in for Wind flow Numerical Smooth hill validation in s wind resource module using 5 th Symposium on in Wind Energy (SOWE 2017), Javier Magdalena Saiz, José Marín Palacios, Jesús Matesanz García and Laura Cano Criado SOLUTE Ingenieros, Avda. Cerro del guila 3, 28703 - San Sebastin de los Reyes (Madrid) April 27 th, 2017
Ingenieros in for Railway Wind & renewable energies Auto space Energy Civil engineering Wind flow Numerical Activities in the following fields: Structural & mechanical analysis Advanced numerical simulation Wind turbine load analysis and dynamics Wind turbine certification Engineering & design of wind energy projects SODAR & mast measurement campaign Wind resource evaluation & micro-siting
For Uncertainty Reduction Of Wind in for Elliott Bache Wind flow Numerical Data analysis Download of MERRA, MERRA2 & ERA-I data Manual flagging & flagging with rules MCP analysis with several reference masts Micrositing Wind resource assessment Wind resource calculation using multiple masts Extreme wind maps Use of atmospheric stability parameters in simulations Wake ing using up to 8 wake s Site compliance to check class and subclass for each position of the wind farm
s linear wind flow in for Wind flow Numerical Model derived from UPMORO code & based on Jackson & Hunt (1975) 1 potential flow theory As is WAsP Atmospheric stability is modified through Monin-Obukhov length Forests are led with canopy heights and forest density 1 Jackson, P.S. and Hunt, J.C.R. (1975), Turbulent wind flow over a low hill, J. Royal Met. Soc. 101, pp 929-955.
implementation in in for Wind flow Numerical For more complex terrains, more precise (& longer) simulations are necessary Linear calculation is used to initialize & impose BC s for Reynolds-averaged Navier-Stokes (RANS) simulations (Source: Backpacker magazine, Image by Mark Goodreau) Wind Wind Wind 40m 80m 120m
implementation in in for Wind flow Numerical Linear solution mapped to CFD domain solution Wind Wind Wind 40m 80m 120m
in for Wind flow Numerical is validated with experimental results 2 z y experiment with bump Inlet velocity = 27.5 m/s Recirculation bubble & velocity profiles downstream of bump are compared x z/h [ ] 10 0 10 1 10 2 U/U ref k/k ref 10 3 0 0.2 0.4 0.6 0.8 1 2 G. Byun, R.L. Simpson, C. Long, AIAA journal 42,754 (2004)
RANS Turbulence s in for Wind flow Numerical Different turbulence s are used High Re Spalart-Allmaras Low Re Spalart-Allmaras High Re k-ε Low Re Launder Sharma k-ε Low Re Lien Leschziner k-ε High Re k-ω Low Re k-ω High Re k-ω SST Low Re k-ω SST Inlet conditions l = 0.22δ ε = C µ 3/4 k 3/2 l ω = ε C µ k
Separation (a) & reattachment (b) pts. in for Wind flow Numerical Model Pt. a Pt. b Experimental 2 0.97 2.00 High Re k-ε - - Low Re LL 3 k-ε 0.64 2.05 Low Re LS 3 k-ε 0.96 1.92 High Re SA 3 - - Low Re SA 3 0.32 2.24 High Re SST 1.15 1.92 Low Re SST 0.192 2.12 High Re k-ω 1.28 1.99 Low Re k-ω 0.51 1.99 Experimental results 2 High Re k-ε & SA 3 have no bubble! 2 G. Byun, R.L. Simpson, C. Long, AIAA journal 42,754 (2004) 3 LL = Lien Leschziner, LS = Launder Sharma, SA = Spalart-Allmaras
Downstream velocity profiles : far from wall in for Wind flow Numerical Velocity profiles reported at x/h = 3.63 downstream of bumptop 2,3 1 Best results Low Re k-ω Worst results Low Re SST High Re SA Low Re SA z/h [ ] 0.8 0.6 0.4 0.2 Experimental High Re k ε Low Re LL k ε Low Re LS k ε High Re SA Low Re SA High Re SST Low Re SST High Re k ω Low Re k ω 0 0 0.2 0.4 0.6 0.8 1 U/U ref [ ] 2 G. Byun, R.L. Simpson, C. Long, AIAA journal 42,754 (2004) 3 LL = Lien Leschziner, LS = Launder Sharma, SA = Spalart-Allmaras
Downstream velocity profiles : near wall in for Wind flow Numerical Velocity profiles reported at x/h = 3.63 downstream of bumptop 2,3 10 0 Best results Low Re k-ω Worst results High Re k-ω High Re SA Low Re SA z/h [ ] 10 1 10 2 10 3 10 4 Experimental High Re k ε Low Re LL k ε Low Re LS k ε High Re SA Low Re SA High Re SST Low Re SST High Re k ω Low Re k ω 0 0.2 0.4 0.6 0.8 1 U/U ref [ ] 2 G. Byun, R.L. Simpson, C. Long, AIAA journal 42,754 (2004) 3 LL = Lien Leschziner, LS = Launder Sharma, SA = Spalart-Allmaras
Two-bump domain in for Two-bump domain is created to study bubble effect on 2 nd bump Wind flow Numerical Large dispersion in results at 2 nd bumptop Turbulence can greatly affect wind resource in complex terrain!
1 st bumptop velocity profiles in for Wind flow Numerical All profiles 3 follow similar (more or less) tendency No bubble High Re k-ε High Re SA Deformed bubble Low Re SA Low Re SST z/h [ ] 2 1.8 1.6 1.4 1.2 High Re k ε Low Re LL k ε Low Re LS k ε High Re SA Low Re SA High Re SST Low Re SST High Re k ω Low Re k ω 1 0.8 1 1.2 1.4 U/U ref [ ] 3 LL = Lien Leschziner, LS = Launder Sharma, SA = Spalart-Allmaras
2 nd bumptop velocity profiles in for Wind flow Numerical Bubble has large effect on 2 nd bumptop profiles 3 z/h [ ] Standard deviation (σ) is much larger Height at which σ 2.5% (excluding deformed bubble s) : 1 st bumptop : 0.15H 2 nd bumptop : 0.30H 2 1.8 1.6 1.4 High Re k ε Low Re LL k ε Low Re LS k ε High Re SA Low Re SA High Re SST Low Re SST High Re k ω Low Re k ω z/h [ ] 2 1.8 1.6 1.4 1 st bump 2 nd bump 1.2 1 0.8 1 1.2 1.4 U/U ref [ ] 1.2 1 0 0.02 0.04 0.06 0.08 σ U/Uref [ ] 3 LL = Lien Leschziner, LS = Launder Sharma, SA = Spalart-Allmaras
in for Wind flow Numerical in wind tunnel No bubble : High Re k-ε & Spalart-Allmaras Velocity profile downstream of bubble Closest to experiment : Low Re k-ω Furthest from experiment : Spalart-Allmaras (high & low Re) Dispersion at 2 nd bumptop is much larger than at 1 st Larger bubble makes more diffuse boundary layer downstream For a 200m hill, 5% differences may occur up to 60m at 2 nd hilltop
Future work in for Wind flow Numerical Current studies Periodic hills case Full-scale domain Detached eddy simulation (DES) Future studies Large eddy simulation (LES) Modify wall functions More complex Function of p +
The End in for Thank you! Wind flow Numerical Any questions? Any comments? elliott.bache@solute.es