Delegates are invited to meet and discuss with the poster presenters in this topic directly after the session 'Whole-life foundation and structure integrity' taking place on Wednesday, 12 March 2014 at 14:15-15:45. The meet-the-authors will take place in the poster area.

**Byeong Cheol Kim**Korea Institute of Construction Technology, Korea, Republic of

Co-authors:

Byeong Cheol Kim (1)

^{F}

^{P}Youn Ju Jeong (1) Young Jun Yoo (1) Min Su Park (1) Du Ho Lee (1)

(1) Korea Institute of Construction Technology, Goyang-si, Korea, Republic of

Printer friendly version: Print

## Presenter's biography

*Biographies are supplied directly by presenters at EWEA 2014 and are published here unedited*

Dr. Kim, Byeong Cheol majored in Civil Engineering at University of Seuol. His doctoral dissertation's main topic is wind induced response analysis of bridge using CFD. Now, he is working at Korea Institute of Construction Technology. He has been involved in hybrid support structure development project of offshore wind turbine. His research is mainly focused on the wind load estimation and the interation analysis between wind and the support structure.

## Abstract

**Analysis of ultimate wind induced vibration of hybrid support structure of NREL 5MW offshore wind turbine using 3-dimensional CFD
****Introduction**

South-western coast of Republic of Korea which is one of the most suitable site to construct offshore wind turbine has economical indistinct between mono-pile and jacket type support structures because of the shallow depth as 20~30m. To solve this problem, gravity-jacket hybrid support structures have been considered, but there is few studies related on the safety of the hybrid structures. This study developed dynamic analysis tools including an open source CFD library and analyzed the support structureâ€™s responses under ultimate wind conditions considering fluid structure interaction (FSI) effect to clarify the wind resistance safety.

**Approach**

The FSI effect of wind turbine structure should be analyzed to estimate exact responses of the structure because the loads transmitted to the support are varied by wind fluctuation and structural deformation. But 3D FSI analysis needs huge amount of computational resources so that makes it hard to perform. Recent development of high performance computing technology makes the full scale CFD analysis possible. This study develops fluid structure interaction code that contains open source library, Elmer, and simulates an ultimate wind state offshore turbine considering wind speed fluctuation.

The CFD library used in this study (RÃ¥back, ForsstrÃ¶m, Lyly, & GrÃ¶hn, 2007) stabilize the solution while using equal degree of shape function orders to the pressure and velocity, by introducing streamline upwind/Patrov-Galerkin (SU/PG) method which adds a square term of governing equationâ€™s residual to Galerkin weighted function. The moving mesh was emulated by arbitrary Lagrangian-Eulerian (ALE) method. The governing equation discretized by Glaerkin Least Square method by using the open source library.

FSI software developed in this study (Kim, 2013) is consists of three parts which are fluidic, structural and interaction part as shown at Figure 1. If n is the number of parallel analysis node. The n number of fluidic modules give same number of parted pressure results. Interaction part transfers these results to the structural part in a concept of modal analysis. Structural part calculates modal displacement at each time step and the interaction part turns it back to the fluidic parts. In case of analysing 3.4 million tetrahedron elements using 64 nodes at once, it takes about 5~20sec at each time step. The mesh of a blade was configured as Figure 2.

Figure 1: Flowchart of developed fluid structure interaction program.

Figure 2: Divided mesh for parallel computing.

**Main body of abstract**

The object of this study was to analyze response differences between the case of considering FSI effect and the other case of not considering the effect for a support structure of a wind turbine under the 50-year return period ultimate wind speed. The geometries of the structure should be exactly defined to perform the CFD analysis. NREL 5MW reference turbine (Lago, Ponta, & Otero, 2013) is selected by target upper structure. The hybrid support structure was developed to construct at the 20m depth south-western coast shown as Figure 3. The hollow sectioned concrete base has 0.5m wall thickness and 10m height, the jacket structure of concrete filled hollow steel tube has 21m height.

Figure 3: Developed Hybrid Support Structure

The full model of the wind turbine shown as Figure 4 is established to analyze FSI using 3D CFD. The mass and stiffness of the turbine and the blade was referenced from the NREL 5MW turbine and the blade shape is generated using the airfoil geometries of DOWEC 6MW turbine (Kooijman, Lindenburg, Winkelaar, & Van der Hooft, 2003). The length of the blade is 63.333m and the chord is 4.557m. The model of ambient fluid region of one blade shown as Figure 5 is constituted using 358,652 nodes and 1,985,609 tetrahedron elements.

Figure 4: NREL 5MW Wind Turbine Model including Developed Hybrid Support.

Figure 5: Generated Mesh of a Blade of the Analysis Model.

The fluidic region of the analysis model is divided by 64 parts and calculated at each computing node. However, the structural region is calculated at only one computing node, all calculated surface pressures from the fluidic part must be collected and converted to the structural loads. Because this pressure-load converting process needs relatively long time, this study adopts modal analysis that considers only 20 lower natural mode shapes to calculate dynamic responses of the structure. The number of pressure vectors are converted by only 20 or more modal force vectors using property matrix of the structure, and the structural responses are calculated from the converted modal force vectors. The structural displacement is transmitted to the fluidic region using ALE method.

Generated wind velocity turbulent as a boundary condition of CFD model is shown as Figure 6. The time history records of wind velocity substituted at the inlet boundary conditions are generated apart 10m interval along the lateral line to considering time and space correlations, the other non-generated points used linear interpolated ones.

Figure 6: Generated turbulence from Kaimal spectrum (Time and space correlated).

In case of using blade momentum theory to calculating the wind turbines responses, the process of calculating aerostatic coefficients of the airfoil section is needed. Furthermore, many of commercial codes could not consider the FSI effect. However, 3D CFD analysis of this study calculated the responses directly without any aero-static and dynamic coefficients. The time history loads transmitted to the support structure is shown as Figure 7. The two results shows relatively similar tendency but the differences are getting larger in process of time. From the results, in case of considering FSI effect, the ultimate loads transferred to the support structure have a tendency of decrease 3~7%.

Figure 7: Transmitted loads to support structure of wind turbine.

**Conclusion**

This study calculates the load of a support structure of a static state wind turbine under the ultimate wind condition, and compares the results of considering FSI or not. The results of this study are listed below.

1. Supercomputer executable 3D CFD code which can consider FSI effect has been developed. The program developed in this study is constituted by CFD, structure and interaction modules, can use 30~1000 computing nodes at a time. It reduces analysis time from 3 more month to less than 2 days.

2. Even in that case of using wind tunnel test, the detailed flow of the wind cannot be verified. This study calculates wind turbine ambient wind velocity and pressure, and analyzes the detailed wind flow and surface loads on the structure to estimate the structural responses.

3. The blade momentum theory which is used to calculate wind load on the blade cannot consider the FSI effect. In the case of using flutter derivatives to consider FSI effect, the accuracy of the response will be limited. But 3D CFD can directly calculate the FSI responses.

4. As the results of the load on the support structure, the responses contains FSI effects shows relatively small value. If consider the FSI effect to calculate ultimate load on the support structure, the structure can be designed more economically.

Three-dimensional CFD analysis has lesser error factor than wind tunnel test and can analyze the aerodynamic responses and detailed wind pressure distribution on a structure. To investigate the wind induced responses more exactly, 3D CFD considering FSI effect should be executed. If the numerical results are verified by the measurement and the computation equipment are developed continuously, the proposed CFD method can be one of the good alternative for the wind tunnel tests or field measurements.

**Learning objectives**

The results of full 3D CFD analysis of a 5MW wind turbine on the ultimate wind speed. FSI effect should be considered to estimate exact responses and the results will be used to develop the other types of hybrid support structures.

**References**

Kim, Byeong Cheol. (2013). Buffeting Response of Cable-stayed Bridge using 3-Dimensional Computational Fluid Dynamics. (Doctoral Dissertation), University of Seoul, Seoul. Retrieved from http://www.riss.kr/link?id=T13073609

Kooijman, HJT, Lindenburg, C, Winkelaar, D, & Van der Hooft, EL. (2003). DOWEC 6 MW pre-design: Aero-elastic modelling of the DOWEC 6 MW pre-design in PHATAS. DOWEC Dutch Offshore Wind Energy Converter 1997â€“2003 Public Reports.

Lago, Lucas I., Ponta, Fernando L., & Otero, Alejandro D. (2013). Analysis of alternative adaptive geometrical configurations for the NREL-5 MW wind turbine blade. Renewable Energy, 59, 13-22. doi: http://dx.doi.org/10.1016/j.renene.2013.03.007

RÃ¥back, Peter, ForsstrÃ¶m, Pirjo-Leena, Lyly, Mikko, & GrÃ¶hn, Matti. (2007). Elmer-finite element package for the solution of partial differential equations. Paper presented at the EGEE User Forum.

Follow EWEA on: