VALEOL Thirty Month Review Meeting 23 rd - 24 th November 2017, - - PowerPoint PPT Presentation

Funded by the European Union

1 AEROGUST M30 PROGRESS MEETING

VALEOL

Thirty Month Review Meeting

23rd - 24th November 2017, INRIA, Bordeaux Presented by Claire Taymans

Funded by the European Union

2 AEROGUST M30 PROGRESS MEETING

Summary

WP 2 Task : Investigation of predicted non-linear behaviour of wind turbine blades subjected to gust using incompressible flow model

• 3D natively parallel CFD code on Octree grids under development for my PhD Thesis.

WP 4 Task : Collection of in-service wind turbine data

• Meteorological mast and sensors on the blade will stay until April 2018

WP 5 Task : Comparison between experimental data and numerical simulations

• Coupling with structural model
• Analysis of the experimental data ongoing

Funded by the European Union

WP 2 3 AEROGUST M30 PROGRESS METING

3D CFD Code on Octree grids

Methods developed ▪ Natively parallel by using PABLO library (Optimad) ▪ Finite volume solver with semi-Lagrangian method ቐ 𝐸𝒗 𝐸𝑢 = 1 𝑆𝑓 Δ𝒗 − 𝛼𝑞 𝛼𝒗 = 0 ▪ Fractional step method ▪ Penalization is used to take into account the obstacles ▪ LES has been implemented with a Vreman subgrid model ▪ Order of convergence in space of the Laplacian solver : 2 ▪ Order of convergence of the whole NS solver : 1 for L฀ 1.8 for L1 and L2 ▪ Interpolations when a jump in the refinement level occurs : with Radial-Basis Functions ➢ Some problems exist and should be resolved by changing the method

Figure : Explanation of semi-lagrangian method Figure : Example of Octree mesh

Funded by the European Union

WP 2 4 AEROGUST M30 PROGRESS METING

3D CFD Code on Octree grids

Validation with literature Test Case : Flow around a sphere at Re = 500

Funded by the European Union

WP 2 5 AEROGUST M30 PROGRESS METING

3D CFD Code on Octree grids

Validation with literature Test Case : Flow around a sphere Re Present work [1] [2] [3] [4] [5]

300 0.6268 0.675

• 0.657

0.653 500 0.5488 0.52 0.4818 0.476

• 0.555

Table : Comparison of drag coefficients with data from literature at different Reynolds Number

References :

[1] Campregher (2009) [2] Fornberg (1988) [3] Fadlun et al. (2000) [4] Kim et al. [5] Correlations found in Subramanian (2003)

Funded by the European Union

WP 2 6 AEROGUST M30 PROGRESS METING Test Case : Flow around a cylinder with diameter D at Re = 3900 ▪ Spanwise extension of the domain = 40 D ▪ Inlet at 3D from cylinder. Outlet at 37D. ▪ No periodic BCs ▪ Number of cells : 25.5 Millions ▪ 6 different levels of refinement ▪ Minimum grid size : 0.0195 D -> 50 points along D ▪ 960 processors were used ▪ 11 wall clock hours needed for 30 convective times ▪ Turbulence with Vreman subgrid model

3D CFD Code on Octree grids

Validation with literature

Figures : Vorticity of flow around cylinder (10 levels) without LES (top) and with LES (bottom)

Funded by the European Union

WP 2 7 AEROGUST M30 PROGRESS METING Test Case : Flow around a cylinder with diameter D

3D CFD Code on Octree grids

Validation with literature

Figure : Wake profile at different positions obtained by averaging Velocity after a preliminary simulation Figure : Evolution of aerodynamic coefficients along time after a preliminary simulation

▪ CD mean : 1.4 instead of 1. in the literature ➢ Study of the effect of eddy viscosity coefficient needed

Funded by the European Union

WP 2 8 AEROGUST M30 PROGRESS METING

3D CFD Code on Octree grids

Other preliminary results

Figure : Flow around a cylinder at Re = 106 Figures : Flow around a fixed wind blade at Re = 400. Horizontal view on the top, section view on the bottom Figure : Penalization of the apx48 wind blade

Funded by the European Union

WP 4 9 AEROGUST M30 PROGRESS MEETING

Collection of experimental data

Meteorological Mast ▪ 49 days of data have been lost in end of March and between mid-April and mid-May ➢ The cause is not well-known, the 1s data have not been sent because of communication problems ➢ An intervention occurred to solve the problem ➢ 81% of data is available ▪ The mast will stay at least 1 complete year to well know the wind resource of the site Sensors on the wind blade ▪ Periods have been lost ➢ The inquiry system is connected to the plug of the wind turbine but power cuts occurred ➢ The clock had problem to synchronize with the network because of poor 3G ➢ An intervention has been done to add a battery and a proper clock to the system ➢ 78.3% of data is available ▪ The sensors will probably stay until next March or April

Funded by the European Union

WP 5 10 AEROGUST M30 PROGRESS MEETING

Analysis of experimental data

Filtering of experimental data

▪ Availability of both met mast and sensors on wind blade ▪ When the time has a reasonable uncertainty ➢ Evaluated at < 7 s ▪ With a wind direction lined up with the met mast and the wind turbine ➢ ± 10 ° around ▪ Blade rotational speed representing a normal behaviour of the turbine ▪ > 11 rpm and < 25 rpm

Figure : Evolution of the wind velocity and the deformation measured by a sensor on the wind blade along time (during 9h 30 min)

Funded by the European Union

WP 5 11 AEROGUST M30 PROGRESS MEETING

Analysis of experimental data

Influence of upstream wind direction

Figure : Deformation sensors at different locations as a function of wind direction

▪ The next figures have been plot for a period of 1 whole month of data after filtering ▪ A first check : relevance of the wind direction range ± 10°

Funded by the European Union

WP 5 12 AEROGUST M30 PROGRESS MEETING

Figure : Deformation sensors at different locations as a function of blade angle

Analysis of experimental data

Influence of blade angle

Funded by the European Union

WP 5 13 AEROGUST M30 PROGRESS MEETING

Analysis of experimental data

Influence of wind velocity

Figure : Deformation sensors at different locations as a function of wind velocity

▪ Work in progress

Funded by the European Union

14 AEROGUST M30 PROGRESS MEETING

Future Work

WP 2 ▪ Implementation of wall functions. ▪ Fixing totally the interpolation problems. WP 5 ▪ Identification of gusts in the experimental data. ▪ Implementing the interface between the CFD code and the ROMs developed by INRIA. ▪ Running the case of a blade in a rotational domain with simulation of fluid-structure interactions.

Funded by the European Union

15 AEROGUST M30 PROGRESS MEETING The research leading to this work has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement number 636053.