance (or maximum value for tip deflection), the following graph can be seen (see figure 8).

-tr+5-5 -tt+5 -tt+5-5 -tt-5+5

Tower clearance -3 % +37 % + 18 % + 25 %

Tip deflection +1% -11 % -5% -8%

Table 2

Also here clearly can be seen that variation tt+5 is most favourable (-11%). It must be stated however that almost all variations have a reduction in tip deflection (tt+5-5 has 5% less tip deflection and tt-5+5 has 8 % reduction). Only the variation when the rotation around the root is fixed, has a somewhat larger tip deflection.

DOWEC 10070 rev 2

30-09-2003 Page: 7 of 19

Minimum tower clearance

Maximum tip deflection

140%

105%

130% 100% 120% 110%

95%

100% 90%

90% 80% baseline

-tr+5-5

85% -tt+5

-tt+5-5

-tt-5+5

Figure 7: Minimum tower clearance and maximum tip deflection

2.3.2

Maximum bending moment

The bending moment can be divided into three directions: lead-lag direction (edgewise), flap direction and the resulting moment. The bending moment is represented as the maximum bending moment of all load cases (which mostly is load case 234). The results are shown in table 3 (for the root location and at largest chord) and in figure 9 (only for root).

Location -tr+5-5 -tt+5 -tt+5-5 -tt-5+5

Mlead Root Largest chord -4 % +4 % -12 % -11 % -4 % 0% +10 % +10 %

Mflap Root Largest chord 0% -1 % +1 % -1 % +3 % +2 % +2 % +2 %

Mres Root Largest chord +1% -1 % -1 % 0% +2 % +2 % +2 % +2 %

Table 3

DOWEC 10070 rev 2

30-09-2003 Page: 8 of 19

115% 110%

baseline

-tr+5-5

-tt+5

-tt+5-5

-tt-5+5

105% 100% 95% 90% 85% 80% Mlead

Mflap

Mres

Figure 8: Maximum bending moment root

From these results can clearly be seen that the lead lag moment is significantly reduced for variation tt+5 (> 10 %) and slightly for tt+5-5. The other two have an increase for the lead lag moment. All variations are comparable to the baseline for the flap moment as well as the resulting moment. All variations in one way or another have a reduction except tt-5+5, which is considered to be a non-favourable solution. Also variation tr+5-5 is not very favourable concerning lead and lag moments, but is less “bad” than tt-5+5. 2.3.3 Fatigue loading Fatigue loading is only calculated for the load cases with nominal production without yaw misalignment (12 m/s, 18 m/s and 24 m/s). In table 4 the results for equivalent flap- and edgewise bending moment at the root are shown.

-tr+5-5 -tt+5 -tt+5-5 -tt-5+5

12 m/s Flap Edge +5.6 % -1.5 % +0.7 % -0.4 % -0.1 % +1.4 % +4.1 % +2.3 %

18 m/s Flap Edge +0.4 % +1.4 % +0 % +1.3 % -1.5 % -5.1 % +2.5 % +5.5 %

24 m/s Flap Edge +0.3 % -0.8 % +0.8 % +0.8 % -0.4 % -12.3 % +1.1 % +0.8 %

Table 4

In figure 10 this is shown for each variation with the three wind speeds. Most remarkable result is that the edgewise bending moment is significantly reduced for variation tt+5-5 above rated wind speed. Variation tt-5+5 has both higher flap- and edgewise bending moments and seems most unfavourable. Variation tt+5 is most comparable with the baseline as the difference is nowhere higher than 1 %. Variation tr+5-5 has a higher flapwise bending moment below rated wind speed.

DOWEC 10070 rev 2

30-09-2003 Page: 9 of 19

e dge wise be nding mome nt blade root 110%

flatwise be nding mome nt blade root 110% 12 m/s

12 m/s 105%

18 m/s

18 m/s

105%

24 m/s

24 m/s 100%

100%

95%

95%

90%

90%

85%

85%

80%

80% tr+5-5

tt+5

tt+5-5

tt-5+5

tr+5-5

tt+5

tt+5-5

tt-5+5

Figure 9: 1 Hz equivalent bending moment at root

2.4

Conclusions and recommendations

When evaluating all the results, it can be concluded that there is not a significantly better variation. One variation however can be identified as not favourable and that is tr+5-5 (rotation of main spar fixed at root what results in a rotation of principal stiffness axis towards feathering). The reason for this is that the maximum (leadwise) bending moment and fatigue (flapwise) bending moment below rated wind speed both increases. Besides this there is no reduction of the tip deflection. The most favourable variation can be said to be tt+5 (rotation of main spar fixed at tip what has no influence on the principle stiffness axis, but only moves the centre of gravity towards trailing edge) as the tip deflection (-10 %) as well as the leadwise bending moment (-12 %) reduce significantly. The fatigue loading does not change much though. As it is very difficult to precisely tell that one variation is better than the other, more investigations need to be done into the effect of structural pitch.

DOWEC 10070 rev 2

30-09-2003 Page: 10 of 19

3 Carbon blades 3.1

Introduction

Rotorblades nowadays are normally built of glass fibre reinforced polyester or epoxy. For large blades also carbon seems to be an option as stronger and lighter designs are expected. To evaluate the influence of carbon fibres in a blade, two different designs are made besides the reference design. This results in three different variations: • Reference design, consisting only of glass fibres. • Carbon fibre main spar, a design where only the main spar consists of carbon fibres and the rest of the blade of glass fibres. • Full carbon fibre blade, a design where all glass is replaced by carbon. All three designs are optimised with Focus and Finstrip. They have been analysed to have sufficient static, fatigue and stability strength. The results, which will be compared, are stiffness, weight, tip deflection, eigenfrequencies, static moment, inertia and the costs. The used data were obtained in a EU study. The main characteristics were: • Stiffer • Better fatigue • Higher tensile strength • Lower compressive strength LM Glasfiber Holland derived an own load spectrum for the 6 MW (wind conditions derived from according to IEC IB norm) turbine using Focus and Flexlast. The first load spectrum was a glass epoxy blade. The first versions of the glass/carbon fibre model were also derived with this load spectrum. But for later versions LM has made a new load spectrum. 3.2

Comparison of model properties

3.2.1 Mass If the glass fibre model is considered as the reference model, than there is a relative reduction of mass as shown in figure 11.

100%

87%

71%

Reference blade

Carbon fibre main spar

Full carbon blade

Figure 10: Mass of different models

As can be expected the blades containing carbon fibres have a lower mass. When glass was replaced one on one by carbon, the mass would already be lower due to lower specific mass of DOWEC 10070 rev 2

30-09-2003 Page: 11 of 19

carbon fibre. Besides this, the carbon fibres are better in fatigue and comparable in static strength, so the design can also be made lighter. The reduction of mass however is not as spectacular as one would expect. This is mostly due to the fact that the static strength of carbon fibres is not better than glass, the compression strength (available at that time) is even worse that the glass fibres. 3.2.2 Stiffness The stiffness in both directions (edgewise and flapwise) is shown in figure 13 and 14.

Stiffness

0

Reference blade Carbon fibre main spar Full carbon blade

10

20

30

40

60 radius [m]

50

Figure 11: Edgewise stiffness distribution

Stiffness

Reference blade Carbon fibre main spar Full carbon blade

0

10

20

30

40

50

60 radius [m]

Figure 12: Flapwise stiffness distribution

3.2.3 Tip deflection The tip deflection of the three models is different because of the differences in stiffness. Their deflection is represented in figure 13.

DOWEC 10070 rev 2

30-09-2003 Page: 12 of 19

100%

58%

55%

Reference blade

Carbon fibre main spar

Full carbon blade

Figure 13: Tip deflections of different designs

3.2.4 Eigenfrequencies Also the eigenfrequencies depend on the stiffness of the blade. If the stiffness increases, the frequency will be higher. As seen in the stiffness figures, the stiffness of the full carbon blade is higher than the stiffness of the reference blade and the blade with a carbon fibre main spar of which the reference blade has the lowest stiffness. The eigenfrequencies follow the same pattern. The frequencies of the blade are shown relative at two different rotor speeds in the next diagram.

zero rotor speed

173%

172%

nominal rotor speed

120% 100%

119%

100%

Reference blade

Carbon fibre main spar

Full carbon blade

Figure 14: Frequencies of different designs

DOWEC 10070 rev 2

30-09-2003 Page: 13 of 19

3.2.5 Static moment and inertia Due to the differences in mass a difference in static moment and inertia occurs. More mass means a higher static moment and higher inertia. This is especially of importance for the bolts of the blade-bearing connection. The proportions of these aspects are compared in figure 15.

100%

stat ic moment

100%

inert ia

81%

78% 73%

Reference blade

Carbon fibre main spar

71%

Full carbon blade

Figure 15: Static moments and inertia of different designs

3.3

Costs

Another important issue is the costs of the blade. This can be divided into material costs, production costs and total costs: • The material costs go hand in hand with the materials that are used. Almost all used materials in the blades are familiar in the existing blades nowadays. • The production costs are based on an linear relation between thickness and production costs. Density has to be used instead of mass, because of the fact that the time of producing depends on how many layers glass fibre or carbon fibre are used, not how much mass that is. Aspects like production halls, tools, engineers and profits are not taken into account. • To obtain the total costs the material and production costs need to be combined. In figure 16 this is shown.

DOWEC 10070 rev 2

30-09-2003 Page: 14 of 19

M at erial costs

258%

Product ion costs Tot al cost s

204%

164% 141%

100%

100%

100% 83% 65%

Reference blade

Carbon f ibre main spar

Full carbon blade

Figure 16: Total costs of different designs

As can be seen the total costs for a full carbon blade are twice as high as for the reference blade. Due to the high price of carbon fibres the carbon fibre blade is twice as expensive as the reference blade. As production costs drop slightly due to less layers of composite, the total costs compared to the reference are less than the material costs. 3.4

Conclusions

It can be concluded that the carbon fibre blade has many advantages except its price. This price has become so high because of the high material price of carbon fibre. If the price would decrease with 80%, the carbon fibre blade becomes just as expensive as the glass fibre blade and might be a realistic option. It is to be expected that the price of carbon fibres will decrease in the future, but probably not this much. On the other hand a higher price is affordable, because of the possibility to make a light design with carbon fibre. If the compression strength of carbon fibre increases to the level of glass fibre (or higher) due to better tests, many advantages will be achieved. This can result for example in lower stresses / higher reserves or a reduction of carbon fibres, which will make the blade lighter and less expensive. As a conclusion, the carbon fibre blades are more expensive, but they can have an added value for the turbine. This added value can be: • Shorter main shaft due to less tip deflection, • Thinner main shaft due to less dead mass, • Smaller hub, pitch bearing and cheaper bolts due to smaller fatigue loads.

DOWEC 10070 rev 2

30-09-2003 Page: 15 of 19

4 Aerodynamic blade root At modern LM blades the aerodynamic profiles start at 15 % so this first part does not have a contribution in the power performance. It is investigated if it is possible to modify this part of the blade so that it can have a larger contribution to the aerodynamic power performance of the blade. In this chapter some variations are suggested to improve the losses at the root. The expected payoff is approximately 1 – 2 %. 4.1

Possible modifications

Some possible modifications and ideas will be listed here, with their possible contribution and if it is expected to give a contribution. The possible ideas are the following: • Include vortex generators near root Principle: Vortex generators (VG’s) give a delay of separation and this results in an elongation of the linear part of the lift curve until the VG location has a separated flow (same influence as a high c/r value). Improvement: None as the circular root has no camber and due to the low Reynolds number produces no lift. • Include zig-zag tape near root Improvement: None as this will only have influence near the tip of the blade due to the low Reynolds number (low wind speeds) • Extend the aerodynamic profiles towards root Principle: The root has an aerodynamic shape and is able to produce lift. Improvement: A root, which produces lift, could give an improvement in performance and also could reduce losses. As only the latter modifications is expected to give an improvement this will be further investigated. 4.2

Input

The input for the blade is based on the preliminary design of LM64.5 (see also ref. [3]). The same aerodynamic characteristics as mentioned in the reference are used for the power calculation. To obtain properties of the profiles that need to be extrapolated towards the root, a first estimated profile with a 50% profile thickness has been calculated with Rfoil. The profile is obtained from the DOWEC structural predesign at a section between root and largest chord (this profile is based on a DU profile). The aerodynamic properties of the rotating case need to be used because near the root high values of c/r exist (now used c/r = 0.263). The results of the Rfoil calculations can be seen in figure 17. As can be seen the maximum lift coefficient increases largely due to the delay in separation caused by the rotational effects. This also results in a decrease of the drag coefficient.

DOWEC 10070 rev 2

30-09-2003 Page: 16 of 19

Aerodynam ic characteristics of 49% profile 1.6 Cl

1.6 Cl

1.4

1.4 1.2

Without rotation

1.2

1

With rotation

1 0.8

0.8

0.6

0.6

0.4

0.4

0.2

0.2 0

0 0

0.01

0.02

0.03

-0.2

Cd

0.04

0.05

-5

-0.2

0

5

10 α (deg)15

20

-0.4

-0.4

Figure 17

Besides extrapolating the aerodynamic profiles towards the root, also the twist has to be adjusted. TU Delft has performed this adjustment of twist. In figure 18 the chord and twist distribution for the first 30 m has been shown. The rest of the chord and twist distribution is completely the same as the reference blade and will therefore not be shown. Conventional root Aerodynamic root

Twist distribution

18

5.0

100%

4.5

90%

4.0

80%

3.5

70%

3.0

60%

2.5

50%

2.0

40%

1.5

30%

1.0

20%

0.5

10%

2

0%

0

twist (deg)

chord (m)

Chord distribution

16 14 12 10 8

Radius (m)

0.0 0

5

10

15

20

25

30

6 4

Radius (m) 0

5

10

15

20

25

30

Figure 18

DOWEC 10070 rev 2

30-09-2003 Page: 17 of 19

4.3

Results

First of all the Cp – λ curve is calculated to investigate at with λ the maximum Cp appears (this could be different for both designs. The results of this calculation are shown in figure 19. Cp-λ for LM63.2 blade

0.6

0.5

Conventional root Aerodynamic root

Cp

0.4

0.3

0.2

0.1

0.0 0

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

λ

Figure 19

For both designs the maximum Cp lies at λ = 8 and the corresponding Cp are respectively 0.491 for the conventional root design and 0.509 for the aerodynamic root design. This illustrates that the aerodynamic root design gives a maximum pressure coefficient that lies closer to the Betzoptimum. Reason for this is a reduction in aerodynamic losses near the root. The calculated annual yield for both designs can be seen in figure 20. The absolute increase in annual yield at all wind speeds is approximately 400 MWh (this is 1.5 % of annual production for k=2 at 9m/s). As this factor is almost constant the difference in percentage between the two designs goes towards each other. Assuming a price of 0.09 €/kWh this implies an increase of 36,000 €/year extra yield.

DOWEC 10070 rev 2

30-09-2003 Page: 18 of 19

Annual Yield for LM63.2 blade 35

103.5% Aerodynamic root Conventional root

Yield (GWh)

30

103.0%

Difference aero/conv root

25

102.5%

20

102.0%

15

101.5%

10

101.0%

5

100.5% 6

7

8

9

10

11

12

Windspeed (m/s)

Figure 20

4.4

Conclusions and recommendations

From this evaluation can be concluded that modifying the root to reduce losses can have an effect on the annual yield. In the previous sections only a 50% airfoil has been used and as a start it was stated that the root must not have a larger diameter. In this case the root length has increased with 50% due to the use on this profile. To have a more realistic model more sections need to be known and also the profile thickness has to go to approximately 65-70%. The loads due to storm on the blade root also need to be investigated as it is expected that the larger chord will give higher loads for the blade and the turbine. As it is intended to make this aerodynamic shape by a form piece, the loads cannot be too high. Besides this, the aerodynamic root makes it is more difficult to pitch due to the presence of the nose cone. Therefore, a close investigation has to be done together with the turbine manufacturer if this is to be further developed.

References [1] [2] [3] [4]

Bulder, B.H.: Dowec 6 MW pre-design. ECN-C-01-???, ECN, August 2001. Elenbaas, M.: Seminar proceedings Annex 1. Descriptions of one-dimensional variations. DOWEC-F1W0-ME-02-054/00-C, 25-1-02. Kooijman, H.J.T. et al.: DOWEC 6 MW pre-design. ECN-CX-01-135, ECN, January 2002 Kooijman, H.J.T.: DOWEC 6 MW “structural pitch analysis”. Calculations by Stentec. Note received by email form ECN, 22-07-02.

DOWEC 10070 rev 2

30-09-2003 Page: 19 of 19