WAsP Results depending on Size of the digital map

[Petros]

Dear WAsP team,

Recently i received some past WAsP projects done with some really small sized maps. This triggered my curiosity to see how small sized maps affect the results on WAsP.
What i did was that i clipped a map of a recent project, that was primarily cut according to WAsP best practices (minimum distance between the WTGs and the borders about 5km), to a smaller area where distance between the WTGs and the borders was not bigger than 1,5km.
The results where really close on all variables (AEP, wind speed, wake losses). Net AEP difference 0,5%.
So my question is, why should we have larger digital maps (meaning extra cost) when we get almost the same results with much smaller ones?
Is there any rule (apart from the 5km one) or any explanation how the map affects the calculations?
I am talking about complex terrain sites, with varying roughness.

Thank you in advance
Petros

[Rae]

I really wonder how the WAsP Team would be answered also.

[Duncan]

I expect you'll get a proper answer from one of the scientists, but in the meantime I can say something from the technical point of view.

Reducing the size of the map is by far the most effective way of speeding up calculations. But it's not easy to say how much information can be safely removed without affecting the results.

It's the number of data points in the map, not the area covered by the map which is significant. For any reasonably large map, if you reduce the number of points by half, the calculation speed will double.

In terms of distance from the sites, I understand that roughness change information should extend as far as possible. It's possible to construct a map in which the roughness information layer is much larger than the orographic layer, and that's fine for WAsP.

The safe distance to the map edge for both orography and roughness is a function of the hub height, I think.

If you want to maintain safe distances but minimise the data volume, then you can construct a circular map. (The map editor supports this). That can reduce the amount of redundant information by about 20%.

But most of the redundant information will usually be contained in high-resolution elevation contour data far from the site. More distant orgraphic features are less significant for the site effects (all other things being equal), so it's worth concentrating the data around the sites. Further away, you can thin the lines: removing contours and removing points. Again, the Map Editor can help here.

If you've got high-resolution roughness information, it's not so easy to thin out, but a small stand of bushes 20 km from the site is unlikely to be significant. A shoreline 15 km away is relevant.

[Petros]

Hello Duncan and Rae,

What made me reduce the size (surface) of the map was the results that i would get and not the speed of calculations.

Duncan if you mean the resource grid area, then i would agree that the extent of the grid affects the calculation time, but when it comes to Wind Farm calculations the calculation time is insignificant.

I just found out that there were only minor changes on the results while i expected higher discrepancy.

Roughness lines did not change when i changed the height contour map, and were covering the same area (10km away from the WTGs)as in the primary condition.

The height contour intervals did not change (20m).

Now for high complex terrain sites (which is the rule in Greece) i don't really see that even in closer distances (3-5km from WTGs) roughness can affect dramatically the results. When the slopes are 40,50% or even higher i suppose that turbulence causes high mixture of the air, so roughness is much less significant than it would be in a flat terrain site.

[Niels Gylling Mortensen]

Our general recommendations for elevation and landuse information (roughness) are given in the help files and summarised in the WAsP Best Practices; the latter document can be downloaded from http://www.wasp.dk/Support/FAQ.html.

Both elevation and roughness maps can be made coarser (i.e. fewer contours and fewer points) far away from the sites in order to speed up calculation time. Circular (or other non-square) maps may be used to decrease the number of points and lines in a map.

In some cases, the results may not change much when you clip the map (like described above), but this is not generally true! So, it's recommended to stick to the guidelines offered by the modellers in the best practices.