Showing papers by "Andreas Bechmann published in 2012"

Benchmarking of Wind Turbine Wake Models in Large Offshore Windfarms

Abstract: Quantifying accurately wind turbine wakes is a key aspect of wind farm economics in large wind farms. This research compares three engineering wake models with power production data from the Horns Rev and Lillgrund offshore wind farms. Single and multiple wake cases are investigated to verify the performance of the models in different conditions. The simulations reveal that the three wake models have similar behaviours for both wind farms although the turbine spacing and the turbulence intensity are different. The results prove the robustness of the models to provide accurate power predictions when the simulations are averaged over wind direction sectors of 30◦. However, all models significantly underpredict the power production of a single row of wind turbines using narrow sectors of 3◦ or 5◦. This discrepancy is discussed and justified by the wind direction uncertainty included in the datasets.

How fine is fine enough when doing CFD terrain simulations

Abstract: The present work addresses the problem of establishing the necessary grid resolution to obtain a given level of numerical accuracy using a CFD model for prediction of flow over terrain. It is illustrated, that a very high resolution may be needed if the numerical difference between consecutive refinements should be of the order of one percent for all flow directions. For the present terrain case, resolution in the order of 1 billion grid points is needed.

In the wake of Bolund: Benakanahalli - Stratification and complex terrain

Abstract: ID: 36 Theme: RESOURCE ASSESSMENT Topic: Advances in measuring techniques for resource assessment IN THE WAKE OF BOLUND: BENAKANAHALLI STRATIFICATION AND COMPLEX TERRAIN Jacob Berg(1) (F) (P) Andreas Bechmann(1) Michael S. Courtney(1) TIlman Koblitz(1) Yavor V. Hristov(2) (1) DTU Wind Energy, Roskilde, Denmark(2) Vestas Technology R&D , Randers, Denmark Introduction Micro-scale numerical flow models are the cornerstones of wind energy site assessment. With more and more new sites in complex terrain the model requirements have increased dramatically in the recent years. This has been followed by a demand of full scale measurement campaigns designed to benchmark the models in complex terrain. The Bolund measurement campaign performed near Risø DTU, in 2009 (Boundary Layer Meteorology 141, p. 219 & p. 245, 2011) provided significant insight into flow in complex terrain and has proven very suitable for benchmarking of models. Even though up-scaling is permitted to some degree, the 12 m tall peninsula, the Bolund Hill, is still too simple for many purposes; it lacks atmospheric stratification and the Coriolis effects. In addition the wake of Bolund was very poorly resolved. Though an optimal test case for micro-scales model Bolund is not a realistic wind turbine site. With this in mind a new measurement campaign was planned and conducted as a joint collaboration between Risø DTU and Vestas Technology R&D. Approach Main body of abstract In this presentation we will present the first results of a new large scale measurement campaign where the flow is altered by both the presence of complex terrain and atmospheric stratification. Whereas studies of complex terrain have been done mainly in neutral conditions and studies of atmospheric stratification in primarily flat terrain, a study where also the combined non-linear effects are studied should be much welcomed in the wind energy community, since a large part of turbines today are erected exactly at places where these are non-negligible. The name of the campaign is Benakanahalli. The measurements were conducted from February to April 2010 in the province of karnataka in India close to the village of Benakanahalli. Five 80m masts were erected near and on a long almost two-dimensional 120m natural ridge with slopes of around 30 degrees. The hill type is very common in wind energy, which can easily be verified by a short drive in the area where a huge amount of wind turbines are erected on similar hills. Equipped with sonic anemometers in five different heights, both the mean flow and the turbulence are well resolved. In addition temperature sensors were mounted on the upstream mast. Together with measurements of heat fluxes from the sonic anemometers a good estimate of the thermal stratification is thus obtained. Three mast were positioned in the wake of the hill giving us the opportunity to estimate its size and hence the recovery length of the wind speed. Conclusion We will show the first comparison with a numerical micro-scale model, the Risø DTU EllipSys3D RANS k-epsilon model, which manage to capture many of the interesting features happening as a result of the interplay between complex terrain, thermal stratification and Coriolis force.

Atmospheric stability in CFD N Representation of the diurnal cycle in the atmospheric boundary layer

Abstract: For wind resource assessment, the wind industry is increasingly relying on Computational Fluid Dynamics (CFD) models that focus primarily on modeling the airflow in a neutrally stratified surface layer So far physical processes that are . , ifi t th t h i b d l (ABL) f l th C i li f spec c o e a mosp er c oun ary ayer , or examp e e or o s orce, buoyancy forces and heat transport are mostly ignored in state of the art CFD , models. In order to decrease the uncertainty of wind resource assessment, especially in complex terrain the effect of thermal stratification on the ABL should , be included in such models. The present work examines the influence of stability on the whole ABL using the modified in-house CFD code (DTU Wind Energy) EllipSys3D Typical diurnal . l i l t d d d i t i i l ti d cyc es are s mu a e an compare aga ns prev ous s mu a ons an measurements from the GABLS II model intercomparison [6]. Obj ti ec ves T dif th i ti Elli S 3D CFD t t i t d i ti f • o mo y e ex s ng p ys o ge a more appropr a e escr p on o the wind flow in the ABL during non-neutral conditions • To validate the model against previous simulations and measurements T l th i fl f di l t t i ti th ABL • o ana yze e n uence o urna empera ure var a ons on e To set the starting point for non neutral simulations over complex terrain • Stable stratification Figure 2: (a) surface temperature (b) geostrophic wind (c) time series of wind speed at 10m 400 m a.g.l. (d) time series of turbulent kinetic energy at 55m a.g.l. (e) vertical profiles of wind speed at 1400 23 October (f) vertical profiles of potential temperature at 1400 23 October , , . 10 U t bl t tifi ti u (m/s) ns a e s ra ca on 0 The above modifications of the 3D version of EllipSys enable us to apply the developed model for flows over complex terrain Figure 1 shows the horizontal Figure 1: non-neutral simulation over complex terrain: horizontal wind speed for (a) stable and . (b) t bl diti th B k h lli it i I di F d t il b t th wind speed for non-neutral conditions over the Benakanahalli site in India uns a e con ons over e ena ana a s e n n a. or e a s a ou e . Benakanahalli measurement campaign see poster ID 107. C l i onc us ons Methods The general features of the typical diurnal cycle and its representation by the The present study considers the simulation of the diurnal cycle in the ABL The model are presented. Furthermore the sensitivity of the numerical results on the . f i fl fl i (h i ll h fl ) bj d f i d th i iti l diti i d Th h th d l t ocus s on ow over at terra n or zonta y omogeneous ow , su ecte to orc ng an e n a con ons are exam ne . e c osen me o o ogy o temporally varying surface temperatures To model the ABL more appropriately the implement stability effects into the CFD code EllipSys3D represents a promising . h d i fi t t t t d th li ti f th d l t t tifi d fl effect of the Coriolis forcing and buoyancy are included in the CFD code approac an s a rs s ep o ex en e app ca on o e mo e o s ra e ow Elli S 3D Th f ti f th i t f th t ti l over complex terrain p ys . ere ore an equa on or e energy n erms o e po en a . temperature is solved in addition to the RANS equations To close the given set of The developed model is able to reproduce the general flow pattern of a non-neutral . i difi d i f h k b l d l i d i h ABL fl bj t d t di ll i f t t equat ons a mo e vers on o t e -ε tur u ence mo e s use : n contrast to t e ow su ec e o a urna y vary ng sur ace empera ure. standard formulation we use a limiter on the resulting length scale [1 2] and , , The present simulations and the measurements from [6] show rather good additional buoyancy terms [3 4] Also ambient floor values for the turbulence t f l t ft til l i D i th i t iti , . i bl i d i d t id i l i d t t b l l agreemen rom a e a ernoon un ear y morn ng. ur ng e morn ng rans on, var a es are mpose n or er o avo numer ca ssues ue o ur u ence va ues however the growth of the convective ABL and the turbulence level are too weak close to zero [5] With these modifications the model is capable of representing , , hi h i l i ibl i th d ti t d l l l i d d N f th . l di i d h difi d i f h b l w c s a so v s e n e un eres ma e oweve w n spee . one o e non-neutra con t ons, an t e two mo e transport equat ons or t e tur u ent models intercompared in [6] was able to capture the morning transition Modelling kinetic energy k and the dissipation ε read: . the diurnal cycle presents a big challenge, and the present model shows the bi t d i ti ft th i t iti d th b t t d i l t (1) gges ev a ons a er e morn ng rans on, an e es agreemen ur ng a e afternoon and early evening. C i i t b ti i th i f i iti l d b d (2) ompar son aga ns o serva ons ra ses e ssue o n a an oun ary conditions of numerical experiments because perfect test cases do not occur in , lit Th d ll d lt iti t th i iti l t t fil Al (3) rea y. e mo e e resu s are sens ve o e n a empera ure pro e. so large scale atmospheric variations influence measured statistics as for example h B i th d ti ( l) f t b l t ki ti b b , w ere s e pro uc on or remova o ur u en ne c energy y uoyancy apparent in the non-constant geostrophic wind during the GABLS test case (see forces and depends upon the local temperature gradient The coefficient C in (2) fi 2b) Th i l t d i d fi ld i i fl d i ifi tl b th h i f th , . ε1 i l d b (3) i d li i h i i l h [1 2] All h ffi i gure . e s mu a e w n e s n uence s gn can y y e c o ce o e s rep ace y n or er to m t t e m x ng engt , . ot er coe c ents are geostrophic wind as the model’s forcing chosen according to [3 4] . , . References Results 1. Apsley, D.D., Castro I.P., 1997. A limited-length-scale k–e model for the neutral and stably-stratified atmospheric boundary layer. Boundary-Layer Meteorol, 83:75–98. Surface winds temperature stratifications and TKE values resulting from a non2. Blackadar, A.K., 1962. The Vertical Distribution of Wind and Turbulent Exchange in a Neutral Atmosphere. J. Geophys. Res, 97: , f f 3095–3102.. neutrally strati ied ABL low are compared against modelled and observed data 3. Sogachev, A., Panferov O., 2006. Modification of two-equation models to account for plant drag. BoundaryLayer Meteorol., from the GABLS II experiment from Kansas USA [6] shown in figure 2 The 121:229–266. , , . 4. Sogachev, A., 2009. A note on two-equation closure modelling of canopy flow. Bound.-Lay. Meteorol., 130: 423–435. development of the ABL throughout the day is driven by a prescribed surface 5. Spalart, P.R., Rumsey, C.L., 2007. Effective Inflow Conditions for Turbulence Models in Aerodynamic Calculations. AIAA Journal, t t th t i i d ith ti d t t t hi i d (fi 2 b) Vol. 45, No. 10, pp. 2544-2553. empera ure a s var e w me an a cons an geos rop c w n gure a, . 6. Svensson, G. et al., 2011. Evaluation of the Diurnal Cycle in the Atmospheric Boundary Layer Over Land as Represented by a Details concerning model set-up and forcing are given in [6] Variety of Single-Column Models: The Second GABLS Experiment. Boundary-Layer Meteorology 140, no. 2: 177-206. . EWEA 2012 Copenhagen Denmark: Europe’s Premier Wind Energy Event , , 0000 060