Sunday, July 4, 2010

An unorthodox (and maybe crazy), but Really Simple idea for the CRC (and beyond)

It's time to take a little time off from the Independence Day (in the US) festivities to make a suggestion--that will likely be dismissed as crazy.  It's a proposal for the Columbia River Crossing that will strike many as absurd on its face--especially those who believe that a key goal of the project is to keep the freeway moving at "freeway speeds" for as many hours of the day as humanly possible.  It's an idea that some might not only consider foolish--but unpatriotic.  (The fact that certain ideas for bettering the country are considered treacherous in some quarters is itself idiotic in the extreme, but that's another topic...)

It's an idea that I can't take credit for... but I'll make the suggestion anyway.

A few weeks back, I had an email conversation with Jarrett Walker of fame on the Columbia River Crossing, a conversation which led to this post on the controversial project.  Most of the content of the email exchange went into the posting (after much editing and expansion), but one exchange was left out.  That exchange is the subject of this post.

Why is it so expensive, anyway?

One reason the project cost is so high is not the cost of the new bridge itself, but the cost to essentially rebuild I-5 and all its interchanges from SR500 on the Washington side, down to the Interstate Avenue interchange on the Oregon side.  That's a distance of almost five miles.  Why are ODOT and WSDOT so eager to pour $4 billion worth of concrete?

I-5, as currently designed, is functionally obsolete--there are numerous ramps with short distances between them, and ramps with very short merging lanes--particularly the two (from Hayden Island and SR14) immediately preceding the current bridge in both directions.  There's about seven onramps and offramps in each direction over the stretch, including an infamous 400 degree ramp (!!!) from OR99E northbound onto I-5 north.  (You cross over I5,  turn right, pass back under the freeway, and then do a 270 back onto I-5--at which point you have to merge quickly if you're not headed to Jantzen Beach).

Through the magic of such things as collector/distributor lanes and braided ramps--and a vastly widened bridge, the two state DOTs (or should that be DsOT?) intend to make I-5 once again compliant with Modern Design Standards.  Of course, it goes without saying that this sort of freeway construction requires a big footprint, a lot of concrete--meaning it does a lot of damage to surrounding urban fabric and costs a lot of money.

Whose design standards?

On my recent trip overseas, I noticed that many freeways in Hong Kong and China--both places are building them crazy, aren't built to anything remotely resembling US design standards--yet appear to be reasonably safe and functional.  A six lane footprint (3 per direction) was common, ramps were frequently closer than the 2 miles (3.2km) that DOTs and traffic engineers here prefer for ramp spacing on arterial freeways.  So I asked Jarrett, who's been in far more places than I have, about this:

Which gets me thinking:  Are US standards for freeway design to strict?  Is safety over-emphasized--a common issue with wealthy countries?  (It's interesting to compare China and India--for example, see this article).  Or is this an example of mobility values (in the highway sense of the term) trumping access?  For a highway which penetrates the downtown core of a major city, it seems to me that access concerns ought to predominate--which means more ramps and lower speeds, not designing the thing so that through traffic can blast through at 55MPH (90 km/h) during rush hour, but with a tremendous footprint that wholly disrupts neighboring communities.  Should highway engineers distinguish between "through" freeways" and "access" freeways--right now, pretty much all freeways are classified as throughput-focused arterials. 

And Jarrett responded:

Re road standards for merges, etc, I think you're on the right track.  Most standards for these things presume a "design speed," and you can reduce the amount of facility that the standards require by reducing the design speed.  For example, there is some speed (of the entire freeway) at which the southbound ramp from SR 14 would be workable.  What is it?  30 mi/hr?  It seems to me that one no-project solution -- as long as the thing isn't really in danger of collapse -- is just to lower the speed limit to whatever the road geometry can serve.  Throughout the CRC debate, any reference to highway design standards should be met with queries about design speed, and whether that speed is appropriate.
While I'm not one that favors a "no-project" solution--no project means no transitway (whether bus or rail), no decent pedestrian or bike facility, and continued interference with the Columbia navigation channel.  I'm in the "do something" camp, just not the "build a 12-lane behemoth" camp.

One foot on the brake, and one on the gas.... 

So here's a thought experiment:  What if I-5 were signed for a lower speed for a couple of miles; and the DOTs could just do a bridge replacement without having to rebuild the entire freeway as currently proposed?  (Or could build less?) What affects would that have?  Would the existing freeway, sans the current bridge, function adequately at the lower speed?

Some of the consequences would be:
  • Through trips on the freeway through the 35MPH section would take about 50% longer, assuming in either case, motorists routinely travelled 5MPH over the posted speed.
  • Highway throughput (per lane) might decrease slightly.  A good approximate value for the throughput of a highway is 1800 vehicles/lane-hour, which is easily derived when one considers the two-second rule.  If one vehicle passes by a point every two seconds, that's 3600/2, or 1800 per hour.  A slightly more accurate formula (accomodating for the fact that cars take up room) is 1800*x/(x+5), where x is the speed in MPH--if x is 60, that comes to 1661 vphpl, at x=40 we get 1600 vphpl.  More detailed analysis of this requires traffic engineering chops I don't possess (I'm not that kind of engineer).
  • Lower speeds may turn marginal ramps configurations into acceptable ones.  The length of ramps (and of merging lanes) is dictated by the need to provide room for vehicles to safely accelerate to (or decelerate from) freeway speed, and in the case of onramps, successfully merge with freeway traffic.  For loop ramps such as ramp from SR14 to I-5 south, and any ramp with with metering signals (essentially, all of them in the project area), the portions of the ramp which precede the signal and the straightaway don't count.
  • Lowering the speed limit wouldn't help the fuel economy of combustion-powered passenger cars much, and for some models may hurt slightly.  Two important factors for fuel economy of a combustion-powered vehicle are its aerodynamic properties, and its transmission.  Most passenger cars are sufficiently aerodynamic that drag doesn't dominate fuel economy until speeds exceed 60MPH (97km/h) or so--and given that, transmission gear ratios are chosen so that the car performs well in the highest gear at that speed.  A highway speed which prevents the use of high gear would negatively impact the fuel economy of combustion-powered vehicles.
  • On the other hand, diesel-electric and electric powered vehicles, as well as some types of hybrids, have different considerations  Internal combustion engines generate low torque at low RPMs (and stall outright if engine speed drops too low), and thus need mechanical transmissions or torque converters to keep the engine turning at adequate speed regardless of the vehicle's speed.  Electric motors can generate high torque at low RPMs (includng zero)--thus low-speed operation can occur at correspondingly low engine speeds, with the result that fuel economy doesn't suffer from the need to downshift or idle.
  • For large vehicles, regardless of the powertrain, drag will dominate fuel economy at far lower speeds.  Fuel economy for pickups and SUVs tends to peak at 40-45MPH (65-72 km/h); and larger vans, trucks, and busses peak at even slower speeds.  Any reduction in highway speeds will improve fuel economy for these vehicles.
  • Speaking of such vehicles, they often take a long time to accelerate to typical freeway speeds, causing potentially more disruption when merging in and out of traffic at higher posted speeds.
  • At lower speeds, the storage capacity of a highway increases (the number of vehicles which can be accommodated on a given chunk of concrete).  Storage capacity is maximized at low speeds, which is one reason traffic jams are so popular.

Where else?

Lowering the speed limit might help with some other stretches of substandard highway in the Portland area as well.  OR 217, linking US26 in Beaverton to I-5 in the Tigard/Lake Oswego area, is a notorious parking lot, owing to having about 10 interchanges along its 7-mile (11km) length. In Beaverton, there's a stretch of four consecutive interchanges (Denney, Allen, OR 8/10, and Walker) where one ramp's onramp and the following offramp are each about 1000' (300m) apart.   Last year, ODOT considered closing numerous ramps along the freeway during rush hour, to improve the highway's performance--an idea which was unsurprisingly unpopular with Beaverton and Tigard residents.

The problem with ODOT's proposal is that it focuses too much on mobility, and not enough on local access. While 217 serves an important role connecting the high tech Beaverton/Wilsonville corridor with I-5, it also serves much local traffic as well.  The area doesn't have anything resembling a fully-connected street grid, and the parallel local transit service (WES, and the 76 bus which is well-used, but infrequent and slow) leaves much to be desired. 

Plans to widen the freeway are in the works; although it is rather early in the process--and simply widening the highway will not fix the problems, as the study report indicates.  The report in question also calls for "interchange improvements"--braided ramps and such--if and when funding becomes available.

If instead of doing this--could simply slowing down the highway, to make the existing ramp configurations safe(r)--improve traffic outcomes at a fraction of the cost?  End-to-end commute times between I-5 would increase by several minutes in the worst case; but in times of congestion, the effective speed limit on the highway drops dramatically due to the numerous merge/weave conflicts--and that's without an accident occurring.


It is widely assumed, in the United States, that 55MPH (90km/h) or faster is the default minimum speed for freeways, without good reason otherwise.  Mitigating dangerous road conditions due to geography is often considered acceptable; Interstate 5 is signed at 50MPH (80km/h) through the Terwilliger Curves (a notorious set of S-curves on a fairly steep grade, where the freeway passes through the Tualatin Mountains just south of downtown).  However, the idea of mitigating dangerous road conditions due to excessive access seems to be out of scope--many take it as an article of faith that the proper response in these conditions is to redesign or eliminate the access in question--even if it costs billions of dollars and greatly increases the highway's cross-section.

But there are plenty of examples of freeways with lower speed limits, many of them labeled as "parkways".  In Washington DC, for instance, one finds the George Washington Memorial Parkway and the Clara Barton Parkway--both controlled access, divided highways which look and act like freeways; yet are signed with speed limits ranging from 25-50 MPH (40-80 km/h).  Even the Baltimore/Washington Parkway, a major freeway between DC and Baltimore, is signed at 45MPH for quite a bit of its length.  Yet all three roads are widely used (and widely accepted) by Washington-area commuters.

Bottom line

The bottom line for this post is the following:  Freeways perform fine at lower speeds.  The most important attribute of a freeway is not a high speed limit, but controlled access and grade separation--the free flow of traffic.  These attributes make safe travel at high speed possible--highways in urban areas which have high limits but at-grade intersections or unlimited property access tend to acquire reputations as deathtraps--or as ODOT likes to call them, "safety corridors".   However, high speeds are not necessary for a freeway to function as a freeway--and eliminating things like stoplights and driveways will go a lot longer to improving the speed of a highway than simply raising the limit.

And in the urban context, where the regional mobility function of a highway frequently comes into conflict with the access functions, attempts to simultaneously improve mobility while keeping speeds high, generally are expensive--and interfere greatly with the surrounding community.

Trading speed (or other performance measures) of for reliability is a common solution in many engineering domains, not just traffic engineering. Yet the tradeoff seems to be something which is regarded as unthinkable by many.

Which is unfortunate. We could vastly slow down our need to engage in expensive and destructive highway re-designs... if we could simply slow down.

[edited for typos]


  1. Yes! Our state DOT is doing the same thing with four different projects in NYC. Building forgiving highways. I'm told by some transit advocates that the feds insist on "Interstate standards."

  2. State DOTs love to blame overdesign on "fededral standards", I'm told--when in truth, many standards can be waived. It's easier to get a waiver when grandfathering an existing design (such as the Interstate Bridge; a drawbridge built in 1918), but the rules aren't quite as strict as they are frequently made out to be.

  3. I believe there is a two-pronged problem with this proposal. Well, three-pronged if you subscribe to the "it makes perfect sense, so it will never work" school of thought. :)

    1) I believe that, to be considered an Interstate Highway (and eligible for Federal Interstate Highway funds), a highway must have a speed limit of at least 55mph. I recall as a wee tyke learning that the reason there was a long stretch of I-84 in Hartford, Connecticut that didn't have the traditional roadside reminder signs that it was I-84 was the presence of a large 45mph curve: the road stopped being I-84 for a couple of miles, and was paid for entirely by the state, because there was no other way to thread the road through the city.
    If that is true, then reducing the speed limit on the crossing may actually increase the cost to the states, as the federal funding will go away.

    2) As I understand it, once something is part of the Federal Highway system, it is virtually impossible to get it out of that system. Quite reasonably, the Federal Government figures that if they paid for part of the construction cost of a road with Interstate Highway funds, that road should remain part of the Interstate Highway system until they have received full value for their investment. Slightly less reasonably, they interpret that to mean "forever". So reducing the speed limit on the existing road might be extremely difficult, as the Federal Government will contend that it was built with Interstate Highway funds and thus needs to remain part of the system (and, as I said above, that means a minimum speed limit of 55mph).
    It isn't actually impossible: I know that a stretch of road that was built as part of I-84 in Connecticut got re-designated as US-6, but I have no idea what sort of voodoo rituals might have been involved in that deal.

  4. As noted in the article, I-5 in SW Portland, through the Terwilliger Curves, is presently signed at 50MPH; it used to be signed at 55MPH IIRC but was downgraded in speed as a safety measure. This isn't an advisory (yellow sign) speed, it's a mandatory (white sign) speed.

    In that case, geography is an issue; the freeway descends through a ravine, hangs a sharp left, and then hugs the side of a cliff as it descends into downtown Portland. Google map here.

    Now whether "excessive urban fabric" or "functionally obsolete design" can be used to obtain a similar reduction in speed is an interesting question; given that we've established that "challenging terrain" will work. A lot depends on the desire of the DOT; and in the case of the CRC, both DOTs really really want to rebuild the freeway.

  5. Works for me. Your moles in various DOTs may be able to point out other examples where freeway speed limits drop dramatically at especially dangerous curves. I don't see any justifiable distinction between a dangerous curve and a dangerous merge.

    I also suspect that the design speed ODOT is using is higher than the speed limit. Because driving 10-15 over the limit is "normal" in the US, DOTs tend to design for 70 if they're going to post the speed limit at 55. So to post at 35 they might design for 50. I'm not a highway engineer, but I bet that would make a difference.

  6. I'm assuming that the usual distinction between "design speed" and "posted speed" is observed. That said, as the link mentioned above by Cap'n Transit points out, that practice is often an invitation to speeding...

  7. The Central Freeway in San Francisco was intended to be part of Interstate 80. The tight curve where it joins the Bay Bridge approach (still Interstate 80) is signed for 35 MPH and it really isn't safe to go much faster than that.

    The Western Freeway was never built, but it was intended to be Interstate 280 and had a design speed of 45 MPH in an attempt to do less property damage than a typical Interstate design.

  8. Another interesting example in Oregon--albeit not an Interstate--is the Bend Parkway (the present alignment us US97 through Bend). The Parkway is a highway which isn't particularly scenic, looks and acts like a freeway for much of its length (it has a few signalized intersections in the southern part of town, and a few "RIRO" intersections)--and is signed for 45MPH.

    The lower speed limit seems to be tolerated--evne though it is a vast improvement over 3rd Avenue (the old alignment of US97 through the city).

    ODOT seems determined to build a continuous freeway (or something resembling it) pretty much from Redmond to LaPine.

  9. I'm not sure that design speed is the only issue. There is still a lot of room to debate design standards at a given design speed, and there's also room to debate the level of congestion that is appropriate to plan for.

    On standards, there are differences between states (states set design standards, not the feds). If I understand correctly, ODOT has a 12' standard for shoulders, while WSDOT requires 8-10'. There seems to be some room for negotiation there.

    The big question on I-5 - from my point of view - is why it's necessary to combine local and regional travel onto the same facility, Adding traffic just before the bridge that exits just after crossing the river is the major reason for all the added traffic and weaving going on in this project, Why does it all have to be combined into a single massive bridge? Why can't there be a local facility for the local trips and a regional facility that's sized to carry the regional trips (which are limited by bottlenecks to the south)?

    This seems to be a question of aesthetics more than anything else. Massively scaled bridges are hard to stomach. There is always the temptation to add traffic just before a bridge crossing, and to put in tons of access points in close proximity nearby. But that leaves a huge footprint, A 35 mph design speed won't reduce that much, since lanes are still 12', and you won't reduce the number of lanes needed just by lowering the speed. It makes more sense to me to separate the local and regional movements onto separate facilities, and to reconsider what level of congestion is acceptable.


Keep it clean, please