Professional Documents
Culture Documents
DOWEC-F1W2-MZ-02-063/02-P
DOWEC 63 rev. 2
version Date No of
pages
0 08/05/02 New document
1 14/05/02 Include fatigue analysis; Editorial changes
2 20/08/03 Converted to PDF
DOWEC-F1W2-MZ-02-063/02-P
Preface
Results are reported for a contribution to the single objective concept alternatives in WP 2 Task 1 of
the project ‘DOWEC’ commissioned by EET. The following persons have contributed to this work:
• TU Delft
Michiel Zaaijer (editor)
Jan van der Tempel
• NEG-Micon
Peter Fish
• Ballast Nedam
Ton Topper
Arjen Vos
The report has been published by TUDelft, DUWIND, under number 2002.008.
Page 2 of 21
DOWEC-F1W2-MZ-02-063/02-P
Contents
1 INTRODUCTION.........................................................................................................................4
8 REFERENCES............................................................................................................................21
Page 3 of 21
DOWEC-F1W2-MZ-02-063/02-P
1 Introduction
The baseline support structure design is a tubular tower on a monopile foundation for site 1 of the
Terms of Reference [4]. For the 6 MW DOWEC and the environmental conditions at the selected site
it is expected that this is a very economic solution, when only procurement costs of the support
structure are considered. However, this concept constrains some other alternatives, such as the suction
pile foundation and installation procedures. Therefore, a space frame or tripod foundation must be
designed to obtain reference values of loads, dimensions and costs.
Furthermore, the stiffness/natural frequency of the support structure must be designed such that
resonance due to wave excitation and wind excitation at the rotor and blade-passing frequencies are
avoided. This appeared to be critical for the monotower design, particularly with respect to
uncertainties and variations due to scour. The design for stiffness/natural frequency is expected to be
easier for a space frame or tripod.
Last but not least, the monopile may become technically/economically less feasible at other sites, in
which case a good reference for a space frame or tripod is convenient.
The overall objective is the design of a cost effective steel space frame or tripod, without scour
protection. This objective is narrowed down to the design of a tripod, which incorporates most
characteristics of a (sub-sea) space frame. Within this work, cost effectiveness relates to the
manufacture costs of the tripod and foundation. In principle, the baseline design of the tower above the
pile-tower-flange will be maintained.
In a first step the main dimensions of the tripod will be optimised with an engineering model that gives
an automated design solution. In the second step some small adaptations will be made to the optimal
design solution for details that are not modelled adequately in the engineering model. In the last step
of the design the hydrodynamic fatigue in a 20-year lifetime is checked.
The definition of installation procedures and an estimate of installation costs are not treated in this
document. For general reference the topology of the tripod is illustrated below:
Central column
Node height
Brace
Base radius
Base (centre)
Base (outside)
Page 4 of 21
DOWEC-F1W2-MZ-02-063/02-P
The engineering model is particularly suited for the early design phase, where concept selection and
general optimisation must be performed. The engineering model and some examples of its use are
described in [11] and [10]. These descriptions are particularly useful to get an understanding of the
limitations of the model.
Turbine data
Turbine data is based on the information provided in [5].
Nominal power: 6 MW
Rotor diameter: 129 m
Hub height: 91.4 m above MSL
Rotor-nacelle mass: 272·103 kg
The engineering model assumes a uniform cohesionless soil. A friction angle of 32° was used, as an
average of the different soil layers.
Operating conditions:
Rated wind speed (hub height) [5] 12 m/s
Wave height at rated wind speed [3] 2.1 m
Page 5 of 21
DOWEC-F1W2-MZ-02-063/02-P
Design data
The following diameter to wall thickness ratios are used as input for the model:
The diameter to wall thickness ratio of the tripod braces is not predefined, but analysed in the
optimisation process.
No dynamic amplification has been used for hydrodynamic loading and for aerodynamic loading a
gust response factor of 1.5 has been applied to all static aerodynamic responses.
Pile design
The dimensions of the piles are compared in Table 1.
The differences are (at least qualitatively) explained by the different assumption in seabed level. The
baseline is designed for a seabed level up to 36 m below MSL, instead of 21 m. The extra water depth
results in a larger maximum moment in the baseline pile and corresponding slightly larger diameter
and extra penetration of 4 m in the seabed.
By choice, the pile design in the engineering model stops at the seabed, instead of at the tower flange.
Tower design
The dimensions of the tower are compared in Table 2. Only values above the flange of the baseline, at
30 m above the seabed, can be compared directly. Below the flange the tower design of the
engineering model and the pile design of the baseline use a different diameter to wall thickness ratio.
Page 6 of 21
DOWEC-F1W2-MZ-02-063/02-P
The tower top diameter of the engineering model is smaller than that of the baseline. This is probably
mainly due to negligence of the static bending moment due to the eccentric centre of gravity of the
rotor and nacelle, local effects of the yaw system connection and torsion. Furthermore, the baseline is
designed as a linearly tapered tower, whereas the engineering model designs a segmented tower. At
the tower flange the dimensions of the baseline and the engineering model are nearly the same and the
total mass above the flange is also nearly equal.
Loads
For a seabed level of 21 m below MSL the loads at the mudline are compared in Table 3.
The loads correspond sufficiently well for the purpose of an initial design.
Page 7 of 21
DOWEC-F1W2-MZ-02-063/02-P
3 Optimisation
Figure 1 shows the total mass of tripod, tower and piles as a function of node height, base radius and
diameter to wall thickness ratio, D/t. The first observation is the relatively large area of possible tripod
solutions with less than 10% deviation from the minimum mass. It is also observed that the diameter to
wall thickness ratio of the braces has only a small influence.
45 45
40 40
35 35
Node height
Node height
30 30
25 25
20 20
15 15
10 10
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
Base radius Base radius
Figure 1 Total mass as function of node height and base radius in 103 kg
Figure 2 gives the mass of tripod and piles separately, for D/t = 100. The tripod mass is up to the
flange at 30 m above the seabed. The tower above the flange is equal for all tripod dimensions and has
a mass of 226·103 kg. The two major aspects that determine the tripod mass as a function of base
radius are the change in brace length and normal forces in the braces. There is an optimum between 10
and 25 m, depending on the node height. Pile mass increases dramatically at small base radius, due to
the excessive axial pile loads to withstand the overturning moment. For very small base radii pile
masses are even underestimated, because group effects are not modelled.
Page 8 of 21
DOWEC-F1W2-MZ-02-063/02-P
45 45
40 40
35 35
Node height
Node height
30 30
25 25
20 20
15 15
10 10
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
Base radius Base radius
Tripod 3 Piles
Figure 2 Mass of tripod and piles as function of node height and base radius in 103 kg
The node should be outside the splash zone, to avoid excessive hydrodynamic loading and corrosion
of that sensitive part of the structure. Furthermore, inspection is easier completely below or above the
water surface. The splash zone is determined from the following data in [2]:
Maximum height above MSL for tide plus storm surge (return period 50 year): 2.39 m
Minimum height above MSL for tide plus storm surge (return period 50 year): -1.23 m
Extreme wave height (return period 50 year, see Section 2.2): 11.8 m
For a linear wave the top-top wave height is equal to a wave amplitude of 5.9 m. In reality the non-
linear wave would have a larger positive water elevation. Combining the above data with a MSL of
21 m results in a splash zone between 13.9 m and 29.3 m above the mudline. Figure 1 shows that a
node above the splash zone can give the lowest total mass of the structure.
A node height of 30 m above the mudline is selected, which is near the node height of the minimum
achievable total mass. With this node height the flange with the tower can be at the same level as in
the baseline design, which facilitates comparison of dynamic behaviour.
At the selected node height minimum total mass is obtained at a base radius of 20 m. The resulting
tripod and pile dimensions for D/t equal to 100, node height equal to 30 and base radius equal to 20
are given in Table 4. For the given pile penetration depth it is reasonable that pile group effects are
ignored. It should be kept in mind that total mass is not very sensitive around the selected parameters,
providing room for optimisation for other criteria. For instance, a narrow base could be preferred with
respect to installation or a wide base could be preferred with respect to loads on suction piles.
Page 9 of 21
DOWEC-F1W2-MZ-02-063/02-P
Several modifications are made to the engineering model design, before the hydrodynamic fatigue and
natural frequency analyses. The modifications are described below and the dimensions of the resulting
initial design are given in Table 4.
The piles were originally dimensioned assuming no scour. Pile diameter and wall thickness are
adapted to be able to withstand 7 m local scour. These piles are more realistic when the effect of scour
on the natural frequencies is assessed. The engineering model is based on a different topology of the
tripod base and the diameter and wall thickness of the engineering model are used for the members of
the base that is illustrated in the introduction. This results in a higher total mass of the base members.
The optimisation process resulted in a tapered central column of the tripod with a constant diameter to
wall thickness ratio. Such a central column gives large flexibility and is difficult to manufacture.
Instead, a constant diameter of 4 m is taken, which was the result at the tripod top for D/t = 50. The
wall thickness decreases in steps from 0.08m at the top to 0.02m at the base.
Transition from tripod to tower, tripod member connections and pile sleeves are not analysed. Some
margins are used for instance to account for transfer of tower moment to the braces (in the engineering
model a hinged connection was assumed).
The foundation loads of the initial design are (without safety factor):
All extreme loads were observed for the load case with the turbine in operation (case 3). These
foundation loads are used in the feasibility study of suction buckets [13]
Page 10 of 21
DOWEC-F1W2-MZ-02-063/02-P
All wave loading and dynamic amplification of wind and wave loading is considered as dynamic
loading. The (quasi-) static solution of wind loading during extreme and operating conditions is
considered as static loading.
Page 11 of 21
DOWEC-F1W2-MZ-02-063/02-P
Long-term wave conditions are taken from the scatter diagram of wave height and zero crossing period
in [2]. The parameters of the tripod and tower of 3.2 are used to model the structure above the seabed,
with the addition of the nacelle inertia parameters of Section 5.1. The foundation is modelled with a
stiffness matrix, whose elements are determined using the pile-foundation model in ANSYS.
Using SESAM the fatigue damage during the 20-year lifetime is calculated. Fatigue damage with a
value lower than 1, indicates that the structure is likely to survive the dynamic loading during its
lifetime. A safety-factor equal to 2 is used, meaning that a fatigue damage of 0.5 is considered an
acceptable maximum value for all elements of the tripod.
Table 6 shows the hydrodynamic fatigue damage of the members of the tripod, calculated with
SESAM.
Evidently, the tower and central column of the tripod have sufficient margin against hydrodynamic
fatigue damage. However, the braces and base need improvement. To keep the fatigue damage below
0.5 the wall thickness of the tripod members is adapted. When a wall thickness above 80 mm is
required the diameter is increased. The modified tripod members are given in Table 7.
Page 12 of 21
DOWEC-F1W2-MZ-02-063/02-P
Table 7 Dimensions of adapted tripod with sufficient resistance against hydrodynamic fatigue
Braces
Outer diameter (m) 1.60
Wall thickness (m) 0.030
Base (centre)
Outer diameter (m) 1.00
Wall thickness (m) 0.081
Base (outside)
Outer diameter (m) 1.00
Wall thickness (m) 0.016
Central column
Overall diameter (m) 4.0
Top wall thickness (m) 0.080
Base wall thickness (m) 0.020
Because the base appears to be very susceptible to fatigue, a different topology of the base is
considered. A top view of the alternative base is illustrated in Figure 3. The base segments have a
maximum fatigue damage of 0.09, when they have a diameter of 0.8 m and a wall thickness of
0.016 m. This topology has a smaller mass of the base segments than the original topology after the
fatigue analysis. Therefore, the final pre-design of the tripod uses the topology of Figure 3.When the
braces have the same dimensions as the initial design, maximum fatigue damage for the braces is 1.54.
The wall thickness of the braces is increased to get the damage below 0.5. The dimensions of the final
pre-design are given in Table 8.
Page 13 of 21
DOWEC-F1W2-MZ-02-063/02-P
Page 14 of 21
DOWEC-F1W2-MZ-02-063/02-P
5.1 Introduction
For civil structures often stiffness is used as a typical property of foundations. For wind turbines, the
natural frequency of the entire support structure is a more useful property, since it relates more directly
to the dynamic behaviour. A broad study of the sensitivity of the natural frequency to variations in
foundation parameters was performed in the OWTES project and summarised in [14] and [12]. The
study confirmed the dominant influence of general and local scour. For the DOWEC tripod the natural
frequency is calculated for several scour situations, using the same modelling approach as [12].
To enable comparison, first the natural frequencies of the baseline are determined. These baseline
results are also compared with the results of [5], to check consistency within the project team. At the
end of this chapter the natural frequencies of some variations in the design are assessed.
The support structure dimensions of Section 3.2 are used in this chapter, without the modification
suggested in Chapter 4. In addition, the following rotor-nacelle inertia properties are used, based on
[5]:
Table 9 Natural frequencies of tower, clamped at pile-tower transition 9 m above MSL (CD+10)
ECN [5] TUD (current study)
st
1 fore aft (Hz) 0.356 0.360
2nd fore-aft (Hz) 2.41 2.40
The difference between the two models is quite significant. The models of the pile foundation differ in
the following aspects:
1. The foundation flexibility in the ECN study is determined at extreme loading conditions and in the
current study at the neutral position. Due to plastic deformations during extreme loading the flexibility
of the soil is larger in that situation.
2. In the ECN study the frequency analysis uses a flexibility matrix to represent foundation behaviour,
based on a deflection analysis performed in [9]. In the flexibility matrix the column relating to lateral
forces is neglected. The current study uses a full FEM representation of the foundation pile and
springs based on py-curves for pile-soil interaction.
3. Pile-soil interaction is modelled differently. In the current study no interpretation of the
automatically generated py-curves has taken place, whereas in [9] calculated very large lateral soil
resistance forces were reduced manually [15].
Page 15 of 21
DOWEC-F1W2-MZ-02-063/02-P
The difference is further analysed in [6]. For comparison of natural frequencies of the tripod with
those of the baseline the difference must be considered.
Due to the increased first natural frequency the risk of tower resonance at 1-P excitation is greatly
reduced and there is still sufficient margin with 3-P excitation. Additionally, the first natural frequency
of the tripod is further away from the high-energy frequencies in the wave spectrum. In case the
baseline natural frequency obtained by ECN is the most realistic, the tripod has a clearly beneficial
reduction of resonance risk and expected hydrodynamic and aerodynamic fatigue.
The effect of local and general scour on the first two natural frequencies of the tripod and baseline are
shown in Figure 4. It is expected that the absolute local scour depth around the tripod piles is smaller
than around the monopile. However, the tripod structure may also cause some additional scour.
General scour is not considered for the tripod, since this has also not been considered in the design
phase.
1.0
Baseline general scour
Relative natural frequency
0 1 2 *D Baseline (-)
0.5
0 5 10 15 (m)
Scour depth
Figure 4 Variation of natural frequency with local and general scour
The first natural frequency of the tripod is far less sensitive to local scour than that of the baseline,
particularly when the scour depth is smaller. However, the second natural frequency of the tripod is
very sensitive to scour. It is expected that the large sensitivity is caused by the lateral flexibility of the
unsupported pile section in the scour hole. As was shown in Section 3.2 the lateral load on the pile top
Page 16 of 21
DOWEC-F1W2-MZ-02-063/02-P
The baseline is more sensitive to global scour than to local scour, due to the remaining part of the
overburden pressure in case of local scour.
45 0.289
0.287 0.81
0.92 40
35 0.267
0.269 0.90
1.05
Node height
30
25
20
0.249
1.32
15
0.254
10
1.39 5 10 15 20 25 30 35 40
Base radius
Figure 5 1st and 2nd natural frequency of several tripod alternatives in Hz (see Figure 1)
• The variation of the first natural frequency without large increase of material use is limited to
a range of approximately 10%. For the second natural frequency the range is approximately
20%.
• Increase of node height results in increase of first natural frequency and decrease of second
natural frequency.
• Increase of base radius results in decrease of second natural frequency, but is not directly
related to increase or decrease of first natural frequency.
• For small node heights the first natural frequency can even be below that of the baseline.
Finally, several variations in the topology of the tripod with node height of 30 m and base radius of
20 m are assessed, aiming at a stiffer tripod. The results are shown in Table 12.
Page 17 of 21
DOWEC-F1W2-MZ-02-063/02-P
The alternative topologies result in a small increase of first natural frequency of less than 5% and a
larger increase in second natural frequency up to 30%.
The fictitious rigid tripod has a first natural frequency very close to the first natural frequency of the
tower clamped at its base (see Section 5.2). This shows that the piles act very stiff in the first natural
mode shape. The second natural frequency deviates more from that of the tower clamped at its base,
due to the lateral flexibility of the pile foundation. The natural frequencies of the design alternatives
are still far away from those of the rigid tripod.
Page 18 of 21
DOWEC-F1W2-MZ-02-063/02-P
6 Indication of costs
6.1 Assumptions
This chapter gives an indication of the costs of the tripod support structure. The costs are estimated for
a wind farm of 80 turbines.
For the baseline the costs of the monopile are estimated at € 2.25 per kg. This value is also used for the
foundation piles of the tripod. For the tripod (up to the flange at 9 m above MSL) a slightly higher
price of € 3,50 per kg is used, due to the increased complexity and workspace requirements. The tower
above the flange is equal to that of the baseline and hence, costs are the same. The costs of 80 towers
are 74 M€, omitting the transition pieces of 31 M€ [16].
Table 13 Cost estimation of tripod support structures for 80 turbines in 21 m water depth
Mass per turbine (kg) Costs of 80 turbines (M€)
Tower, including coating (above (Price based on baseline result) 74
MSL + 9 m)
Tripod 306·103 86
Foundation piles 82·103 15
Total - 175
Table 14 Cost estimation of tripod support structures for 80 turbines in 36 m water depth
Mass per turbine (kg) Costs of 80 turbines (M€)
Tower, including coating (above (Price based on baseline result) 74
MSL + 9 m)
Tripod 445·103 125
Foundation piles 80·103 14
Total - 213
Page 19 of 21
DOWEC-F1W2-MZ-02-063/02-P
7 Conclusions
A preliminary design of a tripod for the 6 MW DOWEC at Site 1 of the Terms of Reference was
made. A water depth of 21 m was assumed, consistent with the monopile design used in the baseline
cost model. Since large sand waves and scour are expected at this site a second preliminary design was
made for a water depth of 36 m (the reported design of the monopile is actually for this deeper water).
However, analysis of hydrodynamic fatigue and natural frequency are only performed for the tripod in
21 m water depth.
A relatively large variation of node height and base radius around the optimum values has only
moderate effect on total mass of the support structure. However, exclusion of the splash zone for the
node height reduces the region of possibilities to a large extent. At 21 m water depth the node of the
tripod should be above the splash zone, to obtain the lowest support structure mass. The base radius
can still be optimised with respect to e.g. installation or (low) foundation loads, without large mass
increase. In the initial design the node height is set at 30 m above the seabed and the base radius is
20 m.
The application of scour protection, or not, affected pile mass by a factor 2, mainly due to change in
the pile diameter to withstand the bending moment due to lateral forces. However, even when scour is
allowed to occur pile mass is still only approximately 15% of the total support structure mass.
After some small modifications to the engineering model design and an analysis of the hydrodynamic
fatigue during the lifetime of the support structure, the mass of the tripod, between the seabed and
tower flange, increased with approximately 40%. The tower mass remained the same. Including the
increased pile mass, the total mass of the final pre-design of the support structure was approximately
25% higher than the mass of the design solution of the engineering model. The tower and central
column of the tripod appeared to be insensitive to hydrodynamic fatigue, which is remarkable, since
monopiles can experience significant hydrodynamic fatigue up to the tower top.
At 0.282 Hz, the first natural frequency of the tripod is further away from 1-P and hydrodynamic
excitation than for the baseline, resulting in decreased resonance risk and less fatigue. Furthermore, the
first natural frequency of the tripod is less sensitive to scour than that of the baseline. However, the
second natural frequency is more sensitive to scour, due to the lateral flexibility of the unsupported top
sections of the piles.
A preliminary study of the possibilities to change the first natural frequency of the tripod by design
variation was less promising than expected. For the assessed alternatives the variation of first natural
frequency was below 10%.
The final preliminary design solution for 21 m water depth, has a total mass of 613·103 kg per unit and
costs 175 M€ for 80 turbines. For 36 m water depth the mass equals 750·103 kg per unit and costs
213 M€ for 80 turbines.
Page 20 of 21
DOWEC-F1W2-MZ-02-063/02-P
8 References
[1] API, RP 2A-LRFD: API Recommended Practices for Planning, Designing and Constructing
Fixed Offshore Platforms – Load and Resistance Factor Design, First Edition, July 1, 1993.
[2] Bierbooms, W.A.A.M., Wind and Wave Conditions - DOWEC Work Package 1 Task 5,
DOWEC-F1W1-WB-01-047/00-C, Section Wind Energy, Delft, November 2001.
[3] Germanischer Lloyd, Rules and Regulations IV – Non-Marine Technology, Hamburg, 1999.
[4] Goezinne, F., Terms of Reference DOWEC, 176-FG-R0300 V1, Neg Micon Holland bv,.
Bunnik, September 2001.
[6] Kooijman, H.J.T., Tower analysis 6 MW, Forthcoming, ECN, Petten, May 2002.
[7] Kooistra, A., DOWEC Soil data, 999470\00C0003.AKO, Ballast Nedam Engineering bv,
Amstelveen, June 2001.
[8] Kühn, M., Dynamics and Design Optimisation of Offshore Wind Wind Energy Conversion
Systems, PhD thesis, Delft University of Technology, Delft, 2001.
[9] Nehal, R.S., DOWEC Foundation Design Monopile - 3.6 MW & 6 MW Wind Turbines,
100130.R01 rev. a, Ballast Nedam Engineering bv, Amstelveen, November 2001.
[10] Zaaijer, M.B., Properties of Offshore Support Structures for Large Scale Wind Turbines, In:
Proceedings of Offshore Wind Energy Special Topic Conference (CD-rom), Brussel, Belgium,
December 2001.
[11] Zaaijer, M.B., Broek, W. van den, Bussel, G.J.W. van, Toward Selection of Concepts for
Offshore Support Structures for Large Scale Wind Turbines, In: Proceedings of MAREC 2001,
Newcastle, UK, March 2001.
[12] Zaaijer, M.B., e.a., Design methods for offshore wind turbines at exposed sites (OWTES) -
Sensitivity analysis for foundations of offshore wind turbines, Section Wind Energy, WE 02181,
Delft, March 2000.
[13] Zaaijer, M.B., e.a., Suction bucket foundation - Feasibility and pre-design for the 6 MW
DOWEC, DUWIND 2002.007, Delft, Mei 2002.
[14] Zaaijer, M.B., Vugts, J.H., Sensitivity of Dynamics of Fixed Offshore Support Structures to
Foundation and Soil Properties, In: Proceedings of the European Wind Energy Conference and
Exhibition 2001, Copenhagen, Denmark, July 2001.
[15] Personal communication with Arjen Vos, Ballast Nedam Engineering bv, April 2002.
Page 21 of 21