(PDF) Fully coupled aero-hydrodynamic analysis of a semi-submersible FOWT using a dynamic fluid body interaction approachArticlePDF AvailableFully coupled aero-hydrodynamic analysis of a semi-submersible FOWT using a dynamic fluid body interaction approachJuly 2016Renewable Energy 92:244-261DOI:10.1016/j.renene.2016.02.021Authors: Toan Thanh TranNational Renewable Energy Laboratory Dong-Hyun KimGyeongsang National University Download full-text PDFRead full-textDownload full-text PDFRead full-textDownload citation Copy link Link copied Read full-text Download citation Copy link Link copiedCitations (50)References (82)Figures (15)Abstract and FiguresIn the design phase of a floating offshore wind turbine, the influence of aero-hydro-structure dynamic coupling needs to be fully considered to yield reliable analysis results. In this study, a highly elaborated computational model based on a dynamic fluid body interaction method with a superimposed motion and catenary mooring solver is applied and compared with common engineering approaches. An overset-based technique is also utilized to effectively handle large movements of a full floating wind turbine body due to the coupled influence of wind-wave loads. The DeepCwind semi-submersible floating platform mounted by the NREL 5-MW baseline wind turbine is used to obtain validation and verification of the new computational model with the experimental test data and the NREL FAST code. Various computational results for unsteady aerodynamics, hydrodynamics, and fully coupled aero-hydrodynamics including mooring line loads are compared stage by stage with the test data and numerical results calculated by the NREL FAST code. Overall, the predicted results of the aerodynamic performances, platform dynamic responses, and mooring line tensions show good agreements with the presented numerical solutions and the FAST solutions. In addition, multi-phase unsteady flow fields with complex inference effects in the blade-tip vortices, shedding vortices, and turbulent wakes are numerically visualized and investigated in detail. Platform gross properties.… Mooring system properties.… The computational mesh domain for the fully configured OC4 DeepCWind semi-submersible wind turbine: (a) Full grid domain, (b) entire turbine model with overset regions, (c) closed-in view of the platform surface mesh, (d) closed-in view of hub and nacelle, and (e) closed-in view of the blade surface.… Unsteady convergence test of the time-step size for the prescribed surge and pitching motions of the FOWT platform.… +10(continued.)… Figures - uploaded by Dong-Hyun KimAuthor contentAll figure content in this area was uploaded by Dong-Hyun KimContent may be subject to copyright. Discover the world s research20+ million members135+ million publications700k+ research projectsJoin for freePublic Full-text 1Content uploaded by Dong-Hyun KimAuthor contentAll content in this area was uploaded by Dong-Hyun Kim on Dec 07, 2017 Content may be subject to copyright. Fully coupled aero-hydrodynamic analysis of a semi-submersibleFOWT using a dynamic fluid body interaction approachThanh Toan Tran, Dong-Hyun Kim*Research Centre for Offshore Wind Turbine Technology (ReCOWT), Graduate School of Mechanical and Aerospace Engineering, Gyeongsang NationalUniversity (GNU), 900 Gajwa-dong, Jinju 660-701, South Koreaarticle infoArticle history:Received 10 July 2015Received in revised form4 February 2016Accepted 7 February 2016Available online xxxKeywords:Dynamic fluid body interaction6-DOF solverFully coupled aero-hydrodynamicsOC4 DeepCWindOverset gridFAST codeabstractIn the design phase of a floating offshore wind turbine, the influence of aero-hydro-structure dynamiccoupling needs to be fully considered to yield reliable analysis results. In this study, a highly elaboratedcomputational model based on a dynamic fluid body interaction method with a superimposed motionand catenary mooring solver is applied and compared with common engineering approaches. Anoverset-based technique is also utilized to effectively handle large movements of a full floating windturbine body due to the coupled influence of wind-wave loads. The DeepCwind semi-submersiblefloating platform mounted by the NREL 5-MW baseline wind turbine is used to obtain validation andverification of the new computational model with the experimental test data and the NREL FAST code.Various computational results for unsteady aerodynamics, hydrodynamics, and fully coupled aero-hydrodynamics including mooring line loads are compared stage by stage with the test data and nu-merical results calculated by the NREL FAST code. Overall, the predicted results of the aerodynamicperformances, platform dynamic responses, and mooring line tensions show good agreements with thepresented numerical solutions and the FAST solutions. In addition, multi-phase unsteady flow fields withcomplex inference effects in the blade-tip vortices, shedding vortices, and turbulent wakes are numer-ically visualized and investigated in detail.©2016 Elsevier Ltd. All rights reserved.1. IntroductionOne of the common challenges in designing a floating offshorewind turbine (FOWT) is the ability to accurately predict criticalloads due to various turbulent-wind and stochastic-wave condi-tions. It is very hard to accurately calculate these critical loadsbecause of the complex multi-physical phenomena in realisticoperating conditions. Thus, careful validations of numerical anal-ysis methods are needed to build confidence in the design process.Because of challenges in predicting the loads in the design re-quirements, various experimental floating substructures in bothwave basin facilities and on ocean coasts are either routinely per-formed or in the planning stages all over the world [1e3]. On theocean coast, several intermediate-scale FOWT models have beendeployed, such as the 100 kW tension-leg platform by BlueH(Italian coast), Poseidon wind/wave prototype with three 11 kWturbines (Denmark coast), 1:5 scale test tension-leg spar design bySWAY (Norwegian coast), 2 MW semi-submersible platform off thecoast of Kabashima Island (Japan coast), and 1:8 scale prototype of a6 MW turbine platform called VolturnUS (US), etc. [3,4]. Theseintermediate-scale tests are essentially required prior to the full-sized construction of commercial-scale projects [4]. However,they are expensive and difficult to perform under controlled con-ditions. They may also be unrealistic in waiting for extreme weathersituations to occur. Therefore, a wave basin test for a scaled-downFOWT model is desirable to reduce the risk and cost, which al-lows the dynamic characteristics of a floating system to be accu-rately evaluated. The scaled-down FOWT model tests in wavebasins have been initiatively conducted by research groups utilizingdifferent concepts, such as the WindFloat concept by PrinciplePower, Inc., U.S. [5,6], GustoMSC Tri-Floater concept by GustoMSC[7], TLPWT concept by CEHINAV-UPM [8], HYWIND concept byHydro Oil Energy, Norway [9], SPAR-type FOWT concept byYokohama National University [10], three DeepCWind concepts bythe University of Maine [11,12], TLP and SPAR concept by WorcesterPolytechnic Institute [13], and the semi-submersible concept byUniversity of Strathclyde [14], etc. These experiments aimed to*Corresponding author.E-mail addresses: dhk@gnu.ac.kr,dhk0521@gmail.com (D.-H. Kim).Contents lists available at ScienceDirectRenewable Energyjournal homepage: www.elsevier.com/locate/renenehttp://dx.doi.org/10.1016/j.renene.2016.02.0210960-1481/©2016 Elsevier Ltd. All rights reserved.Renewable Energy 92 (2016) 244e261 validate and investigate the global dynamic characteristics ofvarious FOWT types. Although most of the previous experimentalworks showed interesting and practical results, these tests in thescaled-down FOWT models have underlying limitations forsimultaneously satisfying essential scaling laws (Froude and Rey-nolds number scaling). That is, when the Froude scaling law isapplied, the Reynolds number scaling law is not guaranteed andvice versa. An experimental test of a scaled model normally costsmuch more than ordinary numerical analysis, and in numericalanalysis, the influence of a scaled model does not need to beconsidered due to unlimited scale-up capability. Therefore,improving the development and application of a sophisticatednumerical analysis method, which can fully consider the complexmulti-physical phenomena due to aero-hydro-multibody dynamicsof a designed FOWT model under realistic operations, is still animportant matter.Fully coupled aero-hydro-servo-elastic dynamic approaches[15e22] and a simplified aero-hydrodynamic method [23] havebeen used practically to calculate the dynamic responses of anFOWT. The FAST code [15], which has been developed by the Na-tional Renewable Energy Laboratory (NREL), is a comprehensivesimulation tool capable of predicting the aero-hydro-servo-elasticeffects of both land-based and offshore wind turbines. There arealso other numerical analysis codes for simulating FOWT, such asSIMO, HAWC2, 3Dfloat, DeepC, Bladed, etc. [19e21]. However,almost all of the previous analysis codes are based on the con-ventional blade element momentum (BEM) theory to calculateunsteady aerodynamic forces. It is well known that the BEMmethod is based on empirical models with correction factors (e.g.,Glauert correction, skewed wake correction, etc.) [24e27].AsMatha et al. [28] importantly noted, the large low-frequency plat-form motions experienced by FOWT result in flow conditions whichare considerably more complex than those experienced by con-ventional onshore or fixed-bottom offshore wind turbines. Inparticular, there are different interactions between the turbinerotor and its wake, with the rotor in some cases traversing backover its own wake. Sebastian and Lacker [29] also indicated that theBEM method is still lacking and questionable in the prediction ofthe aerodynamic loads of FOWT. Therefore, numerical simulationtools are necessary to accurately and efficiently carry out themultiple physical models of FOWT.Furthermore, the effect of viscous flow in the hydrodynamicsimulation of a FOWT cannot be directly calculated by thepotential-based panel approach [15,30e32]. An additional dampingcoefficient based on experimental test data or the Morisonformulation based on a strip-theory is usually applied for the ac-curate prediction of hydrodynamic loads [30,31]. These approacheshave been well applied by the HydroDyn module of the NREL FASTcode [33]. It is well known that the famous Morison formulation,which is a semi-empirical equation for the inline force on a body inoscillatory flow, is applied to efficiently estimate the wave loads inthe design of offshore structures. However, it also has theoreticallimitations, such as the assumption of uniform flow acceleration,and it is not capable of describing the time-dependent force historyvery well, although the inertial and drag coefficients can be tunedto give the correct extreme values of the force. Thus, these co-efficients essentially need to be produced from the correctedexperimental data or the calculated data when the Morrisonequation is applied. In case of a potential flow solution or a hybridsolution, it is noted that the HydroDyn module still requires third-party software such as WAMIT [34], which is an external programbased on the potential method to generate hydrostatic forces,added mass, added damping, and wave forces for the platformstructure. This code has also been used in the calculations of floatercoefficients required by floater-mooring coupled dynamicprograms, such as TimeFloat [5] and CHARM3D [35]. Some physicalphenomena, such as wave run-up against semi-submersible col-umns [36,37] and viscous flow separation on the floaters, cannot befully captured by the potential-based panel approach and theMorison equation. In addition, the results of such simulations canonly be trusted if these are validated by the full-scale measure-ments of an experiment model [7,38]. On the other hand, the un-steady computational fluid dynamic (CFD) approach can directlyinclude all physical effects (e.g., flow viscosity, hydrostatic, wavediffraction, radiation, wave run-up, and slamming, etc.) of a floatingplatform [39e41]. Several CFD simulations have been performed toinvestigate the hydrodynamic characteristics of FOWT platforms[42e47]. The OC3 Hywind platform model, which has a relativelysimple geometry compared to a semi-submersible platform, wasconsidered and analysed in previous studies [42,43]. However, theheave degree-of-freedom (DOF) motion was constrained to preventnumerical instabilities during the first iterations of the coupledsimulation [42]. Full-system simulations considering the excitationload of wind-wave coupling were successfully performed for theOC4-DeepCwind semi-submersible FOWT model. However, theaerodynamic loads exerted on the wind turbine were applied bythe averages of the forces and the moments of the wind turbine atdifferent wind speeds, and mooring lines were not added to thesemi-submersible support system [44]. There have also beenfundamental CFD application studies using the OpenFOAM solver.These studies were performed to compare the hydrodynamic loadsto the FAST solution. In addition, the authors wanted to understandhow to set the hydrodynamic coefficients for the OC4-DeepCwindsemi-submersible model when an engineering tool such as theFAST code [45e46] was applied. The hydrodynamics of the OC4-DeepCWind semi-submersible model was also studied using acommercial package which was capable of solving multiphasefluids with six degrees-of-freedom (6-DOF) motion [47]. The resultsof the free-decay analysis and regular wave cases were validatedand verified by the Maritime Research Institute Netherlands(MARIN) data and the potential-based panel approach with andwithout the Morison equation, respectively. The influence of dy-namic mooring lines was also carried out by the potential-basedpanel approach.Recently, CFD simulations have been applied to analyse theFOWT model under combined wind-wave conditions [48e50]. Dueto advances in high-performance computing (HPC) technologies,the high fidelity but computationally expensive CFD methods haveshown great potential for the accurate aerodynamic predictions ofwind turbines [48], especially for FOWT simulations. Wake andturbulence effects have also been well predicted. Li et al. [48]proposed an advanced numerical method of coupling aero-dynamics and multi-body dynamic code for OC3 Hywind aero-elastic simulations. Compared to the publicly available simulationresults of OC3, they showed good agreements in the results for theaerodynamic loads and blade tip deflections in both the timedomain and the frequency domain. However, the aerodynamicpower showed a large discrepancy in the case of the couplingregular wave-steady wind condition [43], and this has attractedmany researchers to discuss in the view of aerodynamic analysis[51]. The unsteady aerodynamic effects of the rotating blades,considering the typical pitching, yawing, and surge motions of theplatform, have also been extensively studied using unsteady CFDmethods with a dynamic moving grid technique. The numericalresults for different oscillating amplitudes and frequencies werecompared to those calculated by conventional methods [52e55].The fully coupled aero-hydrodynamic analyses of a semi-submersible FOWT have been conducted using the RIAM-CMENcode [49,50], which contains a CIP-based free surface flow solver,a wind turbine model and a mooring line model. As a result of theseT.T. Tran, D.-H. Kim / Renewable Energy 92 (2016) 244e261 245 works, reasonably good agreements between numerical simula-tions and experiments have been obtained for floating body mo-tions in large waves [49].The main purpose of the present study is to conduct a virtualtest of a real scale 5-MW semi-submersible FOWT using theadvanced numerical methodology of dynamic fluid body interac-tion (DFBI). A full-scale semi-submersible FOWT configuration withrotating blades, hub, nacelle, and tower shapes was utilized in themultiphysical simulation to lead a virtual test simulation (VTS) ofthe FOWT with the 6-DOF floating motions exposed in air andwater. Herein, these major components were considered withoutsurface structure deformations. The volume of fraction (VOF)method, which is a free-surface modelling technique, was appliedto adequately investigate the aero-hydrodynamic behaviours of thesemi-submersible platform with mooring lines. In addition, thepresented DFBI method used a moving overset grid technique toeffectively solve the large movement behaviours of the FOWT,which consist of simultaneous time-dependent motions of thesemi-submersible platform and the rotating blades due to thewind-wave coupling exerted on the FOWT system. Using the pre-sented coupling method, almost all of the complex flow physicsaround the FOWT were involved. The results of the DFBI simulation,considering the free-decay test and regular wave conditions, werevalidated and verified by MARIN data and FAST code. The results ofthis paper showed good agreements with them. In addition, theaerodynamic performance, platform dynamic responses, and themooring line tension, considering fully coupled aero-hydrodynamics including catenary line loads, was calculated andcompared to the current DFBI model and the FAST code.2. Floating wind turbine modelThe floating semi-submersible wind system of the OffshoreCode Comparison Collaboration Continuation (OC4 Phase II) project[30,31,56] was chosen to perform the unsteady simulation underwind-wave coupling. Design parameters of the full-scale OC4DeepCWind semi-submersible platform are presented in Table 1,and detailed data related to the 1/50th scale model can be found ina previous study [30]. In the previous study, the properties of thetop-tower wind turbine were modified to achieve consistency inthe design of the scale-test model. The 1/50th scale test model ofthe OC4 DeepCWind semi-submersible platform performed at theMaritime Research Institute Netherlands is shown in Fig. 1. Here,the OC4 DeepCWind semi-submersible platform is a triangularthree-column design possessing a smaller fourth column centrallylocated to mount the horizontal-axis of the NREL 5-MW baselinewind turbine [57]. The diameters of the base column, upper col-umn, and main column of the full-scale model are 24 m, 12 m, and6.5 m, respectively. The diameters of pontoons and cross-braces are1.6 m. The draft of the OC4 DeepCWind semi-submersible platformis 20 m, and its centre of gravity (CG) is 14.4 m below the sea waterlevel (SWL). The water depth was assumed to be 200 m in thecurrent simulation. In addition, three slack, catenary mooring linesare attached at 14 m below the SWL. They normally provide theprimary global restoring forces to suppress the motions of thesurge, sway, and yaw. The attachment angle between two adjacentmooring lines is 120, and a wave heading angle of 0is assumed tobe parallel to the direction of mooring line 2, which is also parallelto the platform surge direction in this study. Major properties of themooring line system are given in Table 2.3. Numerical modelling3.1. Physics settingsIn the present study, basic flow equations, which are integralforms of the unsteady NaviereStokes equations, have been appliedand solved using the STAR-CCM þsoftware coupled with devel-oped in-house codes. A segregated flow model was applied to solvethe flow equations (one for each component of the velocity and onefor pressure). All simulations employed a semi-implicit method forpressure-linked equations (SIMPLE) solution algorithm in whichthe linkage between the momentum and the continuity equation isachieved with a predictor-corrector approach. A second-order up-wind scheme was set up for the convection terms. In the unsteadysimulation, a second-order central difference scheme was used fortemporal time discretization. In addition, the overset grid tech-nique was utilized to effectively handle complex moving bodieswith relative motions in the simulations of coupled fluidestructureinteractions. In general, the computational domain was composedof a background grid domain and smaller overset refine grid do-mains. Their simulations can be employed by interpolating active,inactive or acceptor cells depending on their positions inside thecomputational domain. Detailed discussions of overset meshmethodologies can be found from a previous study [58].The overall computational process for the DFBI analysis usedhere is conceptually illustrated in Fig. 2. In this study, speciallydeveloped user-field functions were used to conduct the fullsimulation of FOWT. Blade rotation was resolved by a superposedrotation technique, which superimposes a fixed body rotation inaddition to the floating motion. The VOF model in conjunction withthe 6-DOF solver was used to efficiently solve the fluid-induceddynamic motion of FOWT in a multi-phase flow composed of airand water. Here, a kinematic model for the rigid body of motions,including the effect of head wind, was used to solve the 6-DOFequation of motion of the FOWT model by applying the third or-der Runge-Kutta method [59]. The VOF model effectively simulatessurface gravity waves on a light fluid-heavy fluid interface. It is alsoan efficient numerical technique for free-surface modelling basedon the Eulerian method. Therefore, the same set of basic governingequations describing momentum, mass, and energy transport in asingle-phase flow can be solved. The equations are solved for anequivalent fluid whose physical properties are calculated as func-tions of the physical properties of its constituent phases and theirvolume fractions. The discretization of the transport equation in theVOF approach requires a special treatment because it must bebound between zero and unity. The sharpness between immisciblefluids is achieved by limiting the cell-face value to fall within theTable 1Platform gross properties.Depth to platform base below SWL (total draft) 20.0 mElevation to platform top (tower base) above SWL 10.0 mPlatform mass, including ballast 13,444,000 kgDisplacement 13,986.8 m3Center of mass (CM) location below SWL along platform centerline 14.4 mPlatform roll inertia about CM 8.011 109kg m2Platform pitch inertia about CM 8.011 109kg m2Platform Yaw inertia about platform centerline 1.391 1010kg m2T.T. Tran, D.-H. Kim / Renewable Energy 92 (2016) 244e261246 shaded area of a normalized variation diagram (NVD) [60]. Moreinformation about these approaches can be found in a previousstudy [61]. In addition, the wave damping model [62] was imposedto minimize the effect of wave reflections on the far downstreamboundary. Thus, the VOF wave can be damped in the vicinity ofselected boundaries to reduce wave oscillations near thoseboundaries. This damping introduces vertical resistance to thevertical motion of the wave. To consider the effect of the mooringlines, the catenary coupling element technique [63] was used tomodel three mooring lines attached to the FOWT model in the timedomain. The catenary coupling models an elastic, quasi-stationarycatenary (such as a chain or towing rope) hanging between twoendpoints, which is being subjected to its own weight in thegravitational field and can yield the coupled mooring line tensionduring the simulation.3.2. Computational domainThe full-scale OC4 DeepCWind semi-submersible platform wasanalysed using the overset grid method. Fig. 3 shows the oversetgrid topology constructed in this study. The entire hexahedralcomputational domain had the dimensions of 1550 (x) 600(y) 540 (z) m, and it was extended up to 350 m and 1200 m in thenegative (upstream) and positive (downstream) x-directions fromthe wind turbine, respectively. The trimmed cell mesh techniquewas used to generate a high-quality grid for the complex geome-tries of the FOWT model. Refine mesh regions were also createdaround the blades and the supporting platform to capture accu-rately the complex unsteady flow behaviours. Near the wall sur-faces of the entire platform, we generated 10 layers of boundarylayer mesh with a first layer thickness of 0.8 mm and a progressionfactor of 1.2. Fig. 3(b) shows the refine overset grid regions aroundthe full FOWT model, and Fig. 3(c)~(e) illustrate closer views of theplatform, hub, nacelle-tower, and blades. The grid domain wascomposed of three finite volume regions; the number of cells in theFig. 1. DeepCWind semi-submersible model for 1/50th scale tests [61].Table 2Mooring system properties.Number of mooring lines 3Angle between adjacent lines 120Depth to anchors below SWL (water depth) 200 mDepth to fairleads below SWL 14 m 14 mRadius to anchors from platform centerline 837.6 mRadius to fairleads from platform centerline 40.868 mUnstretched mooring line length 835.5 mMooring line diameter 0.0766 mEquivalent mooring line mass density 113.35 kg/mEquivalent mooring line mass in water 108.63 kg/mEquivalent mooring line extensional stiffness 753.6 106N/mFig. 2. Computational procedures for the coupling approach.T.T. Tran, D.-H. Kim / Renewable Energy 92 (2016) 244e261 247 background region, overset blade-hub region, and overset nacelle-tower-platform region were 1,125,510, 5,829,297, and 5,179,429,respectively. Thus, the total number of cells of the full configurationFOWT model was 12,134,236. In the case of wave modelling, muchdenser grids were vertically allocated to accurately capture theunsteady behaviours of free surface waves. The generated grid sizewith respect to the wave direction was about 1/80th of the VOFwavelength, whereas the vertical grid size of the ocean waveregime was set up approximately 1/20th of the wave amplitude. Ithas been shown that this kind of grid generation is enough toaccurately capture the VOF waves, according to the best practiceguidelines [61].3.3. Boundary conditionsFig. 3(a) shows the general definitions of the boundary condi-tions implemented by the current DFBI analysis. The inlet velocitywas set at the upstream inlet boundary, seabed, and top surface ofthe computational domain. Here, the inlet surface of the CFDdomain was filled with water or air according to the VOF parameter.A non-slip wall condition was defined on the geometric surfaces ofthe FOWT model. A symmetric boundary condition was imposed onthe sides of the far field regions of the entire computationaldomain. To avoid wave reflections from the far field boundaries, adamping wave reflection parameter was also defined in this study.For the layer mesh generation of the FOWT wall surfaces, themaximum y þvalue was approximately 250, while the maximumyþvalue on the turbine blade tip surfaces was approximately 130.In this simulation, a hybrid treatment of near-wall modelling(referred by all of the yþwall treatments) attempting to emulatethe high-yþwall treatment for coarse meshes and the low-yþwalltreatment for fine meshes was applied. The desirable characteris-tics for producing reasonable results for meshes of intermediateresolutions was also formulated (i.e., when the wall-cell centroidfalls within the buffer region of the boundary layer). All of theyþwall treatments shared a common need to specify the profiles ofthe mean flow quantities in the near-wall region of the turbulentboundary layers. These profiles are termed wall laws, whichincluded standard wall laws and blended wall laws [61]. In thepresent study, almost all of the computations were carried out on aparallel computing server with an Intel®Xeon®CPU E5-2687W v2@ 3.4 GHz (16 core) processor and 128 GB of RAM. During the CFDcomputations, 10 sub-iterations for each unsteady time step wereimposed to reduce the convergence order and increase the timeaccuracy of the unsteady CFD solutions.All structural dynamic properties of the three blades, tower,nacelle, drivetrain system, and supporting platform werecompletely considered in the DFBI analysis model. This means thatthe current analysis model contained the total mass and the massmoment of inertia (MOI) for both the wind turbine system and thefloating platform. Therefore, the z-coordinate of the centre ofgravity (CG) of the current analysis model was approximatelylocated at 10.5534 m below SWL. In addition, the x-coordinate ofthis CG point was changed to 0.0103 m because of the assembledconfiguration of the rotor, drivetrain system, tower and nacelle onthe centre column of the DeepCWind OC4 platform. In the presentanalysis, the exact CG location of the fully assembled model (windturbine system plus supporting platform) was considered byincluding the downwind distance, lateral distance, and verticaldistance of the CG point from the mean sea level (MSL). However, itis noted that the FAST FOWT model had an underlying limitationfor the location of the total CG point. It can only consider the ver-tical distance of the CG point along the vertical centreline of thetower because of its adopted theory.4. Results and discussion4.1. Comparison of unsteady aerodynamicsIn this section, the numerical convergence test and verificationof the unsteady aerodynamic thrust and power of the NREL 5-MWoffshore wind turbine were conducted for the prescribed periodicmoving conditions with respect to surge translation and pitchrotation. Almost all simulations in this paper were based on theEulerian-Lagrangian approach in order to accurately consider theindependent motions of bodies in the air flow or multiphase flow(including air and water). Due to the lack of experimental data, theaerodynamic analysis implemented by the present couplingapproach was compared with different aerodynamic modelscalculated by the FAST code. To show consistent and objectivemutual verifications, some assumptions were made for all of thecomputations in this paper: (1) The effect of structural deformationwas ignored, thus all of the structural parts of FOWT were assumedto be rigid bodies, (2) the rotating speed of the blade and the bladepitch angle were assumed as constants, (3) the Beddoes-Leishmandynamic stall model was used in the FASTcode simulation, and (4) auniform inflow model was used, unless otherwise stated. Theconvergence test for the grid quality of the NREL 5-MW blade shapewas fundamentally studied in a previous paper by the authors [52].Fig. 4 shows the results of the convergence test of the time-stepsize for the prescribed surge and pitching motions of the platform.The purpose of this simulation was to determine the proper time-step size which can guarantee the convergence of unsteady aero-dynamic performance when the blade rotated at close to the ratedrotation speed. As one can see in the results, the unsteady aero-dynamic computations using the present method tended to giveconverged solutions after one cycle of the blade revolution. Theunsteady calculation of the surge motion was conducted at a fixedcondition for 20 s (s), and then the platform surge motion wasimposed. The unsteady computation of the platform pitch motionwas continuously conducted after obtaining the steady aero-dynamic solution. At the given conditions described in Fig. 4,itisshown that the time-step size ofubdt ¼4 deg/iteration(dt ¼0.056 s) also gave a good convergence in the predictions of theunsteady aerodynamic thrust and power for both the surge andpitch motions, where the termubis the rotation speed of theblades. It is noted that the default thrust output of the FAST codeincluded the contribution of gravitational and inertial forces. Thus,these effects were excluded when we compared the net aero-dynamic thrust calculated by the FAST code. All inflow conditionswere assumed to be uniform wind speeds. A large time-step sizewas effectively applied in the aerodynamic simulation of prescribedpitching motion of FOWT in a previous study [28]. The selectedtime-step size should be small enough to account for the unsteadyaerodynamic interference effects of the FOWT. In addition, thistime-step size should also be able to achieve a good convergence ofthe hydrodynamic simulation considering wave conditions. In thepresent simulation, the time-step size of 0.056 s (ubdt ¼4 deg/iteration) is smaller than the recommended time-step size of a VOFwave simulation [61], which can be evaluated by the numerical ruleof a wave period (12.1 s) per a given value (2.4 80). Thus, a time-step size of 0.056 s was chosen to perform the unsteady couplingsimulations considering wind-wave coupling.Fig. 5 compares the results of different aerodynamic methods,including the present CFD, the FAST code (Ver.7.02), the actuatordisc model (ACD) [64], BEM [64], and the unsteady blade elementmomentum (UBEM) method [54], for the surge motion with an 8 moscillation amplitude and various frequencies. Overall, the resultsfor the cases of the blade alone and the full configuration modelwith tower interference showed good agreements except for theT.T. Tran, D.-H. Kim / Renewable Energy 92 (2016) 244e261248 Fig. 3. The computational mesh domain for the fully configured OC4 DeepCWind semi-submersible wind turbine: (a) Full grid domain, (b) entire turbine model with oversetregions, (c) closed-in view of the platform surface mesh, (d) closed-in view of hub and nacelle, and (e) closed-in view of the blade surface.T.T. Tran, D.-H. Kim / Renewable Energy 92 (2016) 244e261 249 case of the relatively high oscillation frequency of 0.77 rad/s. Thediscrepancy between the present CFD solution and other numericalsolutions increases moderately as the oscillating frequency of theplatform surge increases. It is noted that the results from the pre-sent CFD, the FAST code, and UBEM code were obtained by thepresent computations. The effect of tower interference for the un-steady aerodynamic thrust and power can be clearly observed inFig. 5(d). As clearly seen in the zoomed-in view, variations of theunsteady aerodynamic thrust and aerodynamic power magnitudespredicted by the conventional methods tended to be higher thanthose of the present CFD method when each blade passed thetower. The tower interference model in the BEM approach used apotential flow solution around a cylinder as the base flow field,along with a downwind wake model dependent on the tower dragcoefficient (based on the diameter), and a tower dam model for theupwind influence [25]. The model provided the tower influence onthe local velocity field at all points around the tower. However, theflow interference due to tower movement during the prescribedsurge motion was not considered because the apparent wind ve-locity of the surge motion was added to the velocity term of thefreestream flow. The details of the numerical simulations imple-mented by the FAST code can found in a previous study [54]. Here,FAST-BEM indicates the selection of the ‘EQUIL’option and FAST-GDW indicates the selection of the ‘DYNIN’option as the inflowmodel in the AeroDyn input data of the FAST code. It can beconcluded based on the numerical background that the unsteadyCFD method yielded much more accurate results because it candirectly consider unsteady viscous flow separation, free wake,vortex shedding, and complex interference effects in the rotatingblades, hub, nacelle, and the tower.4.2. Free-decay and RAO analysisBefore conducting the fully coupled aero-hydrodynamic anal-ysis of the full FOWT, numerical validations of the hydrodynamicanalysis of the platform alone were fundamentally conducted.Here, the platform model of the OC4 DeepCWind wind turbine wasconsidered because some comparable experimental data existed.The initial positions of the platform were based on the experi-mental test conditions at the grid generation stage and thenreleased to give a free motion to monitor the hydrodynamicresponse in the free-decay analysis. The additional surge stiffness of7.39 kN caused by the effect of the cable bundle was included, asindicated in a previous study [30]. The free-decay analyses wereconducted using the present DFBI method considering both the 6-DOF platform motions and the constraint effects of the mooringcables. The present approach has the capability of three differentwave models, including the 1st order wave, 5th order wave, andFig. 4. Unsteady convergence test of the time-step size for the prescribed surge and pitching motions of the FOWT platform.T.T. Tran, D.-H. Kim / Renewable Energy 92 (2016) 244e261250 irregular wave. Here, the 1st order wave model was imposed at theinlet boundary conditions. The initial wind and wave speeds wereexcluded in this case. An air density of 1.225 kg/m3was consideredat the sea level, and the water density was assumed to be 1025 kg/m3. The static pressure at the far-field boundary was assumed to be101,325 Pa. The effect of air damping during the free-decay motionwas considered here, although it can be neglected in the free-decayanalysis. In addition, the FAST code (Ver.8.10) was applied to verifythe present DFBI approach. Second-order wave loads and a com-bination of the two solutions of the potential-flow theory and thestrip-theory were chosen to calculate the hydrodynamic forces ofthe platform. The FAST code with the use of the mooring analysisFig. 5. Comparisons of unsteady aerodynamic power and thrust coefficients among different aerodynamic methods for the prescribed platform surge motion with differentoscillating frequencies.T.T. Tran, D.-H. Kim / Renewable Energy 92 (2016) 244e261 251 program (MAP) option was applied to solve the constraint of themooring line system.The experimental test and analysis data were obtained from aprevious study [30]. Drag and dynamic effects were neglected inthe catenary mooring lines. For a semi-submersible platform, thedrag force due to the flow viscosity seemed to be necessary for thecalculation of total hydrodynamic damping. The HydroDyn moduleof the FAST code [66] utilized linear time-domain radiationdamping and omitted nonlinear, second-order wave diffractioneffects, etc. To account for the viscous damping forces and momentsof the platform, a quadratic damping coefficient was required [30].The FAST code adopted an efficient equivalent modelling techniqueto compute drag forces based on Morison s equation even for athree column semi-submersible platform [30]. Thus, it was difficultto accurately consider the transient complex flow effect. The nu-merical accuracy in the presence of strong waves, wake in-terferences, and vortex shedding phenomena in the columns of thesemi-submersible FOWT during dynamic motions on water stillseemed to be unclear. In addition, although the hydrodynamicloads on multiple interconnected members of a substructure can becalculated by the extended HydroDyn module of the FAST code [33],the drag coefficient should be carefully estimated, considering thepresence of a free surface, free ends, and/or multimember ar-rangements [45,46]. As Benitz and her co-author indicated, theMorison s equation implemented by HydroDyn and other engi-neering codes do not capture time varying loads, which occur dueto vortex shedding in both the inline and transverse flow directions.These oscillatory loads could have a significant impact on the fa-tigue of semi-submersible and other similar offshore structures. Onthe contrary, the advanced CFD-based approach can simulta-neously take into account of the full viscous interferences andhigher order wave diffraction and radiation.The results of the numerical convergence test using differenttime-step sizes are presented in Fig. 6. In this DFBI analysis, theshear stress transport (SST) keuturbulence model was used toaccurately analyse the transient flow of a viscous incompressiblefluid around the moving platform. The free-decay responses fortime-step sizes of 0.3 s and 0.5 s are in good agreement. This meansthat an unsteady convergence solution can be achieved using atime-step size of 0.5 s or smaller. The effect of different turbulencemodels on the free-decay dynamic response was also investigatedin order to select a proper model. In Fig. 6, the laminar, SST keu,and Spalart-Allmaras turbulence models showed very goodagreements with each other, while the standard keumodelshowed a much more damped response compared to the others. Itappears that even the disturbed water flows around the platformneeds to be considered as laminar because the average Reynoldsnumbers experienced during the free-decay motions are smallenough. Because the SST keuturbulence model was applied topredict the unsteady aerodynamic behaviours in a previous study[52e55] and the hydrodynamic behaviours in the present study, itwas chosen to further investigate the fully coupled aero-Fig. 5. (continued.)T.T. Tran, D.-H. Kim / Renewable Energy 92 (2016) 244e261252 hydrodynamic analysis of the FOWT using the present DBFImethod.Fig. 7 shows the comparison of different solvers in the free-decay responses with respect to the pitch and surge directions.Overall, good agreements were observed with each other [67].However, there were slight differences regarding the phases of thefree-decay dynamic responses, which can be caused by the exis-tence of nonlinear wave loads [14]. This led to the somewhatdifferent predictions of natural periods, as presented in Table 3.Slight discrepancies of the damping ratios of the free-decay mo-tions among the different numerical simulations were alsoobserved. The maximum differences of each natural period withrespect to the scale model experiment are different depending onthe natural mode direction, but the biggest difference betweenthem is less than 8.4%, except for the results obtained from thenewest version of the FAST code (Ver.8.10). Due to a geometricFig. 6. The effects of the time-step size and different turbulence models on the free-decay response with respect to the surge direction (OC4 DeepCWind model and initial surgecycle amplitude ¼22 m).Fig. 7. Comparison of the free-decay responses among different solvers (OC4 DeepCWind model with initial pitch and surge cycle amplitudes ¼8 deg and 22 m, respectively).T.T. Tran, D.-H. Kim / Renewable Energy 92 (2016) 244e261 253 approximation of the NREL 5-MW turbine [30], small differences inthe natural frequencies are expected for the current FAST simula-tion. The Simo/Riflex code, in which the Morison s equation wasapplied as shown in the previous study [68], obtained higher pre-dictions than the MARIN data and the present DFBI approach. Theslight differences between the model-scale flows and the full-scaleflows may be due to effects caused by the relatively differentboundary layer used in the previous study [69]. The damping ratiospredicted by the present DFBI method showed better agreementswith the experimental data than those from the previous FAST codewith a tuning drag damping model [30] (see Fig. 8).The characteristics of the OC4 DeepCWind platform consideringa regular wave condition were investigated by calculating theresponse amplitude operator (RAO), which is the normalized valueof the amplitude of a periodic response of a field variable by theamplitude of the regular wave [30]. Here in this simulation, thefreestream wind speed was ignored, and a regular wave conditionwas considered. The selected wave condition was from theprevioustest of the 1/50thscale model [30] with an amplitude of 3.79 m anda period of 12.1 s. Fig. 9 shows the comparison of RAO among thescale model test, the FAST code, and the present DFBI method forthe real-scale model. Overall, there are good agreements for thesurge and the heave RAOs; however, a large difference in the RAO ofthe mooring tension lines existed. As a previous study [30] indi-cated, this difference may be caused by the excluding the effect ofthe dynamic mooring line in the simulation. Additionally, it issuspected that possible uncertain scale model effects existed in theexperimental test; further investigation is required to confirm thishypothesis. The results of a previous FAST code analysis [30] wereobtained using an additional tuning process based on the experi-mental test data.4.3. Fully coupled aero-hydrodynamic simulationThe influences of coupling between the unsteady aerodynamicsand hydrodynamics play a significant role in the FOWT simulation.The loads generated by both wind and wave give dominant effectson the motion of FOWT. It is very important that the energy feed-back mechanism between the rotor blade aerodynamics and theplatform hydrodynamics should be fully considered by the mostaccurate methods to correctly predict the performance and thedynamic behaviour of FOWT during the coupling analysis. It shouldbe emphasized that small differences of predicted unsteady aero-dynamic and hydrodynamic forces can induce a large deviation ofthe platform motion because of the tall wind turbine tower and viceversa. Thus, the different predictions of the aerodynamic loads byFig. 9. Comparison of RAO for a regular wave case (wave amplitude ¼3.79 m and waveperiod ¼12.1 s).Table 3Comparison of the rigid-body natural periods between the experimental test and the simulation (unit: s).DOF Experiment 1/50 scale [30] FAST [30] (with tuning drag damping model) Simo/Riflex þTDHMILL [68] FAST (Ver.8.10) (present) Present DFBISurge 107.0 107.0 115.9 120.0 108.1Sway 112.0 113.0 117.3 120.0 114.5Heave 17.5 17.3 17.1 17.1 17.8Roll 26.9 26.7 26.0 25.0 25.3Pitch 26.8 26.8 25.8 25.0 25.2Yaw 82.3 82.7 80.2 75.0 83.3Fig. 8. Comparison of natural periods and damping ratios for the free-decay motions of the OC4 DeepCWind platform.T.T. Tran, D.-H. Kim / Renewable Energy 92 (2016) 244e261254 Fig. 10. Comparison of the platform dynamic responses and the mooring line tensions between the present DFBI method and the FAST code.T.T. Tran, D.-H. Kim / Renewable Energy 92 (2016) 244e261 255 the numerical methods lead to different hydrodynamic loads andmotions.The fully coupled aero-hydrodynamic simulation for the full-scale DeepCWind OC4 model was finally conducted using theadvanced DFBI method and the FAST code (Ver.8.10). The windspeed (V), the wave height (H), and the wave period (T) wereassumed to be 11 m/s, 7.58 m, and 12.1 s, respectively. The time forone revolution of the blades with a rotation speed of 11.89 rpm(u¼1.245 rad/s or 71.34 deg/s) is 5.046 s. The time-step size (dt)of0.056 s utilized here corresponds to a 4.0increment of the azimuthangle of the blade for each global time-step during the time-marching iterations of the present DFBI simulation. A wave head-ing angle of 0was assumed to be parallel to the direction ofmooring line #2, which is also parallel to the platform surge di-rection. In the DFBI computation, the FOWT was released to allow6-DOF motions after running unsteady flow analysis for a durationFig. 11. Comparison of aerodynamic performance, the platform dynamic responses, and the mooring line tension.T.T. Tran, D.-H. Kim / Renewable Energy 92 (2016) 244e261256 equal to two times the wave period. The blade pitch angle of 0andthe rotating speed of the rotor blades were kept constant during thesimulation to give the same boundary conditions for different so-lution methods. The implementation of the blade pitch control isstill under-developed in the present coupling approach. Thesimultaneous consideration of the blade pitch control mechanismleads to the overall complexity and difficulty of an objective com-parison in the present model. Thus, this is beyond the scope of thepresent study and will be demonstrated in a future study. However,it is still obvious that the use of the present computational methodmay increase the reliability of the development of the controllers ofFOWT and sufficiently reduce their risks in the design stage. Allcomputations of FOWT considering the wind-wave coupling wasperformed by a server with an Intel®Xeon®CPU E5-2687W v3@3.1GHz processor and 128 GB of RAM. The parallel processing elapsedCPU time per global time-step with 10 sub-iterations was approx-imately 3.8 min using 32 CPUs. The total number of required globaltime-marching iterations for a simulation run time of 500 s wasapproximately 8930. The total elapsed simulation time using 32CPUs to obtain the results in Fig. 10 was approximately 566 h (~24days).The full configuration FOWT model shown in Fig. 3 was analysedusing the present DFBI method. Fig. 10 shows the comparison re-sults of the dynamic responses and the mooring line tensions ofFOWT for a given operating condition. It seems that there areoverall good correlations of the response trends in Fig. 10(a), (c),and (e), although this simulation contains various multi-physicalphenomena in the flow around the FOWT. The average values ofthe heave and the pitch responses of the FOWT platform seems tobe nearly the same, but the amplitudes of the responses aresomewhat different. In the FASTcode (Ver.8.10) analyses, the hybridsolution of the potential-flow theory and the Morison s equationwas applied instead of using an additional quadratic drag. Thedifferences of the average values of the surge (Fig. 10(a)) and themooring line tensions (Fig. 10(g)) are relatively large. The compar-isons of the remaining responses, such as sway, roll, and yaw mo-tions, are presented in Fig. 10(b), (d), and (f), respectively. It isobserved that these responses show minor effects to the globalmotion of the FOWT platform. The average roll responses calcu-lated from the FAST code are approximately 0.2 deg, while theaverage responses of yaw and sway from the FAST code are nearlyzero. Although the amplitude values are very small compared tomajor responses, such as the surge and heave motions, the relativedifferences between the FAST code and the present DFBI methodbecome very large due to the complex unsteady flow phenomenaaround the FOWT. Further investigations need to be conducted todistinctly confirm this hypothesis in the future.To investigate the characteristics of the aerodynamic perfor-mance and the dynamic responses in detail, enlarged responseresults with respect to time are shown in Fig. 11.Fig. 11(a)~(b)show the comparison of the operational aerodynamic thrust andpower responses due to the 6-DOF motions of the FOWT in anunsteady multiphase flow condition, in which the wind and thewave mutually interact with each other on the free surface waveboundary. Aerodynamic performance responses are sinusoidal,and there is somewhat a difference in the phase angle among theresponses. FAST code analyses were performed using both theconventional BEM theory and generalized dynamic wake (GDW)Fig. 12. Side-view of instantaneous iso-vorticity contours for the full configuration FOWT with 6-DOF motions.T.T. Tran, D.-H. Kim / Renewable Energy 92 (2016) 244e261 257 model. The average aerodynamic thrust predicted by the FAST-GDW is approximately 2.6% higher than that of the FAST-BEM,whereas the oscillation amplitude of aerodynamic thrust differs5.3% between FAST-BEM and FAST-GDW. In Fig. 11(a),threedifferent responses are obtained from the present DFBI approach.First, the aerodynamic load only refers to the computed netaerodynamic thrust, which was obtained by integrating theinstantaneous unsteady pressures on the blade surfaces withoutconsidering structural forces such as gravity and inertial forces. Itis noted here that the thrust force calculated by the FAST codealready include all structural forces. The second DFBI calculation(dash-dot-dot-dash line) was obtained only by considering theadditional thrust component of the gravity force due to the statictilt angle (5 deg) of the rotors. The third DFBI result (short dash)was obtained by considering all thrust components, i.e., bothpurely aerodynamic and structure components, including gravityand acceleration inertial forces due to the tilt and the dynamicplatform pitch angles. Thus, the third result of the DFBI showsgood agreements with those of the FAST code. The oscillationamplitudes of the thrust by the present DFBI approach and FAST-BEM code are approximately 141 kN and 38 kN, respectively. Themaximum thrusts predicted by the FAST-GDW and the FAST-BEMfor this case are approximately 895 kN and 860 kN, respectively.However, it is shown that the current DFBI method presents asignificant variation of the aerodynamic power from 3.95 MW to5.29 MW, whereas the FAST-GDW predicts a much smaller vari-ation of the aerodynamic power from 4.83 MW to 5.20 MW. Thesmallest variation was observed when the FAST-BEM approachwas used. The amplitudes of the oscillating aerodynamic powercomputed by the FAST-BEM and the FAST-GDW are 5.3% and 7.6%,respectively. However, the amplitude of the dynamic powerpredicted by the present DFBI approach significantly increases upto 34.0%.Furthermore, local drops of the thrust responses from thepresent DFBI approach indicate the aerodynamic interference ef-fect between the blade and the tower. This interference effect canbe clearly observed in the power response, as shown in Fig. 11(b).It is interesting that the tower interference effect shows differentcharacteristics (Cycle A ~ Cycle D) in the blade rotation cycleswhile experiencing platform motion. Three times of the passinginterference effects of blades exist per one revolution of the threerotor blades, whereas no tower inference effects are shown in theresults by the FAST code. Although the FAST code is definitely veryuseful, well developed and continuously improved throughenormous efforts by the NREL, it still has some technical aspectswhich are being investigated and improved. Thus, the resultspresented in this paper can be interesting and helpful for theimprovement of the current version of the FAST code. It needs tobe reminded here that the aerodynamic drag loading exerted onthe turbine tower is not yet available for FOWT in the present FASTcode (Ver.8.10).Here, the dynamic behaviours of the FOWT platform will beinvestigated in detail. A comparison of the platform surge re-sponses in a given time range (470 s ~ 490 s) is presented inFig. 11(c). It can be observed that the surge responses haveoscillatory forward and backward motions with respect to theiraverage surge displacements. The oscillating surge amplitude ofthe present DFBI approach is approximately 1.8% larger than thatof the FAST-BEM approach, and the maximum surge response bythe present DFBI approach is approximately 16.0% higher thanthat of the FAST-BEM. The average values of the dynamic surgeresponses by the present DFBI approach and the FAST-BEM areFig. 13. Iso-parametric view of instantaneous iso-vorticity contours for the full configuration FOWT with 6-DOF motions.T.T. Tran, D.-H. Kim / Renewable Energy 92 (2016) 244e261258 9.62 m and 8.02 m, respectively, which indicates a 20.0% differ-ence between them. A comparison of the platform heave re-sponses is shown in Fig. 11(d). The upper and lower bounds andthe amplitudes of the heave responses show small differences,and the average value of the heave response can be considered asnearly zero. It is also observed that the solver option of the FASTcode does not give any effect to the heave response. The dynamicpitch responses of the platform are compared in Fig. 11(e),showing almost the same average pitch angles from the presentDFBI approach (3.24 deg) and the FAST-BEM (3.21 deg). Theoscillatory platform pitch amplitudes computed by the presentDFBI approach and the FAST code are 2.27 deg and 2.44 deg,respectively. The maximum and the amplitude values of the pitchresponse obtained by the FAST-BEM are slightly higher thanthose by the DFBI approach. Nevertheless, the differences in themaximum pitch responses and the amplitudes of the FOWTplatform are 1.4% and 7.5%, respectively.Finally, the tension force comparison of mooring line #2 isshown in Fig. 11(f). Unlike the previous responses, there are largedifferences in this result. First of all, the maximum and averagecable tension obtained by the present DFBI approach is 34.5% and32.2% higher than that by the FAST-BEM, respectively. This indicatesthat much more attention should be placed on designing themooring cable strength. The difference of the amplitude of thetension response is 54.0% for the considered operation condition. Itis reminded here that mooring line #2 is arranged parallel to thewave direction in this study; thus, a higher tension load is inducedhere than in the other mooring lines (see also Fig. 10(d)). The majorreason for this large difference in the mooring cable can be physi-cally explained. Comparing Fig. 10(a) and (g), one can observesimilar dynamic behaviours in these results. This means that thedynamic tension of mooring cable #2 is dominantly influenced bythe platform surge motion, and the magnitude of the cable tensionis affected by the average magnitude of the platform surgedisplacement.Because there are critical discrepancies in some parts of thecomparison results obtained by the different numerical ap-proaches, more discussion on these results will be presentedhere. There has been no uniquely accurate numerical methoduntil now, and every numerical method has its own merits anddemerits. The present study used a quasi-steady mooring cablemodel [47] that is quite similar to that of the FAST code; theprimary differences are in the computational parts of the un-steady aerodynamics and hydrodynamics in the simulationmethods. In view of the numerical analysis theory, there is stillmuch more uncertainty in conventional aerodynamics to calcu-late unsteady aerodynamic forces considering platform motion.There were previous numerical tests to investigate the limitationsof the conventional unsteady aerodynamic analysis methodsconsidering platform motion by the authors [52e55]. Inciden-tally, the FAST-GDW method, which is based on a potential flowsolution to the Laplace s equation, can be modelled with moreflow states and a fully nonlinear implementation to account forthe turbulence and the spatial variation of the inflow [25].However, it should be noted that the wake model used in theFAST-GDW approach was not developed to fully account for theblade tipewake interaction at the interference regimes when awind turbine moved downstream due to surge motion. Whenrapid changes in the blade angle-of-attack occur during forwardmotion, the dynamic-inflow effect is significant. Another weakpoint of the GDW model is that it does not account for wakerotation [25]. In addition, the standard released version of Aer-oDyn does not include enough finite states to provide a goodprediction of induced velocities at the blade tip and root [65].According to Chaney et al. [70], poor results were obtained usingthe GDW model, while Hansen and co-workers obtained goodresults [71,72]. As mentioned by Laino et al. [73], the currentdynamic inflow model using GDW has not been extensivelytested. On the other hand, BEM theory may not include inherentmodelling of the dynamic wake effect, tip losses, and skewed-wake aerodynamics. These effects are addressed by the additionof specific correction factors or empirical models [25,73].Itisnoted again that the current version of the FAST code (Ver.8.10)has not yet included the influences of tower shadow andaerodynamic loads on the tower of FOWT. This feature is stillbeing overhauled by the NREL. Thus, the unsteady CFD-basedapproach can be theoretically considered as the most accuratemethod because it can directly simulate full three-dimensionalunsteady viscous flow fields (air and water), as illustrated inFigs. 12~13.Fig. 12~13 show the instantaneous iso-vorticity and wavecontours generated by both the wind turbine with rotating bladesand the floating platform with 6-DOF motions due to wind andwave loads. One can clearly see that the wind and wave inducedthe motion of the FOWT simultaneously considering the rota-tional motion of the three blades. These instantaneous plots withrespect to time show the existence of strong flow interactions,including the complex generation of wake flows from the bladeroot, hub and tower regions. The bound vortices of the blade areshed downstream from the rotor in individual vortex tubes. Thegenerated turbulence wakes around the tower and nacelleconfiguration are diffused out differently during the motion of theFOWT. Because of the existence of the tower shape, there arestrong unsteady flow interactions near the tower root region be-tween the blade-tip vortex tubes and the tower wakes down-stream. This kind of highly unsteady flow around the fullconfiguration wind turbine cannot be simulated by a semi-empirical model, which is often numerically implemented in aconventional analysis. In addition, time-dependent propagatedwave behaviour and water flow interference with the semi-submersible platform structure can be seen here. Regardingthese figures, wind turbine designers can closely investigate themulti-physical characteristics of the FOWT under realistic oper-ating environments, including complex unsteady flows, aero-dynamic performances, induced dynamic behaviours, criticaldesign loads, and possible dynamic instabilities. Thus, they canimprove the reliability of the FOWT design for various extremedesign conditions.5. ConclusionsIn this study, an advanced computational methodology, usingthe DFBI method with overlapping moving grid techniques, wasproposed for a 5-MW semi-submersible FOWT with elasticmooring lines. A full FOWT configuration with rotating blades, hub,nacelle, and tower shapes was successfully considered in themultiphysical simulation to lead a computational virtual test (CVT)for a FOWT with 6-DOF floating motions exposed in air and water.The validation and verification of the presented DFBI model wasconducted stage by stage, i.e., the unsteady aerodynamics, the free-decay motion and RAO, and the fully coupled aero dynamic-hydrodynamic analyses, by comparing both the experimental testdata and numerical simulations using the NREL FAST code. In thecomparison of the unsteady aerodynamics of the prescribed plat-form surge and pitch motions, aerodynamic thrust and power be-tween the DFBI method and the FAST code generally showed goodagreements except for the high oscillation case. The maximumdifference of the natural periods of the FOWT platform bycomparing with the previous experimental test data was 6.3% in thesway mode. The comparison of the pitching damping ratio of theT.T. Tran, D.-H. Kim / Renewable Energy 92 (2016) 244e261 259 present numerical simulation for different initial cycle amplitudesindicated better agreements with the experimental test data thanthose obtained from previous analyses using the FAST code. In theRAO comparison with the experiment for a regular wave case, therewere relatively good agreements in the surge and heave motions;however, there were large differences in the mooring cable forcessimilar to a previous study. Finally, a comparative study of the fullycoupled aero-hydrodynamic simulation by the present DFBIapproach and the FAST code was conducted in detail. The unsteadyaerodynamic thrust, aerodynamic power, dynamic behaviours, andcable tension forces of the full configuration FOWT were simulta-neously computed and compared with those predicted by the FASTcode. The comparison of the results revealed that there wereoverall good agreements; however, some large discrepancies exis-ted for the same operating condition. The maximum values of thepredicted unsteady aerodynamic thrusts showed less than a 4.1%difference between the DFBI approach and the FASTcode. However,it was observed that there was an approximate four-fold differencein the maximum oscillating amplitude of the aerodynamic powerbecause of the existence of very complex multi-physical charac-teristics of multiphase flow due to the 6-DOF motions of the FOWT.In addition, there was also a 32.2% difference in the predictedaverage tension of the mooring cable. This assessment significantlyindicates the importance of the application of an accurate unsteadyaerodynamic theory in conjunction with hydrodynamic theory forFOWT simulation. Unsteady flow fields around the FOWT weresuccessfully visualized through instantaneous contours so thatcomplex flow characteristics, such as unsteady blade-tip vortexwakes and downstream wave interference effects, could be inves-tigated in an ocean operating condition. The present results alsoindicate that much more attention needs to be placed on using theFAST code in designing the blade pitch controller and the mooringcable strength, etc. Furthermore, the DFBI method may produceinteresting results for conducting mutual validation of the scalemodel test, as well as numerical analyses for various horizontal-axis and vertical-axis FOWT designs. Although the present papersuccessfully showed some interesting simulations, further in-vestigations for various design load conditions (DLCs) of the FOWTmodel need to be conducted in the future.AcknowledgementsThis work was supported by the Human Resources Develop-ment Program (No. 20124030200140) of the Korea Institute ofEnergy Technology Evaluation and Planning (KETEP) Grant fundedby the Korea government Ministry of Trade, Industry and Energy.This research was also supported by a grant (code 12 TechnologyInnovation E09) from the Construction Technology InnovationProgram funded by the Ministry of Land, Infrastructure andTransport(MOLIT) of the Korea government. The authors would liketo express our gratitude to the NREL for the public use of the FASTcode and Dr. J. Jonkman for his useful advice.References[1] European Wind Energy Association report, Deep Water: the Next Step forOffshore Wind Energy, July 2013. www.ewea.org/report/deep-water.[2] European Wind Energy Association report, The European Offshore WindIndustry-key Trends and Statistics 1st Half 2013, July 2013. http://www.ewea.org/fileadmin/files/library/publications/statistics/EWEA_OffshoreStats_July2013.pdf.[3] Maine International Consulting LLC report, Floating Offshore Wind Founda-tions: Industry Consortia and Projects in the United States, May 2013. Europeand Japan.[4] A.M. Viselli, A.J. Goupee, H.J. Dagher, Model test of a 1:8-scale floating windturbine offshore in the gulf of maine, ASME J. Offshore Mech. Arct. Eng. 137 (4)(2015) 041901. http://dx.doi.org/10.1115/1.4030381.[5] D. Roddier, C. Carmelli, A. Weinstin, A floating foundation for offshore windturbines, part I: design basis and qualification process, in: ASME 2009 28thInternational Conference on Ocean, Offshore and Arctic Engineering, Hono-lulu, Hawaii, USA, May 31eJune 5, 2009. http://dx.doi.org/10.1115/OMAE2009-79229.[6] D. Roddier, C. Cermelli, A. Aubault, A. Weinstein, WindFloat: a floating foun-dation for offshore wind turbines, J. Renew. Sustain. Energy 2 (3) (2009)033104. http://dx.doi.org/10.1063/1.3435339.[7] F. Huijs, E.J. Ridder, F. Savenije, Comparison of model tests and coupled sim-ulations for a semi-submersible floating wind turbine, in: Proceedings of theASME 2014 33rd International Conference on Ocean, Offshore and ArcticEngineering, San Francisco, California, USA, June 8-13, 2014.[8] R. Zamora-Rodriguez, P. Gomez-Alonso, J. Amate-Lopez, V. De-Diego-Martin,P. Dinoi, A.N. Simos, A. Souto-Iglesias, Model scale analysis of a TLP floatingoffshore wind turbine, in: ASME 2014 33rd International Conference onOcean, Offshore and Arctic Engineering, San Francisco, California, USA, June8e13, 2014. http://dx.doi.org/10.1115/OMAE2014-24089.[9] B. Skaare, T.D. Hanson, F.G. Nielsen, R. Yttervik, A.M. Hansen, K. Thomesn,T.J. Larsen, Integrated dynamic analysis of floating offshore wind turbines, in:European Wind Energy Conference and Exhibition, 2007, pp. 1e3 (Milan,Italy).[10] Md Nur-E-Mostafa, A Study on Dynamic Response of Spar-type Floating WindTurbine in Waves Using 3-D Green Function Method Considering Rotation ofWindmill Blades, Ph.D. thesis, Yokohama National University, Japan, 2012.[11] A.J. Goupee, B.J. Koo, R.W. Kimball, K.F. Lambrakos, Experimental comparisonof three floating wind turbine concepts, ASME J. Offshore Mech. Arct. Eng. 136(2) (2014) 020906. http://dx.doi.org/10.115/1.4025804.[12] B.J. Koo, A.J. Goupee, R.W. Kimball, K.F. Lambrakos, Model tests for a floatingwind turbine on three different floaters, ASME J. Offshore Mech. Arct. Eng. 136(2) (2014) 020907. http://dx.doi.org/10.1115/OMAE2012-83642.[13] S.K. Naqvi, Scale Model Experiments on Floating Offshore Wind Turbines,Master thesis, Worcester Polytechnic Institute, Worcester, MA 01609, UnitedStates, 2012.[14] L.R. Perez, Design, Testing and Validation of a Scale Model SemisubmersibleOffshore Wind Turbine Under Regular/Irregular Waves and Wind Loads,Master thesis, University of Strathclyde, Glasgow, UK, 2014.[15] J.M. Jonkman, Dynamics of offshore floating wind turbines-model develop-ment and verification, Wind Energy 12 (2009) 459e492. http://dx.doi.org/10.1002/we.347.[16] S. Shim, M.H. Kim, Rotor-floater-tether coupled dynamic analysis of offshorefloating wind turbines, in: Proceedings of the Eighteenth InternationalOffshore and Polar Engineering Conference, July 6-11, 2008. Vancouver,Canada.[17] C. Cermelli, D. Roddier, A. Aubault, WindFloat: a floating foundation foroffshore wind turbines. Part II: hydrodynamics analysis, in: Proceedings of theASME 28th International Conference on Ocean, Offshore and Arctic Engi-neering OMAE, Honolulu, Hawaii, USA, May 31eJune 5, 2009. http://dx.doi.org/10.1063/1.3435339.[18] L. Marshall, Buhl Jr., A. Manjock, A Comparison of wind turbine aeroelasticcodes used for certification, in: 44th AIAA Aerospace Sciences Meeting andExhibit, 2009. Reno, Nevada, USA.[19] A. Crozier, Design and Dynamic Modeling of the Support Structure for a 10MW Offshore Wind Turbine, Master Thesis, Norwegian University of Scienceand Technology, Trondheim, Norway, 2011.[20] A. Cordle, State-of-the-art in Design Tools for Floating Offshore Wind Tur-bines, Deliverable D4.3.5 (WP4: Offshore Foundations and Support Struc-tures), 2010.[21] E.A. Bossanyi, GH Bladed Theory Manual, Garrad Hassan, 2007.[22] L. Li, Z. Hu, J. Wang, Y. Ma, Development and validation of an aero-hydrosimulation code for an offshore floating wind turbine, J. Ocean Wind Energy2 (1) (2015) 1e11.[23] M. Karimirad, T. Moan, A simplified method for coupled analysis of floatingoffshore wind turbines, J. Mar. Struct. 27 (2012) 45e63. http://dx.doi.org/10.1016/j.marstruc.2012.03.003.[24] H. Glauert, The Elements of Aerofoil and Airscrew Theory, Cambridge ScienceClassics, Cambridge University Press, 1983.[25] P.J. Moriarty, C.A. Hansen, AeroDyn Theory Manual. Tech. Rep, NationalRenewable Energy Laboratory, Golden, CO, 2005. NREL/EL-500e36881.[26] M.O.L. Hansen, J.N. Sorensen, S. Voutsinas, N. Sorensen, H.A. Madsen, State ofthe art in wind turbine aerodynamics and aeroelasticity, Prog. Aerosp. Sci. 42(2006) 285e330. http://dx.doi.org/10.1016/j.paerosci.2006.10.002.[27] M.O.L. Hansen, Aerodynamics of Wind Turbines, Science Publishers. Jamesand James Ltd, 2000.[28] D. Matha, A. Cordle, J. Jonkman, R. Pereira, M. Schlipf, Challenges in Simulationof Aerodynamics, Hydrodynamics, and Mooring-line Dynamics of FloatingOffshore Wind Turbines, ISOPE, Maui, Hawaii, USA, 2011.[29] T. Sebastian, M.A. Lackner, Characterization of the unsteady aerodynamics ofoffshore floating wind turbines, Wind Energy 16 (3) (2012) 339e352. http://dx.doi.org/10.1002/we.545.[30] A.J. Coulling, A.J. Goupee, A.N. Robertson, J.M. Jonkman, D.J. Dagher, Validationof a FAST semi-submersible floating wind turbine numerical model withDeepCWind test data, J. Renew. Sustain. Energy 5 (2013) 023116. http://dx.doi.org/10.1063/1.4796197.[31] M. Masciola, A. Robertson, J. Jonkman, A. Coulling, A. Goupee, Assessment ofthe importance of mooring dynamics on the global response of the DeepC-Wind floating semisubmersible offshore wind turbine, in: Proceedings of theT.T. Tran, D.-H. Kim / Renewable Energy 92 (2016) 244e261260 Twenty-third International Offshore and Polar Engineering, June 30-July 5,2013. Anchorage, Alaska, USA.[32] S. Gueydon, S. Weller, Study of a floating foundation for wind turbines,J. Offshore Mech. Arct. Eng. 125 (3) (2013). http://dx.doi.org/10.1115/1.4024271.[33] J.M. Jonkman, A.N. Robertson, G.J. Hayman, HydroDyn User s Guide andTheory Manual, NREL, 2014. Available online. Assessed on Dec 2014: https://nwtc.nrel.gov/system/files/HydroDyn_Manual_0.pdf.[34] C. Lee, J. Newman, M. Kim, D. Yue, The computation of second-order waveloads, in: Proceedings of the 10th International Conference on Offshore Me-chanics and Arctic Engineering, 1991, pp. 113e123. Stavanger, Norway.[35] Y.H. Bae, M.H. Kim, Rotor-floater-tether coupled dynamics including second-order sumefrequency wave loads for a mono-column-TLP-type FOWT(floating offshore wind turbine), Ocean. Eng. 61 (2013) 109e122. http://dx.doi.org/10.1016/j.oceaneng.2013.01.010.[36] T. Shan, J. Yang, X. Li, L. Xiao, Experimental investigation on wave run-upcharacteristics along columns and air gap response of semi-submersibleplatform, J. Hydrodyn. Ser. B 23 (5) (2011) 625e636. http://dx.doi.org/10.1016/S1001-6058(10)60158-8.[37] S. Diaconu, Wave run-up simulations and comparison with experimental dataon a semisubmersible, Mech. Test. Diagn. 1 (2013) 32e37.[38] F. Huijs, E.J. Ridder, F. Savenije, Concept design verification of a semi-submersible floating wind turbine using coupled simulations, Energy Proce-dia 53 (2014) 2e12. http://dx.doi.org/10.1016/j.egypro.2014.07.210.[39] D.E. Christensen, H. Bredmose, A.E. Hansen, Extreme Wave Forces and WaveRun-up on Offshore Wind Turbine Foundations, Copenhagen Offshore Wind,2005.[40] J. Ramirez, P. Frigaard, T.L. Andersen, D.E. Christensen, Numerical modeling ofwave run-up: regular waves, in: Proceedings of the Twenty-first InternationalOffshore and Polar Engineering, June 19-24, 2011. Maui, Hawaii, USA.[41] C.H. Chen, K. Yu, CFD simulations of waveecurrent-body interactionsincluding green water and wet deck slamming, Comput. Fluids 38 (5) (2008)970e980. http://dx.doi.org/10.1016/j.compfluid.2008.01.026.[42] F. Beyer, M. Arnold, P.W. Cheng, Analysis of floating offshore wind turbinehydrodynamics using coupled CFD and multibody methods, in: Proceedings ofthe Twenty-third International Offshore and Polar Engineering, June 30-July 5,2013. Anchorage, Alaska, USA.[43] S. Quallen, T. Xing, P. Carrica, Y. Li, J. Xu, CFD simulation of a floating offshorewind turbine system using a quasi-static crowfoot mooring-line model,J. Ocean Wind Energy 1 (3) (2014) 143e152.[44] W. Zhao, D. Wan, Numerical study of interactions between phase II of OC4wind turbine and its semi-submersible floating support system, J. Ocean WindEnergy 2 (1) (2015) 45e53.[45] M.A. Benitz, D.P. Schmidt, M.A. Lackner, G.M. Stewart, J. Jonkman,A. Robertson, Comparison of hydrodynamic load predictions between engi-neering models and computational fluid dynamics for the OC4-DeepCWindsemi-submersible, in: ASME 2014 33rd International Conference on Ocean,Offshore and Arctic Engineering, San Francisco, California, USA, June 8-13,2014.[46] M.A. Benitz, D.P. Schmidt, M.A. Lackner, G.M. Stewart, J. Jonkman,A. Robertson, Validation of Hydrodynamic Load Models Using CFD for theOC4-DeepCWind Semisubmersible, 2015. NREL/CP-5000e63751.[47] T.T. Tran, D.H. Kim, The coupled dynamic response computation for a semi-submersible platform of floating offshore wind turbine, J. Wind Eng. Ind.Aerodyn 147 (2015) 104e119. http://dx.doi.org/10.1016/j.jweia.2015.09.016.[48] Y. Li, A.M. Castro, T. Sinokrot, W. Prescott, P.M. Carrica, Coupled multi-bodydynamics and CFD for wind turbine simulation including explicit wind tur-bulence, Renew. Energy 76 (2015) 338e361. http://dx.doi.org/10.1016/j.renene.2014.11.014.[49] C. Liu, C. Hu, CFD simulation of a floating wind turbine platform in rough seaconditions, in: Proceedings of the Twenty-fourth International Ocean andPolar Engineering Conference, June 15-20, 2014. Busan, Korea.[50] C. Hu, M. Sueyoshi, C. Liu, Y. Liu, Hydrodynamic analysis of a semi-submersible-type floating wind turbine, J. Ocean Wind Energy 1 (4) (2014)202e208.[51] D. Matha, T.A. Nygaard, Discuss. J. Ocean Wind Energy 1 (3) (2014) 185e188.[52] T.T. Tran, D.H. Kim, Computational fluid dynamic analysis of a floatingoffshore wind turbine experiencing platform pitching motion, Energies 7(2014) 5011e5026. http://dx.doi.org/10.3390/en/7085011.[53] T.T. Tran, D.H. Kim, The platform pitching motion of floating offshore windturbine: a preliminary unsteady aerodynamic analysis, J. Wind Eng. Ind.Aerodyn. 142 (2015) 65e81. http://dx.doi.org/10.1016/j.jweia.2015.03.009.[54] T.T. Tran, D.H. Kim, B.H. Nguyen, Aerodynamic interference effect of a hugewind turbine blade with periodic surge motions using overset grid-basedcomputational fluid dynamics approach, J. Sol. Energy Eng. 137 (6) (2015)061003. http://dx.doi.org/10.1115/1.4031184.[55] T.T. Tran, D.H. Kim, The aerodynamic interference effects of a floating offshorewind turbine experiencing platform pitching and yawing motions, J. Mech.Sci. Technol. 29 (2) (2015) 549e561. http://dx.doi.org/10.1007/s12206-015-0115-0.[56] A. Robertson, J. Jonkman, M. Masciola, H. Song, A. Goupee, A. Coulling, C. Luan,Definition of the Semisubmersible Floating System for Phase II of OC4,Technical Report NREL/TP-5000e60601, 2014.[57] J. Jonkman, S. Butterfield, W. Musial, G. Scott, Definition of a 5MW ReferenceWind Turbine for Offshore System Development, Technical Report NREL/TP-500e38060, 2009.[58] H. Hadzic, Development and Application of Finite Volume Method for theComputation of Flows Around Moving Bodies on Unstructured, OverlappingGrids, Ph.D. Thesis, Technical University Hamburg-Harburg, Germany, 2005.[59] H.J. Mørch, S. Enger, M. Peric, E. Schreck, Simulation of lifeboat launchingunder storm conditions, in: 6th International Conference on CFD in Oil Gas,Metallurgical and Process Industries, SINTEF/NTNU, Trondheim, Norway,2008.[60] S. Muzaferija, M. Peric, Computation of free surface flows using interface-tracking and interface-capturing methods, Chap. 2, in: O. Mahrenholtz,M. Markiewicz (Eds.), Nonlinear Water Wave Interaction, ComputationalMechanics Publications, WIT Press, Southampton, 1999.[61] B. Leonard, Bounded higher-order upwind multidimensional finite volumeconvection-diffusion algorithms, Adv. Numer. Heat Transf. 1 (1997) 1e57.[62] J. Choi, B.Y. Sung, Numerical simulations using momentum source wave-maker applied to RANS equation model, Coast. Eng. 56 (10) (2009)1043e1060. http://dx.doi.org/10.1016/j.coastaleng.2009.06.009.[63] User manual, STAR-CCMþVer.9.06, CD-adapco, 2014.[64] J.B. de Vaal, M.O.L. Hansen, T. Moan, Effect of wind turbine surge motion onrotor thrust and induced velocity, Wind Energy 17 (1) (2014) 105e121. http://dx.doi.org/10.1002/we.1562.[65] F. Meng, P. Masarati, M. Tooren, Free/open source multibody and aerodynamicsoftware for aeroelastic analysis of wind turbines, in: 47th AIAA AerospaceSciences Meeting Including the New Horizons Forum and Aerospace Exposi-tion, January 5 e8, 2009 (Orlando, Florida).[66] J. Jonkman, Definition of the Floating System for Phase IV of OC3, TechnicalReport NREL/TP-500e47535, 2010.[67] SIMPACK News Edition July 2013, Accessed on 22 May 2014: http://www.simpack.com/fileadmin/simpack/doc/newsletter/2013/Jul_2013/SN-2013-Jul_all_full-issue_Spreads_HQ150dpi.pdf.[68] C. Luan, Z. Gao, T. Moan, Modelling and Analysis of a Semi-Submersible WindTurbine with a Central Tower with Emphasis on the Brace System, in: Pro-ceedings of the ASME 2013 32nd International Conference on Ocean, Offshoreand Arctic Engineering, June 9-14, 2013. Nantes, France, http://dx.doi.org/10.1115/OMAE2013-10408.[69] K. Hochkirch, B. Mallol, On the importance of full-scale CFD simulations forships, in: The 12th International Conference on Computer Applications andInformation Technology in the Maritime Industries (COMPIT 2013), April 15 e17, 2013. Cortona, Italy.[70] E. Chaney, A. Eggers, P. Moriarty, W. Holley, Skewed wake induction effects onthrust distribution on small wind turbine design, ASME J. Sol. Energy Eng. 123(2001) 290e295. http://dx.doi.org/10.1115/1.1410109.[71] D. Davis, A. Hansen, Operation and Load Measurements During Extreme WindEvents for a Southwest Windpower Whisper H40. Windpower, AmericanWind Energy Association, 2002.[72] D. Corbus, A. Hansen, J. Minnema, Effect of blade torsion effects on modelingresults for the small wind research turbine (SWRT), in: The 44th AIAA Aero-space Sciences Meeting, American Institute of Aeronautics and Astronautics,2006.[73] D.J. Laino, A. Craig Hansen, AeroDyn User Guide, Version 12.50, WindwardEngineering, 2002.T.T. Tran, D.-H. Kim / Renewable Energy 92 (2016) 244e261 261Supplementary resource (1)10kW FOWT Picture (Production Level)DataJune 2018Dong-Hyun KimDownloadCitations (50)References (82)... Such methods are also applied to floating wind turbines under various platform motion problems. [40][41][42][43][44][45][46] Unsteady CFD simulations based on an advanced overset moving grid method were adopted by Tran et al. [41][42][43][44][45] for the study of floating wind turbines under surge and pitch motions. The results indicated a strong interaction between the blade tip vortices and the rotating blades causing unsteady aerodynamic loads on the blades during oscillating platform motions. ...... Such methods are also applied to floating wind turbines under various platform motion problems. [40][41][42][43][44][45][46] Unsteady CFD simulations based on an advanced overset moving grid method were adopted by Tran et al. [41][42][43][44][45] for the study of floating wind turbines under surge and pitch motions. The results indicated a strong interaction between the blade tip vortices and the rotating blades causing unsteady aerodynamic loads on the blades during oscillating platform motions. ...Load control and unsteady aerodynamics for floating wind turbinesArticleFull-text availableFeb 2021P I MECH ENG A-J POW Xin ShenZhaohui Du Xiao cheng ZhuUnlike fixed-base offshore wind turbine, the soft floating platform introduces 6 more degrees of freedom of motions to the floating offshore wind turbine. This may cause much more complex inflow environment to the wind turbine rotors compared with fixed-base wind turbine. The wind seen locally on the blade changes due to the motions of the floating wind turbine platform which has a direct impact on the aerodynamic condition on the blade such as the angle of attack and the inflow velocity. Such unsteady aerodynamic effects may lead to high fluctuation of the loads and power output. The present work aims to study the high unsteady aerodynamic performance of the floating wind turbine under platform surge motion. The unsteady aerodynamic loads are predicted with a lifting surface method with a free wake model. A preview predict control algorithm is used as the pitch control strategy. A full scale U.S. Department of Energy s National Renewable Energy Laboratory (NREL) 5 MW floating wind turbine is chosen as the subject of the present study. The unsteady aerodynamic performance and instabilities have been discussed in detail under prescribed platform surge motions with different control targets. Both minimizing the power output and rotor thrust fluctuation are set as the control objectives respectively. The theory analysis and the simulation results indicate that the blade pitch control can effectively alleviate the variation of the rotor thrust under platform surge motions. Larger amplitude of the variation of blade pitch is needed to alleviate the variation of the wind turbine power and this leads to high rotor thrust fluctuation. It is also shown that negative damping can be achieved during the blade pitch control process and may lead the floating platform wind turbine system into unstable condition.ViewShow abstract... The additional entrainment velocities due to the relatively large FOWT tower motion results in higher and different unsteady aerodynamics of FOWT rotors, and this introduces significant uncertainty in the FOWT analysis making use of legacy low-fidelity aerodynamic codes. To investigate and, ultimately, reduce this uncertainty, several research groups began cross-comparisons of FOWT aerodynamics using lowfidelity codes and high-fidelity Navier-Stokes (NS) Computational Fluid Dynamics (CFD) [4], and these analyses are increasingly becoming of a multi-disciplinary nature [5,6], adopting CFD to resolve both rotor aerodynamics and floater hydrodynamics in a fully coupled fashion. ...... The resonant pitching frequency during operation is expected to be higher, although still below 0.1 Hz, due to metocean condition-dependent wind, hydrodynamic and mooring line damping [17]. The pitching frequency of 0.1 Hz has been used in other recent FOWT CFD studies [3,4,5], and, to enable cross-comparison of this study s results with other in the literature, it has been decided to continue using this same frequency. ...Cross-comparative analysis of loads and power of pitching floating offshore wind turbine rotors using frequency-domain Navier-Stokes CFD and blade element momentum theoryArticleFull-text availableSep 2020J Phys Conf Andrea Ortolani Giacomo PersicoJ. DrofelnikM.S. CampobassoView... On the other hand, the Reynolds similitude is usually achieved by using the same lift and drag coefficients of blade or ensuring that the measured Reynolds number is basically at the same magnitude [42,43]. While the computational models reported for offshore HAWT in recent years are large-scale wind turbines [44,45], the results from the upscaled model should be the same as that of the small-scale wind turbine if the scaling laws are consistent [43]. But it should also be acknowledged that the ratio of the pitching period to the revolution period of a small-scale wind turbine is not exactly the same as that of the large-scale wind turbine. ...Aerodynamic performance assessment of ϕ-type vertical axis wind turbine under pitch motionArticleFeb 2021ENERGY Jie SuYu Li Yaoran Chen Yan BaoThe floating vertical axis wind turbine (VAWT) is considered as a competitive device in the utilization of offshore wind energy. However, the platform pitch motion would affect its aerodynamic behavior. In this paper, the aerodynamic performance of a floating ϕ-type VAWT under pitch motion is investigated by using the Improved Delayed Detached Eddy Simulation SST k−ω turbulence model. After verifying the feasibility of the numerical model, the effects of pitch motion amplitude and period on the aerodynamic characteristics were evaluated, and the impacts of these observations were elucidated. The results showed that the averaged net power coefficient increment of about 1.5% to 15% could be obtained under platform pitch motions, and the fluctuation of aerodynamic loads was found to increase. Besides, the pitch motion pattern could be regarded as the combination of surge and heave motions, which explained the similarity of their effects on the wind turbine aerodynamics. Furthermore, it was found that the frequency of the peak torque coefficient would change under different periods of pitch motion, which should be noticed in the design of floating wind turbine. Finally, it was concluded that the current study provided additional information about the effect of pitch motion on wind turbine aerodynamics.ViewShow abstract... Del Águila Ferrandis et al. computed the response amplitude operators of an unconventional FOWT platform in regular waves using CFD and compared the results to both frequency-and time-domain boundary element methods [16]. More recent works incorporated fully coupled aero-and hydrodynamics to investigate the effect of wave-induced motion on the performance and operation of the mounted wind turbines [17][18][19][20]. ...Uncertainty Assessment of CFD Investigation of the Nonlinear Difference-Frequency Wave Loads on a Semisubmersible FOWT PlatformArticleFull-text availableDec 2020Lu WangAmy RobertsonJason Jonkman Yi-Hsiang YuCurrent mid-fidelity modeling approaches for floating offshore wind turbines (FOWTs) have been found to underpredict the nonlinear, low-frequency wave excitation and the response of semisubmersible FOWTs. To examine the cause of this underprediction, the OC6 project is using computational fluid dynamics (CFD) tools to investigate the wave loads on the OC5-DeepCwind semisubmersible, with a focus on the nonlinear difference-frequency excitation. This paper focuses on assessing the uncertainty of the CFD predictions from simulations of the semisubmersible in a fixed condition under bichromatic wave loading and on establishing confidence in the results for use in improving mid-fidelity models. The uncertainty for the nonlinear wave excitation is found to be acceptable but larger than that for the wave-frequency excitation, with the spatial discretization error being the dominant contributor. Further, unwanted free waves at the difference frequency have been identified in the CFD solution. A wave-splitting and wave load-correction procedure are presented to remove the contamination from the free waves in the results. A preliminary comparison to second-order potential-flow theory shows that the CFD model predicted significantly higher difference-frequency wave excitations, especially in surge, suggesting that the CFD results can be used to better calibrate the mid-fidelity tools.ViewShow abstract... The Hywind floater, which is a spar type floater, was evaluated using CFD by Beyer et al. [5] and Quallen et al. [6]. The DeepCwind semisubmersible floater adopted by OC5 and OC6 was simulated by Tran and Kim [7] under free decay motions and several regular wave conditions. Liu et al. [8] performed simulations of the same floater under rough sea conditions. ...IOWTC2020-3558 VERIFICATION STUDY OF CFD SIMULATION OF SEMI-SUBMERSIBLE FLOATING OFFSHORE WIND TURBINE UNDER REGULAR WAVESConference PaperFeb 2021 Yu WangHamn-Ching Chen Guilherme Vaz Simon MewesUtilization of Computational Fluid Dynamics (CFD) codes to perform hydrodynamic analysis of Floating Offshore Wind Turbines (FOWTs) is increasing recently. However, verification studies of the simulations that quantifying numerical uncertainties and permitting a detailed validation in a next phase is often disregarded. In this work, a verification study of CFD simulations of a semi-submersible FOWT design under regular waves is performed. To accomplish this goal, Response Amplitude Operators (RAOs) are derived from the computational results of the heave, surge and pitch motions. Four grids with different grid sizes with a constant refinement ratio are generated for verification of spatial convergence. Three different time increments are paired with each grid for verification of temporal convergence. The verification study is performed by estimation of the numerical errors and uncertainties using procedures proposed by Eca and Hoekstra [1].ViewShow abstractStability of Floating Wind Turbine WakesArticleFull-text availableMay 2021J Phys ConfV G KleineL FranceschiniB S CarmoDan S. HenningsonViewFloating Spar-Type Offshore Wind Turbine Hydrodynamic Response Characterisation: a Computational Cost Aware ApproachArticleApr 2021 Andrea Coraddu Luca OnetoMiltos Kalikatzarakis Maurizio ColluViewNonlinear effects and dynamic coupling of floating offshore wind turbines using geometrically-exact blades and momentum-based methodsArticleFull-text availableJun 2021OCEAN ENG Shanran Tang Bert SweetmanJu GaoNonlinear blade effects and overall dynamic coupling of floating offshore wind turbines (FOWTs) are investigated using the latest developments in nonlinear beam theory and multibody dynamics. An aero-hydro-servo-elastic coupled model is developed combining momentum-based beam theory (MBBT) and the momentum cloud method (MCM) based on a multi-time-scale coupling scheme, in which geometrically-exact blades are coupled with a highly-compliant floating platform design that allows large angular motions. Comprehensive simulations are performed and the results are analyzed to demonstrate the importance of nonlinear effects and dynamic coupling of FOWTs. Nonlinear blade dynamics are shown to be critical in assessments of turbine performance and blade fatigue. The highly-compliant floater design is found to be feasible in realistic operating conditions. A strong one-way dynamic coupling is observed from platform motions to blade vibrations. Axi-asymmetric response of the blades is found to induce significant imbalanced aerodynamic loading onto the turbine and the platform.ViewShow abstractMulti-phase simulation of semi-submersible platform with pencil column using CFDArticleFull-text availableDec 2020J Phys ConfK RohitV SharanM RavishankarR. HarishViewComputational Fluid Dynamics Analysis of Floating Offshore Wind Turbines in Severe Pitching ConditionsArticleDec 2020J ENG GAS TURB POWER Andrea Ortolani Giacomo Persico Jernej Drofelnik Sergio CampobassoThe unsteady aerodynamics of floating wind turbines is more complex than that of fixed-bottom turbines, and the uncertainty of low-fidelity predictions is higher for floating turbines. Navier–Stokes computational fluid dynamics (CFD) can improve the understanding of rotor and wake aerodynamics of floating turbines, and help improving lower-fidelity models. Here, the flow field of the NREL 5 MW rotor with fixed tower, and subjected to prescribed harmonic pitching past the tower base are investigated using blade-resolved CFD compressible flow COSA simulations and incompressible flow FLUENT simulations. CFD results are also compared to predictions of the FAST wind turbine code, which uses blade element momentum theory (BEMT). The selected rotor pitching parameters correspond to an extreme regime unlikely to occur without faults of the turbine safety system, and thus relevant to extreme aerodynamic load analysis. The rotor power and loads in fixed-tower mode predicted by both CFD codes and BEMT are in very good agreement. For the floating turbine, all predicted periodic profiles of rotor power and thrust are qualitatively similar, but the power peaks of both CFD predictions are significantly higher than those of BEMT. Moreover, cross-comparisons of the COSA and FLUENT predictions of blade static pressure also highlight significant compressible flow effects on rotor power and loads. The CFD analyses of the downstream rotor flow also reveal wake features unique to pitching turbines, primarily the space- and time-dependence of the wake generation strength, highlighted by intermittency of the tip vortex shedding.ViewShow abstractShow moreValidation of Hydrodynamic Load Models Using CFD for the OC4-DeepCwind SemisubmersibleConference PaperFull-text availableMay 2015Maija A. Benitz David P. Schmidt Matthew Lackner Amy RobertsonComputational fluid dynamics (CFD) simulations were carried out on the OC4-DeepCwind semisubmersible to obtain a better understanding of how to set hydrodynamic coefficients for the structure when using an engineering tool such as FAST to model the system. This study focussed on the drag behavior and the effects of the free surface, free ends and multimember arrangement of the semisubmersible structure. These effects are investigated through code-to-code comparisons and flow visualizations. The implications on mean load predictions from engineering tools are addressed. This study suggests that a variety of geometric factors should be considered when selecting drag coefficients. Furthermore, CFD simulations demonstrate large time-varying loads caused by vortex shedding that FAST’s hydrodynamic module, HydroDyn, does not model. The implications of these oscillatory loads on the fatigue life needs to be addressed.Copyright © 2015 by ASME Country-Specific Mortality and Growth Failure in Infancy and Yound Children and Association With Material StatureUse interactive graphics and maps to view and sort country-specific infant and early dhildhood mortality and growth failure data and their association with maternalViewShow abstractComparison of Hydrodynamic Load Predictions Between Reduced Order Engineering Models and Computational Fluid Dynamics for the OC4-DeepCwind Semi-SubmersibleArticleFull-text availableJun 2014Maija A. Benitz David P. SchmidtMatthew Lackner Amy RobertsonHydrodynamic loads on the platforms of floating offshore wind turbines are often predicted with computer-aided engineering tools that employ Morison s equation and/or potentialflow theory. This work compares results from one such tool, FAST, the National Renewable Energy Laboratory s wind turbine computer-aided engineering tool, and the high-fidelity computational fluid dynamics (CFD) package, OpenFOAM, for the OC4- DeepCwind semi-submersible analyzed in the International Energy Agency Wind Task 30 project. Load predictions from Hydro- Dyn, the offshore hydrodynamics module of FAST, are compared with results from OpenFOAM. HydroDyn uses a combination of Morison s equation and potential-flow theory to predict the hydrodynamic forces on the structure, at a small computational cost compared to CFD. The implications of the assumptions in Hydro- Dyn are evaluated based on this code-to-code comparison.ViewShow abstractThe Elements of Aerofoil and Airscrew TheoryBookJun 1983H. GlauertViewBounded higher-order upwind multidimensional finite-volume convection-diffusion algorithmsArticleJan 1997B.P. LeonardViewComputation of free surface flows using interface-tracking and interface-capturing methodsArticleJan 1998 Samir Muzaferija Milovan PericViewComparison of Model Tests and Coupled Simulations for a Semi-Submersible Floating Wind TurbineConference PaperJun 2014Fons HuijsErik-Jan de Ridder Feike SavenijeThe GustoMSC Tri-Floater is a slender and robust three-column semi-submersible supporting an offshore wind turbine. Model tests were performed for a Tri-Floater equipped with an operational wind turbine and mooring system exposed to wind and waves in the offshore basin at MARIN. A high quality wind setup and special low Reynolds number blades were used, aiming at delivering the Froude scaled thrust. The base scope of experiments was performed with fixed blade pitch angle and generator speed. Some of the experiments were repeated with active blade pitch and generator torque control using a dedicated algorithm developed by ECN. The experiments covered typical operational and survival design conditions. Numerical simulations for the same wave and wind conditions were performed using ANSYS-AQWA coupled with PHATAS. The paper describes the setup and results of both the model tests and the simulations.From the comparison of the numerical and experimental results, it is concluded that coupled aero-hydro-servo-elastic simulations can be used to predict the response of the floating offshore wind turbine to a sufficiently accurate level for design purposes. Furthermore, it is shown that the Tri-Floater motion response is very favorable and that the nacelle accelerations, air gap and mooring loads comply with the design requirements.Copyright © 2014 by ASME Country-Specific Mortality and Growth Failure in Infancy and Yound Children and Association With Material StatureUse interactive graphics and maps to view and sort country-specific infant and early dhildhood mortality and growth failure data and their association with maternalViewShow abstractWave run-up simulations and comparison with experimental data on a semisubmersibleArticleJan 2013S. DiaconuViewBounded high-resolution upwind multidimensional finite-volume convection-diffusion algorithmsArticleJan 1991B.P. LeonardViewCFD Simulation of a Floating Wind Turbine Platform in Rough Sea ConditionsArticleJan 2014 Cheng Liu Changhong HuIn this paper new improvements are presented on development of CFD model for simulation of a floating offshore wind turbine (FOWT) platform in large waves and strong winds. The CFD model is a Cartesian grid method for multi-phase free surface flows which contains a CIP based free surface flow solver, a wind turbine model and a mooring line model. A comparison computation against an experiment with a semi-submersible type FOWT model has been performed for validation of the CFD model. Reasonably good agreement between the numerical simulation and the experiment has been obtained on floating body motions in large waves. Copyright © 2014 by the International Society of Offshore and Polar Engineers (ISOPE).ViewShow abstractAssessment of the importance of mooring dynamics on the global response of the DeepCwind floating semisubmersible offshore wind turbineArticleJan 2013Marco Masciola Amy RobertsonJason Jonkman Andrew GoupeeThis paper studies the influence of mooring line dynamics on the response of a coupled floating offshore wind turbine against an equivalent uncoupled model. The semisubmersible modeled in this paper is based on a design developed by the DeepCwind program and uses the National Renewable Energy Laboratorys (NRELs) 5-megawatt (MW) baseline wind turbine to represent the tower, nacelle, and blade properties. The uncoupled model was formed using FAST, an open-source program that models the wind turbine aerodynamics, control, motion, tower/blade flexure, and wave forces, but with the mooring line forces treated using a quasi-static approximation. In contrast, the coupled model was enabled by pairing FAST with OrcaFlex. OrcaFlex replaces FASTs wave force and quasi-static cable model with an equivalent subsea fluid-structure representation and a lumped-mass cable system to capture the mooring line dynamics. This analysis revealed that an uncoupled model using the quasi-static mooring approximation can underestimate peak mooring line loads versus a coupled model using a dynamic mooring line. Copyright © 2013 by the International Society of Offshore and Polar Engineers (ISOPE).ViewShow abstractShow moreAdvertisementRecommendationsDiscover moreProjectStatic and Dynamic Loads Analysis of WIG Ship Dong-Hyun KimStatic and dynamic loads analyses of designed WIG ship models have been conducted using advanced CFD-MBD coupling approach. View projectProjectFE Modeling, Structural, Vibration, and Aeroelastic Analysis of a Whole Configuration Composite WIG Ship Dong-Hyun Kim- Complete static and dynamic FE modeling of a full configuration composite WIG ship.- Structral strength analysis of a designed composite WIG ship model based on design loads calculated by 2-way coupled CFD-MDB approach- Buckling safety analysis analysis of the the designed composite WIG ship model - Natural vibration analyses of the designed composite WIG ship model - Transient buffeting response analysis of the T-tail due to the downstream wake of the propeller.- Aeroelastic analysis of the the designed composite WIG ship model using DLM and CFD aerodynamics- Bird strike analysis of the conopy, wing, and tail wing- Design certification of the designed WIG ship by KR (Korean Register) ... [more]View projectProjectPrototype development and realistic ocean test of commercializing 10~20 kW class FOWT Dong-Hyun Kim- Fundamental design of a 10~20 kW FOWT model- Fundamental moorling lines and sea-bed anchor designs- Conventional and advanced desing load analyses of the designed FOWT model- Detailed moorli ng lines and sea-bed anchor designs- Detailed structural desing of the FOWT model- Construction of 3D FE model- Structural, vibration, and fatigue analyses- Structral design improvement- Construction of manufacturing drawings- Prototype manufacturing for structural parts (Under the maximum size limit requirements of parts by the ground transformation laws)- Design of remote performance and stability monitoring system (wireless) including GPS and CCTV.- System assembly of the full system at a harbor.- 1st test on the harbor ground - Ocean installation with mooring lines- 2nd test on the windy sea near harbor (shor term test)- Test site approval by a local government- Remote moving and ocean installation with mooring lines- 3rd test on the sea near remote island (mid and long term test) (under extream wave and wind conditons including typhoon)- Certifications
