HomeUpSearchMail
NEW

Simuleren van wind bij De Hoop, Den Oever

Simuleren van wind bij De Hoop, Den Oever by Victor Reijs is licensed under CC BY-NC-SA 4.0

Introductie

• Informatie over de molen omgeving
  
• Eigenschappen van obstakels
• Maken 3D model in SketchUp
  
• Editing 3D model in SIMSCALE
• Configureren van de CFD
• Uitvoering van de CFD
• Conclusies
• Referenties
• Acknowledgements

Sorry for the two langauges gebruikt in deze webpagina!

Vette paarse tekst heeft nog extra aandacht nodig.

Informatie over de molen omgeving

• De Hoop in Den Oever.
• Height of the wind shaft 9.7m from ground/street level
  
• Sail span is 17m
• The ABL wind speed is 7.09m/sec@10m and z₀ of 0.5m (parkland aanstroom gebied).
• CFD wordt gedaan tussen 170 en 260deg (zodat een goede [weighted] vergelijking gedaan kan worden met windmetingen tussen 180 en 250deg).
  

Eigenschappen van obstakels

• Gebruikt de DSM van Nederland (AHN: systematische fout: 8cm, 1 sigma: 5cm)
• Voor een glooiende omgeving volgt de wind verdeling (ABL) het terrein, dus de hoogte is ten opzichte van lokale grondniveau.
  
• Bepaal hoogte/grootte van gewenste/invloedrijke (huizen/bomen zeg hoger dan 7m) obstakels t.o.v. de lokale grond ('Minimum') niveau: 'Analyse' -> 'Lijn' of 'Polygon' (zorg ervoor dat 'Lijn' of 'Polygon' ook de lokale grond bevatten, zodat men dat kan compenseren ('Maximum' - 'Minimum').
  
• De grootte van bomen veranderd in de tijd. Check de datum van de DSM en compenseer eventueel (0.5m/jaar voor de meeste bomen bij De Hoop).
  
• Bepaal diameter van bomen m.b.v. 'Analyse' -> 'Lijn'.
  
• Voor (elke) bomen: bepaal wat voor boomsoort het is (dit in verband met bladverliezend en porositeit).
  Vlakbij De Hoop staat een treurwilg (Boom1) en drie Esdoorns (Boom2, 3 en 4). Alle zijn bladverliezend.
  Verder zijn het Cat. 1B bomen (tussen 12 en 20m grootte).
  
  LAI: Leaf Area Index (see for some examples here [Section 2 'Tree Model'], but the LAI Woody Plant Database has been used)
  f: Forchheimer coefficient (somthing related to porosity)
  De verhouding stam hoogte en boom grootte is voor volwassen bomen 1 op 1.  Maar er zullen meestal ook struiken (tot 6m) onder bomen, in een tuin, staan. Dit verdient nog aandacht.
  Remark: The value of C_d is not known with certainty, but SIMSCALE's default has been used of 0.2 as default.
  
• Effort: 2 hours.

Maken 3D model in SketchUp

• Zet in 'Model Info' -> 'Length Units' op 'Meter' -> 'Display Precision' op 0.01m -> 'LengthSnapping' ook op 0.01m
  
• 'Add location' -> 'Select Region' -> Selecteer de juiste regio rondom molen De Hoop (een regio van 800*800m is standaard, met de molen in het midden) -> 'Import'
• In het gebied van de windrichtingen (170 tot 290deg) komen alle gewenste/invloedrijke obstakels. Maar ook obstakels in de tegenovergestelde richting, over een afstand van op zijn minst 5*hoogste obstakel hoogte/grootte.
  Voor deze simulatie zal een beperkt windrichtingen en gebied meegenomen worden: zo'n 150m tegen de windrichting (in west-zuid sector) en 50m met de windrichting (in oost-noord sector).
  
• Zet gebouwen neer door contouren (polygon) met 'Line' te volgen in 'Location Snapshot' (zorg dat de lijn op 'On face in Location Snapshot' komt). Gebruik 'Push/pull' om de hoogte van gebouw in te stellen; makkelijkste door de hoogte (in meter) intypen.
  Remark: de effectieve hoogte van gebouwen: is dat de nok hoogte of halverwege dak hoogte, of zou het dak gemodelleerd moeten worden? Dit laatste is hoogstwaarschijnlijk nodig voor gebouwen dicht bij de molen.
  Remark: welke gebouwen?: alle binnen de 400m straal of alleen de meest invloedrijke(n). In STworkflow (3 meest invloedrijke); DHMworkflow (1 meest invloedrijke); VR workflow (alle gewenste gebouwen in 400m straal).
  
• Zet het molenlichaam als een cirkel (Circle), en type de Radius (straal in meter) van beneden vloer in. Gebruik 'Push/pull' om de hoogte van molenkap in te stellen, makkelijkste door de hoogte (in meter) intypen. Gebruik 'Select' om bovenste vlak(dak) van molen te selecteren; en daarna 'Scale' en gebruik 'CTRL' - 'Uniform Scale about Center' om de scaling naar kapvloer straal te maken ('Scale' = kapvloerstraal/benedenvloerstraal).
• Plaats de bomen (als rechte cilinders) door: 'Circle' met straal en 'Push/pull' voor (grootte-stamhoogte). Daarna verhef ('Move') de cilinder met de stamhoogte.
  Bomen zijn dus cilinders, die zo'n 2m (stamhoogte) verheven zijn boven de grond.
  Remark: welke bomen?: alle binnen de 400m straal of alleen de meest invloedrijke(n). In STworkflow (3 meest invloedrijke); DHMworkflow (1 meest invloedrijke); VR workflow (alle gewenste bomen in 400m).
• Het 3D-model ziet er nu zo uit:
  

• Sluit af als klaar: 'Save' en 'Download' -> 'STL'
• Effort: 4 hours.

Editing 3D-model in SIMSCALE

We are making use of SIMSCALE. Make sure you make a Community Plan account. This Community Plan allows only for 10 simulations (but one can do more; but with more manual work). Anyway make sure you have a good CAD-model and that you train/educate yourself around SIMSCALE.
The following steps can be done:

• Het simulatie project staat hier.
  
• Import het STL model
• Edit this (not yet rotated) model by: Edit a copy
  
  
• Delete the 'Location Snapshot' (a Sheet in SIMSCALE)
• Select All and Transform -> Rotate om de Z-hoek (say 50deg; windrichting 230 [=180+50] deg) om de heersende windrichting te veranderen.
  
  
• Make a Flow Volume -> External waarbij de poreuze objecten (bomen: 'Excluded parts') niet deelnemen aan het 'External flow volume'.
  
  Zie voor richtlijnen mbt  vrije afstanden rondom het model hier [Franke, 2007]:
  
  • Xmin, Xmax: minimaal 5*hoogte/grootte hoogste obstakel [Franke, 2007, page 17] extra aan linker en rechter kant van 3D-model. In dit geval ~5*~20m (molen): ~ 150m
  • Ymin: Distance in front of model (in Y-direction) minimaal 8*hoogte/grootte hoogste obstakel [Franke, 2007, page 18] extra aan voorkant van 3D-model. In dit geval ~8*~20m (molen): ~ 150m
  • Ymax: Distance at back of model (in Y-direction) minimaal 15*hoogte/grootte hoogste obstakel [Franke, 2007, page 18] extra aan achterkant van 3D-model. In dit geval ~15*~20m (molen): ~ 300m
  • Zmin: Ground (in Z-direction) minimaal om invloed van modelerings afrondingen te voorkomen:  ~ 0.1m
  • Zmax: Height (in Z-direction) minimaal 6*hoogte/grootte hoogste obstakel [Franke, 2007, page 17]. In dit geval ~6*~20m (molen): ~ 150m
  • By the way, the default values in SIMSCALE were always somewhat larger than the above, so its defaults are also ok. Except for Zmin; I would put that always at 0.1m.
• Delete alle gebouwen (dus niet de bomen)
  
• Save
• Effort: 2 hours.

Configureren van de CFD

Take the following steps (if not default, it is explicitly mentioned in below steps):

• Goto SIMULATIONS +
• Incompressible -> Turbulence-model -> Realizable k-epsilon [Franke, 2007, page 14]
  
  
• Materials -> Air -> Apply
  Assigned Volumes -> Flow region
  
  
• Boundary conditions
• Velocity Inlet
  Assigned Faces -> the wind side
  (U) Velocity -> U_y -> ABL Formula (7.09m/sec at 10m and z₀=0.5m [the ABL of the incoming wind: 'parkland']]
  (0.41*7.09/log((10+0.5)/0.5))/0.41*log((z+0.5)/0.5)
  Turbulence -> Fixed value
  (k) Turb. kinetic energy -> ABL derived Formula
  ((0.41*7.09)/(log((10+0.5)/(0.5)))^2/(0.09)^0.5)
  (ε) Dissipation rate -> ABL derived Formula
  (0.41*7.09/log((10+0.5)/0.5))^3/(0.41)/(z+0.5)
  Save
  
  
• Pressure outlet
  Assigned Faces -> opposite inlet side
  Save
  
  
• Wall
  Assigned Faces -> two sides and top
  (U) Velocity -> Slip
  Save
  
  
• Custom (ground)
  Assigned Faces -> bottom side
  Wall roughness -> On
  Roughness height -> 1m (values of 0.1 and 5m have been tried to get a stable ABL at start of Flow Volume: but this parameter has no significant influence, so default is used)
  Roughness constant -> 0.5
  Save
  
  
• Advanced concepts -> Porous media -> Darcy-Forchheimer
  In the below grab the trees are more or less not existing (very low f [=0.01m^-1 ]). For leafed trees an f=0.3, 0.6 and 0.9m^-1 have been used. Darcy coeficient (d) is zero [SIMSCALE, 2023, section TreeModel].
  
  
• Mesh
  Fineness = 5, 8.2, 8.9 and 9.5 have been used.
  
  
• Mesh -> Refinements -> Inflate boundary layer
  
• Effort: 2 hours.

Uitvoering van de CFD

Analysis for wind direction at 230deg

• Simulation Runs +
• The windspeed some 240m in front of the windmill (some 100m from the first buildings) is around 6.9m/sec. This is lower than the inlet ABL (7.1m/sec); don't know why the ABL is not stable in the first part of the Flow Volume.
  

•  The (square) plane of the sails is provided in below pictures. The middle cross is the location of the windshaft. The sail plane is 1m from the mill cap. The colouring is between 0 (dark blue) and 10 (red) m/sec in steps of 1m/sec.
  
• Using Forchheimer f=0.01m^-1 (virtually no trees, Fineness=5)
  
  Speed at horizontal sail tips is around 3.6 to 5msec: on average 4.3m/sec (for Fineness=8.9, is the average tip speed 4.7m/sec)
• Using Forchheimer f=0.3m^-1 (all leafed trees, Fineness=5)
  
  Speed at horizontal sail tips is around 2.8 to 2.9msec: on average 2.8m/sec (for Fineness=8.9, is the average tip speed 2.5m/sec)
• For Fineness=5: The windspeed is reduced to 62% (=4.3/6.9) with no trees and reduced to 42% (=2.8/6.9) with leaved trees (comparing it with the wind speed in front of the buildings).
  If we compare the influence of all trees with the influence of existing buildings; the reduction is 68% (=42/62).

• The f for the trees has been differentiated (instead of all 0.3m^-1) and the stems are for some at 4m (instead of 2m); they now have f= 0.3, 0.6 and 0.9m^-1 (based on uncertainty about the C_d, the f is 3 times the value in this table).
• Using Forchheimer f_VR=0.3, 0.6 and 0.9m^-1 (Fineness=8.9)
  
  Speed at horizontal sail tips is around 2.4 to 2.6msec: on average 2.5m/sec
• For Fineness=8.9 and more realistic (less porous) trees:  The windspeed is reduced to 62% (=4.3/6.9) with no trees and reduced to 36% (=2.5/6.9) with leaved trees (comparing it with the wind speed in front of the buildings).
  If we compare the influence of all trees with the influence of existing buildings; the reduction is 58% (=36/62).
• It looks better to use SIMSCALE's default of C_d=0.2 (as defined in this table). The f for the trees is this f_VR/3 (=0.2/0.6);  now f = f_VR/3 =f_SIM= 0.1, 0.2 and 0.3m^-1
  
• Using Forchheimer f_SIM=0.1, 0.2 and 0.3m^-1 (more realistic leafed trees, Fineness=8.9)
   
  Speed at horizontal sail tips is around 2.8 to 3msec: on average 2.9m/sec
• For Fineness=8.9 and reduced C_d:  The windspeed is reduced to 62% (=4.3/6.9) with no trees and reduced to 42% (=2.9/6.9) with leaved trees (comparing it with the wind speed in front of the buildings).
  If we compare the influence of all trees with the influence of existing buildings; the reduction is 68% (=42/62).
• Influence of f (or better C_d) on the average tip speed (Fineness=8.9 and only looking at a 230deg direction):

  trees' f used [m-1]	average tip speed [m/sec]	compared with regard to free space	compared with regard to built
  0.01	4.3	68%	100%
  0.3	2.5	36%	58%
  0.3,0.6,0.9 fVR	2.5	36%	58%
  0.1,0.2,0.3 fSIM	2.9	42%	68%

• Varying the meshing:
  The number of used CPUhours is collected below (it also provides an idea if a PWC [Pedestrian Wind Comfort] of 360degrees [in steps of 10deg] would be done)
  

  Fineness	Mesh + Simulation [CPUhours]	Mesh MCells	estimated duration PWC of 36 sectors and 128 CPUs [hours]
  10	#N/A	#N/A	#N/A
  9.5	92	20	26
  8.9	14.7	6.8	4
  8.2	7	3.2	2
  5	2.6	1.2	0.7

  
  The speed results (in a plane 1 m from the windshaft head) differ somewhat, see the below front view (the mill is at the vertical line):
  

  Fineness = 9.5 (11Mcells)
  Speed difference between Fineness= 9.5(black) and 5(green)	Speed difference between Fineness=9.5(black) and 8.2(blue)	Speed difference between Fineness= 9.5(black) and 8.9(red)

  
  <for the Difference row: collored/black speed contour-lines are in steps of 1/msec; max. is 9m/sec>
  The speed results for higher Fineness' tend to be lower. The resuotuion of the results is higher for higher Fineness (as expected).
  The speed differences between Finenesses are determined for three speed contour-lines: 1m/sec (happens close to wind shaft location); 5m/sec (happens near highest sail point) and 7m/sec (above mill):
  

  Fineness comparison	Speed contour [m/sec]	Difference along countour-line			Diff. at mill location
  		Min diff. [m/sec]	Max diff. [m/sec]	Largest |diff.| [m/sec]	Diff. [m/sec]
  9.5 - 5	1	-1.2	0.3	1.2	-0.2
  	5	0.3	0.6	0.6	0
  	7	-0.2	0.3	0.33	0.1
  9.5 - 8.2	1	-0.8	0.4	0.8	-0.2
  	5	-0.1	0.3	0.3	-0.1
  	7	-0.1	0.2	0.2	0
  9.5 - 8.9	1	-0.3	0.5	0.5	-0.1
  	5	-0.2	0.2	0.2	0
  	7	-0.4	0.2	0.4	-0.3

• All mesh metric stay within the defined min and max range, at least for the tested Fineness, between 5 and 9.5.
• Effort: 4+3*1= 7 hours.

Analysis for wind direction at 180 to 250deg

• Weighted relative speeds of different wind directions (each sector of 10deg used some 20CPUh, Fineness=8.9).
  The wind speed/energy is averaged over 8 positions of the long and short rest sail positons.
  The standard deviation on relative speed measurements and simulations is assumed to be (like for Impington) not less than 6%. The relative energy results migth have a standard deviation of some 15%.
  The simulated speeds are relative to the above ABL (for a 'parkland'). They are weighted by including the neighboring sector's speed based on the distribution of the measured wind direction (with a standard deviation of around 15deg).
  Red curve is: built without trees. Blue curve is: built and with porous trees (using f_SIM: SIMSCALE's default C_d=0.2). Yellow curve is: built and with porous trees (using 3*f_SIM: thus a more or less three times higher Cd than SIMSCALE).
  Here is a comparison of the relative wind speeds:
  
  And here a comparison of relative energy:
  
• If using formula (instead of simulations) the following power increase can be seem (this is based on the Beljaars' method, created by V. Reijs, see also here regarding the way the Effective Useable Energy [EUE] is determined):
  
  So there is a maximum increase over the sectors 210 to 240deg of aorund 17%. This is not that far off from the simulations (22%).
  This is also within the important wind direction (SW).
  

• Again, it shows that a variation of porosity of the trees does not have that much influence. This is nice, as up to now the theory behind the tree's value of porosity is also poor/unclear.
  
• Effort: expected 3*13*0.5=20 hours.

Conclusions

• Extensive sensitivity analysis has been performed.
  
• Make sure one understands this document.
• Het simulatie project kan hier bekeken worden.
  
• Only a few default SIMSCALE settings need to change for this De Hoop simulation:
  
• Size of External Flow Volume
• Incompressible -> Turbulence-model -> Realizable k-epsilon
• Increase of mesh Fineness
  
• Near this simulated mill: the speed does not change much (on average around 0.2m/sec for the wind regime of 1 to 7m/sec) when having 1.2Mcell or 20Mcells in the mesh.
  
• Although the mesh size has been increased (from 1.2 to 20Mcells), no attention has been given to specific (possible important?) areas where the mesh really/only needed to be increased.
  This needs to be investigated further.
• The result of SIMSCALE CFD are quite close (within standard deviation) to another CFD's results.
  
• Make sure to understand/solvie the above bold purple points.
  Some important points to solve:
• The modelling of roof might be important for buildings nearby a mill.
• Fully understand the influence of the mesh for the mill environment.
• An overview of outstanding issues/questions/etc is here.
  

References

Franke, Jörg  et al. COST Action 732: Best practice guideline for the CFD simulation of flows in the urban environment. Brussels, COST Office 2007.
EAGM, Rapport nr 523006 dd 20mrt2023 WieringerMolens_Windonderzoek_DeHoopAMSTELDIEP, 2023
SIMSCALE, Advanced Modelling PWC, 2023

Acknowledgements

Disclaimer and Copyright
HomeUpSearchMail