Structural Design and Analysis of Offshore Wind Turbines from a System Point of View


Published on

Offshore wind turbines are relatively complex structural and mechanical systems located in
a highly demanding environment. In this study, the fundamental aspects and major issues
related to the design of such structures are inquired. The system approach is proposed to
carry out the design of the structural parts: in accordance with this philosophy, a
decomposition of the system (environment, structure, actions/loads) and of the structural
performance is carried out, in order to organize the qualitative and quantitative assessment
in various sub-problems. These can be faced by sub-models of different complexity both for
the structural behavior and for the load models. Numerical models are developed to assess
the safety performance under aerodynamic and hydrodynamic actions. In the structural
analyses, three types of turbine support structures have been considered and compared: a
monopile, a tripod and a jacket.

Published in: Design
1 Like
  • Be the first to comment

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Structural Design and Analysis of Offshore Wind Turbines from a System Point of View

  1. 1. Structural Design and Analysis of Offshore Wind Turbines from a System Point of View byFrancesco Petrini, Sauro Manenti, Konstantinos Gkoumas, Franco Bontempi R EPRINTED FROM WIND ENGINEERING VOLUME 34, N O . 1, 2010 M ULTI -S CIENCE P UBLISHING C OMPANY 5 WATES WAY • B RENTWOOD • E SSEX CM15 9TB • UK T EL : +44(0)1277 224632 • FAX : +44(0)1277 223453 E-MAIL: • WEB SITE:
  2. 2. W IND E NGINEERING VOLUME 34, N O . 1, 2010 PP 85–108 85Structural Design and Analysis of Offshore WindTurbines from a System Point of ViewFrancesco Petrini1, Sauro Manenti2, Konstantinos Gkoumas3, Franco Bontempi*,41Department of Structural and Geotechnical Engineering, University of Rome “La Sapienza”, viaEudossiana, 18 - 00184 Rome, Italy ( of Hydraulics Transportation and Roads, University of Rome “La Sapienza”, viaEudossiana, 18 - 00184 Rome, Italy ( of Hydraulics Transportation and Roads, University of Rome “La Sapienza”, viaEudossiana, 18 - 00184 Rome, Italy ( of Structural and Geotechnical Engineering, University of Rome “La Sapienza”, viaEudossiana, 18 - 00184 Rome, Italy ( ABSTRACT Offshore wind turbines are relatively complex structural and mechanical systems located in a highly demanding environment. In this study, the fundamental aspects and major issues related to the design of such structures are inquired. The system approach is proposed to carry out the design of the structural parts: in accordance with this philosophy, a decomposition of the system (environment, structure, actions/loads) and of the structural performance is carried out, in order to organize the qualitative and quantitative assessment in various sub-problems. These can be faced by sub-models of different complexity both for the structural behavior and for the load models. Numerical models are developed to assess the safety performance under aerodynamic and hydrodynamic actions. In the structural analyses, three types of turbine support structures have been considered and compared: a monopile, a tripod and a jacket.1. INTRODUCTIONOffshore wind turbines (OWT) emerge as an evolution of the onshore plants for which theconstruction is a relatively widespread and consolidated practice providing a renewablepower resource [1]. In order to make the wind generated power more competitive withrespect to conventional exhaustible and high environmental impact sources of energy, theattention has turned toward offshore wind power production [2]. Besides being characterized by a reduced visual impact, since they are placed far awayfrom the coast, OWTs can take advantage from more constant and intense wind forcing,something that can increase the amount and regularity of the productive capacity and makesuch a resource more cost-effective if the plant is lifelong and operates with minimuminterruption through its lifespan. From a general point of view, an OWT is formed by both mechanical and structuralelements. Therefore, it is not a “common” civil engineering structure; it behaves differentlyaccording to different circumstances related to the specific functional activity (idle, power*Corresponding author
  3. 3. 86 S TRUCTURAL D ESIGN AND A NALYSIS OF O FFSHORE W IND T URBINES FROM A S YSTEM P OINT OF V IEW Blades – Rotor – Nacelle Tower Transition platform MONOPILE TRIPOD JACKET Water level Sub- Support structure structure Sea floor Foundation Seabed Foundation Figure 1: Main parts of an offshore wind turbine for different support structures.production, etc), and it is subject to highly variable loads (wind, wave, sea currents loads, etc.).In the design process, different structural schemes for the supporting structure can beadopted (Figure 1), mainly depending on the water depth, which determines thehydrodynamic loads acting on the structure and drives the choice of the proper techniques forthe installation and maintenance of the support structure. Moreover, since the structural behavior of OWTs is influenced from nonlinearities,uncertainties and interactions, they can be defined as complex structural systems [3]. The above considerations highlight that a modern approach to study such structures hasto evolve from the idea of “structure” itself, intended as a simple device for channeling loads,to the one of “structural system”, intended as “a set of interrelated components which interactone with another in an organized fashion toward a common purpose” [4]. This systemapproach includes a set of activities which lead and control the overall design,implementation and integration of the complex set of interacting components [5,6]. In this study, the original definition by NASA [4] has been extended in such a way that the“structural system” organization contains also the actions and loads. The latter derive from,and are strictly related to, the environment (Figure 2). A certain amount of complexity arises from the lack of knowledge and from the modelingof the environment in which the turbine is located. In this context two main design issues canbe individuated: the consideration of the uncertainty deriving from the stochastic nature ofthe environmental forcings (in particular aerodynamic and hydrodynamic) and the proper
  4. 4. W IND E NGINEERING VOLUME 34, N O . 1, 2010 87 PERFORMANCE Structural system Interaction ENVIRONMENT STRUCTURE ACTIONS Figure 2: Structural system organization. Wind and wave flow ENVIRONMENT ZONE EXCHANGE ZONE Aerodynamic and Wind site basic aeroelastic parameters phenomena Structure Wind, wave and Structural (non- Wave site basic Site-specific sea current environmental) parameters environment actions system Non- Other environmental environmental solicitations agents Hydrodynamic phenomena Types of uncertainties Propagation Propagation 1. Aleatoric 1. Aleatoric 1. Aleatoric 2. Epistemic 2. Epistemic 2. Epistemic 3. Model 3. Model 3. Model Figure 3: Generic depiction of the uncertainties and the interaction mechanisms in the design of an offshore wind turbine structure.modeling of the possible presence of non linear interaction phenomena between the differentactions and between the actions and the structure. In general, uncertainties can spread during the various analysis phases that are developedin a cascade. The incorrect modeling of the involved uncertainty can lead to an incorrectcharacterization of the structural response from a stochastic point of view and, thus, to animproper quantification of the risk for a given structure subjected to a specific hazard. Having as a goal the schematization of the problem and the individuation of theuncertainty propagation mechanisms, reference can be made to the Figure 3, where theprocess of the environmental actions generation is qualitatively represented also withconsiderations on the involved uncertainties. Following the wind and the hydrodynamic flows impacting on the structure, it is possibleto distinguish two zones: • Environment zone: it is the physical region sufficiently close to the structure to assume the same environmental site parameters of the structure, yet far enough to neglect the flow field perturbations (in terms of particle’s trajectories, pressure field, etc.) induced by the presence of the structure itself. In the environment zone, the wind and the hydrodynamic flows can interact with each other and with other environmental agents, changing their basic parameters. The physical phenomena and uncertainties in the environment zone propagate in the neighborhood regions.
  5. 5. 88 S TRUCTURAL D ESIGN AND A NALYSIS OF O FFSHORE W IND T URBINES FROM A S YSTEM P OINT OF V IEW • Exchange zone: it is the physical region adjoining the structure. In this zone, the structure itself, the wind and the hydrodynamic field experience the mechanical interchange (aerodynamic and hydrodynamic phenomena) from which the actions arise. In the exchange zone, some non-environmental solicitations are present; these solicitations may change the dynamic or aerodynamic characteristics of the original structure; so the actions are generated considering this structural sub-system (original structure combined with non-environmental solicitations) instead of regarding only the original structure itself. By definition, physical phenomena and uncertainties cannot propagate from the exchange zone to the environment zone. In general, the uncertainties can be subdivided in three basic typologies: • aleatory uncertainties (arising from the unpredictable nature of the magnitude, the direction and the variance of the environmental actions); • epistemic uncertainties (deriving from the insufficient information and the errors in measuring the previously mentioned parameters); • model uncertainties (deriving from the approximations in the models). Regarding for example the wind model and considering the turbulent wind velocity fieldas a Gaussian stochastic process, an uncertainty related to the hypothesis of Gaussianity isintroduced. The aleatory uncertainties can be treated by carrying out a semi-probabilistic (looking forthe extreme response) or a probabilistic analysis (looking for the response probabilisticdistribution) analysis. A possible way to reduce the model uncertainties is given by differentiating the modelinglevels. This can be carried out not only for the structural models, but also for the action andinteraction phenomena models; for this reason different model levels are adopted (for thesake of simplicity, the epistemic uncertainties are not considered in this study). A suitable tool to govern the complexity is given from the structural systemdecomposition, represented by the design activities related with the classification and theidentification of the structural system components, and by the hierarchies (and theinteractions) between these components. As mentioned before, the decomposition regards not only the structure, but also theenvironment and the actions and loads, and it is the subject of the first part of this study. Furthermore, due to the complexity, the design of these structures has to be carried outunder a Performance-Based Design philosophy: different aspects and performance underdifferent loading conditions (with reference to all possible system configurations that can beassumed by the blades and the rotor) have to be investigated for this type of structures.Additional design issues related to the structural aspects are mentioned below with someproper references: • Aerodynamic optimization [7]. • Foundation design and soil-structure interaction [8, 9, 10]. • Fatigue calculations [11, 12]. • Vessel impact and robustness [13]. • Life Cycle assessment [14, 15]. • Marine scour [16, 17]. • Possible floating supports [18, 19]. • Standards certification [20, 21, 22, 23, 24, 25].
  6. 6. W IND E NGINEERING VOLUME 34, N O . 1, 2010 89 Figure 4: Different views of the jacket support structure adopted after the numerical analyses. Finally, an important issue for offshore wind turbines is the choice of the support structure,related principally to the water depth, the soil characteristics and economic issues. If thewater depth (h) is considered as the principal parameter, according to the DNV-OS-J101 [22],the following rough classification can be made: monopile, gravity and suction buckets(h < 25m); tripod, jacket and lattice tower (20m < h < 40–50m); low-roll floaters and tension legplatform (h > 50m). In this study focus is given to monopile, tripod and jacket supportstructures. The paper starts with the description of the system approach applied to OWT design: whilethe system point of view is a consolidated practice in many engineering fields (e.g. aerospaceengineering), in the case of OWTs, it is not fully established and represents an ongoing process.In the second part of the paper, the system point of view is applied to the numerical modelingof a case study. More precisely, numerical analyses are carried out on different OWT supportstructures. The obtained results justify the adoption of a jacket structure for the specific case(Figure 4).2. STRUCTURAL SYSTEM DECOMPOSITIONAs previously stated the decomposition of the structural system is a fundamental tool forthe design of complex structural systems, and it has to be performed together with thedecomposition of the performance the structure has to fulfill (Figure 5). Thedecomposition is carried out focusing the attention on different levels of detail: startingfrom a macro-level vision and moving on towards the micro-level details (for more detailssee Bontempi et al. [26]).2.1. Decomposition of the EnvironmentThe first step of the structural system decomposition concerns the environment. This is dueto the fact that, in a global approach, the structure is considered as a real physical objectplaced on an environment where a variety of conditions, strictly related to the acting loads,should be taken into consideration. Their decomposition is performed in the first column ofFigure 5.
  7. 7. 90 S TRUCTURAL D ESIGN AND A NALYSIS OF O FFSHORE W IND T URBINES FROM A S YSTEM P OINT OF V IEW ENVIRONMENT STRUCTURE ACTIONS/LOADS Wind conditions Main structure Gravitational / Inertial Normal wind conditions Rotor–nacelle assembly Gravity Extreme wind conditions Nacelle Braking Marine conditions Junctions/bearings Aviation Waves Rotor Seismic activity Normal wave conditions Junctions/bearings Aerodynamic Extreme wave conditions Blades Hydrodynamic Sea currents Junctions/bearings Wave Water level Support structure Current Marine growth Tower Actuation Seabed movement and scour Junctions Torque control Other conditions Substructure Yaw and pitch actuator loads Air temperature Junctions Mechanical breaking loads Humidity Foundations Other Solar radiation Junctions Wake loads Rain, hail, snow, ice Secondary structure Impact loads Chemically active substances Energy production Tsunami Mechanically active substances Energy transfer Environmental aggressiveness Auxiliary structure Seismicity Operation Water density Maintenance Water temperature Emergency Lighting Maritime traffic PERFORMANCE Serviceability Safety Reliability Robustness Service Limit States –SLS Ultimate Limit States –ULS_1 Ultimate Limit States –ULS_2 Accidental Limit States –ALS Deflections/Displacements Stress limit Degradation effects Impact Vibrations Strain limit Fatigue Limit States –FLS Explosion Buckling Fire Figure 5: Structural system and performance decomposition of an offshore wind turbine.2.2. Decomposition of the StructureThe second step of the decomposition relates to the offshore wind turbine structure. This isorganized hierarchically, considering all the structural parts categorized in three levels(second column of Figure 5): • Macroscopic (macro-level), related to geometric dimensions comparable with the whole construction or parts with a principal role in the structural behavior; the parts so considered are called macro components which can be divided into: – the main structure, that has the objective to carry the main loads; – the secondary structure, connected with the structural part directly loaded by the energy production system; – the auxiliary structure, related to specific operations that the turbine may normally or exceptionally face during its design life: serviceability, maintainability and emergency.
  8. 8. W IND E NGINEERING VOLUME 34, N O . 1, 2010 91 Focusing the attention on the main structure, it consists in all the elements that form theoffshore wind turbine. In general, the following segments can be identified: a. support structure (the main subject of this study); b. rotor-nacelle assembly. • Mesoscopic (meso-level), related to geometric dimensions still relevant if compared to the whole construction but connected with specialized role in the macro components; the parts so considered are called meso-components. In particular the support structure can be decomposed in the following parts: a. foundation: the part which transfers the loads acting on the structure into the seabed; b. substructure: the part which extends upwards from the seabed and connects the foundation to the tower; c. tower: the part which connects the substructure to the rotor-nacelle assembly. • Microscopic (micro-level), related to smaller geometric dimensions and specialized structural role: these are simply components or elements.2.3. Decomposition of the Actions and LoadsThe next step of the structural system decomposition is the one regarding the actions relatedto the environmental conditions. These are decomposed as shown in the third column ofFigure 5, from which the amount of the acting loads can be comprehended. It is important to underline that, since the environmental conditions in general are ofstochastic nature, the magnitude of the actions involved is usually characterized, from astatistical point of view, by a return period TR: lower values of TR are associated with the socalled “normal conditions”, while higher values of TR are associated with “extreme conditions”.2.4. Performance DecompositionAs a final step, the performance requirements are identified and decomposed as follows(lower part of Figure 5). • Assurance of the serviceability (operability) of the turbine, as well as of the structure in general. As a consequence, the structural characteristics (stiffness, inertia, etc.) have to be equally distributed and balanced along the structure; • Safety assurance with respect to collapse, in plausible extreme conditions; this is applicable also to the transient phases in which the structure or parts of it may reside (e.g. transportation and assembly), and that have to be verified as well; • Assurance of an elevated level of reliability for the entire life-span of the turbine. As a consequence, a check of the degradation due to fatigue and corrosion phenomena is required; • Assurance of sufficient robustness for the structural system, that is to ensure the proportionality between an eventual damage and the resistance capacity, independently from the triggering cause, ensuring at the same time the survival of the structure under a hypothetical extreme condition. The following performance criteria can be identified for the structural system, leadingeventually to the selection of appropriate Limit States: • Dynamic characterization of the turbine as defined by the functionality requirements, regarding the: – natural vibration frequencies of the whole turbine, including the rotor-nacelle assembly, the support structure and the foundations;
  9. 9. 92 S TRUCTURAL D ESIGN AND A NALYSIS OF O FFSHORE W IND T URBINES FROM A S YSTEM P OINT OF V IEW – compatibility of the intrinsic vibration characteristics of the structural system with those of the applied forces and loads; – compatibility of the displacement and the acceleration of the support system with the functionality requirements of the turbine. • Structural behavior with respect to serviceability (SLS- Serviceability Limit State), regarding the: – limitation of deformations; – prevention of any loosening of the connections. • Preservation in time of the structural integrity, regarding the: – durability for corrosion; – structural behavior with respect to fatigue (FLS-Fatigue Limit State). • Structural behavior under near collapse conditions (ULS-Ultimate Limit State), regarding the: – assessment of the stress state for the whole structure, its parts, elements and connections; – assessment of the global resistance of the structural system; – assessment of the resistance for global and local instability phenomena. • Structural behavior in presence of accidental scenarios (ALS-Accidental Limit State), regarding the: – provisions for the decrease in the load bearing capacity proportional to the damage (the concept of structural robustness- see for example Starossek [27] and Bontempi et al. [28]); – provisions for the survival of the structural system in presence of extreme and/or unforeseen, situations; these include the possibility of a ship impacting the structural system (support system or blades), with consequences accounted for specific risk scenarios.3. ENVIRONMENT AND ACTION MODELFrom all the loads indicated in Paragraph 2.3, in this study attention is focused on theaerodynamic and the hydrodynamic ones. Typically, an environmental action, when observed during a short time period, is made oftwo components: a mean (or slowly variable) component, and a stochastic one. For theaerodynamic and the hydrodynamic actions, the mean component is generated respectivelyby the mean wind velocity and by the sea current, while the stochastic component isgenerated respectively from the turbulence wind velocity and from the linear waves. The definition of “mean” has to be specified with reference to a specified “short timeperiod” (usually less than 1 hour); in contrast, the so called “mean component” varies in astochastic manner during long time periods. For this reason, in what follows the meancomponents will be considered as constant only for short periods of analyses. The generic environmental configuration is shown in Figure 6, where the macro-geometric parameters defining the problem are also represented. These are the local meanwater depth (h), the hub height above the mean water level (H) and the blade length (or rotorradius, R). Correct prediction of the structural response under extreme and normal loadconditions requires the definition of their probability distribution and statistical
  10. 10. W IND E NGINEERING VOLUME 34, N O . 1, 2010 93 WP (t) P uP (t) VP (t) Turbulent wind Vm (zP) P W ate r le ve Mean l (m R ea wind n) level Hub Waves Mu z dl H ine y Vw (z′) y′ x, x′ Current z′ n) Vcur (z′) vel (mea Wa ter le h e d lin Mu Figure 6: Problem statement and configuration of the actions.parameters; these are site specific, and have to be estimated by carrying out statisticalanalyses of the measurements database. In particular two kinds of investigations areusually carried out: short term statistics for fatigue analysis, and long term statistics, forultimate limit state analysis. Finally, the definition of the extreme load cases requires an estimation of the probabilitydistribution for: (i) the extreme 10-min average wind velocity at the reference height, and (ii)the significant wave height estimated in a 3-hour reference period along with the associatedspectral peak period. When no information is available for defining the long term joint probability distribution ofextreme wind and waves, it shall be assumed that the extreme 10-min mean wind speed witha 50-year recurrence period occurs during the extreme sea state with a 50-year recurrenceperiod (IEC 64100-3 [25]) adopting appropriate reduced values.3.1. Aeolian and Hydrodynamic Fields ModelConcerning the wind modeling for the computation of the aerodynamic action, a Cartesianthree-dimensional coordinate system (x,y,z), with origin at the mean water level and the z-axis oriented upward is adopted. Focusing on a short time period analysis, the threecomponents of the wind velocity field Vx ( j ), Vy ( j ), Vz ( j ) at each spatial point j (the variationwith time is omitted for the sake of simplicity) can be expressed as the sum of a mean (time-invariant) value and a turbulent component u(j ), v(j), w(j) with mean value equal to zero.
  11. 11. 94 S TRUCTURAL D ESIGN AND A NALYSIS OF O FFSHORE W IND T URBINES FROM A S YSTEM P OINT OF V IEWAssuming that the mean value of the velocity is non-zero only in x direction, the threecomponents are given by: Vx (j) = Vm(j) + u(j ); Vy ( j ) = v(j ); Vz ( j ) = w(j) (1)The mean velocity Vm(j) can be determined by a database of values recorded at or near thesite, and evaluated as the record average over a proper time interval (e.g. 10 minutes), whilethe variation of the mean velocity Vm with the height z over a horizontal surface ofhomogeneous roughness can be described by an exponential law. Finally, the turbulentcomponents of the wind velocity are modeled as zero-mean Gaussian ergodic independentprocesses. By using the proposed model, it is possible to generate samples of the wind actionexerted at each point j of the structure. Concerning the hydrodynamic actions, as previously stated, they are due to currents andwaves. For what concerns the sea currents induced by the tidal wave propagation in shallowwater condition (i.e. the ratio between water depth h and wave length L is lower than 0.05), ingeneral they are characterized by a sub-horizontal velocity field, while their intensitydecreases slowly with the depth. Waves act on the submerged structural elements and on thetransition zone above the water-air interface; in the first case actions are the result of thealternative motion of fluid particles, induced by the fluctuating perturbation of the liquidsurface; in the second case the action is the consequence of the breaking waves, which mayoccur in shallow water condition. In general the short period water surface fluctuation, withrespect to the mean sea level, is a time-dependent stochastic variable, and can be described bymeans of statistical parameters: • the significant wave height HS ; it is defined as four times the mean square root of the sea elevation process. It represents a statistical measure of the intensity of the wave climate as well as of the variability in the arbitrary wave heights. • the spectral peak period TP ; it is related to the mean zero-crossing period of the sea elevation process. For extreme events analysis, in general the significant wave height is defined with respectto a proper return period TR (DNV-OS-J101 [22]). For fatigue analysis the sea state is characterized through the wave energy spectraldensity, defined upon the domain of frequency and geographic direction of the wavecomponents: usually this is obtained by multiplying the calculated one-dimensional spectrumS(f) by a function of directional spreading, symmetric with respect to the principal directionof the wave propagation [29].3.2. Aerodynamic and Hydrodynamic Actions ModelIn general, the components of the actions could be calculated separately for all structuralelements adopting a common frame of reference and then superimposed by a vector sum in aphase-correct manner. The aerodynamic force can be decomposed, as usual, in a drag (parallel to the mean windvelocity) and a lift (orthogonal to the mean wind velocity) component, while moments havebeen neglected in the present paper. These can be computed for each structural componentfrom the specific wind velocity field and for each structural configurations (for example,extreme wind and parked turbine configurations), by using well known expressions, as shownin Bontempi et al. [30] and Petrini et al. [31]. The equivalent static load can be derived throughpeak factors, based on the probabilistic characteristics of the wind velocity modeled as astochastic process [32].
  12. 12. W IND E NGINEERING VOLUME 34, N O . 1, 2010 95 Concerning the hydrodynamic loads on a structural slender cylindrical member (D/L < 0.2,with: D member diameter normal to the fluid flow, L wave length), both wave and (stationary)current generate the following two components: • A force per unit length acting in the direction perpendicular to the axis of the member and parallel to the orthogonal (with respect to the member) components of the water particle velocity (wave vw plus current Vcur induced) and acceleration (wave only); it can be estimated by means of Morison equation:  ρ πD 2 · 1  dF ( z ,t ) = ci wat vw ( z ,t ) + cd ρwat D vw ( z ,t ) +Vcur ( z ,t )⋅ vw ( z ,t ) +Vcur ( z ,t ) dz   (2)   4 2   where ρwat is the water density, ci and cd are the inertia (including added mass for a moving member) and drag coefficient respectively, which are related to structural geometry, flow pattern and surface roughness: superposed dot indicates the time derivate, in the Eq. (2). Periodic functions are adopted for both the wave velocities and accelerations [33]. • A non-stationary (lift) force per unit length acts in the direction perpendicular both to the axis of the slender member and to the water current. This component is induced by vortex shedding past the cylinder and inverts its direction at the frequency fl of eddies separation which is related to flow field and structural geometry through Strouhal number St = Dfl /Vcur ; fl should be kept far from the structure’s natural frequency to avoid resonances.In the case of static analysis, equivalent static forces are applied considering the amplitude ofthe fluctuating actions and, eventually, applying proper load amplification factors.4. NUMERICAL MODELING OF THE STRUCTUREAs stated in Section 1, a differentiation of the modeling level is adopted to reduce theuncertainties. The level of a generic model of the structure is here identified by means of twoparameters: the maximum degree of detail and the scale of the model; if the finite elementmethod is adopted, at each model level it is possible to associate a certain typology of finiteelement, which is mainly used to build the model. In general, four model levels are defined for the structure: 1. System level (S): the model scale comprises the whole wind farm and can be adopted for evaluating the robustness of the overall plant; highly idealized model components are used in block diagram simulators. 2. Macro level or Global modeling (G): in these models, the scale reduces to the single turbine structure, neglecting the connections between different structural parts. The component shapes are modeled in an approximate way, the geometric ratios between the components are correctly reproduced; beam finite elements are mainly adopted; 3. Meso-level or Extended modeling (E): these models are characterized by the same scale of the previous level but with a higher degree of detail: the actual shape of the structural components is accounted for and the influence of geometrical parameters on the local structural behavior is evaluated. Shell elements are adopted for investigating the internal state of stress and strain (e.g. for fatigue life and buckling analysis) inside the structure extrapolated from previous models;
  13. 13. 96 S TRUCTURAL D ESIGN AND A NALYSIS OF O FFSHORE W IND T URBINES FROM A S YSTEM P OINT OF V IEW 4. Micro level or Detail modeling (D): this kind of models are characterized by the highest degree of detail and used for simulating the structural behavior of specific individual components, including connecting parts, for which a complex internal state of stress has been previously pointed out e.g. due to the presence of concentrated loads. Shell or even solid finite elements are used. The features for different structural model levels are resumed in Table 1; a similardistinction can be made regarding the specification of the external loads. According to what said above, at the initial stage of investigation structural analyses havebeen carried out with macro-level and meso-level models of the three offshore wind turbinestructure types previously described. With reference to Figure 7 some of the developed macro-level structural models are shownfor the monopile (left part), tripod (middle) and jacket (right part) support structure. Table 1: Definition of the model levels Model Level Scale Maximum Detail Level Main Adopted Finite Elements System level wind farm approximate shape of the BLOCK elements structural components Macro-level single turbine approximate shape of the BEAM elements structural components, correct geometry Meso-level single turbine detailed shape of the SHELL, SOLID elements structural components Micro level individual detailed shape of the SHELL, SOLID elements components connecting parts (a) (b) (c) Emergent Transition Z X Y Immersed Foundation Figure 7: Macro-level finite element models: monopile (a), tripod (b) and jacket (c).
  14. 14. W IND E NGINEERING VOLUME 34, N O . 1, 2010 97 The effect of foundation medium should be simulated with a full non-linear model in orderto account for possible plastic effects and load time-history induced variation of themechanical properties. At this level of investigation, an idealized soil has been simulated bymeans of both: • Linear springs: such technique has been adopted for the macro-level models. Springs are applied at the pile surface and acts in the two coordinate horizontal directions: the corresponding mechanical parameters have been set up on the basis of available soil characteristic and simulates its lateral resistance at the pile interface; • Solid elements: used for meso-level models. These three-dimensional elements simulate the linear mechanical behavior of the soil. The extension of the foundation medium included in the model has been selected in order to minimize boundary effects. Both kinds of models have been used for evaluating the modal response of the structuralsystem. The decomposition of both the structural system and the performance, and thedifferentiation of the model levels can be used to guide and optimize the numerical analysisefforts in this design phase. In this sense, focusing on a certain structural component andselecting the specific performance, the choice of both model level and type of analysis toadopt can be done, in such a way, to give the best efficiency of the analysis (deriving from asuitable balance between the required detail level of outputs and the computational effortsneeded). For example, focusing the attention on the tower with a steel tubular section, themaximum stresses for Ultimate Limit State analysis can be preliminary obtained by adoptinga macro-level model and by carrying out a static extreme analysis (characterized by extremevalues of the environmental loads). However, if the local buckling phenomena need to beassessed, a more detailed meso-level structural model and a static incremental analysis isrequired. These considerations are summarized in Table 2. Table 2: Model and analysis type selection Structural Component Performance Model Level Analysis Type Macro Stress safety (ULS) Static extreme Meso Macro Global Buckling (ULS) Static incremental Meso Tower Meso Local Buckling (ULS) Static incremental Micro Macro (poor) Fatigue (FLS) Meso Dynamic Micro
  15. 15. 98 S TRUCTURAL D ESIGN AND A NALYSIS OF O FFSHORE W IND T URBINES FROM A S YSTEM P OINT OF V IEW5. NUMERICAL ANALYSESThe numerical analyses have been conducted for three different support structures:monopile, tripod and jacket. The principal geometrical and structural features adopted for theanalyses are as follows: • hub height positioned 100 m above the mean sea level (m.s.l.); • tower with a steel tubular section, with a diameter of 5 m and a thickness of 0.05 m; • water depth of 35 m; • foundations depth of 40 m; • foundation diameter of 6 m (monopile), 2.5 m (tripod and jacket). For the tripod substructure, the tubular tripod arm diameter and thickness is respectivelyof 2.5 m and 0.05 m. For the jacket substructure, the diameter (thickness) of the vertical,horizontal and diagonal tubular members, is respectively of 1.3 m (0.026 m), 0.6 m (0.016 m)and 0.5 m (0.016 m). Finally, the tower supports a Vestas-V90 turbine [34] with a rotor diameterof 90 m.5.1. Modal AnalysisThe preliminary task of the dynamic analysis is to assess the natural modes of vibration inorder to avoid that non-stationary load (e.g. wind and wave induced) could cause the systemresonance when excitation and natural frequencies are closer. Geometrical parameters of the three support structures have thus been selected with theaim of maintaining the corresponding natural frequency far from that of the non-stationaryexternal forcing (wind and wave). The finite element modal analysis provided the deformed shapes given in Figure 8, whereonly odd modes are displayed since modes i and i +1 (with i = 1, 3) have the same frequency butvibration occurs in orthogonal planes, according to the axial symmetry of the tower (theeccentric mass of the blades is neglected). In Figure 9, the two x-parallel dashed lines correspond, respectively, to the mean rotorfrequency (1P) and the frequency of a single blade passing (3P), which is triple with respect tothe former one for a three bladed turbine. These frequencies determine the sampling period of the wind turbulent eddies and, as aconsequence, the characteristics of the induced non-stationary actions. Therefore, they(a) (b) (c) M0 M0 1st M0 3rd M0 1st M0 3rd 1st M0 3rd Z Z X X Y Y Figure 8: Modal analysis (macro-level models). Natural mode shapes for the monopile (a), tripod (b) and jacket (c) support structures.
  16. 16. W IND E NGINEERING VOLUME 34, N O . 1, 2010 99 2.5 Freq. [Hz] Monopile Tripod Jacket 2.0 1.5 1.0 3P 0.5 1P 0.0 1 3 5 Mode number Figure 9: Comparison of the natural frequencies.assume importance when performing dynamic analysis and are generally compared withrespect to the first natural frequency fnat in order to classify the structural behavior: • if fnat falls below 1P the structure is called “soft-soft”; for this type of structure the wave load could be dominant with respect to the wind load, and the fatigue effects can be significant; • if fnat is between 1P and 3P the structure is called “soft-stiff”; for this type of structure the wind action frequency could be considerable higher than the one due to waves, and the fatigue effects can be still significant; • if fnat is greater than 3P the structure is called “stiff-stiff”; for this type of structure the fatigue effects in general are not significant. From the results plotted in Figure 9 it can be seen that the structural system falls in the soft-stiff range only if the jacket support type is adopted. In the same figure, it can be noted that forthe first couple of modes the dynamic behavior of the jacket is stiffer than the one of the othertypes, but the trend inverts from the third mode on.5.2. Static Analysis Under Extreme LoadsSteady loads have been assumed for the principal environmental actions and no functionalloads are present (parked condition). The external forcing has been characterized byassuming prudentially a return period larger than the one prescribed by Codes and Standards. The numerical analysis for the selected support structure types has been carried outconsidering the three load cases summarized in Table 3, where: • Vhub represents the wind velocity at the hub height; • VeN (with N = 1 or 100) represents the maximum wind velocity with a return period TR equal to N years, derived from the reference wind velocity associated with the same return period VrefN multiplied by a certain peak factor; • VredN represents the reduced wind velocity with a return period TR equal to N years and it is derived from the previous one by applying a reduction factor;
  17. 17. 100 S TRUCTURAL D ESIGN AND A NALYSIS OF O FFSHORE W IND T URBINES FROM A S YSTEM P OINT OF V IEW Table 3: Load cases Design Combination Wind Marine Load Factors γF Situation Name Condition Condition Environmental Gravitational Inertial 6.1b Vhub=Ve100 H=Hred100 1.35 1.1 1.25 Parked (standstill 6.1c Vhub=Vred100 H=Hmax100 1.35 1.1 1.25 or idling) 6.3b Vhub=Ve1 H=Hred100 1.35 1.1 1.25 In the same table HmaxN and HredN represent respectively the design maximum wave height and the design reduced wave height associated whit a return period TR equal to N years. Steady wind field has been assumed along with stationary and regular wave actions; both actions have been assumed to act in the same direction. The design wind exerts a force distribution that is dependent on the undisturbed flow pattern: the resultant action on the rotor blades has been concentrated at the hub height while the drag forces acting on the support are distributed along the tower and the exposed piece of the substructure (jacket type only). The immersed part of the support structure is subject to combined drag and inertia forces induced by the undisturbed wave and the current induced flow field. In Figure 10, the calculated vertical profiles of the aerodynamic and hydrodynamic actions induced per unit length on the tower and the substructure respectively are shown for the monopile case. The analyses carried out through macro-level models allowed for evaluation of both the reactions at the mud line (shear and overturning moment) and the induced displacement at the hub height. Results obtained with macro-level models are summarized in Figure 11. The maximum shear stress at the mud line is reached for the load case 6.1c, i.e. the one characterized by maximum wave height and reduced wind speed (see Table 3); on the other hand, the combination giving the maximum bending moment at the mud line corresponds to extreme wind and reduced wave height (combination 6.1b). From what above follows that wave and current exert much more influence on the resultant shear force, while the wind appears to be more critical for the overturning moment, being distributed at a higher distance from the base. Aerodynamic Hydrodynamic Hydrodynamic120 40 40 Height 35 35100 above sea 30 30 level [m] 80 Height Height 25 above 25 above mud line [m] mud line [m] 60 20 20 15 15 40 10 10 20 5 5 Action [N/m] Action [N/m] Action [N/m] 0 0 0 0 5000 10000 0 100000 200000 0 100000 200000 Comb 6.1b Comb 6.1c Comb 6.1b Comb 6.1c Comb 6.1b Comb 6.1c Figure 10: Environmental actions (monopile type support).
  18. 18. W IND E NGINEERING VOLUME 34, N O . 1, 2010 101 400000 7000 350000 6000 300000 5000 250000 4000 200000 3000 150000 100000 2000 50000 1000 0 0 [KN*m] Monopile Tripod Jacket [KN] Monopile Tripod Jacket 6.1b 6.1c 6.3b 6.1b 6.1c 6.3b 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 [m] Monopile Tripod Jacket 6.1b 6.1c 6.3b Figure 11: Overturning moment, total shear reaction at the mud line and hub displacements, for three different load cases. Moreover, from the same figure it can be seen that the three structural types experienceapproximately the same resultant shear and moment under each load combination.Concerning the horizontal displacement at hub height, it can be seen an increasing stiffness ofthe support structure moving from the monopile to the jacket type under each loadcombination. Maximum displacement occurs always for load case 6.1b giving rise to the higheroverturning moment; for the jacket type it is almost one-third the one of the monopile. In Table 4 the applied loads and the numerical results obtained for the more severecombination (6.1b) are reported, where the maximum stress in the tower has been computedby the combination of compression (or tension) and bending stresses. Table 4: Applied loads and the numerical results (loads combination 6.1b) Monopile Tripod Jacket Actions Wind on rotor [KN] 1663 1663 1663 Wind on tower [KN] 740 740 428 Wave and current [KN] 3372 3372 3500 Overturning moment [KN*m] 350456 350456 337087 Reactions at Shear reaction at mud 5775 5775 5591 mud line line [KN] Vertical reaction at mud 10714 10356.3 13768 line [KN] (max in pile (max in pile =15018) =9929) Structural Maximum stress in the 286 230 151 checks tower [N/mm2] Nacelle displacement [m] 4.66 3.72 1.82
  19. 19. 102 S TRUCTURAL D ESIGN AND A NALYSIS OF O FFSHORE W IND T URBINES FROM A S YSTEM P OINT OF V IEW Figure 12: Detailed jacket support structure meso-level model. From the previous results, it can be deduced that the jacket support type is the best choicefor what concerns the structural response under extreme loads (above all for the maximumstress in the tower and for the nacelle displacement). A meso-level model has been prepared for this type of support, after the exploration of anumber of tentative models (the model is shown in detail in Figure 12). The meso-level model of the OWT structure is shown in Figure 13 (left part), while the rightpart of Figure 13 shows the foundation medium (five substrates with different mechanicalcharacteristics), modeled using brick finite elements. This level of detail allows the designer to investigate the internal state of stress for criticalparts (Figure 14). The connection between the tower (shell elements) and the jacket ismodeled using rigid beams elements (middle part of Figure 14). The meso-model is subjectedto the load case referred to as 6.1b in Table 3 (most severe); the result gives a nacelledisplacement equal to 2 m and a maximum stress in the tower equal to 178 MPa (at the jacket-tower connection). This is in good agreement with the result of the macro-level. The smalldifferences are probably related to the variation in the tower diameter (ranging from5.0 meters at the tower base to 3.4 meters at the top) along the vertical direction and to the
  20. 20. W IND E NGINEERING VOLUME 34, N O . 1, 2010 103 Figure 13: Meso-level structural model of the jacket and corresponding deformed shape under static aerodynamic and hydrodynamic loads. Figure 14: Elastic internal state of stress at critical zones, jacket-tower connection and tower thickness transition.varying thickness of the tubular member at a fixed transition section (right part of Figure 14).These features are properly reproduced in the meso-level model, while in the macro-levelmodel they are set equal to their maximum values.5.3. Buckling AnalysisAnother important aspect concerns the stability problem. A static incremental analysis hasbeen conducted in order to assess the buckling load; in this case, the hydrodynamic actions
  21. 21. 104 S TRUCTURAL D ESIGN AND A NALYSIS OF O FFSHORE W IND T URBINES FROM A S YSTEM P OINT OF V IEW Figure 15: Results of the buckling analysis.have been schematized by using of single force acting on the jacket at the mean water level(Figure 15). The analysis gives a multiple of 1.17 for the extreme load case referred to as 6.1b in Table 3.It is important to outline that the first buckling mode shows a local instability of the towertubular section, an effect that cannot be accounted for with the macro-models.6. CONCLUSIONSIn this paper, the system approach has been proposed as a conceptual method for the designof offshore wind turbine structures. In this sense, a structural system decomposition has beenperformed, with a specific view on the structural analysis and performance. The presentedconsiderations aim at the organization of the framework for the basis of design of offshorewind turbines, as a support to the decision making, with specific reference to the structuralsafety, serviceability and reliability for the entire lifespan. Furthermore, numerical analyseshave been performed to compare the safety performance of three different support structuretypes, generally adopted for a water depth lower than 50m: monopile, tripod and jacketsupport structures. Extreme loads with a recurrence period of 100-years have been applied atthis stage of investigation. Well-known analytical formulations have been summarized for correct characterizationof both the aerodynamic and hydrodynamic actions, whose contribution is crucial forassessing the structural behavior. An early analysis has been carried out for the investigationof the dynamic response for each one of the three support structures. Thus, the natural modesof vibration have been determined in relation with the principal geometrical designparameters. This is essential for avoiding the occurrence of resonance when the frequenciesof the external forces could excite the structure’s natural modes. A subsequent static analysishas been carried out simulating three different load combinations as prescribed byInternational Standards: the relative influence of aerodynamic and hydrodynamic loads hasbeen assessed, focusing on the resultant shear force and the overturning moment at the mudline, and on the horizontal displacement at the hub height. This step is introductory for theselection of the jacket structure as the appropriate support type. Moreover, the internal state of stress under the abovementioned steady extreme loads hasbeen evaluated by means of two different levels of detail for the numerical models (macro-and meso-level). The analyses have confirmed that macro-level model results can predict thebasic aspects of the structural response, yet the meso-level model provides an additional andmore detailed picture of the structural behavior due both to the major capabilities of the
  22. 22. W IND E NGINEERING VOLUME 34, N O . 1, 2010 105adopted finite elements (shell and brick instead of beam elements) and to the highergeometrical resolution of the models. Finally, an incremental analysis has been carried out to assess the buckling load of theexamined offshore wind turbine: this occurs in the tower tubular section for a multiplier equalto 1.17 for the more severe extreme loads. Starting from the results presented here, future and more refined studies can take intoaccount for other relevant effects influencing the dynamic response of the structure (e.g.scour, coupling with foundation medium, non-stationary loads, non-linear interactions etc.) byperforming transient analyses.ACKNOWLEDGEMENTSThe present work has been developed within the Wi-POD Project (2008-2010) and otherresearch projects in the field of wind engineering, partially financed by the Italian Ministry forEducation, University and Research (MIUR). Fruitful discussions with Prof. Pier GiorgioMalerba of the Politecnico di Milano, Prof. Marcello Ciampoli of the Sapienza – Università diRoma, Professor Hui Li of the Harbin Institute of Technology and Dr. Ing. Gaetano Gaudiosi ofthe OWEMES association are gratefully acknowledged. Finally, Prof. Jon McGowan isacknowledged, for inspiring part of this work.REFERENCES[1] Hau, E., Wind Turbines: Fundamentals, Technologies, Application, Economics, 2nd edn., Springer-Verlag Berlin, Heidelberg, 2006.[2] Breton, S.-P. and Moe, G., Status plans and technologies for offshore wind turbines in Europe and North America, Renewable Energy, 2009, 34 (3), 646–654.[3] Bontempi, F., Basis of Design and expected Performances for the Messina Strait Bridge, Proceedings of the International Conference on Bridge Engineering – Challenges in the 21st Century, Hong Kong, 1-3 November, 2006.[4] NASA (National Aeronautics and Space Administration), Systems Engineering Handbook, 1995. Available online on 10/2009 at:[5] Simon, H.A, The Sciences of the Artificial, The MIT Press, Cambridge, 1998.[6] Bontempi, F., Gkoumas, K. and Arangio, S., Systemic approach for the maintenance of complex structural systems, Structure and infrastructure engineering, 2008, 4, 77–94.[7] Snel, H., Review of Aerodynamics for Wind Turbines, Wind Energy, 2003, 6 (3), 203–211.[8] Westgate, Z.J. and DeJong, J.T., Geotechnical Considerations for Offshore Wind Turbines, 2005, Report for MTC OTC Project, Available online on 10/2009 at: OWC.pdf.[9] Ibsen, L. B. and Brincker R., Design of a New Foundation for Offshore Wind Turbines, Proceedings of the IMAC-22: A Conference on Structural Dynamics, Michigan, 26–29 January, 2004.[10] Zaaijer, M. B., Foundation modelling to assess dynamic behaviour of offshore wind turbines, Applied Ocean Research, 2006, 28 (1), 45–57.[11] Veldkamp, D., A probabilistic approach to wind turbine fatigue design, Proceedings of the European wind energy conference and exhibition, Milan, 7–10 May, 2007.[12] Tempel, J. van der, Design of support structures for offshore wind turbines, PdD Thesis, Technische Universiteit Delft, 2006.
  23. 23. 106 S TRUCTURAL D ESIGN AND A NALYSIS OF O FFSHORE W IND T URBINES FROM A S YSTEM P OINT OF V IEW[13] Biehl, F. and Lehmann, E., Collisions of Ships and Offshore Wind Turbines: Calculation and Risk Evaluation, Proceedings of the International Conference on Offshore Mechanics and Arctic Engineering, Hamburg, 4–9 June, 2006.[14] Martìnez, E., Sanz, F., Pellegrini, S., Jiménez, E. and Blanco, J., Life cycle assessment of a multi-megawatt wind turbine, Renewable Energy, 2009, 34 (3), 667–673.[15] Weinzettel, J., Reenaas, M., Solli, C. and Hertwich, E.G., Life cycle assessment of a floating offshore wind turbine, Renewable Energy, 2009, 34 (3), 742–747.[16] Sumur, B.M. and Fredsøe, J., The mechanics of scour in the marine environment, World Scientific Publishing Co., Singapore, 2002.[17] Tempel, J. van der, Zaaijer, MB. and Subroto, H., The effects of scour on the design of offshore wind turbines, Proceedings of the 3 rd International conference on marine renewable energy, Blyth, July 6–9, 2004.[18] Henderson, A.R. and Patel, M.H., On the modeling of a floating Offshore Wind Turbines, Wind Energy, 2003, 6 (1), 53–86.[19] Jonkman, J.M. and Buhl Jr., M.L., Loads Analysis of a Floating Offshore Wind Turbine Using Fully Coupled Simulation, Proceedings of the WindPower Conference and exhibition, Los Angeles: California, June 3–6, 2007.[20] API (American Petroleum Institute), Recommended Practice for Planning, Designing and Constructing Fixed Offshore Platforms–Load and Resistance Factor Design (RP 2A-LRFD), 1993 (suppl. 1997).[21] BSH (Bundesamt für Seeschifffahrt und Hydrographie), Standard: Design of Offshore Wind Turbines, 2007.[22] DNV (Det Norske Veritas), Offshore Standard: Design of Offshore Wind Turbine Structures (DNV-OS-J101), 2004.[23] GL (Germanischer Lloyd) Wind Energie GmbH, Richtlinie zur Erstellung von technischen Risikoanalysen für Offshore-Windparks, 2002.[24] GL (Germanischer Lloyd) Wind Energie GmbH, Guideline for the Certification of Offshore Wind Turbines, 2005.[25] IEC (International Electrotechnical Commission), Wind Turbine Generator Systems – part 3: Design requirements for offshore wind turbines (IEC 61400-3), 2009.[26] Bontempi, F., Li, H., Petrini, F. and Gkoumas, K., Basis of Design of Offshore Wind Turbines by System Decomposition, Proceedings of the 4th International Conference on Advances in Structural Engineering and Mechanics, Jeju, Korea, 26–28 May, 2008.[27] Starossek, U., Progressive Collapse of Structures, Thomas Telford Publishing, London, 2009.[28] Bontempi, F., Giuliani, L. and Gkoumas, K., Handling the exceptions: dependability of systems and structural robustness, Proceedings of the 3rd International Conference on Structural Engineering, Mechanics and Computation, Alphose Zingoni (Ed.), Millpress, Rotterdam, 2007.[29] Polnikov, V.G., Manenti S., Study of relative role of nonlinearity and depth refraction in wave spectrum evolution in shallow water, Journal of Engineering Applications of Computational Fluid Mechanics, 2009, 3 (1), 42–55.
  24. 24. W IND E NGINEERING VOLUME 34, N O . 1, 2010 107[30] Bontempi, F., Li, H., Petrini, F. and Manenti, S., Numerical modeling for the analysis and design of offshore wind turbines, Proceedings of the 4th International Conference on Advances in Structural Engineering and Mechanics, Jeju, Korea, 26–28 May, 2008.[31] Petrini, F. Giuliano, F. and Bontempi, F., Comparison of time domain techniques for the evaluation of the response and the stability in long span suspension bridges, Computer & Structures, 2007, 85, 1032–1048.[32] Van Binh, L., Ishihara, T., Van Phuc, P. and Fujino, Y., A peak factor for non-Gaussian response analysis of wind turbine tower, Journal of Wind Engineering and Industrial Aerodynamics, 2008, 96 (10–11), 2217–2227.[33] Brebbia, C.A. and Walker Newnes, S., Dynamic analysis of offshore structures, Butterworth, London, 1979.[34]