1. Ramp meters. This has been on the request list for a while I'm sure, and there have been numerous topics about "faking" or "imitating" ramp meter behavior. But more advanced features should be used - alternating lane release, HOV bypass lanes, etc.

2. Better roundabouts. You've heard me say it before! The current roundabout geometry simulation isn't as useful as it needs to be above a single lane roundabout. Additional settings to include here could be splitter island geometry (length, width), pedestrian crosswalk position (often offset one or two car lengths from the from the yield line), as well as more complex striping/lane assignment in the circulating roadway. I consider this very important - we often need to see the interaction of roundabouts with nearby signals or stop signs. It also helps compare apples to apples, rather than a roundabout simulation from a different software (Vissim, Paramics, or a Sidra analysis) compared with a traffic signal run in simtraffic.

3. *.hst file size. when link curvature was introduced, and in networks where it is used on a lot of links, history files get big, and quick. The relatively small hard drives that the state of california has us using (80GB is our standard PC, if you can believe it...) makes data handling on large or long duration network simulations (that generate history file sizes in excess of 1 GB for one run) problematic. Network storage space is at a premium, and simulating actively over the network is much slower.

Eric Royer
Caltrans