Localized Bottleneck Reduction Program
Every highway facility includes decision points, such as on and off ramps, merge areas, weave areas, lane drops, tollbooth areas, and traffic signals; or design constraints, such as curves, climbs, underpasses, or narrow or non-existent shoulders. In many thousands of cases, these operational junctions and characteristics operate sufficiently and anonymously. However, when the design itself becomes the constricting determinant in processing the traffic demand, then an "operationally influenced" deficiency exists. Operational influences are tangible root causes of "recurring" chokepoint delays. Recurring backups are routine to the point of being predictable in cause, location, time of day, and duration.
The FHWA "Traffic Congestion Reliability" reports define congestion as "an excess of vehicles on a roadway at a particular time resulting in speeds that are slower - sometimes much slower - than normal or free flow speeds. (Congestion is) stop-and-go traffic. Previous work has shown that congestion is the result of six root causes often interacting with one another." The FHWA site Focus on Congestion Relief lists the six contributing sources as:
Only the first and second sources contribute to recurring congestion; i.e., they are tangible in design and function, and therefore, candidates for remediation. The remaining sources of congestion are nonrecurring and random. In this context then, a bottleneck certainly constitutes "congestion," but congestion cannot be said to be universally analogous to a "bottleneck". In much the same way that "paper" can be used to describe either a singular sheet or multiple sheets, "congestion" can be meant to describe the result of a local circumstance or an overarching systemic condition. It is not merely sufficient to explain that "high volumes" represent congestion, for this is a term that is relatively applied. High volumes, or at least higher volumes than designed for a facility, including even higher than the design volume plus a safety factor, will overburden any facility, regardless of whether operational influences exist or not. When too many vehicles compete along all segments of a facility, "congestion" will inevitably result, and is overarching. But when only determinant, subordinate segments of that facility are routinely over taxed, then "operationally recurring bottlenecks" within the facility are said to exist.
Among Webster's dictionary definitions of "bottleneck" are i) a narrow or obstructed portion of a highway or pipeline, and ii) a hindrance to production or progress. Certainly the elemental roots of a traffic bottleneck exist in these descriptions; namely, the narrow portion of highway and the hindrance to progress. However, a road need not "narrow" for a recurring bottleneck condition to result: e.g., witness a weave condition, sun glare, or a vertical climb. For that reason and others, in the context of traffic analysis, there are recurring and nonrecurring causes of bottlenecks, which invariably shapes one's definition. Note: across all FHWA sites "bottleneck" is often used interchangeably and idly as a catch-all definition including sometimes even meaning "congestion." One should recognize the context, and distinguish it appropriately as a "recurring" or "nonrecurring" bottleneck. Specific to this site, the reader is reminded that the LBR Program is dedicated to operationally influenced recurring bottlenecks.
In layman's terms, a bottleneck is distinguished from "congestion" because it occurs on a subordinate segment of a parent facility, and not pervasively along the entire facility. It is mandatory only for recurring bottlenecks that "traffic over-demand" be present. "Rubbernecking" past traffic incidents that do not require a lane closure, or simply driving into sun glare, often results in slowdowns even though excess traffic demand may not be present. The mere act of one or more lead vehicles slowing creates a rippling effect; a shock wave that reverberates back to vehicles that are following. In other words, this slowing could be the result of a traffic confluence or the rubbernecking. The slowing reduces room to maneuver, which self-perpetuates the shock wave. The problem begins to clear once past the incident, as vehicles begin to accelerate away, and maneuvering room downstream of the incident increases. One California study found that the mere optical illusion of a bridge that appears lower than it really is creates a recurring bottleneck on Interstate 880 north approaching 23rd Avenue. Truckers slow in anticipation of the "low" bridge, thereby forcing following vehicles to adjust. The location is independent of a classic operational deficiency and can be independent of volume. Nevertheless, the condition creates a bottleneck.
Good question! The FHWA Office of Freight Management and Operations is tasked with understanding the challenges that come with the increasing demand for freight transportation, and improving freight mobility and productivity. Understanding "freight bottlenecks" is a specialized study within that program area. Freight bottlenecks are both unique to this genre (e.g., steep upgrades, truck operating restrictions or limitations, delays at terminals) but also routine (e.g., stuck in traffic at notorious backups) in that they contribute to higher pass-through costs of goods. Therefore, a freight bottleneck is "freight stuck in traffic" but more to the point, the study of the economic loss of those goods being stuck in traffic. Getting freight moving around or through bottlenecks on trucking routes is a focus of that program. One might argue there is little difference in freight or any other user caught in a bottleneck; free up the bottleneck and everyone benefits. But highway service holds such great import to the trucking industry that "freight analysis" is justified as a unique area of study by the FHWA and peer organizations. Whereas, the LBR Program focuses on localized, operationally influenced bottlenecks, the Freight Analysis program focuses on major freight routes and the impact to the economy. If solving an LBR Program "bottleneck" includes solving a major freight-impacting location, all the better.
The LBR Program focuses on operationally influenced locations; that is to say, those that have an identifiable cause, resulting in recurring delays of generally predictable times and durations. The root cause of traffic flow degradation at the subject point of constriction is a potentially correctable solution. The following conditions either exist or help to identify a recurring bottleneck condition.
Fixing operationally influenced deficiencies applies to the fourth of the following four strategies available to combat congestion.
In layman's terms, "actively managing the traffic" means to make real-time adjustments to the facility to "manage" the speed, density or safety conditions thereon. Active Traffic Management (ATM) or Active Transportation Demand Management (ATDM) are brother and sister terms, wherein, the former is typically applied only to the roadway facilities, and the latter is typically a broader integration of a larger pool of related activities, like transit, parking, and driver-behavior elements. ATM enhancements involve some sort of "smart highway" feature that uses real-time speed, vehicle-count, or even vehicle-occupancy data to open or close certain lanes, adjust the speeds on the mainlines, or vary the candidacy to even be in certain lanes (e.g., HOV, HOT, truck-only, etc.) in the first place. In some cases, the special lane is free to certain qualifying vehicles (e.g. HOV's, special-tagged vehicles) but is variably priced for non-exempt vehicles; that is, the price fluctuates as the density fluctuates. In other cases, even the general purpose lanes may be managed; for example, by using ramp metering, reversible lanes, or allowing shoulder use during peak hours. Speed harmonization is the practice of adjusting speeds when congestion thresholds have been reached and slowing and queuing is imminent. Overhead signs on gantries lower the speeds – sometimes for the entire facility and sometimes in gradual reductions, say, every succeeding mile – to reduce the "shock waves" of stop and go traffic. Safety is improved and congestion can be relieved when the traffic stream is more-or-less traveling at the same speed. Typical accouterments of ATM include real-time traffic condition-recorders, variable pricing, congestion, or metering algorithms, changeable message signs (e.g., speed limits, HOV/HOT permissions, real-time changes in tolls), and "EZ Pass" or "FasTrack" card readers. Agencies operate and maintain these ATM lanes from a traffic operations center (TOC) or similar, which have 24/7, 365-day presence. In the case of ATDM (see above) the broader integration of elements like transit, parking disincentives (to dissuade folks from driving and encourage use of public transit, et al) and trip-choice behavioral changes (telecommuting, ridesharing, etc.) can "shift" or soften peak-travel time periods, and can have a positive effect on congestion mitigation efforts.
If poor planning "causes" congestion, then wouldn't good planning defeat congestion? Yet congestion is pervasive; ergo, just about all planning must be poor!
We can almost hear the planning community gathering their torches and pitchforks at that statement! Like most debates, there are probably many facets to this argument. On the one hand, some pundits may point to the strict geometry of a plan (e.g., cul-de-sac neighborhoods vs. grid streets; mixed-use development vs. single-use zoning) while others may focus on the socio-economic aspects of a plan (e.g., car-free cities vs. downtown activity centers, etc.) to claim that the plan is or isn't working as intended. There will always be planners who favor opposing tenets just like there will always be Republicans vs. Democrats, preservationists vs. industrialists, and paper vs. plastic. In our view, as the end-users and managers of the traffic network, it matters less whether the original plan tenets were met, but whether they were managed properly. Actually (and to some degree, regardless) the planning may have been a resounding success but the management of that planning may have failed. Unless one can definitively assign "poor" planning practices (e.g., negligence, unqualified skills, etc.) then at worst, "planning" occurred which may have contributed to congestion**. Most likely, "planning" occurred which requires better management of demand.
Poor "anything" begets the unintended opposite of the original intention. Plans are vetted by a review board and experts, and then adopted by vote. It would be a stretch to say that the plan they adopted was either intentionally poor or even resultantly poor, for one can't imagine a planning panel that isn't knowledgeable of the local history, dynamics, economics, population and business trends. They obviously are going to adopt a plan that they favor and which presumes trends and forecasts. So why then would that plan fail? Actually, it may not have failed from a business, renewal or growth standpoint, but ironically it may be overly successful as evidenced by the demand that may result. With regards to that demand aspect, congestion may "result" but it is most likely the management challenge of that demand that "causes" congestion.
Be careful not to confuse demand with congestion. The former is inherent in any plan, while the latter assumes a component of overburden. That overburden is known as recurring congestion and may be systemic (i.e., overarching or regional) or local. Systemic congestion overwhelms the entire infrastructure network, and may be a result – but not a cause – of planning, while localized congestion overwhelms unique subordinate facets of that network for short and predictable durations, and is more closely tied to tangible operational and design limitations. Mitigating the latter congestion may seemingly reduce the former, but not necessarily in the reverse, for localized chokepoint problems would exist with or without systemic congestion. The same subordinate location that suffers recurring congestion is congestion-free for the bulk of the day, indicating if you will "not so bad" planning. Pundits are eager to assign blame to the planning, when it may be the infrastructure (and management of same) that is not keeping up with demand; for example, in urban areas, is the transit component fully engaged? Are there capable refuges for deliveries, parking, and interconnectivity? Are signals optimized? Do exclusive turn lanes exist? Is the downtown pattern of one-way streets working? The density of all that planned residential and commercial activity must be managed properly. Often, the resulting congestion is a failure (at worst) or a challenge (at best) in managing that activity. In our estimation, congestion is ascribed to have a supply/demand cause and effect, and less-so a planned/unplanned one, for hopefully no one "plans" congestion. There are a myriad of dynamic forces that shape any plan, including execution over time, sprawl, urbanism, gentrification and sustainable development, or more precisely, "non" sustainable development. The good news is that those locations are a design fix away from being cured; something that can't be said about systemic congestion or nonrecurring congestion. The bad news is that it may take much money or right-of-way to fix it. Being of the supply/demand genre, the demand can be muted by adding more supply, improving the efficiency of that supply which exists, or inducing demand-mitigations. But note that neither systemic nor local congestion is necessarily corrected by mere planning absent management of same, which is an inverse way of saying that "planning" alone probably didn't cause either in the first place.
**One must exclude nonrecurring congestion from this discussion, as the congestion derived by those events – e.g., incidents, weather, work zones, special events – are less predictable and disappears when the event disappears. [ Return to note **. ]
The first step in bottleneck remediation is identifying bottleneck locations and the root causes of the bottleneck. Sometimes, the problem is evident, intuitive, or anecdotal. However, within multi-mile corridor congestion, travel demand models can assist in Identifying, separating, and analyzing bottleneck dynamics within the corridor. Traffic analysis tools can mathematically identify the problem areas by analyzing road segments for congestion or poor levels of service. Freeways with traffic detection use archived data to identify where and how often bottlenecks occur, and how severe they are. Historical data is used to determine if the problem is growing or receding.
Determining the root cause of the bottleneck can be accomplished with a range of tools. Travel-time runs and videos of problem areas can be used to pinpoint and measure deficiencies. Micro simulation tools can provide a detailed analysis of the specific attributes of the bottleneck(s) and can assist in determining the impact of alternative solutions. When conducting bottleneck analysis, care should be taken to ensure that:
Here is a sampling of remediations that would apply to low-cost, quick-fix, operationally influenced bottlenecks. (Note: other solutions not mentioned here might exist for larger bottlenecks or systemic corridor congestion.)
In a 2006 survey of state and local agencies, the most frequently mentioned operational bottleneck improvements were ramp metering, auxiliary lanes, and introduction of high occupancy vehicle (HOV) lanes. To what degree these are "low-cost" is for the agency to decide. Certainly, other remediations may serve other bottleneck problem areas.
The knee jerk reaction might be "lack of money." But that's everyone's first complaint about, well, most every problem! In visiting with many states to ascertain if they have a bottleneck-specific program or similar that targets chokepoint congestion, we have found a sampling of reasons.
Are there also "eventless" causes to initiate slowdowns?
Lately within the academic community there has been some discussion of just where the terms "recurring" and "nonrecurring" divide. Historically one "is" and the other is everything else, making them clearly opposites.
For as long as congestion has been studied, nonrecurring congestion has always been defined as unpredictable, irregular or occasional events, as evidenced by incidents, weather, work zones, and special events. Remove the event and the congestion dissipates.
Conversely, recurring congestion has been described as repeating and predictable events, as typically exemplified by routine commuter peak hour backups. For the most part, contributing operational designs like lane drops, merging or diverging at on- and off-ramps, too-short turn lanes and ramps (that, for instance, cause back up congestion onto the mainline), traffic signal delays, and weaving sections take the brunt of the blame. Remove the overburden and the congestion dissipates; which is to say that at "off" times when the overburden doesn't exist, the operational element is unobtrusively innocent. (If the operational design is inherently unsafe it would require attention on that merit). Whether or not this recurring congestion is locally determinant (i.e., bottlenecks and chokepoints) or systemic (overarching due to a vehicle-dense urban area) is less the issue than are the predictable, resulting slowdowns themselves. To paraphrase a favorite observation of anyone stuck in traffic: "I'm not the problem! The rest of you are!"
Unlike nonrecurring events, which for the most part act outwardly upon the traffic stream (e.g., an external event forcing a situational response) some recurring events are merely the result of internal dynamics attributed to motorists' human factors responses. It was described above that a recurring event typically requires two factors; an operationally influenced design and an overburden of traffic. However, to be fair, a third predominant factor may actually be one that is "eventless" to an outside observer; drivers slow – causing a shock wave of slowing behind them – due to unfamiliarity, distraction, or an overabundance of caution; i.e., human factors. Once slowed, this shock wave becomes self-perpetuating.
Some in the academic community argue that a number of exogenous factors may contribute to drivers slowing by rote or reflex, thus blurring the lines between the two descriptions. For example, narrow lane widths or shoulders, two-way traffic, nominally wet pavement via gentle rain, the number of side-street conflicts (e.g., access management criteria) and even night versus day (or, lit versus unlit) conditions, may contribute in affecting ways. The argument says these factors may, by themselves, contribute to unconscious actions taken by motorists that result in other motorists having to adjust, and thereby perpetrating a slowdown that otherwise wouldn't exist under "perfect" (laboratory?) conditions. Is it even fair to compare the dynamic of driving to the conformity of laboratory testing? Or are these reactions merely akin to being a safe and responsible driver?
The inverse arguments are that motorists see these exogenous conditions regularly and should inure themselves to face them as part of everyday motoring. Put another way, these are background conditions that are wholly static and undeserving of blame. Or are they? Can this list of exogenous factors actually be acted upon such that improved lane definition, improved shoulder definition, improved pavement friction and improved night-driving conditions (e.g., streetlights) combine to reduce congestion? Would transportation agencies be willing or able to afford to undertake "corrections" or improvements to hundreds (thousands? millions?) of lane-miles of mostly innocuous parameters on such a scale that it would justify benefiting any resulting congestion reduction? Or should we continue our efforts on the known "thumbprint" locations of operationally induced local bottlenecks and chokepoints that would seem to have a higher benefit/cost return?
A second argument is that a subordinate accident (deemed nonrecurring) that occurs during a peak hour traffic jam (deemed recurring) is a cross-over incident; that is to say that the former would not have occurred save for the latter. The accident may be an extension, outfall, or implicit result (frustration) of the chokepoint. Should these cause-and-effect events be recorded separately or as part-and-parcel of the original backup? Does it matter to the motorists who are still in the backup?
Nonrecurring work zone merge
Recurring merge (typ.)
Are you an "early" or "late" merger?
Regardless of the cause, motorists find themselves in a backup and invariably find it necessary to either merge (e.g., due to a lane drop) or jockey (e.g., due to weaving) to get past the point of the bottleneck release. Heaven forbid they all simply stay in their lane and wait their turn! Murphy's Law of traffic jams is "the other lane always moves faster" so drivers either forcibly merge or they passively allow others to merge around them – and then do it again . . and again, thus the congestion is self-defeating and perpetuates. Vanderbilt1 calls this either "profiteering" (acting in oneís own interest) or "altruism" (acting for the betterment of one's fellow motorists). He describes the ant colony as an example of how a teeming hive of activity can exist, not seemingly in helter-skelter chaos, but often in orderly single-file uniformity; first, heading out to forage food, and then back to share food, all-the-while acting towards a greater-good. This defies the expectation that humans, at least in the free-for-all on the highways, would rarely, if ever, exemplify this cooperation. In nonrecurring-casual backups there may be some hope in the fact that response arrives in the form of on-site enforcement, cones, dynamic message signs (DMS's) and barrels, et al, combining to direct us how, when and where to merge. It seems that anytime order can be restored, the situation improves somewhat. However in the recurring case, in the absence of anything stronger than a static "Merge" sign/symbol (note: advisory, and not regulatory) or its sister sign "Lane Drop," merging is like so much cat herding. On your own? Good luck.
Vanderbilt's book deals with the fact the drivers are their (our?) own worst enemy. Humorists have long-targeted traffic, or rather the human aspects thereto. Dave Barry once remarked, "There is only one traffic law, which is that no driver may ever be behind any other driver." The late George Carlin had a famous monolog in which he observed that "there are only two camps; everyone driving slower than you is an idiot and everyone faster is a moron." And journalist Dan Rather remarked "Americans will put up with anything provided it doesn't block traffic."
Motorists can merge at-speed early, at-speed late, at-crawl early, or at-crawl late. The terms "early" and "late" refer to leaving one's lane either upstream well in advance of- or moving up in the dropped lane to merge at last possible chance. We take it as significant that in Vanderbilt's book, which is principally about drivers' habits, distractions, and reactions to traffic circles, signals, signs, roads, parking, speeding, vehicle ownership – and yes, ants, et al – that he chooses the topic of merging as the headlining preface to the book; to wit, he states why and how he has come to be a convert to late merging. We think that speaks to how great of a frustration that activity is to all of us. He (now) endorses to drive up to the end of the dropped lane to the last possible point and then "ask" in, or, force-in if need be. The prevailing speed of the continuing lane and the dropped lane help frame the aforementioned four merge conditions. These are choices, and in the absence of absolute enforcement, they are for the most part, predicated solely on personal preference. This latter move (staying in one's lane until the last possible moment) is an enabler of "zippering"; i.e., Vanderbilt's de facto "late merge."
Photo of zipper (typ)
Absent perhaps an on-the-spot police officer directing you so, there is no legal mandate that in a side-by-side merge motorists must take turns; i.e., "zipper". The term is self-defining. For all practical purposes (and undoubtedly legal ones too) engineers, officials or officers will only go so far on-record as to advise to "merge when it is safe to merge." At face value, this leaves it up to you. However, the practice of "zippering" or "alternate merge" is a de facto practice that a few state DOTs and others have nevertheless endorsed on websites and pamphlets. But it is important to note that in each case they are only speaking to the work zone condition or at special recurring "funnel" locations (e.g., two lanes forced into one under a narrow bridge), either of which is at-crawl and in immediate vicinity of the nozzle of the constriction. None of the researched sites endorses zippering upstream and/or when traffic is at-speed. For the record, no "Zipper" sign (message or symbol) exists in the Manual of Uniform Traffic Control Devices (MUTCD). In some cases, state DOTs have taken it upon themselves to install yellow advisory signs indicating "Use Alternate Merge" (or similar) with some sort of arrow-combining symbol at locations, say, at the aforementioned narrow underpass, or on the far side of an intersection where two lanes past the intersection merge into one, but such instruction is non-conforming and is used at the states' volition.
That being said, a few state DOTs and elsewhere are on-record endorsing the practice if not the actual signing of zippering; which is to say they recognize the term, allow teaching it, but at the end of the day, they do not legislate it. The Minnesota DOT website (http://www.dot.state.mn.us/zippermerge) claims safety and durational benefits when motorists endeavor to "merge at the last moment" and understandably adds "this is fair practice too", i.e., in a slow crawl, each driver will take his turn, implying that road rage may be reduced. It even goes so far as to advise "donít worry about being 'Minnesota Nice' (upstream); stay orderly". MN even includes the practice in their Driver's Manual.
In addition to billboards and pamphlets, MnDOT also produced YouTube-based videos to discuss and exemplify the practice. On one video it reminds motorists that it is against the law (up to $100 fine) to intentionally block a lane of traffic just because you may feel it your prerogative to dissuade others from getting past you in a work zone delay. A representative of the Washington state DOT is also on record endorsing zippering as a practice (in the aforementioned work zone condition only) for much the same reasons. There are anecdotal references in other states too.
In all cases that we have researched, zippering is addressed as an ad-hoc practice that is recognized to exist, much like flashing one's headlights2 in a friendly or communicative way, or waving a hand to acknowledge being allowed into a queue, or to be fair, giving that "one finger salute" to respond when one is not allowed in. Zippering was not found anywhere to be an embedded program tenet or functionally mandated practice, nor are we aware of any legislation defining it that would put it on par with, say, seat belt laws, distracted driving laws or having to yield to emergency vehicles and stopped school buses. In short, there may not be mandates to make it happen, but there may be benefits if motorists politely practice it under certain qualifying conditions.
1 "Traffic: Why We Drive the Way We Do (And What it Says About Us)", 2008, by Tom Vanderbilt
2 Some states (varies by state) prohibit this in certain qualifying circumstances, like using high beams against oncoming traffic or to obstruct, obfuscate or replicate a police procedure or response.
Not all bottlenecks – either recurring or nonrecurring – are the result of a "bottle neck" design; i.e., a funnel-type constriction like a lane drop or an accident taking away a lane. Some recurring bottlenecks are not in any way complicit with a reduction in capacity. Some result from confluence (e.g., mixing) and some from slowing (e.g., rubbernecking).
A confluence is generally defined as a coming together, a meeting, or mixing. Using classic examples, this can be a forced merge, as in the case of a lane drop, but also a mixing, as in the case of a weave condition. In the first case there is a loss of capacity. In the second case, there is no loss of capacity but in either case a confluence results.
Examples of Confluence
A bottleneck caused by slowing is one in which motorists more or less do not have to change or fight for lanes, but are effected by mostly-superficial impacts that simply cause them to hit the brakes, and cause motorists behind them to react.
A driver facing sun glare
Examples of Slowing
Sun glare, dusk and wet pavement, et al, could easily be viewed as weather-related nonrecurring events. However, in the context of facing these conditions rather frequently (or in the context of some academics questioning if these aren't in fact nominal occurrences) can one justify that these are normal conditions that drivers should have some expectation and familiarity with, which is to say, should the mere use of headlights or windshield wipers really be the cause of congestion? And we are not speaking of atypical reactions, fears or even phobias. Psychologists and human factors experts remind us that categorical phobias exist; for example, the mere "fear of driving" (a recognized stress disorder with no Latin term per se but comparable to 'hodophobia', the fear of traveling) or the fear of driving across bridges ('gephyrophobia'), etc., and other phobic extremes, like rain-on-windshields, or driving next to jersey walls.