GridLAB-D Taxonomy Feeder Graphs

By Michael A. Cohen

Below are links to visualizations I created of the GridLAB-D "taxonomy feeders" using Graphviz and glm2dot, a ruby script I wrote to translate .glm (GridLAB-D) files into DOT files that can be laid out by Graphviz. In the spirit of the GridLAB-D licensing terms, I'm releasing these under the Creative Commons "CC BY 3.0" license; that is, you're free to use the images in the pdfs for whatever you like (including commercial purposes) as long as you credit me by name (Michael A. Cohen) and/or linking to this page.

I created these images while working on a project with my advisor Duncan Callaway to examine the impact of distributed photovoltaic (solar) power on the electrical distribution system. We're grateful to SolarCity for collaborating with us on this project and the California Solar Initiative RD&D program for their financial support.

Key

Notes and caveats

• I've given these a quick once-over to make sure they look reasonable (basically, to make sure they're well-connected) but they should not be considered validated in any way.
• I've shortened the node labels for the sake of space, but they do correspond to the original .glm node names. So, e.g. R1-12-47-3_node_35 becomes node35. I have not labeled the edge-like elements as the graph felt busy enough with just the node names.
• The labels are stored as searchable text in the pdf documents, so you can use your pdf viewer's "find" function to find a particular node by (shortened) name.
• Graphviz does a quite respectable job, but graph layout is a very difficult problem and no algorithm is perfect. In particular, there's no checking for edge crossings so you'll sometimes see lines cross in a way that's unrealistic and/or confusing.
• Also, sometimes (especially in dense areas) the labels are not readable due to overlapping with nodes. However, you should be able to read some label in the general area that will help you locate the relevant objects in the .glm file.
• As noted above, the lengths of overhead and underground line segments are roughly to scale. However, since .glm files contain no information about angles or absolute positions, the overall spatial arrangement of the nodes is arbitrarily determined by Graphviz (with the constraint that the individual segment lengths are preserved).
• The "edge" components other than overhead and underground lines have a constant, arbitrary length; long enough to be visible, but short enough not to distort distances too much.
• Using the same scale for all feeders involves some trade-offs (e.g. the load boxes in the GC feeder are quite large, some of the rural feeders create very sparse pdfs) but I decided to stick with it for now for simplicity and to aid intuition in comparing one feeder to another.
• If you are interested in using (and ideally contributing to) the ruby script that translates the .glm files to DOT, please contact me at macohen@berk_DropThisPart_eley.edu. As you can see, I've chosen to display a fairly limited set of information from the .glm files in the graphs; there's a lot more that could be done with it.

Graph files

The graph files are presented below in two formats. The .pdf files contain visual representations of the feeders; these are generally the best place to start. The .dot files contain the “Attributed DOT Format” (text) output from Graphviz with coordinates attached to each object (in the pos property); these can be useful if you'd like to use the numerical coordinates in some other software, like GIS. See Drawing graphs with dot for documentation of the file format. Note in particular that the pos coordinates are in units of printers‘ points. There are 72 points in an inch, and the layouts were prepared at a scale of 1 inch = 200 ft, so:

real world feet = point coordinate / 72.0 * 200.0

Of course, the origin point for each layout is arbitrary since we don't know the true physical location (or orientation) of the taxonomy feeders.

Last updated 2013-10-03