Squaring the circle: The problem of over-constrained projects.


How to deal with overly-constrained projects

Over at Human Transit, transit planner Jarrett Walker did an article on the proper role of the transit planner: Is (s)he a dispassionate expert, much like an engineer is expected to be? Or should planners and other professionals serve a more activist role–essentially serving as advocates of the transit-riding public, and defending their interests? Jarrett, who has made numerous remarks about the limits of mixed-traffic streetcar (and has been accused, unfairly in my opinion, of being a “bus fanatic”), noted that his job has elements of both: He does prefer to optimize for mobility outcomes, and streetcar frequently fares poorly as a mobility measure; but when he takes on a project he needs to live within the project’s constraints: If a project which hires Jarrett as a consultant is chartered with building streetcars, then he will help the agency design the best streetcar network that they can afford.

But then, an obnoxious commentor (OK, yours truly) threw a wrench into the gears. While I phrased it more politely over there, I asked the essential question: What if the project requirements are nonsensical to begin with? Jarrett’s answer focused on the role of transit planners in addressing all of this; and I defer to his expertise on such matters. Instead, this article looks at the more fundamental problem: projects with fundamentally conflicting, and often capricious, requirements.
Too many cooks

Many public works projects, especially those in a multi-layer democracy like the United States, have many, many stakeholders. And not all of those stakeholders have the interest of the general public at heart, let’s be honest. Politicians love to show up at ribbon-cuttings, and may have ideological axes to grind. Agencies frequently seek to expand their scope, power, and influence. Developers, vendors, unions, and other parties often want to cash in, and frequently aren’t shy at trying to influence decision-makers (often in ways which are perfectly legal). NIMBYs who want it somewhere else.

Even among those stakeholders who are legitimately concerned about a project’s goals, one can frequently find many demands on a project. Institutions who fall into the “golden hammer” trap, where their job involves swinging hammers and thus view every problem looks as a nail. Professional societies frequently have standards and practices which they view as sacrosanct. Diverse communities of users may impose conflicting requirements. If grants are part of the funding package, the granting agency will often impose conditions of their own. And spools of bureaucratic red tape will surround the project, particularly if Uncle Sam is involved.

All to often, public works projects collect so many differing requirements and constraints, both legitimate and not, that running the project is like squaring the circle. (For the non-mathematically inclined, constructing a square with the same area as a given circle using only straightedge and compass, was proven impossible in the 19th century). And this is without taking into account financial and schedule constraints.

Yet projects which attempt to square the circle–which attempt to satisfy simultaneously many conflicting requirements, often dictated by stakeholders with de facto veto power over the project–still happen way too often, often times with disappointing results. At least two prominent projects in the metro area–one primarily highway, one exclusively transit, exhibit signs of being over-constrained–or at least they do in this blogger’s opinion (an opinion that might not be shared, in its entirety, by the other hosts here).

One of them is the Columbia River Crossing. The other? The Lake Oswego-Portland Transit Project.

(To address a likely comment in rebuttal: Milwaukie MAX does not suffer from being over-constrained, at least on the requirements side–the project stakeholders largely agree on the project requirements and design, and the selection of the LPA and route proceeded in a mostly-orderly fashion. There are, of course, legitimate concerns about the cost and financing of the project, and some objections to the project’s existence, but the sort of acrimony over project fundamentals like are found in the CRC case, simply do not exist with MLR).

The Swiss Army Bridge

I’ll start with the CRC, as that project has been a gigantic clusterfudge from the beginning.

The fundamental goal of the project, ought to be conceptually simple. Modernize (structurally and functionally) the primary crossing of the Columbia, providing multi-modal crossing support, while eliminating the draw span. Straightening out the shipping channel on the river is a bonus. But what has actually happened has been a mess.

The first problem is governance. Given that it’s a bi-state effort, there isn’t any single entity which is an obvious candidate to run the project–thus management was given jointly to the state DsOT, with the other interested governments (the cities of Portland and Vancouver, the counties of Clark and Multnomah, Metro, the SW Washington Regional Transportation Council, TriMet, and C-Tran) having seats at the table. ODOT and WSDOT, being agencies fundamentally charged with building and maintaining highways (they do other things, but highways are their primary concern), drafted purpose-and-need statements that pretty much excluded any solution other than a New Freeway Bridge. Throw in a pile of rules from various highway manuals (some of which, I suspect, aren’t as ironclad as frequently claimed), and poof–rather than simply buildling a bridge, the project now involves rebuilding about five miles of I-5, and a Hayden Island cross section comparable to a football field. But its got to have light rail and other “green” features to satisfy Portland–not to mention an Iconic Design™. Cost? Nearly $4 billion, which means tolling. The folks in Vancouver, who would bear the brunt of tolls, naturally objected. And so on. The result, at this stage, seems to be a design that nobody really likes, that has a murky funding picture, that has cost eight figures to produce nothing but paper so far, and which has no end date in sight.

Did it have to be this way? That’s a hard question. One fundamental issue is that the City of Portland objects to a key design goal of the highway departments on both sides of the river–“modernization” of the freeway (a catchall term which includes widening, ramp reconfiguration, and all sorts of other stuff designed to reduce congestion). While some of Portland’s objections spring from ideological concerns that other stakeholders don’t share (“who cares about the environment”, and/or a belief that the alleged impacts of automobile pollution are overblown, are still popular attitudes in some quarters), Portland has a very legitimate concern that redesigning the bridge simply will move the bottleneck south. Portland has its own pro-transit agenda; which ODOT seems more than willing to hold hostage (an ODOT staffer once reportedly suggested that the agency would block any attempt to extend MAX across the Columbia, unless part of a project to widen I-5). And Vancouver doesn’t want to be stuck with an ever-escalating bill.

Part of the present dynamic seems to involve both sides wishing that someone (Governor Kitzhaber, the feds) would “see the light” and kick the other side to the curb.

The Lake Oswego Quit Calling It Streetcar (At Least For Now) Project

Compared to the CRC, the Lake Oswego project is a model of piece and harmony. The “what” of the project was largely fixed: a streetcar line, running from the current end of the Portland Streetcar in SoWa, along the old Jefferson Branch line to Lake Oswego. The project goals make sense: Use an existing asset (the rail right-of-way) to leverage federal funds, and build a transit service running in exclusive right-of-way which ought to be faster than local bus service on Highway 43. Demonstrate the potential of “rapid streetcar” as a budget alternative to light rail for shorter corridors. A no-brainer, right? Unlike the CRC, where leadership was distributed among a handful of agencies with contrary goals and a decided lack of trust, the involved government agencies (TriMet, Metro, and the cities of Portland and Lake Oswego) aren’t fighting over the project requirements.

But the devil, as is often the case, is in the details.

The most fundamental issue is that the project is promoted as rapid transit–as an upgrade over the existing bus service–a premise which has been undermined by the proposed implementation. The need to be an extension of the existing local-service Streetcar system, which offers local-stop service along the entire route, and which bypasses the main transit corridor downtown (the mall). The plan to use existing Portland Streetcar rolling stock, which are optimized for mixed-traffic application. The desire to provide front-door service to John’s Landing merchants, and not upset local condo owners who don’t want trains (Never mind that the rail line has been there far longer than the condos) whizzing past their front door; both of which require a detour onto OR43–the same highway which is predicted to turn into a parking lot in the near future, justifying the mobility need for the project in the first place.

Unlike the new MAX line directly across the river, which is designed to function well as rapid transit until hitting downtown, the streetcar is not so designed–and actually offers a performance decrease over the #35 bus between Lake Oswego and downtown, and that’s without considering the need to transfer (Bus service will be curtailed at Lake Oswego to provide operating funds for the streetcar line). The streetcar does offer modest capacity improvements over the bus, and has the cachet of being rail (a phenomenon which is oft-debated in transit circles, both as to accuracy and relevance, but will likely have an impact given the demographics along the line). But the mobility improvements of the project are close to nil; and for longer-distance commuters on the 35 and 36, probably a net negative.

(Perhaps land-use improvements alleged to flow from the project will be worth the local investment, though much of the area along the line is already developed or not suitable for development. Perhaps the ability to get a big check from Uncle Sam for a minimal local cash contribution–given that the Feds are willing to consider the value of the right-of-way in calculating their match–makes the project worth doing. This is a difficult case to make, however, to the transit-riding public, who tend to care more about headways and trip times than they do about property values).

Signs of an over-constrained project

Back at HT, another commenter posed an interesting question: How do you know if a project has requirements or constraints that make it difficult to do a good job? The question was posed in the context of bad-faith requirements (such as developers engaging in rent-seeking), but the answers also apply to good faith attempts to square the circle. My response to her is here; the answers are also reproduced below, edited for brevity. (In particular, observations about the CRC and LOTP which are redundant with the criticisms above are excised; if you want to see the original answers, click the link).

  • Overly constrained initial project requirements. It’s useful to distinguish here bona-fide requirements from design/implementation details, the latter of which ought to flow from the former. But sometimes, elements which ought to be details are set forth in the requirements without adequate explanation of why this should be so. Sometimes these requirements aren’t stated explicitly, but still are constrained enough that only the solution preferred by the powerbroker can meet them. [CRC used as example]
  • Bizarre decision criteria which may not match the stated goals of the project. For example, publicly identifying a project as “rapid transit”, then de-emphasizing performance criteria, or basing decisions on highly speculative future estimates. [LOTP used as example]
  • Thee presence of strawman alternatives in the DEIS (Draft Environmental Impact Statement) or equivalent planning/analysis document. By “strawman alternatives”, I mean proposed alternatives which are obviously bad, and included only to satisfy process requirements that multiple alternatives be studied in depth. [LOTC “enhanced bus” used as example; though in fairness, NEPA did require a bus option beyond no-build be studied. That said, a more robust BRT solution was excluded from analysis early in the project..which lead us to the next item…]
  • Viable project alternatives rejected early in the planning phase, often due to being “out of scope” (see the first item concerning overly restrictive requirements), or on the basis of vague or overly-picky technical factors. Look for signs that point to “this would work, but we really don’t want to (or can’t) consider it, so we’ll dispose of it as quickly as we can”. [Many have alleged that the proposed “supplemental bridge” alternative to the CRC is another example].
  • Projects that appear “out of the blue”, rather than the result of organic planning activities, or which are done “out of sequence” compared to their apparent priority. May represent a unique opportunity (such a project eligible for funding that isn’t available for other projects)–or someone with his thumb on the scale.
  • “Economic development” being touted as an advantage is a frequent red flag. It is always touted, of course, as politicians love to boast of bringing home the bacon, but if “economic development” is the main reason for doing a project–and especially if the “development” in question refers mainly to the project’s construction effort itself and not to post-project activities the work will enable–a good response is to ask if there are any places to deploy the “economic development” that will have better post-project outcomes. Spending money on projects and employing hardhats is often a good thing (especially if the money comes from outside the region), but it’s nice if the labor and materials go to something that’s useful. Paying someone to dig ditches and refill them can be considered “economic development” but doesn’t produce something useful; better to pay people to build useful things.
  • And one other, not in the HT article: The use of unproven or untested designs or methodologies in the project, or anything dubbed “experimental”. WES was experimental in many respects, and until recently, the CRC was considering an experimental bridge design, until cooler heads prevailed.

Of course, not all over-constrained projects are failures. Westside MAX had some annoying constraints placed on it (the King Street station comes to mind), but is overall a successful project. OTOH, had ten extraneous stops been sprinkled along the line between Portland and Hillsboro, would the line be as successful?

Dealing with over-constrained projects

What to do about all of this? The hard-and unfortunate–part of the overconstrained project is that often, we have to live with them. It’s easy to fantasize about driving bad actors out of the process, and about having strong visionary leaders who have the foresight and the clout to sweep conflicting requirements out of the way, without losing support for the project–but such individuals are rare, and quite a few of ’em use their political talents for less honorable purposes. But a few suggestions come to mind:

  • Governance matters. It’s hard with multiple stakeholders, but having someone competent in charge does help. In the case of the CRC, it seems to me that the first step to fixing the project is for the stakeholders to jointly hire an outside project leader; one who has no particular ties to either Oregon or Washington, or to the various modal factions, to lead the project. Of course, for such a leader to be effective, the various agencies will need to cede a fair bit of authority to said leader; I’m not sure any of them are willing to do so at this point
  • Sunlight is the best disinfectant. A transparent process, one where decisions can be easily traced to inputs and planning work products are available for inspection, may help cut down on–or at least expose–some of the pettier forms of backroom dealing. Many bad actors don’t like being subject to public scrutiny. Transparency also helps good-faith projects avoid accusations of backroom dealing; virtually every large capital project gets accused of being done in order to grease someone’s palm; an accusation which is frequently not true.
  • Be prepared to say no. The City of Portland has won some concessions on the CRC with this tactic–but if a project is really going of the rails, cancellation should always be an option.
  • Bifurcation and phasing may work. A controversial and difficult project can sometimes be split up into two separate projects. I’m not sure this would work for the CRC, given the lack of trust, but the idea that the LO Streetcar should have two separate lines (a local service to Macadam, and a limited-stop service to Lake Oswego) has some merit.
  • Better advocacy for users. One of the unfortunate parts of transit advocacy in the Portland area is a lack of effective organization of transit users. Freight users of the highway system are well-organized, and often asking government for better freight mobility. The auto lobby is likewise strong and forceful. Even the cycling community in town has levels of organization. Transit users do have some organized advocates, such as OPAL, but many of these activists represent subset of the overall transit community, not transit users as a whole. (Depending on TriMet to adequately represent the interests of its riders is, unfortunately, not a wise thing to do).

9 responses to “Squaring the circle: The problem of over-constrained projects.”

  1. In the past few years I have followed two major bridge projects pretty closely: replacing the Sellwood; TriMet’s new transit span.

    Fred Hansen said he asked Vera Katz to oversee committee work, which she did with discipline, fairness, aplomb. A casual observer like me could observe engineers, architects, maritime lawyers, the Coast Guard, tugboat operators wrestle with real requirements and realistic constraints. I even got a chance to present an alternate design.

    Superbly done!

    On the other hand, Multnomah County’s process on the Sellwood seemed like an exercise in public relations. This was so because Mike Pullen, the County’s PR chief, was the project runner. To be fair to Mike, perhaps he did not relish the job and was doing the best he knew how. But the heavy hand of ODOT clearly was calling the shots, to the tune of a worthless $134 million interchange, the main element bloating an essential $80 million bridge into a $290 million highway project.

    Sellwood has been the CRC writ small.

    Kudos to TriMet. Boos to ODOT.

  2. Is it ODOT’s position that only a grade-separated interchange at the west end of the bridge is acceptable? Given that the east end of the bridge is Tacoma Street, a city street, and that OR43, while a state highway, is one with lots of signals on it–might a lower-impact design, such as a rotary with a bypass lane for northbound OR43 traffic, been acceptable?

  3. Following the prior thought: If ODOT is in the position to demand that the city install (and pay for) expensive interchanges and other traffic-mitigation elements on a city-owned facility which happens to connect to a state highway–shouldn’t the reverse be true? Shouldn’t the city have the right to demand that when ODOT wants to upgrade an intersection or highway within the city limits, that said project include significant enhancements for alternate transport? I’m not just talking about bike lanes painted in the shoulder or other “easy” design elements; I’m also talking about things like queue jump lanes and signal priority for busses at intersection upgrades; new bus lanes if a highway is widened, etc.

    After all, if no expense is spared to try and ensure free-flowing auto traffic, a similar attitude ought to apply to other modes. The best way of reducing auto congestion, after all, is for fewer people to drive.

  4. Scotty,

    You’ve pretty well summed up many of my concerns on the LOTP project. I wish that I could approach it dispassionately. As one who literally will be forced off the bus and onto the tracks, I haven’t been able to do that.

    A serious question: Have we ever had a transit project before which was designed to reduce mobility for most users?

  5. Your legal points are well taken, Scotty, but I do not have answers to them.

    As the site at Sellwood is a smaller scale version of Saint Johns, complete with a numbed highway teeing into a bridge, I cannot see why a similar intersection at its west end would not work perfectly well. Saint Johns is four lanes with high usage by heavy trucks. Sellwood never will carry such traffic.

    Simple and cheap are best.

    Let’s keep this thread going.

  6. I don’t really have personal interest in the LOTP outcome (I rarely travel in that corridor), but I think maybe the best solution is to delay the project for now, and split the streetcar in two when it’s built later.

    I commented on this on another thread, so I won’t go into detail here, but basically, I think the Streetcar is better approached as two separate projects. One is the Lake Oswego Rapid Streetcar, which would have custom-built “long” streetcars (longer than a single MAX unit) with extenders to serve MAX stations and Tri-Met livery. Share the Milwaukie MAX line from Union Station to South Waterfront, and then use the Jefferson Branch line to reach Lake Oswego — very briefly sharing track with the Portland Streetcar.

    Project two is the Johns Landing Streetcar extension, which would be a City of Portland project using Portland Streetcars on or near Macadam for local circulation and development purposes. It would go most of the way to Willamette Park, and would briefly share the Lake Oswego Rapid line at a few places.
    (The Johns Landing Streetcar could later be extended south along the Rapid Track, with no stops, as far as the Sellwood Bridge, and then cross the river to provide local circulator service around Sellwood).

    Seems to me that two intersecting streetcar projects would handle MOST of the demands from stakeholders, except the condo owners who don’t want the rail line used ever again. (Zero sympathy from me; they saw the tracks when they purchased their homes. They had fair warning that rail traffic was a possibility in the future. They should be thankful it won’t be freight trains with loud blaring horns.)

    Also, Tri-Met shouldn’t eliminate local bus service on Macadam to pay for streetcar operations. Better to delay the project until Tri-Met can afford both rapid and local service along the corridor.

    Really, the constraint is the decision to run ONE streetcar line instead of two. A rapid (MAX-lite) and a local streetcar can be funded and built separately, the second piggy-backing on trackway from the first, and opened up as operating funds become available.

    As for the CRC, most of the targets of the project can be met with a MUCH less ambitious bridge than proposed. It can even be “iconic” once it’s scaled down to reasonable size. Simply leaving the existing bridge in place will provide an inexpensive way fro transit and bike traffic, as well as local car traffic, to cross the river. The problem with the CRC all along has been the DOTs insistence that the project HAS to be everything jammed into a single massive bridge, instead of just saying “here’s the functionality we need from this project” and figuring out the best way to do it, preferably in phases.

  7. A little technical history of Saint Johns Bridge:

    Apparently the original plan was for a trussed cantilever. David Steinman, one of this country’s two certifiable genius bridge designers, offered to do a suspension span for less money. He delivered, on time and under budget. Also, he generated the archetype for modern suspension bridges. Saint Johns’s gothic skiamorphs are controversial to this day; appearance alone gets it into nearly every list of the world’s great bridges. But wizzardly engineering is what carries the great loads at low expense. All this in 1931.

    A decade later the other certifiable genius, Conde McCullough, wrote a paper, Technical Bulletin 13, for the Oregon Highway Department, explaining how to design short and medium span suspension bridges so that the stiffening truss carried part of the load to the towers by beam action, making them very economical for such spans. I conjecture that he was codifying what Steinman had done at Saint Johns.

    Suspension bridges look simple, but are very difficult to design, for they are compound bridges: one bridge–the stiffening truss–within another–the main cables and suspenders. McCullough used the preeminent tool of mathematical physics, the calculus of variations, to optimize the balance for short and medium spans.

    David Steinman built the first modern suspension bridge of medium span (1,200 feet) at Saint Johns. Conde McCullough, in Salem, explained how it works.

    The western approaches to Saint Johns are extremely constrained. The bridge is very high, and there is a massive cable anchorage, equal in size to the huge one in Cathedral Park, buried beneath the abutment in steep basalt. Technically, there is no other way to get vehicles up to the western end.

    And both eastern and western approaches work very well indeed. Sit at the stop for the 17 bus at the eastern end and be amazed at the constant flow of heavy industrial vehicles zipping by. Simple designs always are more functional and versatile than overwrought ones. The bridge at Saint Johns is a supremely integrated design: functional, durable, beautiful, and cheap!

    Conditions at Sellwood are far different: landslide at the western end, developed real estate at the eastern end; no possibility of cable anchorages at both ends. But a self-anchoring suspension bridge would finesse both constraints, obviating vacating private real estate to the east and allowing the landslide to proceed at the west. Functionally and visually (but not structurally) it would be a down-scale version of Saint Johns.

    A final word. The concept of “constraints” comes from analytical mechanics, a fundamental branch of physics. One develops a function(al) of energy or power, ascertains boundary conditions–constraints–and applies a criterion to be met. Much of mechanical and structural engineering depends from analytical mechanics. The same sort of procedure can be applied to overall aspects of design.

    The art lies in ascertaining the constraints. Over-constraining leads to over-complicating. This is Scotty’s point.

Leave a Reply

Your email address will not be published. Required fields are marked *