Manually calculating outside views for the LEED IQEc8.2 credit is a tedious exercise. Here are three attempts at automating the LEED views calculation through Grasshopper. More importantly, this is an attempt to measure high quality views, rather than using the pass/fail criteria used for LEED.
Version 0.1
The first version used the View Rose component from Ladybug. I assign walls, and manually set a point at the corners of each window. The View Rose component then displays the area visible to each point.
The View Rose component is great for mapping views from a single point, but is too data-intensive for mapping views from several points. This script worked well for the small test case above, but had a hard time analyzing geometry any more complex than this.
Version 0.2
The next iteration used the Isovist component from Grasshopper. This component sends out a circular array of points from a central node until the points collide with some geometry. In this version, I set a grid of nodes on the floor, and set the isovist radius to 100. It then counts how many nodes sent at least one point that escaped through a window and reached its full radius. It computes the ratio of nodes with views versus nodes without views and outputs a percentage of floor area with views.
This was also rather data-intensive when running at the high resolution needed to get an accurate calculation. Also, it doesn’t provide clean line-of-sight vectors.
Version 0.3
In the latest version, isovists are calculated from the corners of the windows, rather than from a grid of nodes over the floor. In other words, it’s calculating the floor visible from the window, rather than the window visible from the floor. This reduces the number of isovist calculations and makes complex simulations reasonable.
It then draws a polyline connecting the visible points, creating isovist boundaries. When assigned a transparent color, these boundaries overlap to reveal areas with multiple outside views. A solid color results in a pass/fail views diagram.
One issue with calculating views from only the corners of windows is that it doesn’t account for floor area visible from the middle of a window. This results in a slightly shredded view field (on the left) which can be fixed with the trusty paint bucket tool in Photoshop (on the right). These diagrams are only as accurate as the resolution of the isovists. 360 points from each isovist yield fairly accurate results, and 1000 points per isovist is even better. A simulation for a floor plan with this size, complexity and resolution takes a few minutes to run.
I can run the simulation a few times with different radii, and overlay the results in Photoshop to get a better understanding of high quality views. In this case, I’m defining high quality as having a wide angle of outside views, and being close to the glazing.
For the next iteration, I’ll look for a more deliberate way to draw lines-of-sight from window corners to wall vertices, avoiding the blunt-force method of high-resolution isovists and manual selection of window corners. Until then, this is still much faster than calculating views by hand.