Certification of Offshore Wind Farms

Transcription

1 Certification of Offshore Win Farms Silke Schwartz 1 an Kimon Argyriais Germanischer Lloy WinEnergie GmbH, Steinhöft 9, 0459 Hamburg, Germany, 1 tel:+49 (0) fax: -170, ssz@gl-group.com tel:+49 (0) , fax: -170, as@gl-group.com Abstract The paper presents the general way through the certification with emphasis on the loa assumptions. An overview of the state of the art consiering the external conitions for the loa assumptions an methos to improve the instantaneous calculation approach are given. The ifferences between offshore site ata an generic requirements in regulations are presente. Since for the machinery of the win turbine the influence of the wave loas is small, the certification is similar to the certification for onshore turbines for ifferent classes with respect to win park influence. A comparison of the machinery loas for a selective win turbine on- an offshore will be rawn up. The influence of wave loaing on machinery components is analyse. For tower an founation, a generic approach is not possible, etaile knowlege of the site conitions, such as water epth ranges, soil conitions, park influence, wave scatter iagram, extreme win an wave conitions an currents is necessary. Due to lack of available ata, this paper gives ifferent methos to obtain sea states out of the correlation between win spee an wave as well as spectral correlation, incluing shallow water influence. A metho use to combine extreme waves an win gusts is analyse an the influence of braking waves, the variation of soil conitions an water epths are iscusse. Further special offshore influences an bounary conitions like scour, corrosion, maintenance, availability are pointe out. Keywors: Certification, Offshore, sea spectra, loa assumptions, scour 1 Introuction In the growing market for win energy an the limite available space onshore, the evelopment of offshore win farms becomes more an more important. The particular requirements offshore became apparent when planning the first offshore win farms in the UK, Denmark, the Netherlans, Sween an Germany. For reliable preictions of the aeroynamic an hyroynamic loaing for the esign of win turbines the experiences from onshore win technology an the knowlege of the construction of offshore structures (e.g.: oil platform) as well as results from research projects leaing to the publication of Germanischer Lloy (GL) Regulations for the Certification of Offshore Win Energy [1] in 1995 shoul applie. The Regulations were evelope in the JOULE 1 Offshore stuy [] by merging the GL Regulations for the Certification of Win Energy Conversion Systems [3] an the Rules for Offshore-Installations [4]. In the meantime it was possible to apply these guielines in reality an to prove an compare the requirements in the regulations with the conitions at offshore sites. Experience in offshore certification accoring to GL classes as well as site specifications offshore exist through several projects. Further site measurements within the scope of the EU fune research project OWTES (Offshore Win Turbine at Expose Sites) amit to review the requirements in the Regulations an new perceptions for loa assumptions are iscusse.

2 Certification of offshore win farms While onshore win turbines are generally installe with a Type Certificate offshore win turbines nee in general aitional monitoring, verification an certification. All of these activities can be inclue in the Project Certificate. Aitionally to a Type Certificate which can issue for offshore win turbines in the same way as for the onshore (incluing esign assessment, prototype testing (onshore), quality management) the Project Certificate may comprise the following tasks: Type certification of win turbine Site assessment Site specific esign assessment of the founation Risk evaluation Monitoring of fabrication, transport an installation Commissioning witnessing Perioic monitoring Monitoring of ecommissioning From the range of tasks the appropriate moules will be put together, on the basis of the relevant regional requirements an the preferences of the owner/investor. In Type Certification the main task is the evaluation of the esign ocumentation incluing parallel inepenent analysis of loas an structural etails. Regaring offshore projects an the influences on offshore win turbines it became evient that a Type Certification of the whole structure as in the onshore case is not possible. The influences of site specific parameters, like water epths, wave heights an soil conitions, on the support structure (founation an tower) are significant, resulting in site specific esigns. In contrary it can be shown that the influence of the site parameters (except win) on the machinery esign (above yaw bearing) is limite, allowing type classification. Site Assessment is one of the manatory moules in Project Certification. It comprises the assessment of the environmental conitions an the valiation of the esign assumptions on the basis of the site specific conitions. The influence from following items is consiere in the assessment: Site ata (water epth, tial ranges, etc.) Environmental conitions at the site (win, waves, currents, temperature, sea ice, seismic activity etc.) Influence of the turbines to the site an to themselves (scour, wake influence) Geotechnical ata Electrical conitions at the site In many cases the machinery of the turbines to be installe are erive from onshore esigns. In the certification proceure it has to be shown that the measures taken to comply with the requirements of the marine environment are sufficient.

3 3 Requirements for the loa assumptions 3.1 Site ata The site ata shall inclue the co-orinates for all turbines consiere. Special care has to be taken to provie accurate measurements of the water epth for each location reference to a clearly specifie atum (usually LAT = Lowest Astronomical Tie). Water epth is a riving factor for the founation loaing. It has ecisive influence on the turbine ynamics, on the support structure loas an in the case of scour an san movement to the pile penetration epth an loa carrying capacity. For gravity founations the anger of settlements an the appropriate soil preparation have to be taken into account. Thus aitional analysis shoul be performe regaring the soil stability an the occurrence of moving san. The possibility of soil changes uring the whole lifetime of the project has to be consiere. Aitionally, the tial range at the site has to be ocumente. 3. Environmental ata It is of prime importance to obtain appropriate win, wave an current ata for the site. Other ata are to be evaluate as applicable (temperature of air an sea, ice, seismic activity, etc.). Correlation of win an waves The 50 year return perio shall be use for extremes. In the case of combine actions, their joint probability shoul be use. It is consiere that the extreme win spee occurs in the same 50-year storm as the extreme wave. Although uring this storm the two extremes are not correlate. The same may be the case regaring the extreme water level. It is expecte that the extreme storm surge occurs uring the 50-year storm an thus together with the extreme wave. But the 50-year storm is not correlate to the highest astronomical tie. If any statistical information on the extreme water level exists this may be consiere, alternatively the values have to be ae resulting in a conservative approach. Win The external win conitions for an offshore site are given, in the same way as onshore, with annual average win spee, win istribution, the turbulence intensity an the win profile. The win climate offshore is ifferent from the win conitions onshore with lower turbulence intensities, win shear an usually higher mean win spees. It has to be state that at high win spees turbulence intensity increases with increasing win spee. Aitionally, the influence of wakes from neighbouring turbines on turbulence intensity gets more importance than for onshore sites. Since it is ifficult to get measurements at most projecte offshore sites, special attention has to be pai on the numerical moelling of the win conitions at the site. The ifferent win irections an the istance to shore have to be consiere as well as the influence of the sea roughness ue to ifferent fetch an water epths. It is recommene to verify the moels with the most nearby place offshore measurements to get confience in the ata. A long term correction accoring to nearby onshore ata is possible. Waves Wave heights an sea states are highly epenent on site conitions like fetch, time an water epth. The waves are escribe by long term istributions of the sea state in the scatter iagram while the short term escription is given by wave spectral representations. The sea state statistics are given via scatter iagrams presenting the probability of the ifferent significant wave heights (H s ) an zero crossing perios of the waves (T z ) for the given site. To reuce the number of calculations these scatter iagrams are lumpe to some

4 representative mean values. An example of a scatter iagram for a North Sea site is given in Figure 1. WAVE HEIGHT / PERIOD SCATTER TABLE HEIGHT PERIOD (s) (m) TOTALS TOTALS Figure 1: Wave scatter iagram Sea spectra The formulations of wave frequency spectra in marine applications are the Pierson- Moskowitz spectrum (PM-spectrum) for a fully evelope sea an the JONSWAP spectrum (Joint North Sea Wave Project) for a eveloping sea. It is formulate as a moification of the PM-spectrum for a eveloping sea state in a fetch limite situation. For finite water epths the self-similar spectral shape (TMA-Spectrum [9]) was evelope. This spectrum is an extension of the JONSWAP Spectrum an water epth epenent. In offshore inustry the PMspectrum is wiely use for fatigue analysis while the JONSWAP spectrum is use for extreme loa analysis. In many cases the scatter ata are not available or the scatter iagram oes not inclue any information about the win spee to be use in combination with the wave height. In this case theoretical assumptions of the win wave correlation have to be built up. Using the oneparametric wave spectra, or the relations erive for the JONSWAP spectrum in conjunction with the TMA-formulation a relation may be erive for win spee to significant wave height. The JONSWAP spectrum is given in [8]: S η with ( ω) α g = nf 5 ω? = wave frequency ω p g a 5 ω p exp 4 ω = spectral peak frequency = acceleration of gravity 4 γ generalize Phillips constant 1 ω ω p exp σ ω p = nf S PM, η ( ω) γ σ = spectral with parameter (s a 0.07 for ω < ω p ; s b 0.09 for ω > ω p ) γ = Peak shape parameter (Pierson-Moskowitz spectrum for γ = 1) nf = Normalizing factor to ensure same H s to PM-spectrum, i.e. [ ( )] nf = γ 1 ω ω p exp σ ω p

5 For the North Sea ata the parameters γ, s a an s b are not constant but show consierable scatter. Average values from the JONSWAP ata were: γ = 3.3; s a = 0.07; s b = In the ata, the peak shape parameter γ varie between about 1 an 6 an was approximately normally istribute with a mean of 3.3 an a stanar eviation of In ifferent geographical areas the JONSWAP Spectrum form also appears to be capable of representing the observations rather well, provie that the parameters γ, s a an s b are chosen in accorance with the local ata [15]. For not fully evelope sea states the influence of the time an fetch x the win is acting may be consiere with [16]: = / The imensionless time of win: ( g u) time an the fetch: ( g / u ) x θ ξ = with g the acceleration of gravity an u the hourly mean win spee at 10 m above the sea surface. 3 ω p u 0.3 The imensionless peak frequency is = = 7 ν max 0.16;.84 ξ ; 16.8 θ π g the value of the Phillips constant is 3 α = 0.08 ν the peak perio u 1 T p = g ν u 3 an the significant wave height is H s = ν g As state above the P-M an the JONSWAP spectra were evelope for eep water situations an the TMA-spectrum [9],[6] for finite water epth. Its general valiity was checke against measurements (Texel, Marsen, Arsloe). The influence of the water epth on the spectrum may be seen in figure. The resulting significant wave height from the calculation using the TMA- JONSWAP formulation an the measurements from a site are shown in figure 3. an with F k = transformation factor S, TMA ( ω) = S,JONSWAP Φ k ( ω ) η ω η = ω? = imensionless epth epenent frequency = water epth Φ Φ Φ k k k ( ω ( ω ( ω ) = 0.5 ω ) = ( ω ) = 1 ) for for for ω ω ω 1 > 1 > g 5

6 JONSWAP TMA, =10 m TMA, =15 m TMA, =0 m TMA, =30 m TMA, =40 m Omega Figure : JONSWAP an TMA wave spectra In the case shown in figure 3 not only the water epth is limite but the istance to the coast is relatively small. The result will be that the sea state is not fully evelope an the wave heights are limite by the finite water epth. From the external ata consiere it is clear that the importance of water epth, fetch an time shoul be taken into account. TMA full evelope TMA fetch limite JONSWAP fetch limite Measurement Hs [m] Measurement Figure 3: Wave height comparison win spee v [m/s] A further problem in the absence of reliable ata is the combination of probabilities when a wave scatter iagram has to be combine with a win istribution. In most projects examine a wave scatter iagram for the possible project site existe as well as a separate win istribution. It is assume that that shallow water an current effects are inclue in the wave scatter iagram. For the combination a metho was propose by Matthies [10] resulting in a scatter iagram for the combine case. A more simple metho is to correlate the cumulative

7 probability istribution functions of the win spee an the significant wave height. From the pairs of same probability of exceeance the pairs of win spee an significant wave height are erive. In this process caution has to be taken for the existence of win riven sea an swell waves. In this case the analysis has to be performe for two ifferent istributions of wave heights accoring to their overall probability of occurrence. Direction of win an waves Directionality of win, wave an current may be consiere if accurate ata is available. In this case not only the mean win an wave irection have to be consiere, but the istribution of the win an wave misalignment, too. Generally, if no etaile information is available, it is recommene to apply win an wave ata as uniirectional, acting from the same irection for the turbines lifetime. In the regulations the wave irection is always the irection of the mean win spee. In general this applies for open sea conitions an extreme waves. It is site specific to verify if this assumption is correct. If sufficient evience exists that any of the environmental forces are irectional, it may be possible to position the structure in the most avantageous irection. Breaking waves Breaking waves may lea to high loas ue to trigger ringing or ynamic amplifications on the structure. They can be prouce by shoaling (wave moving from eep water into shallow water), by win-wave, wave-wave an wave-current interaction. Shoaling is the most important type for offshore win turbines. In shallow water the empirical limit of the wave height is approximately 0.78 times the local water epth. This has been foun to compare well with ata, although some investigators have foun a weak epenence on beach slope. In eep water, waves can also break with a theoretical limiting steepness of 1/7 (0.14 times wavelength?). Three types of breaking wave may occur in shoaling waters, when the water epth ecreases an the wave height increases. The ifference between the types is not very sharp an their occurrence epens on the eep water wave height, the length ratio an the seabe slope. Galvin [11] characterises breakers with the following ratio: B H ( λ m ) = o r an 0 with: H 0 = wave height in eep water λ 0 = wave length in eep water m = seabe slope B r > 5 spilling breaker 5 > B r > 0.1 plunging breaker 0.1 > B r surging breaker Further information may be foun in [1]. Surging breakers are usually not important for offshore structures. Spilling breakers may be analyse as a limite height regular wave. For plunging breakers, few ata are available on the velocities an accelerations of water particles in waves. Due to the suen immersion of a member the impact force on it lea to a ynamic response of the member, which may increase the force on the member consiere (slap force). There are ifferent engineering methos to calculate wave slap. The physics of the forces are ientical for vertical an horizontal members (slam) so methos to calculate slamming forces are use. In the offshore stuy [1] an the Atkins report [1] a metho for calculating the slap force is escribe. The formula for the slamming/slap force is similar to the rag formula applying a slap coefficient instea of the rag coefficient. The ynamic amplification of the member as a function of the eigenfrequency, angle to the water surface an wave celerity

8 shoul be consiere. In the Danish recommenation for offshore win turbines [13] it is state that the maximum particle velocity in the plunging breaking wave is given by the expression u = max 1. 5 g, an shall be applie as the velocity in a monotonous velocity profile for the entire wave above the still water level. Below the still water level a velocity profile is applie, as in conventional wave theory. A solution use is to calculate the loa using regular wave theory an apply a correcte rag coefficient for the part of the pile affecte by the plunging breaking wave. The application of an increase rag coefficient ue to slap or the correcte rag coefficient for high water particle spee results in high rag coefficients in the magnitue of Loa assumptions 4.1 Extreme loa From the experience in offshore an onshore loa analysis it is generally known that esign riving loa cases for the machinery are epening on the control system of the turbine an the extreme win spee. Aeroynamic loaing ominates the loaing on the machinery. For the support structure the extreme 50-year storm case is often the esign river for the ultimate loaing, with the portion of wave to win loa varying with the water epth. In the following the problem to implement the analysis for the 50-year storm case is aresse. 50 year storm case The assumptions for the extreme win spee an the extreme wave height gain significance in this relation. The extreme wave height to be combine with this win spee is assume to be site specific. Limiting value for the wave height is very often the breaking wave height H B The problem is not only to fin the corresponing extreme wave an the respective win spee, the analysis proceure to combine the loaing implies some problems, too. To take account of the turbine elastic behaviour an the nature of win loaing time omain simulations are commonly performe base on stochastic win spee time series. In this case turbine ynamic behaviour an ynamic amplification of win loas is correctly represente, but several realisations are necessary ue to the stochastic nature of gusts. An often use alternative is to apply a constant win spee an then multiply the loas with a ynamic amplification factor as efine in builing stanars, or erive by separate analysis. In offshore inustry it is recognise that, especially in shallow waters, the wave particle kinematics non-linearity may play an important influence. In this case an analysis has to be performe on the significance of the wave non-linearity an the structure elastic behaviour. Since offshore structures place in shallow waters are ominate by non-linear extreme wave action an not by elastic response the quasi static approach is often use. In combination with the analysis metho using stochastic win spee time series a stochastic wave time series coul be use. This calculation woul result in a correct phasing of win spee an wave elevation uring a storm (provie the simulation length an number is big enough). The turbine elastic behaviour is represente in the loa analysis. A major problem is that only linear wave kinematics can be use for the stochastic wave train. Some methos propose in recent time to inclue wave kinematics non-linearity have foun only scientific application an are still too complicate for engineering use. The alternative form woul be to use the ynamic amplification factor metho for the win loa an nonlinear single wave kinematics for the wave loa. The problematic point is to calculate the ynamic amplification factor for the win loaing an to separate win from wave loaing. The influence of the

9 ifferent analysis methos are shown in table 1 for the mu line overturning moment M xy. The example is base on a generic MW win turbine for a water epth of 0 m. It has to be state, that the principal loa irection of the win an wave loas may be ifferent. Wave only H = 1.1 m M xy [MNm] Win only v = 50 m/s, I =11%, v e = 63 m/s M xy [MNm] Linear (Airy, Wheeler stretche) 56 Constant win, stiff structure 47 Non-linear (Stream function) 95 Turbulent win, elastic structure 61 Difference non-liner/linear 1.7 Difference ynamic/static 1.3 Table 1: Mu line overturning moment for a generic MW offshore win turbine in 0 m water epth. In the analysis it can be seen that the win an wave loas at the mu line are of the same orer of magnitue. At the tower top an for the machinery the wave loas o not have significant influence, except in the cases of resonance between the wave perio an its higher harmonics an the support structure eigenfrequency. In the present case the ynamic amplification of the linear wave loa is below 10% of the maximum loa. It is clear, that a simultaneous combination of the extreme gust win spee an the extreme wave will result in a very conservative approach. It is assume that the extreme wave an the extreme gust occur uring the same 50-year storm but their appearance is not correlate. This can be shown in table, presenting results from stochastic analysis. Win only Wave only Combine win an wave Win + wave Quaratic aition Table : Mu line overturning moment for ifferent combinations. Stochastic, ynamic analysis In the offshore inustry the 1-minute average win spee is usually use in combination with the extreme wave for the global structure analysis. In [] a reuce wave height was erive to be combine with the extreme gust win spee (5s-average), having the same probability of occurrence as the combination of maximum wave height with 1-minute average win spee. Accoring to the offshore stuy [1] the extreme wave height is taken as the most probable highest of 1000 waves uring a storm assuming a Rayleigh istribution. This is a conservative approach an for ifferent sites an extra investigation may lea to ifferent results [7] consiering the wave spectral broaness an the storm uration. To comply with the requirements of the ynamic analysis an the wave kinematics nonlinearity the present approach inclues several realisations of the 50-year storm: Dynamic simulation using stochastic win an waves. This approach uses linear wave theory, but takes full account of the structure ynamics. Simulation using constant (1-minute average) win spee, with correction for ynamic amplification an eterministic non-linear extreme wave (H max = 1.86 H s ). In this case the extreme wave influence for wave loa ominate structures is consiere. Simulation using constant (5-secon average) win spee, with correction for ynamic amplification an eterministic non-linear reuce extreme wave (H max = 1.3 H s ). In this case the extreme non-linear wave influence for win loa ominate structures is consiere. The result for the overturning moment M xy an the shear force F xy at the mu-line is shown in table 3.

10 Win & Wave 5-s av. & reuce wave 1-min. av. & max. wave Turbulent & stochastic F xy [MN] M xy [MNm] Table 3: Mu line loa for ifferent combinations of 50-year storm case. In the present analysis the ynamic amplification is applie as a factor on win spee whereas the amplification ue to wave loa is inclue through the support structure elasticity in simulations using nonlinear waves. Usually not only a single wave but a train with 4 to 5 waves with extreme height is simulate. In this case the ynamic amplification ue to wave loas (an the nonlinear kinematics) is inclue in the analysis, but ue to the train of consecutive extreme waves use a conservative result may be receive. A less conservative way coul be to use a wave train of 4-6 nonlinear waves with linear changing wave height until the maximum wave height is reache. 4. Fatigue loa For the fatigue analysis special care has to be applie in consieration of all aspects of offshore win farms. Different water epths an soil conitions in park configurations may lea to ifferent eigenfrequencies for the iniviual turbines. Furthermore ifferent tial ranges, corrosion allowances, scour influences, higher probabilities of iling loa cases ue to worse availability, wake influences etc. have to be consiere for each turbine resulting in a high number of fatigue analyses. It is essential to evelop methos to etect the turbine with the highest fatigue loaing in an offshore win farm configuration. In the offshore inustry fatigue loas on offshore structures may be calculate using frequency omain analysis [14]. For win loaing time omain simulations are use to cover the non linear behaviour of the structure. In this context the combination of the two methos was trie. A successive consieration of aeroynamic an hyroynamic loaing was carrie an compare with a simultaneous simulation in the time omain. The investigation was carrie out for a MW generic offshore turbine locate at 8 m water epth. In table 4 the equivalent loas for the fatigue analysis for a successive an a simultaneous calculation are compare. The loaing out of the successive investigation is combine with a quaratic superposition. Win only Wave only Win+Wave Win+Wave Deviation [kn] or [kn] or Simultaneous Quaratic [knm] [knm] [kn] or [knm] [kn] or [knm] % Thrust force mu line Thrust force MSL Thrust force tower top Tilting moment mu line Tilting moment MSL Tilting moment tower top Thrust force hub Torque hub Tilting moment hub Blae root flap moment Table 4: Fatigue loa spectra for ifferent section N = 10 8, S/N-curve exponent m = 4 It can be seen that the successive consieration with the quaratic superposition is a goo an conservative approach to obtain loaing on the turbine. It can Also be seen that the influence of the wave loaing on the machine an blae loaing is neglectable in the present case. Design of the main bearing may be influence by an increase thrust at the hub, foun in

11 some cases. This increase thrust force is a factor of the ynamic amplification of the wave forces an epens heavily on first an secon natural frequencies of the support structure. In the example presente a relatively stiff (T s) structure was chosen resulting in minimal influence in thrust force. The metho for the superposition of win an wave generate equivalent loas is a relatively simple solution to gather acceptable an comparable loa spectra of the combine loaing. General experience shows that a superposition of wave loas obtaine from a frequency omain analysis an win loas from a time omain analysis has to be performe very carefully. The ifferent approaches an the iscretisation errors of the wave elevation may lea to consierable errors for sections near the water line. But this metho allows extensive parameter variations to be investigate i.e. covering the whole range of water epths, soil properties etc. within a win farm. This way highly loae win turbines in a farm may be etecte for which more precise time omain analysis nee to be carrie out. 5 Generic Design It is a general view that a generic esign may help manufacturers in the evelopment of offshore win turbines. The turbine classes efine are applicable for win conitions. Wave heights are site specific an the influence of the ifferent external conitions especially on the support structure raises oubts on the applicability of this proceure. In the following an analysis was performe to compare loaing on the machinery an the supporting structure accoring the stanar GL onshore an offshore classes [3][4] an to site specific approaches. The ata were selecte from existing projects GL Win is involve. The environmental conitions are between class 1 an accoring to the GL-Regulations for offshore win turbines. To generalise the results, mean values from the ifferent projects an those erive for generic offshore win turbine esigns are shown. The win conitions accoring to the Regulations an the site assumptions are shown in table min extreme [m/s] 5-sec extreme [m/s] Mean win spee [m/s] Win shear Turbulence intensity [%] Hmax [m] Mean sea level [m] GL offshore class I I = 1+park GL offshore class II I = 1+park 7 8 Site abc (mean) I 15 9+park Site z I = 1+park 7 8 GL onshore class I GL onshore class II Table 5: Environmental conitions for generic classes compare to site ata The wave an current parameters applie in the investigations were site ata. In the case the generic turbine classes o not provie any information on the parameters to be applie an no simplifie metho exists to erive these, the site ata were applie. The results of this analysis are shown in tables 6 an 7. To get a realistic approach on the applicability two cases with ifferent support structures an slightly moifie turbine parameters were analyse. The first case was analyse accoring to GL class an compare with site z which showe environmental ata equal or less severe than class. In the secon case the analysis was performe for a class 1 turbine an the results were compare to the mean of the site specific analysis. As it can be seen in table 6 the fatigue machinery loas (from blae tip to yaw bearing) of the generic offshore win turbine are less or equal to the loas of the equivalent onshore win turbine. Similar results are achieve for most of the extreme loas consiere. From the

12 results it can be seen that the generic approach gives a goo approximation of the average site specific machinery loas. Significant ifferences were seen in the tower top loas ue to the ifferent support structure ynamics (site / generic). The loas at the mu line showe large scatter. This makes clear that the support structure has to be site specifically esigne. Generic moel a GL onshore GL offshore Generic moel b GL onshore GL1 offshore GL offshore site z GL1 offshore mean of sites abc Blae root M ege Blae root M flap Hub fixe F x Hub fixe M x Hub fixe M y Hub rotating M yz Tower top F x Tower top M x Tower top M y Mean Table 6: Comparison of fatigue machinery loas from generic classes an site specific analysis Generic moel a GL onshore GL offshore Generic moel b GL onshore GL1 offshore GL offshore site z GL1 offshore mean of sites abc Blae root M ege Blae root M flap Hub fixe F x Hub fixe M x Hub fixe M y Hub rotating M yz Tower top F x Tower top M x Tower top M y Mean Table 7: Comparison of extreme machinery loas from generic classes an site specific analysis. GL-coorinate system. From the parametric analysis performe following points have been ientifie: Marinize onshore turbines are initially evelope accoring to stanar classes. This leas to a conservative estimate of the loaing for the machinery (at least for the typical sites now in evelopment). The support structure is largely influence by win, wave loaing, water epth an soil parameters. A generic approach for the machine is possible; site specific esign of the substructure is require. Existing classes I an II apply for offshore conitions.

13 6 Further influences on offshore win turbine loaing 6.1 Scour Scour is the removal of seabe soils by currents an waves. Such erosion can be ue to a natural geological process or can be cause by structural elements interrupting the natural flow regime above the sea floor. From observations, sea floor variations can usually be characterize as some combination of the following. Local scour. Steep sie scour pits aroun structure elements. Global scour. Shallow scoure basins of large extent aroun a structure, possibly ue to overall structure effects, multiple structure interaction, or wave-soil-structure interaction. Overall seabe movement of san unes, riges, an shoals that woul also occur in the absence of a structure. Such movements can result in sea floor lowering or rising, or repeate cycles of these. Scour can result in removal of vertical an lateral support for founations, causing unesirable settlements of shallow founations an overstressing of founation elements. Where scour is possible it shall be taken into account in esign an/or its mitigation shall be consiere. In case that scouring may occur, the founation has to be protecte by suitable means, alternatively, the founation has to be consiere partly unsupporte. If no other ata are available for the specific site conitions, the scour epth at pile founations may be estimate as.5 ( = pile iameter) for esign purposes. Inicative analysis using the generic turbine moel for a water epth of 0 m with scour equal to 1 iameter showe consierable influence on loaing. Loa type blae flap [knm] hub thrust [kn] hub tilt [knm] tower top tilt [knm] bottom overturning [knm] Change ratio.5% 7% 3%.5% 68% Table 8: Change of fatigue loaing ue to scour of 1. The changes of the loas ue to scour are not only ue to the moifie wave kinematics but also to the changes in eigenfrequencies of the support structure. In the present example the basic natural frequency of the support structure change by 10%. Due to the relevance of scour criteria lai own for the esign shall be verifie by regular surveys. Countermeasures shall be taken in case of exceeance of the limits establishe in the esign. 6. Soil variation The soil conitions in a win farm may vary consierably. Knowlege of the soil conitions existing at the construction site is necessary to evelop a safe an economic esign. During analysis of the ata from ifferent projects a variation of the first natural frequencies of the offshore win turbines of up to 10% was encountere. Thus geophysical surveys shoul be performe in the first stage with a geotechnical investigation following uring main analysis. Structure founation interaction shoul in general be moelle non-linearly. For pile founations the non-linear behaviour of axial an lateral pile-soil support shoul be moelle explicitly to ensure loa eflection compatibility between the structure an pile-soil system. For pile analysis the effects of geometrical an material non-linearities shoul also be accounte for within the structure-pile-soil system. The effect of cyclic loaing which may cause a reuction of shear strength an bearing capacity of the soil shall be investigate.

14 In practice it is very ifficult to perform time omain simulations consiering the founation non-linearity. In this case the p-y curves are approximate by linear springs which iffer with respect to the applie loa. This means that ifferent natural frequencies are erive for extreme an fatigue loa analysis an for ifferent loa cases. The proceure is not straight forwar since loas are a functions of the eigenfrequencies which in their turn epen on the overall loas. 6.3 Other topics Availability In offshore conitions, more emphasis shall be put on reliability, extene remote control an longer maintenance perios. Whenever a severe fault occurs in an onshore turbine a service team can be at the site within hours. However, if this happens at an offshore location an weather conitions o not allow immeiate access the turbine may remain in the faulty conitions for a long perio. This situation has to be consiere in the loa analysis. The reuce turbine availability will result in a greater percentage with the turbine in the iling moe. Since aeroynamic amping is very low in this case the ynamic amplification of win an wave loas may become consierable. The influence on fatigue loaing may not be neglecte an has to be consiere uring esign. Aitionally longer perios with no gri connection or the turbine iling in a fault moe may lea to the occurrence of extreme storms combine with unfavourable yaw an pitch conitions. Corrosion Control An important esign moification for offshore siting is proper corrosion protection. Even if coatings have a high quality corrosion of the pile at the splash zone has to be consiere. Often a corrosion allowance is applie ranging between 0. mm/year to 0.5 mm/year. The amount is a factor of the protection system applie an the site temperature, salinity, humiity etc. This allowance shoul be consiere in loa analysis by using the most onerous situation for extreme loa cases an an intermeiate mi life situation for fatigue analysis. 7 Conclusion A Type Certification of the whole offshore win turbine structure as in the onshore case, inepenent of the site, is not possible. Since the influence of the wave loaing on the machinery is of minor importance, stanarise machinery esigns may be evelope. The machinery may be certifie accoring to generic classes, as efine in GL s Regulation for the Certification for Offshore Win Turbines. The influence of the site hyraulic an soil conitions on the support structure is significant, resulting in site specific esigns. To comply with the increase requirements for certification of offshore win farms a Project Certification may be carrie out. In this it has to be shown that the win turbines, incluing founation, comply with the requirements of the site environment. Thus one essential part of the Project Certification is the site assessment. In the site assessment effort has to be mae to get a escription of the joint istributions of win an waves an, if possible, their irectional istribution, incluing misalignment. Possible combinations of the ifferent environmental phenomena have to be consiere in a realistic manner. The alternative of the simple aition of the loas from ifferent extreme situations or the simple aition of the loas from win an waves woul result in an extreme conservatism. The structural esign of the offshore win turbine machinery an support structure has to take into account both win loas an wave loas. The structural response of the structure which may result from win an waves has to be analyse an consiere in the efinition of the

15 loa conitions. For the founations currents an the non-linear wave loas shall be inclue in the analysis. The experience from the win inustry as well of the offshore inustry is use, but was aapte to the special shallow water conitions occurring at the project sites. 8 References [1] Germanischer Lloy, Regulations for the Certification of Offshore Win Energy Converter Systems, Hamburg [] Matthies et al, Stuy of Offshore Win Energy in the EC JOULE I (JOUR 007), Verlag Natürliche Energie [3] Germanischer Lloy, Regulations for the Certification of Win Energy Conversion Systems, Hamburg, [4] Germanischer Lloy, Rules for Classification an Construction - Offshore Technology, Part Offshore Installations, Hamburg, [5] H. Söing, V. Bertram, Ships in a Seaway, Hanbuch er Werften XXIV. Ban, [6] V. Müller, Flachwassereinflüsse, TU Hamburg-Harburg, AB Meerestechnik 1. [7] K. Kokkinowrachos, Offshore-Bauwerke in Hanbuch er Werften XV. Ban, [8] CD ISO Petroleum an natural gas inustries Specific requirements for offshore structures Part 1: Metocean esign an operating conitions, 00 [9] E. Bouws, H. Günther, W. Rosenthal, C.L. Vincent, Similarity of the Win Wave Spectrum in Finite Depth Water, Journal of Geophysical Research, vol.90, [10] Matthies H.G., Meyer M., Nath C., Offshore-Winkraftanlagen: Kombination er Lasten von Win un Wellen, Proceeings of the DEWEK 000. [11] Galvin CJ, Waves on Beaches, Acaemic publishing, New York 197. [1] Dept. of Energy, Flui loaing on fixe offshore structures, guiance on esign an construction, WS Atkins engineering sciences, 1987 [13] Danish Energy Agency, Recommenation for technical approval of offshore win turbines, December 001. [14] S. Schwartz, K. Argyriais, Analysis of the Fatigue Loaing of an Offshore Win Turbine Using Time an Frequency Domain Methos, Proceeings of the EWEC 001 [15] W. Blenermann, Umgebungsbeingungen un Lastannahmen in er Meerestechnik Institut für Schiffbau er Universität Hamburg, [16] H. Söing, Bewegungen un Belastungen von Schiffen in Seegang, Institut für Schiffbau er Universität Hamburg, 198.