I've never dealt with anything like this but I've thought about it before.
I think what is more important than an algorithm is the real-life data about traffic patterns, events, and locations of businesses and residences.
Well timed traffic lights and large interchanges help of course but I think each situation needs individual analysis. Sensors (and maybe cameras) would be important to any traffic system to identify where the traffic actually is at any given time. Real time data is what you need and the ability to use that data to manage things like traffic lights and reversible lanes.
I don't know, just some thoughts of mine on the subject.
GStudio could be used to create a traffic simulator but - again - the simulator will only be useful if it uses real-world data. Data that might be hard to estimate 18 years in the future.