Sunday, August 19, 2007

On Traffic

So, I've been doing the same stretch of traffic for a little while now. You'd think, as a chemical engineer, I should be able to figure traffic out. Everyone says, "Yeah, traffic is just like fluid flow." Whatever. For designing highways, I can certainly see why it's just like fluid flow. However, I don't have overly much experience as a water molecule, so I can't really relate my own experiences driving to fluid flow all that well. However, I have been gathering plenty of personal experience, and I hope someday to understand traffic.

So, the first goal, as a driver in traffic, is to be typically moving faster than those around you. If you're beating everyone else, then you're winning. In order to successfully do that, you need to have some sort of insight that the other drivers don't.

Let's start by examining traffic as a fluid, just because that's how everyone starts. Unfortunately, traffic only somewhat follows the concept of viscosity. At higher speeds, it follows it very well, but at low speeds the viscosity is clearly non-linear. At high speeds, there is a tendency for the average lane speed to increase as we move away from the slow lane (which is the slow lane because it technically borders the "wall", which is where cars going significantly slower than the fast moving traffic enters). The fast lane basically gets to be the farthest from the wall, so it typically moves the fastest.

At slow speeds, the "wall" is no longer moving any slower than the fluid and so it can no longer determine the minimum speed. In other words, we typically think of each lane as insulating the lane to its left from the slower moving traffic on its right. In very slow traffic, however, this does not occur because those lanes are not necessarily moving at differing speeds.

Intuitively, this means that we should think of the traffic as block or plug flow. The problem with this assumption is that it treats each lane as moving together. Obviously, this doesn't occur. Oftentimes, there's lanes not moving while other lanes are moving. If we liked dreaming, we could try to assume that over some time or distance, all the lanes moved together (so, a time averaged plug flow). However, this is a poor assumption that is both hard to test in action and would require knowledge about the traffic conditions ahead in order to use to our advantage. Essentially, this would be giving up and saying pick a lane and stick with it (which is not a bad strategy, but we want to do better).

In traffic, we have to extend our analogy beyond just fluid concepts. The lanes are moving slowly, but somewhere up ahead, there is no traffic and we are all slowly moving towards it. We'll have reached it once all the cars in our lane have already reached it (since traffic is stop and go, each time we stop, we're waiting for the cars ahead of us to move). The theoretical number of cars between us and where it opens up determines how soon we get out of traffic (even though, many times, our exit comes before the traffic gets better). This view can provide some important insights.

First of all, it explains why the fast lane (here on, called the number one lane) is not truly the fast lane during traffic. Most people get into the number one lane for the long haul, because they believe it will eventually be the fastest lane (since it's the fastest lane during regular traffic, and they are stuck in their fluid flow conception of traffic). The result is that the number one lane normally has people moving into it, but only occasionally has people coming out. The result is that the theoretical number of cars ahead of you in that lane isn't decreasing as fast as it is in some other lanes.

Next, let's look at the slow lane (or, number four lane). This one experiences a wide variation in speed. Ahead of an exit, it tends to move very fast. The theoretical number of cars ahead of you is decreasing as people get off the freeway. Ahead of a busy entrance, it basically stops. The rate at which cars are being added to the lane is very nearly too much. However, almost no one actually stays in the number four lane, which is how it can recover from entrances. One of the primary reasons a lot of people don't give the slow lane a chance is because it is moving the slowest when they are in it (right after they get on).

Upon consideration, we immediately see something we like about this lane. It is highly predictable. There are signs telling us when exits are approaching and we know entrances typically follow exits. This is something we could, easily take advantage of.

The middle two lanes are fairly similar. They have people moving both in and out. The number three lane tends to behave a little bit more like the number four lane, but they are very similar. In most cases, if you intend to never change lanes, one of these lanes is a good choice. You can count on it not to be the slowest, and it wouldn't be surprising if it came out the fastest (it's just hard to predict what the case will be).

Now, before we talk strategy, let's talk about changing lanes. Changing lanes is inherently evil. Changing lanes is a completely self-absorbed action that causes pain and suffering to a significant number of people behind you. Personally, I'm against doing things that are so evilly evil, so I recommend taking time to very seriously contemplate the magnitude of your decision whenever you are about to change lanes in major traffic.

Always remember that your maximum speed at any time is basically determined by the speed of the car ahead of you. If they slow down, you slow down. However, the other lanes have little to no bearing on your speed. When people change lanes, two things happen. First, they slow down. They have to do this because changing lanes requires some measure of attention and time and slowing down makes it easier. Next, they move between the lanes. During this time, they are determining the speed of both lanes (by technically taking up two lanes). And they have slowed down. The result is that two lanes suffer because of the decisions made by one driver. It's just evil.

The only way to change lanes without causing this kind of suffering is to do it when there is significant space to move into. By moving into what is effectively an empty lane, no slowing is required and neither lane suffers. Of course, such wide open options rarely occur in traffic. The result is that almost all lane changes cause unneeded suffering to others.

So, our goal is to be able to take advantage of the patterns we can find in traffic, while minimizing lane changes. For now, my strategy is simply to be bouncing between the number three and number four lanes. When the number four lane is going into an exit, it speeds up and it is easy to change lanes into it. As I approach an entrance (and I know when they are coming), I find an opportunity to get out. By knowing the slowing ahead of time, I don't need to get stuck, because I have time on my side.

For now, that's the best recommendation I can give you. As time goes on, I hope to explore the art of dealing with traffic in greater detail.

That is all.


No comments: