General Road and Highway Discussion

[danbrotherston]

Those roads were built in the 60s and 70s, before our "modern" traffic engineering. I think more lanes do improve LOS, at least in models, but turn lanes improve LOS at a lower cost while also improving safety.

More lanes on a long segment of roadway obviously increase capacity (almost by definition); but 2 4-lane roads meeting at an intersection need a much larger intersection (turn lanes) if the intersection is to have capacity matching the roads. There is absolutely no point in building roads to carry 5000 vehicles per hour if the intersections where they meet can only handle 2000 vehicles per hour — the extra 3000 vehicles per hour capacity is wasted. Just like you can’t get a more powerful shower by running 2” pipes in your bathroom water supply.

An “engineer” who thinks that adding lanes to a road will make it carry more traffic but doesn’t understand that intersections are different is an insult to the profession. In other engineering disciplines, they are the ones who would blow up steam engines or build bridges that fall down.

As for LOS itself, not sure what you're suggesting? For one, region actually considers LOS B to be acceptable, but when you say how it is used, do you mean that it rules all? There's an old saying (well, I don't know how old it is) but its something along the lines of you build what you measure. This is highly relevant today (and in my technical job) and it's scary because our level of data collection has allowed hyper optimizations of many things in our society. But suffices to say, if you measure LOS, you will optimize for LOS.

My issues with LOS are basically the following:

1. It ONLY measures cars, not people. I think we all understand this problem.
   
2. It measures peak time congestion, and ignores all other times, which are also important and carry the majority of daily traffic.
   
3. It ignores safety, pollution, noise, induced demand, etc.
   
4. Most serious of all, it measures delay at peak time, and ignores entirely total travel time. Some of the cities with lowest LOS have the highest commute times. It is fundamentally the wrong thing to measure.
   

I suspect the reason we measure and optimize LOS is that it is easy. The issues I raise are much harder to measure for.

Yes, I basically mean the attitude that it rules all. I think it’s perfectly reasonable to use LOS as a way of assessing the state of traffic flow, as long as one understands that all it does is provide a convenient way of characterizing different levels of traffic flow.

So for example, I hope that we would all agree that a project that helps 5 cyclists a day but (somehow; this is an extreme example that probably can’t actually happen) reduces a major road from LOS A to LOS F all the time is no good. But on the other hand, if we can fix a major gap in the cycling network and reduce a neighbourhood collector from LOS A to LOS B at rush hour, that is equally clearly totally fine and it would be crazy to object based on the impact on motor vehicle traffic.

I don’t understand LOS to measure only peak time. Rather, one might say that the road is LOS A most of the time and LOS B during rush hour, or whatever. Of course if we only look at peak time then the analysis is no good in the first place, no matter what heuristics we use to express the findings.

Definitely agree that looking at a city and observing with approval that it’s all LOS A (or being concerned that it’s all LOS D) is a bogus way of evaluating things. Arguably, seeing LOS D all the time everywhere says the road network isn’t overbuilt!

My thinking comes out of the sense I got from some articles (can’t even remember when I read them) that seemed to be suggesting that if planners would just switch from LOS to some other way of measuring traffic that they would suddenly start building bike paths all over the place, and I don’t think that’s true at all. If a planner is car centric, they can perfectly well be car centric with or without LOS. I would rather concentrate on what are our goals, then look at the best ways of evaluating the transportation grid in order to move towards reaching the goals.

I mean, I take a far stronger stand against LOS. I agree that it doesn't automatically fix everything, but I think you'd be surprised the degree to which what you measure does affect what you do.

Basically, I'd argue measuring average travel is the metric we should be optimizing for (after safety and equity for all users is achieved), then you're measuring direct value for people. LOS just measures saving a few people a few minutes every day, while everyone else usually has a worse experience. Again, regions with low LOS often have high travel time. Ultimately travel times is what we should care about.