Back to the programme printer.gif Print

Delegates are invited to meet and discuss with the poster presenters in this topic directly after the session 'How does the wind blow behind wind turbines and in wind farms?' taking place on Tuesday, 11 March 2014 at 16:30-18:00. The meet-the-authors will take place in the poster area.

Alvaro Cuerva Universidad Politécnica de Madrid, Spain
Tee-Seong Yeow (1) F Alvaro Cuerva (1) P Javier Pérez (1) Cristobal Gallego (1) Oscar Lopez-Garcia (1)
(1) Universidad Politécnica de Madrid, Madrid, Spain

Printer friendly version: printer.gif Print


On the flow structures over the Bolund Island


We have reproduced the Bolund experiment in our wind tunnel for a 270º wind direction, and we predict that the region of high TKE generated at the edge of the escarpment does not evolve downstream on the plateau uniformly. Instead two longitudinal regions of high TKE and low speed are developed surrounding a central region of high speed and low TKE. At least for one of these two regions, the presence of a longitudinal vortex-like structure has been observed in the mean velocity field. High cross-flow gradients in the mean speed and TKE fields are observed across these three regions.


We used the Bolund experiment, issued by RISø-DTU in 2007 as a case study because it is presently the reference case for benchmarking of numerical and physical modelling of neutral atmospheric flows on extremely complex terrain without Coriolis effects [1,2].

We tested a 1:115 scaled mock-up of the Bolund Island. Our wind tunnel (WT) is a Eiffel type with a 6m length closed test chamber (2.2m-wide, 2.2m-high) and a 18m fetch allowing the natural development of the boundary-layer. No roughness elements were added to neither the wind tunnel floor nor the mock-up. The frontal area of the mock-up was less than 2% of the test section area, so blockage effects were minimised.

We measured the flow velocity at iso-height surfaces of 5m height over the island and in different cross-flow vertical planes using three-components hotwire (3CHW)[9]. Two-components particle image velocimetry (2CPIV)[10] was used to measure the two components of the velocity in vertical longitudinal planes (y=0 in figure 1).

The test Reynolds number based in the free wind speed and the height of the island was Re=41.500, which is obviously lower than the full-scale values ([1020000,4250000]). Neither stratification nor Coriolis effects were reproduced, accordingly to the declared full-scale values for the magnitude of the Obukhov length (>250m) and the Rossby number (667) [1,2]. The WT friction Reynolds number based on the aerodynamic roughness length was 0.21, therefore the scaled inflow was most likely transitionally rough which meant a mismatch of the full-scale conditions (with a value 18.7, certainly in the rough regime). The reproduced ratio of the boundary-layer height to the height of the island was about 2.2, a value smaller than the full-scale one. Implications of these three mismatches are commented in [3].

We previously validated our experimental approaches by comparing our WT results in [3] with full-scale results provided by RISø-DTU [1,2]. We reported mean errors in the range [13.87%,21.05%] for the speed-up and [43.72%,69.38%] for the normalised increase of TKE, depending on the Reynolds number and the experimental technique (2CPIV or 3CHW). Our reported errors were comparable with those reported by other physical modellers involved in the Bolund experiment.

Main body of abstract

One of our main concerns is to facilitate the comparison between our results and the results of other authors following the philosophy of the blind-comparison proposed by RISø-DTU. Pursuing this goal, we performed a detailed characterisation of our inflow boundary-layer. 2CPIV and 3CHW redundancy (but not simultaneous) measurements of the WT inflow boundary-layer were performed and the vertical profiles of speed and TKE are compared to full scale results in figures 2.a.b.

Several relevant characteristics of the flow over the Bolund island have already been described in the full-scale tests [1,2], by WT/water channel measurements [1,2,3,4,5,6] and numerical modelling [2,7]. The topology of the speed-up and normalised increase of TKE as well as direction and tilt angle have been preliminary analysed [3,4,6] for different wind directions.

Considering, for instance, the 270º wind direction (Line B in Bolund jargon), a deceleration region in front of the escarpment (M7 in figure 1) where the TKE increases has been described by different authors [1,2,3,4,6].

The intermittent detachment flow pattern on top of the island, where the largest increase of TKE is measured, has also been analysed [1,3,8]. It is there where both physical and numerical models present the largest bias compared to full-scale measurements (low heights at M6 in figure 1). We have previously quantified the intermittency of the recirculation phenomenon, concluding that no reversed flow region exists in the mean velocity field, and that instantaneous recirculation patterns take place with a probability of less than 10%. A relaxation region (M3 in figure 1) on the plateau follows the intermittent recirculation region. There, the TKE decreases (at least along the line B, y = 0m in figure 1). Finally a wake region is developed in the lee side of the island where the TKE increases again and the mean velocity largely decreases (M8 in figure 1}). All these features have been analysed by the authors, and WT measurements of speed up, flow angles, normalised increase of TKE and spectra/cospectra were successfully compared with full scale results in [3]. Some of these features can be observed in figures 3.a.b where a 2CPIV map of the speed and the TKE are reproduced.

After gaining confidence in the capability of our experimental tools we focus now in the characterisation of additional turbulent flow structures over the island.

Using a three-axis traversing system we have mapped the flow velocity in different points using 3CHW on an iso-height surface at 5m agl, (Coarse-Map from now on) and in five different cross-flow vertical planes (Cross-grids from now on, at different x=cons, schematised in figure 1). The spatial distribution of TKE in the Coarse-Map and the Cross-grid at x=-18.4m is shown in figure 4}. In the figure two longitudinal regions of high TKE surrounding a region of low TKE can be identified at 5m height.

In an attempt to characterise the flow structure behind this TKE distributions, additional higher resolution cross-grids at x=-18.4m, -25.3m, -32.2m, -39.1m and -46m were mapped. The results are showed in figures 5-9. In these figures the [V,W] vector field is plotted together with a colormap of the U component and the TKE.

The plots show that the region of high TKE and low U over the right side of Bolund (y<0m), develops longitudinally, revealing the existence of a longitudinal vortex-like structure in the mean field whose core moves from low to high z positions and towards the plane y = 0m as it is developed streamwise.

The flow approaching the ridge through the central positions (let us say y = [-7m,7m]) experiences the intense disturbance of the island before than the flow approaching through lateral positions (let us say y<-7m). This is why, i.e. for x=-46m, -39,2m -32.2m, the flow is practically longitudinal for y>-7m, but it presents a high up-flow component (W>0) for y<-15m, which contributes to the formation of the longitudinal vortex. This situation is schematised in figure 10.

This structure combining low speed, high TKE and longitudinal vortex-like structure in the mean velocity field leads to very high cross-flow gradients of U and TKE likely challenging for numerical models and potentially harmful for wind turbines.


After gaining confidence in the capability of our experimental tools by comparing our WT results with the full-scale results of the Bolund experiment in previous works, we have focused now on the analysis of the turbulent flow structures over the island of Bolund.

We have predicted some characteristics of the flow topology over the Bolund island for a 270º wind direction. Our analysis is based on three components hot-wire and two components particle image velocimetry, both techniques were previously validated against full scale measurements in the context of the Bolund experiment. The analysis has revealed the existence of two longitudinal regions of high TKE and low speed surrounding a central region of higher velocity and lower TKE. At least, along one of these high TKE regions we have verified the existence of a longitudinal vortex-like structure developing stream-wise in the mean velocity field. The core of the vortex moves upwards and towards the central plane of the island as it develops streamwise. This flow topology leads to high cross-flow gradients of speed and TKE which are potentially harmful for wind turbines and that are likely challenging for numerical models. The predicted flow field is compatible with the geometry of the escarpment for 270º wind directions. Thus, the flow approaching the island through central positions experiences the strong island disturbance firstly, and the high upflow existing at the edge region rapidly disappears leading to an almost longitudinal velocity field in the central region (y close to 0m) of the island. However, the flow approaching the island through the lateral positions, experiences the strong disturbace of the island at a certain distance after. The resulting combination of a central flow region without upflow and a lateral flow region with an intense upflow most likely initiates the formation of the longitudinal vortex detected in the mean flow velocity field.

Learning objectives
High resolution determination of flow structures that are potentially harmful for wind turbines, on a reference case that is presently widely studied by different groups of modellers of wind flow on extremely complex terrain.
Predition of flow structures with a previously validated experimental technique combining 2CPIV and 3CHW in wind tunnel.

[1]. Berg J, Mann J, Bechmann A, Courtney MS, Jørgensen HE. The Bolund Experiment, Part I: Flow Over a Steep, Three-Dimensional Hill. Boundary-Layer Meteorology 2011; 141(2):219_243, doi:10.1007/s10546-011-9636-y.

[2]. Bechmann A, Sørensen NN, Berg J, Mann J, Rethore PE. The Bolund Experiment, Part II: Blind Comparison of Microscale Flow Models. Boundary-Layer Meteorology 2011; 141(2):219243, doi: 10.1007/s10546-011-9637-x.

[3]. Yeow, T.S., Cuerva, A., Pérez-Alvarez, J., Reproducing the Bolund Experiment in Wind Tunnel, Wind Energy, (accepted for publication, 14/10/2013).

[4]. Yeow TS, Cuerva A, Conan B, Pérez J. Wind Tunnel Analysis of the Detachment Bubble on Bolund Island. The Science of Making Torque from Wind, October, 9-11, Oldenburg, 2012.

[5]. Yeow TS, Cuerva A, Pérez J. Pressure Measurements of the Detachment Bubble on the Bolund Island. European Wind Energy Association, April, 16-19, Copenhagen, 2012.

[6]. Conan B. Wind Resource Assessment in Complex Terrain by Wind Tunnel Modelling. PhD. report. Technical Report, Von Karman Institute for Fluid Dynamics and Université d'Orléans 2012, ISBN: 978-2-87516-056-0.

[7]. Prospathopoulos JM, Politis ES, Chaviaropoulos PK. Application of a 3D RANS Solver on the Complex Hill of Bolund and Assessment of the Wind Flow Predictions. Journal of Wind Engineering and Industrial Aerodynamics 2012; 107:149159, doi:10.1016/j.jweia.2012.04.011.

[8]. Mann J, Angelou N, Mikkelsen T, Hansen KH, Cavar D, Berg J. Laser Scanning of a Recirculation Zone on the Bolund Escarpment. The Science of Making Torque from Wind, October, 9-11, Oldenburg, 2012.

[9]. Bruun HH. Hot-Wire Anemometry. Oxford University Press, 1995.

[10]. Raffel M, Willert C, Wereley S, Kompenhans J. Particle Image Velocimetry. A Practical Guide. 2ed. Springer-Verlag, 2007.