Table of Contents - 10th EAWE PhD seminar on Wind Energy in

215 Pages·2014·34.65 MB·English

Table of Contents
Damage tolerant design: failure and crack propagation in composites., G. Pereira [et al.] ....................1
Convex Optimization of Space Frame Support Structures for Offshore Wind Turbines, K. Sandal [et al.]
..................................................................................................................................................................6
Effect of laminate thickness on the static and fatigue properties of wind turbine composites, F. Lahuerta
[et al.] .....................................................................................................................................................10
Development of a wind farm tool using advanced actuator disk models, M. Moens [et al.] .................15
The right size matters: Investigating the offshore wind turbine market equilibrium, N. Ederer............19
DISTRIBUTED CONTROL OF WIND FARMS USING A FLOW INTERACTION MODEL AND A
MULIT-AGENT APPROACH, M. Vali [et al.] .....................................................................................31
COMBINING MODEL-BASED AND DATA-DRIVEN OPTIMIZATION OF WIND FARM
OPERATION IN A LEARNING DATABASE, A. Rott [et al.] .............................................................35
AGGREGATE WIND FARM POWER PERFORMANCE CURVES, A. Elmontaser..........................39
Towards Implementation of an Optimization Tool for Rotor Blades based on the Adjoint Method in
OpenFOAM, L. Vorspel [et al.] ..............................................................................................................43
CFD analysis of a 2-bladed multi-megawatt turbine, L. Klein [et al.] ...................................................47
Numerical investigations of a passive load alleviation technique for wind turbines, A. Fischer [et al.] ...
51
LARGE-EDDY SIMULATIONS OF S826 AIRFOIL WITH DISCONTINUOUS GALERKIN
METHODOLOGY, A. Frère [et al.] .......................................................................................................55
Increased Order Modeling of the Aerodynamic Characteristics of Flexible Blades, Z. Wang...............59
Hybrid aerodynamic analysis of wind turbines, M. Schwarz [et al.] .....................................................63
UNSTEADY AERODYNAMICS OF AIRFOILS FOR SMALL HAWT AT LOW REYNOLDS
NUMBERS, D. Holst [et al.] ..................................................................................................................67
CFD coupled with WRF for Wind Power Prediction, E. Leblebici [et al.] ............................................71
Aerodynamic Modeling for Equal Fidelity Aeroelastic Analysis, T. Hegberg........................................76
AERODYNAMIC DAMPING OF WIND TURBINES UNDER CONSTANT AND TURBULENT
WIND, S. Schafhirt [et al.] .....................................................................................................................80
The Low Induction Rotor, P. Mills [et al.] .............................................................................................84
Aerodynamic Performance Losses due to Ice Buildup in Wind Turbines, O. Yirtici [et al.] .................88
Implementation of Passive Control Strategies through Swept Blades, C. Pavese [et al.] ......................92
Integrated aero-structural optimization of wind turbine rotors, P. Bortolotti [et al.] .............................96
Design of a Floating Wind Turbine to Measure the Coupled Response to Wind and Wave Action, C.
Gilmour.................................................................................................................................................100
EXPERIMENTAL METHODS FOR THE OPTIMAL DESIGN OF SMALL WIND TURBINES, N.
Bartolini [et al.] ....................................................................................................................................105
FLAPS FOR WIND TURBINE APPLICATION: NOISE SOURCE LOCALIZATION ON A TEST
AIRFOIL, C. Brand [et al.] ..................................................................................................................109
Numerical Investigation of an Airfoil with Morphing Trailing Edge for Load Reduction, T. Wolff [et al.]
I
...............................................................................................................................................................114
CFD studies of a 10 MW wind turbine equipped with active trailing edge flaps, E. Jost [et al.] ........119
Wake Flow Model for Wind Farm Control, T. Ahmad [et al.] .............................................................123
Effects of Tip-Injection on the Flow Downstream of a Model Wind Turbine Rotor Blade Tip, A.
Abdulrahim [et al.] ...............................................................................................................................127
Determining the Wind Speed Distribution within a Wind Farm considering Site Wind Characteristics and
Wake Effects, T. Ahmad [et al.] ...........................................................................................................131
Feedback control of blades trailing edge flap for blade root load mitigation, R. Ungurán [et al.] ......135
Investigate derivation of a wind turbine dynamics from measured data, L. Reguera [et al.] ..............139
Controlling Large Wind Turbines ? The effect of wind turbine size on controller design, C. Siddons [et
al.] .........................................................................................................................................................143
CLARIFIFYING THE PERFORMANCE OF CORDINATED CONTROL FOR LARGE WIND
TURBINE LOADS, D. Danzerl [et al.] ...............................................................................................147
Simulation of wind turbines in complex terrain by means of direct CFD, C. Schulz [et al.] ..............151
3D stochastic gusts as an alternative to the Mexican hat wavelet, R. Bos [et al.] ...............................155
Modelling and Evaluation of Wind Speed Time Series for Reliability Analysis of Offshore Wind Farms,
C. Smith [et al.] ....................................................................................................................................159
Investigating the interaction between wind turbines and atmospheric flow with a coupling of the aeroelastic
code FAST and the LES code PALM, M. Bromm [et al.] ....................................................................163
UNSTEADY AND TURBULENT ROTOR LOADS, S. Ehrich [et al.] .............................................167
Influence of the atmospheric boundary layer on wind farm control, L. Vollmer [et al.] .....................171
Analysis of Inflow parameters using LiDARs, A. Giyanani [et al.] ....................................................175
Power System Dynamic Responses - Comparison between simple Simulink model and more complex
time-step based dynamic response modelling, M. Argent....................................................................179
Investigating the Effects of Pitch Control Strategy on the Power Electronics Lifetime of a Vertical Axis
Wind Turbine, R. Dawid.......................................................................................................................183
Vertical Axis Wind Turbine Hydraulic Drivetrain Options, E. Macmahon [et al.] ..............................187
MARE-WINT Project: Improving the Reliability and Optimising O&M Strategies for Offshore Wind
Turbines, R. Mehdi [et al.] ...................................................................................................................192
Wind Prediction Enhancement by Supplementing Measurements with Numerical Weather Prediction
Now-Casts., A. Malvaldi [et al.] ...........................................................................................................196
II
- chapter 1 -
Materials and structures
10th PhD Seminar on Wind Energy in Europe
28-31 October 2014, Orléans, France
Damage tolerant design: failure and crack propagation in composites.

G. Pereira1, L. Mikkelsen1, M. McGugan1
1Technical University of Denmark, Wind Energy, Frederiksborgvej
399, 4000 Roskilde, [email protected]
1. Introduction
The most eye-catching trend for wind energy structural components is the up-scaling where new
turbine designs have consistently provided larger towers, rotor diameters, and power ratings. The wind
energy industry must compete with other energy sources by reducing the cost of energy, and the most
cost effective way of increasing the power produced by a wind turbine is to increase the rotor diameter
[1]. The industry relies on advances in materials technology and design philosophy to deliver the most
cost-effective light-weight structures.

The historical design philosophy for reinforced polymer structures (main material of wind turbine
blades) is based on conservative analysis methods, with large safety factors, underestimating the
material properties, and considering only the linear behaviour of the material. As knowledge about the
material and structure behaviour increased it became possible to safely adopt more advanced design
philosophies, such damage tolerant design, where the material capability is fully exploited. This trend
to more advanced structural design is described by Braga[2].
To achieve this, some research groups are working on a Multi-physics Global Model [3-7] as
represented by the table 1. A Multi-physics Global Model is defined as a fluid-structural interaction
model, which aims to integrate several phenomena models as aerodynamics, hydrodynamics, aero-
elasticity, structural, vibration, energy output, control, etc. Moreover, damage tolerant design requires
a good understanding of the material behaviour, and models capable to simulate the behaviour of the
structure when damaged. However, this approach will not be achieved until all physical phenomenal
present on the wind energy field are fully understood. Wind turbines are a multi-physics problem, and
the complexity of the structure, the unpredictability of the wind and the lack of understanding of
specific phenomena create challenges for the application of damage tolerance design method.

Table 1: Multi-physics Global Model scale main research topics.
Scale 100-1000 Km Weather Forecast; Environment Conditions; Transport and Assembly; Maritime risk assessment;
Maritime route planning;
Scale 1-10 Km Aerodynamic design- Large Eddy Simulation; Turbulence; Wake effect between towers; Maritime
risk assessment: ship collision;
Aerodynamic design- Blade profile shape; Floating structuring; Hydrodynamics; Aero-Hydro-elastic
Scale 10-500 m
coupling: Interaction between wind, waves and the structural; Maintenance planning; Electric

components; Gearbox; etc.
Scale 1-80 m Structural design; Vibrations; Fatigue; Aero-Hydro-elastic coupling; Multibody analysis;
Detail sub-structure design; Vibrations; Fatigues; Non-Linear Materials; Delamination;
Scale 1mm-2m
Bonded/connection joints; Manufacturing Plan;
Scale 1µm-10mm Micro-mechanics; Material resistance; Sensor integration; Fibre-Matrix interface;

“Problem!! We don’t fully understand the input, so how can we rely on the output?”
Is already accepted that a global model that compiles all the theory required and predicts when the
damage will occur and how it will propagate is practically impossible to create. The solution starts by
accepting the presence of damage and unpredictability, but still ensuring the structural health of each
turbine.
“Solution?? If we can’t model the structure, we should monitor the material and understand damage
propagation.”
Detectable changes in response must exist between damaged and undamaged states, thus implying
damage tolerance. Damage tolerance is a property emerging from the particular combination of
structure design, loading environment, and material characteristics. Accepting that a distribution of
damage types and locations can exist across a group of operating wind turbine blades, it follows that
1
1/199
each structure must be characterized individually with a unique "damage map" for that structure.
Evaluating the severity of the particular combination of damage types requires models that describe
the progression parameters for each type under the full range of operating conditions. Only in this way
can condition based maintenance be effectively implemented.
2. Damage tolerant materials and structures
A damage tolerant behaviour is obtained when the stress-strain relationship is initial linear-elastic, but
possesses significant non-linearity before failure. The structure will be designed to be loaded below
the stress-value corresponding to the onset of non-linearity, however if the structure at some point is
loaded beyond the linear-elastic regime, the resulting changes in stiffness will result in a measurable
change in the local compliance of the structure. With respect to the propagation of a crack, damage
tolerance implies that the crack growth should be stable, requiring that the applied load level for
unstable crack growth should be significantly higher than the load level that initiates crack growth. A
way to create damage tolerance is thus to make the crack propagation stable. For instance in the
composite material/adhesive the delamination is accompanied by the formation of a crack bridging
zone, where intact fibres connect the crack faces behind the tip, thus increasing the energy required for
crack propagation (Damage tolerance mechanism) [8].

Figure 1: a) Conventional (Linear) material design philosophy vs Damage tolerant design; b) Damage tolerance mechanism- Fibre bridging.

With damage tolerant design philosophy the designers have the opportunity to create structures that
can operate safely without propagating damage present in that structure, in this way they can fully
exploit the material capability leading to structural optimisation.
3. Failure and crack propagation
In this study, the damage tolerance approach in wind turbine blade sub-structures was addressed,
focusing on the crack growth mechanisms and detection methods. The trailing edge of the blade can
develop damage in the composite material and adhesive interface. The delamination is accompanied
by the formation of a crack bridging zone (Damage tolerance mechanism) [8]. A finite element model
of the crack growth mechanisms in a double cantilever beam (DCB), representative of the trailing
edge, was developed, where different fracture modes were addressed. Experimental tests were
conducted in order to fully characterize this structure and support the model. Then a crack monitoring
technique was implemented using Fibre Bragg Grating (FBG) sensors in order to track the crack and
its propagation. This monitoring approach was incorporated into the finite element model (that was
developed before) in order to predict the sensor output and extrapolate to a real trailing edge case. This
sensor-structure makes possible to study the application of this monitoring technology in different
components/ locations, with the objective of tracking different types of damage, as showed in figure 2.
2
2/199
Figure 2: Modeling scheme of crack growing mechanisms and detection methods.
a. Finite element model and Sensor technology:
A 2D and 3D double cantilever beam (DCB) finite element model was developed in order to represent
the crack growth phenomenon, based on a real trailing edge configuration used by the DTU 10MW
Reference Wind Turbine [9]. It was used cohesive elements that describe the cohesive law that governs
the crack growth mechanism. In table 2 the materials properties used in the DCB model is shown.

Table 2: Scheme and materials properties used in the DCB model.

Composite Material Interface (Cohesive Law) Adhesive
Triaxial Fabric Uniaxial Fabric Elastic Damage Damage Elastic
(Composite) (Composite) (Quadratic stress) Evolution
E =44.3 GPa; E =E =12,9 E =23.8 GPa; E =E =15.05 K=4.2E12 σ = 2.64 MPa ; δ =1.4 ; δ2 E=4.56
1 2 3 1 2 3 n 1
GPa; ν = ν =ν =0.23; GPa; ν = ν =ν =0.513; Pa; σ= 22.15 MPa =0.37 ; GPa;
12 13 23 12 13 23 t
G =G =G =4393 G =G =G =4.393 GPa ν=0.35
12 13 23 12 13 23

Where E is the Young’s modulus, ν is the Poisson’s ration, G the shear modulus, K Penalty stiffness, δ
displacement (opening) at failure, σ and σt the normal and shear traction.
n
b. Sensor technology:
Fibre optic sensors, such Fibre Bragg Gratings (FBG), can be embedded into the composite
materials/adhesive, are virtually non-intrusive to the structure, and have the possibility to measure
several points in a single fibre (multiplexing). This makes FBG’s the perfect type of sensor to track the
growing of certain damage types. A Fibre Bragg Grating (FBG) is formed when a permanent periodic
modulation of the refractive index is induced along a section of an optical fibre, by exposing the
optical fibre to an interference pattern of intense ultra-violet light[10]. The photosensitivity of the
silica exposed to the ultra-violet light is increased, so when the optical fibre is illuminated by a
broadband light source, the grating diffractive properties are such that only a very narrow wavelength
band is reflected back.
3
3/199
When any external phenomenon creates a change on the grating, like temperature, strain, compression,
non-uniform strain fields, etc. this will create a change in the reflected light. However, different
phenomena acting on the grating will make different changes to the sensor response, like a fingerprint,
so it will be possible to track specify phenomena, which are characteristic of damage.
c. Experimental validation:
After the FEM model been successful setup in order to represent the crack growth on the DCB
specimen under the different loading conditions (Mode I/II). A dedicated algorithm predicted the
sensor output, which allowed us to determine the presence of damage and it growth.

Figure 2: FEM model Sensor output for a Mode I loading case.

Then the same material/sensor configuration was tested in order to validate the pair structural-sensor
model. The test was conducted using a double cantilever beam, as described by Sørensen [11] loaded
in order to produce pure Mode I, Mode II and Mixed Mode fracture.
4. Results:
A good agreement between the FEM model and the experiments was found. The FEM model was able
to represent the crack grow under the different loading cases. Also the sensor output model match the
experiments, showing a specific sensor response (“fingerprint” ) when under the effect of a crack.

Figure 3: Experimental results for a Mode I loading case. a) Digital Image Correlation technic, b) FBG reflected
peak, c) FBG sensor response.
4
4/199
5. Conclusion
In this article we present an approach where the use of damage tolerant structural design and damage
tolerant materials combined with an embedded FBG can detect damage evolution. This concept
eventually lead to a condition monitoring maintenance, which consists of the detection of damages by
sensors, characterization of damage (type and size), model predictions of residual life, giving
information that enables decision-making with respect to whether a damage blade should be repaired
or replaced.

The crack growth phenomenon on the trailing edge of the blade was successfully modelled,
representing with good accuracy the fracture mechanisms present. A good agreement between the
sensor output prediction through the FEM model and experiments was found. This demonstrated the
presence of specific fracture features near the crack, which the algorithm and model was cable to
predict and translate into a sensor response change.
These experiments validate the coupled structure/sensor model, so it becomes possible to study the
application of this monitoring technique in other locations, predict the sensor output and track different
damage features. The application of damage tolerant materials and structural monitoring can lead to
safe operation of loaded components even when in damage condition.
6. Acknowledgment
The author acknowledges the Seventh Framework Programme (FP7) for funding the project MareWint
(Project reference: 309395) as Marie-Curie Initial Training Network, Fibersensing® for providing the
FBG sensors and hardware, and SSP-Technology® for providing the material tested.
7. References
[1] Jacobsen TK. 2011, Materials technology for large wind turbine rotor blades - limits and
challenges, Risø Symposium-2011.
[2] Braga D.F.O, Tavares S.M.O., Silva L.F.M., Moreira P.M.G.P., Castro P.M.S.T, Advanced design
for light weight structures: Review and prospects. Progressin Aerospace Sciences (2014),
http://dx.doi.org/10.1016/j.paerosci.2014.03.003i
[3] Qian C., Multi-scale modelling of fatigue of wind turbine rotor blade composites, PhD
dissertation, Delft University of Technology, 2013.
[4] Bauchau O.A., Modeling rotorcraft dynamics with finite element multibody procedures,
Mathematical and Computer Modelling, Volume 33, Issues 10-11, May-June 2001, Pages 1113-1137.
[5] Bottasso C.L., Campagnolo F., Petrovia G.V. , Wind tunnel testing of scaled wind turbine models:
Beyond aerodynamics, Journal of Wind Engineering and Industrial Aerodynamics, Volume 127, April
2014, Pages 11-28
[6] EERA-DTOC Seventh Framework Programme (FP7)- European Energy Research Alliance -
Design Tool for Offshore Wind Farm Cluster, http://www.eera-dtoc.eu/
[7] MAREWINT- Seventh Framework Programme (FP7)- new MAterials and REliability in offshore
WINd Turbines technolog, http://www.marewint.eu/
[8] Bent F. Sørensen, Cohesive laws for assessment of materials failure: theory, experimental
methods and application. Doctor of technices thesis, Risø-R-1736(EN)
[9] The DTU 10MW Reference Wind Turbine Project, http://dtu-10mw-rwt.vindenergi.dtu.dk/
[10] Meltz G., Hill K.O., Fiber Bragg grating technology fundamentals and overview. Journal of
lightwave technology, 15(8):1263-1276, 1997.
[11] Sørensen B. F., Jørgensen K., Jacobsen T.K, Østergaard R.C., DCB-specimen loaded with uneven
bending moments, Int. J. Fract., vol. 141, no. 1–2, pp. 163–176, Sep. 2006.
5
5/199
10th PhD Seminar on Wind Energy in Europe
28-31 October 2014, Orléans, France
CONVEX OPTIMIZATION OF SPACE FRAME SUPPORT
STRUCTURES FOR OFFSHORE WIND TURBINES

K. Sandal, D. Zwick, M. Muskulus

Norwegian University of Science and Technology (NTNU),
Department of Civil and Transport Engineering, 7491 Trondheim,
[email protected], [email protected], [email protected]

ABSTRACT
The aim of the present project is to reduce the cost of support structures for offshore wind turbines by
minimizing their total steel mass. Basic considerations for an iterative optimization approach were
presented by Zwick, Muskulus and Moe (2012), and these have been improved with a convex problem
formulation and faster convergence. Simplified fatigue load assessments have been studied, and
computational expenses of site-specific optimization has been reduced with a factor of 66 compared to
complete analysis. This has been accomplished by using load histories from the initial design to
compute correction factors for each member, which enables a single load case of 10 minutes to
represent 11 load cases of 60 minutes. A jacket has been optimized with this approach, and a
benchmark with the full-height lattice tower concept is presented.

1. INTRODUCTION
Offshore wind turbines are mounted on costly bottom-
fixed support structures such as monopiles and jackets.
It is expected that significant cost reductions can be
achieved by design optimization, particularly for space
frame structures where members can be sized
individually. Due to the large amount of vibrations that
are being induced from both rotor and waves, it has
been observed that fatigue is the dominant failure
mode for such structures. Fatigue assessment requires
comprehensive time-domain simulations, which makes
optimization computationally expensive. In this paper,
considerations for fast and accurate optimization of a
full-height lattice tower (FLT) are presented.

2. METHODOLOGY
Figure 1: Illustration and terminology of
An iterative optimization approach was presented by
space frame support structures
Zwick et al [1], with the objective of reducing weight
while maintaining a fatigue life of 20 years in all parts of the structure. Welded K-, X, and Y-joints are
connecting legs and braces as shown in fig.1, and though the method was developed for a FLT, it can
be easily adapted to any space frame with similar structure. In this paper, a 10 MW turbine with a FLT
support structure has been optimized for fatigue during power production load cases (3-25 m/s, DLC
1.2 in [6]). Fatigue damage is estimated by processing stress history from time-domain simulations
with rainflow counting, SN-curve and Miners rule, which is the recommended practice [3]. An
important assumption of this method is that the sections are uncoupled, meaning that the members in
one section can be optimized without regard to changes in other sections. This is a bold claim, but
numerical results have indicated that it works well enough to give convergence [1], [2] and [4].
Consequently, all sections can be optimized simultaneously, and the problem is split into
(number of sections) problems, each with four variables (thickness and diameter of legs and braces).
𝑁𝑠𝑒𝑐
6/199
3. PROBLEM FORMULATION
Stress concentration factors (SCFs) are used to account for the extra loading experienced in the joints,
and are computed from the DNV standard Fatigue design of offshore steel structures [3]. SCFs are
evaluated at both sides of all welds, as it is normally not the weld that fails, but the nearby material in
one of the connected members. With only thickness and diameter as variables, the formulas from the
standard can be simplified. Note that D and T refer to legs, while d and t refer to braces.

0.9 0.5
𝑡 𝐷
𝑆𝐶𝐹𝐿𝑒𝑔 ∝ 1. 4
𝑇
The scaling laws above are obtained from the dominat𝑑ing stress contribution in the critical hot spot.
𝑆𝐶𝐹𝐵𝑟𝑎𝑐𝑒 ∝
Since the dominating stress contribution will scale wit𝑡h its respective SCF ( ), it can
also be argued that the total stress will scale with this SCF. Stress will of course also scale with the
cross sectional area ( ), which for thin walled pipes scales with thickn𝜎e𝑖ss= tim𝑆𝐶es𝐹 𝑖d∙ia𝜎m0eter.
𝜎0 = 𝐹/𝐴
𝑆𝐶𝐹
𝜎 ∝
𝐴𝑟𝑒0𝑎.9
𝑡
𝜎𝐿𝑒𝑔 ∝ 2.4 0.5
𝑇 𝐷
The SN-curve relation scales stress with maximum num1ber of cycles, and Palmgren Miners rule scales
𝜎𝐵𝑟𝑎𝑐𝑒 ∝ 2
maximum number of cycles (N) with fatigue damage (𝑡U).

log𝑁𝑖 = log𝑎 −𝑚log∆𝜎𝑖
𝑛𝑖
𝑈 = �
𝑖 𝑁𝑖
𝑚
-ratio of both legs and braces should 𝑈Op∝tim𝜎iza (ti𝑚on= pro5b)lem
b𝐷e kept fixed at a minimum value
(𝑇proved in [4]), limited by the validity T, t 2 2
𝑚𝑖𝑛𝑖𝑚𝑖𝑧𝑒: 𝑓 = 𝑇 +𝑟𝑡
range of the SCF formulas (

𝐷 4.5
). The objective function is then a
function only of and 1, 6w<ith𝑇 <a 𝑠𝑢𝑏𝑗𝑒𝑐𝑡 𝑡𝑜 𝑔1 = 𝑈𝐿𝑒𝑔 −1 ≤ 0 �𝑈𝐿𝑒𝑔 ∝ 𝑡14.5�

w64eighting constant which must 𝑇
be included since the𝑡re are m𝑇ore braces 1
𝑔2 = 𝑈𝐵𝑟𝑎𝑐𝑒 −1 ≤ 0 �𝑈𝐵𝑟𝑎𝑐𝑒 ∝ 10�
than legs. The pro𝑟bl=em2 .3(text box & 𝑡
fig.2) is convex, as the design space, the o𝑤 bℎje𝑒c𝑛ti v e 𝑡𝑚𝑖𝑛 ≤ 𝑡 ≤ 𝑇 ≤ 𝑇𝑚𝑎𝑥
function and the two constraint functions are convex

(their Hessians are all semi definite on the intervals

described by the variable bounds). Given that

fatigue damage for both legs and braces have
been calculated fo𝑗r design iteration j, analytical
expressions for 𝑈 and can be derived by setting
.
∆𝑡 ∆𝑇
𝑗+1
𝑈 = 1
1
𝑗 𝑗 10
∆𝑡 = 𝑡 ��𝑈𝐵𝑟𝑎𝑐𝑒� −1�
4.5
1
𝑗 14.5
𝑗 𝑗 14.5 𝑡 +∆𝑡
∆𝑇 = 𝑇 ��𝑈𝐿𝑒𝑔� � 𝑡𝑗 � −1�
Figure 2: Convex optimization problem
7/199