Memorandum

Date       December 20, 2012

To           Boston Region MPO

From      William Kuttner
MPO Staff

Re            Screening Regional Express Highways for Possible Preferential Lane Implementation

Executive Summary

Purpose of the Study

This memorandum presents the findings of the Regional HOV Lane System Planning Study. This study represents the initial phase of a two-part investigation, the second phase of which is to be undertaken during federal fiscal year (FFY) 2013.

Gradually increasing traffic congestion has imposed both economic and quality-of-life burdens on the region, and strategies to limit or reduce congestion are regularly evaluated by the MPO. The congestion relief strategies being envisioned do not include building new express highways, and the widening of express highways is only being considered in a few specific locations.

The implementation of high-occupancy-vehicle (HOV) lanes has been a successful congestion relief strategy in the Boston region, and both HOV lanes and other kinds of preferential lanes have been successful across the country. The value of these lanes, however, depends critically on the traffic conditions of the highway where they are constructed. This study evaluates all express highways in eastern Massachusetts and identifies sections that might be suitable for preferential lane implementation. Conceptual plans for selected suitable sections will be developed in this study’s second phase.

Planning Assumptions and Criteria

This study is informed by three basic planning assumptions:

• Lane eligibility
  It is assumed that MassDOT may choose to apply eligibility criteria other than vehicle occupancy for use of either existing or future preferential lanes. Optimally, these lanes should be used by 1,500 vehicles per hour, a level difficult to realize by way of an occupancy rule alone. Variable-price tolling and low vehicle emissions are lane entry criteria used in other American jurisdictions, and it is assumed that Massachusetts may choose to adopt similar practices. For this reason, the new facilities considered in this study are referred to as “preferential lanes,” implying that users who meet the eligibility criteria can enjoy a travel advantage, but not specifying how eligibility might be established in the future.
• User benefits
  Any envisioned preferential lane implementation should benefit both users of the preferential lane and users of the adjacent general-purpose lanes, and the expected traffic flow improvement for a preferential lane should exceed that for the associated general-purpose lanes. Relative traffic benefits must be in this “sweet spot” for two reasons. First, the higher benefit for the preferential-lane users provides an incentive to meet whatever eligibility criteria the operating agency sets. Second, providing some benefit to traffic in the general-purpose lanes will help develop broad-based support for the not-inconsiderable investment required to build a preferential lane.
• Standard lane design
  The type of preferential lane generally assumed in the analysis is a single, reversible lane operating inbound in the morning and outbound in the afternoon. These facilities would be constructed in the median of existing express highways and would entail extensive reconstruction. Some amount of land taking is considered acceptable, depending on nearby land uses. Lanes would meet all applicable design standards, including sufficient width for breakdowns and enforcement activity.

Conclusions and Recommendations

Most of the regional express highway system was found to be unsuitable for the type of preferential lane envisioned. Either these highway sections failed the user benefit tests in some respect, or there were land use or traffic flow issues specific to particular locations that eliminated them from further consideration.

Important stretches of express highway were, however, found to potentially offer significant user benefits and have tractable construction challenges. The express highway elements recommended for further analysis are shown in Figure 5, found on page 56 of this memorandum. These highway elements relate to each other geographically in such a way that they can be viewed as an integrated preferential lane system focused on I-93. The sections recommended for further consideration include:

• The current inbound HOV facility on I-93 in Somerville could be extended as a reversible lane north to Methuen near the New Hampshire state line. Conceptual planning for this preferential lane north of Boston is the subject of the Phase II study.
• The section of I-95/Route 128 between I-93 and Winter Street in Waltham has sufficiently pronounced peak-period directional traffic that it is a candidate for preferential-lane implementation.
• South of downtown Boston, HOV lanes constructed as part of the Central Artery/Tunnel (CA/T) project could be connected by a new reversible preferential lane to the existing moveable-barrier “zipper” lane in Dorchester and Quincy. This potential new lane was the subject of an MPO report, Improving the Southeast Expressway: A Conceptual Plan, completed in February 2012.  
• Preferential lane opportunities also exist that could improve the flow of traffic between State Highways 24 and 3 through the Braintree Split and connecting with the zipper lane.

Introduction

Background

This memorandum presents the initial findings of the Regional HOV Lane System Planning Study, which was approved by the MPO on August 18, 2011. In recent years extensive data about traffic and level of service on the Boston region limited-access expressway system have been gathered, refined, and systematized. In this initial phase of the investigation, these data are analyzed and specific quantitative measurements are presented that can indicate the appropriateness of having a preferential lane within a section of the regional express highway system.

HOV (high-occupancy vehicle) lanes, commonly referred to as “carpool” lanes, may be used by autos with two or more occupants, depending on the rules of a particular HOV facility. Buses also use HOV lanes, sometimes offering a significant time advantage to bus users as compared with driving alone. With new vehicle and other technologies, highway authorities have been experimenting with additional eligibility criteria for these facilities, such as low- or no-emission vehicles, or willingness to pay tolls collected using a transponder.

In this investigation, the term “preferential lane” has been adopted to make explicit the recognition that any Massachusetts agency responsible for such lanes in the future will be able to establish eligibility as deemed appropriate at that time. Vehicle occupancy data are used in this investigation, however, to the extent that such data are available and relevant to the analysis.

Overview of the Study

This study is has five major sections. The introduction reviews some history of the regional express highway system. The difficulties of adding capacity to a congested highway are discussed, as well as the types of benefits that might reasonably be expected from a successful preferential lane implementation.

In the second major section, the regional express highway system is divided into parts to facilitate congestion analysis. Several measurements of congestion severity, based on recently obtained regional traffic data, are presented. These measurements are applied to each part of the regional express highway system to identify which parts of the system experience congestion that rises to the level of a “problem.”

The next section looks more closely at the highway segments identified as having problem congestion. The regional traffic data are analyzed to determine whether a preferential lane would provide meaningful user benefits in the congested highway sections. Only if the analysis predicts meaningful user benefits is a preferential lane deemed to be suitable.

The express highway sections that emerge from the traffic-based screening process are then evaluated individually. How congested highways interact with nearby and interconnected highway sections is examined. Improvements to the regional express highway system, whether under construction, planned, or envisioned, must be considered when evaluating implementation of a preferential lane anywhere in the region.

The last section presents all the parts of the regional express highway system that emerged from the screening process. These highway sections fit together and function as a regional preferential lane system, which was a desired outcome of the evaluation in the preceding section.

Express Highway Development

Construction of the regional express highway system began in earnest in the 1950s. Most Massachusetts expressways at that time were planned as either four- or six-lane facilities, and were constructed with then-standard breakdown lanes. As traffic volumes increased, some express highways were widened and some of the newer ones were built as eight-lane roadways.

Continued traffic growth and the cancellation of several major expressways in the early 1970s have made maximum utilization of existing highway corridors a critical planning objective. Some expressways have been fully widened with newly constructed breakdown lanes. In other instances the breakdown lane has been converted into a travel lane, with some pullouts provided. Another strategy has been to allow use of the breakdown lane only during peak periods, though this is discouraged as a long-term operating practice.

It needs to be acknowledged that the Central Artery/Tunnel (CA/T) project added significant new capacity and important connections in the core of the regional expressway system, and no new preferential lane facilities are envisioned within the CA/T project area. Outside of the core, however, the regional expressway system continues to struggle with today’s traffic burden. Though future forecasts of traffic growth are moderate and characteristic of a mature travel market, most of the regional expressway system has very little extra capacity to accommodate growth, and crowded highways impose significant congestion and delay penalties on current users.

Potential Express Highway Enhancements

Further widening of regional expressways poses a number of challenges. First, adding two additional general-purpose lanes to an eight-lane section should not be expected to increase capacity by a full 25% because of the increased amount of lane changing that results, especially in a dense urban area such as that of the Boston region. Given the expense and probable land takings that any such widening is likely to require, adding extra lanes to achieve only a modest capacity increase would be unlikely to gain institutional or public support.

An alternative approach to adding capacity in an expressway corridor is to add a lane, or lanes, serving a specialized traffic purpose, and physically separated from the main traffic lanes. A common example of this approach is so-called “collector-distributor,” or “CD” lanes. These lanes, usually built to the right of the main travel lanes, can serve numerous entrances and exits as the expressway passes through a complex of one or more closely-spaced interchanges. A four-lane main travel barrel can operate at full capacity, while the CD lanes accommodate significant amounts of traffic using the various ramps, and rejoin the main travel lanes at a simpler location in the network.

The preferential lanes envisioned in this study would usually constitute specialized, separate lanes, generally constructed in the median of an expressway. The adjacent main travel lanes could continue to accommodate their full lane capacities, up to around 2,200 vehicles per hour, depending on the design standards of the roadway. With preferential lane eligibility set to allow fewer vehicles, perhaps 1,500 per hour, buses and other users of the preferential lane would be able to travel at posted speeds. Depending on traffic conditions, users of the general-purpose lanes would continue to experience delays either as a result of heavy traffic or from queues forming at bottlenecks.

A less common preferential lane implementation is the use of a moveable-barrier contraflow lane, like the so-called “zipper” lane currently in use on the Southeast Expressway. These systems require reversing the direction of one lane during peak travel periods. This increases capacity in the peak direction, but reduces it in the opposite direction. If an expressway section has a sufficiently large disparity in traffic by direction, a zipper lane might be implemented without meaningfully worsening traffic in the off-peak direction. Locations in the express highway system that might benefit from this type of facility will be identified in this study.

Benefiting All Users

The implementation of a preferential lane may offer several benefits. First, the total capacity of an expressway section can be increased. Congestion and queuing in the general-purpose lanes can be reduced, and the preferential lane can still offer its users a travel advantage as compared with users of the general-purpose lanes. This advantage will accrue to important preferential-lane users such as buses, and provide an incentive for other users to meet the eligibility standards, whether set by occupancy, payment of a toll, or other criteria.

It is sometimes suggested that an HOV lane could be implemented by converting a general-purpose lane for the exclusive use of HOVs. Since it is assumed that any preferential lane eligibility rules would result in fewer vehicles in the preferential lane than in the general-purpose lanes, the result would be a reduction in total expressway capacity. Reducing the capacity of a congested expressway would seriously worsen congestion and queuing within and leading to the capacity-reduced corridor, as well as on nearby surface roadways.

This study will identify locations in the regional expressway system where creating a preferential lane could be expected to substantially benefit all or most users. Situations or strategies where the likely benefits would be minimal, or where negative impacts would likely be significant, are identified and eliminated from further consideration.

The appropriateness of adding a preferential lane is considered for all regional expressways, regardless of size. For practical and political reasons, a preferential lane may be the better way of adding capacity to an eight-lane expressway. However, six- and four-lane expressways might at some point be candidates for simple widening as a better option. This study examines the potential benefits of adding a preferential lane to these narrower parts of the expressway system, but will be cognizant of any plans or the possibility of undertaking a roadway widening.

Identifying Express Highways with Congestion Problems

Overview of the Identification Process

This investigation has been able to utilize extensive, recently developed traffic information resources. Narrowing and focusing the available data to develop usable results presents an organizational challenge. This section previews the general approach to the analysis.

For the purposes of this study, regional express highway traffic has been divided into four travel markets:

1. Traffic on radial express highways during the AM peak period
2. Traffic on radial express highways during the PM peak period
3. Traffic on circumferential express highways during the AM peak period
4. Traffic on circumferential express highways during the PM peak period

Each of these express highway travel markets is analyzed separately and in this order, but the structure of the four analyses is basically the same. Each analysis involves dividing the regional express highway system into major components, picking specific highway segments to represent each component, and attaching several congestion measurements to each of the representative segments. These measurements are compared, and some of the segments are identified as having congestion severe enough to be considered a problem.

The analysis begins with a description of how the regional express highway system is divided into major components. The next step is to select a single highway segment within each system component that will represent the component in the comparative analysis.

Each of the representative segments is then characterized by its level of congestion severity, which is represented by four different measurements in this analysis. Each of the four measurements has its own analytical meaning and importance, and each measurement is presented and discussed individually.

It is helpful to point out here that one of the four congestion measurements, vehicles per lane per hour, is also used for selecting representative segments. Although this measure is appropriate for selecting representative segments, it is only one measure used to evaluate congestion severity. In the evaluation of representative segments, all four congestion measurements need to be considered.

Some of the representative segments will be identified as having a congestion “problem,” a level of severity greater than merely observing some congestion. This involves weighing the four measurements, as well as consideration of the overall traffic situation within the given system component.

Congestion on radial express highways during the AM peak period is discussed first. This initial travel market is used to provide numerical examples and network situations with which to illustrate the calculations and concepts applied throughout the analysis. After AM peak radial problem congestion is identified, PM peak radial congestion problems are then identified, but without repeating the illustrative discussion.

After the AM and PM radial congestion problems are identified, the circumferential highway travel markets are analyzed. AM peak conditions are used to illustrate several unique characteristics of circumferential highway analysis. Problem circumferential congestion in the AM peak is identified, followed by problem circumferential congestion in the PM peak.

At this point, the identification of representative segments with problem congestion is complete. In the next major section, the same traffic data that were used here will again be applied to determine which of the segments with problem congestion might be suitable for preferential lane implementation.

Dividing the Regional Express Highway System into Major Components

For the purposes of this study, the regional express highway system has been divided into sections which will be referred to as major “components.” A component will generally be a section of limited-access highway anchored at its two ends by interchanges with other limited-access highways. These major components are in turn divided into two distinct groups: radial components and circumferential components. The analysis and evaluation of these two groups differs in important regards, and as indicated above, the two groups have been evaluated separately, beginning with the radial group.

The analysis begins by using peak hour traffic volumes for every express highway segment to identify the location with the highest number of vehicles per lane per hour for every highway component being evaluated. Figure 1 shows the CTPS model region, as well as several adjoining municipalities. The express highway system is also shown, and radial segments carrying the highest number of vehicles per lane per hour are highlighted in red (AM peak hour) or blue (PM). The position of the highlighting indicates the direction of the peak hour travel, and the length indicates the length of the highway segment.

The segments shown in Figure 1 illustrate some aspects of both component definition and segment selection. There are a total of 26 components analyzed in Figure 1, each of which has both an AM (red) and a PM (blue) segment highlighted. In 19 components, the peak AM and PM segments are between the same pairs of interchanges, with the direction shifting between the AM and PM peaks. In five of the components, the highest volumes per lane per hour are found between different interchanges in the morning than in the afternoon.

Organizing Radial Components into Groups

For analytical purposes, the components in Figure 1 have been organized into three groups: “inner” radials, “outer” radials, and “external” radials. Inner radials are the five express highways components within the I-95/Route 128 circumferential highway. Outer radials are the nine components outside of and connecting with I-95/Route 128. With the exception of Route 3 on the South Shore and Route 128 east of I-95 on the North Shore, outer radial components extend from the I‑95/Route 128 circumferential highway to I-495.

Beyond I-495 there are ten components referred to as external radials. Eight of these highways connect directly with I-495, while I-295 in Attleboro and Route 140 in Taunton are even more distant from Boston, branching from I-95 and Route 24 respectively. Southeast of Route 24, I‑495 is also characterized as an external radial, extending as it does to the edge of the model region.

Route 3 in Plymouth, south of the recently constructed U.S. 44 interchange, completes the group of twelve external radial components. Conforming with the definition of a component used here, most of these external radials extend to an interchange with another express highway outside of the model region. All traffic volume analysis, however, is confined to those segments of the external components that are within the CTPS model region.

Analyzing Radial Segments

AM peak conditions for the 26 radial segments are shown in Table 1. The segments are listed in the left-hand column in their three groups. The demarcations of (N) and (S) for I-93 and I-95 refer to regional system components north or south of Boston respectively, not direction of travel. The components are ordered starting with alignments to the northeast, followed by alignment directions sequencing counterclockwise, ending at the southeasternmost alignment.

The AM peak travel direction for all but three of these segments is toward downtown Boston. The exceptions are indicated in Table 1 with asterisks and include Route 2 in Lexington, Route 128 in Peabody, and I-290 in Northborough. The exit numbers indicating the location of each highway segment are shown, as well as the town where it is located. U.S. 1 has no interchange numbers, and the selected segment is immediately north of Massachusetts Route 60 at Copeland Circle. If only one interchange is shown, the segment is at the edge of the numbering system, or in the case of external radial Route 2 in Littleton, the segment is within the cloverleaf Interchange 40 at I-495.

The selected segment of each component is also characterized by the number of lanes in the peak direction at that point. Segments usually have two, three, or four lanes of general traffic. In the instance of I‑93 between interchanges 46 and 45 north of I‑495, use of the breakdown lane is allowed during the AM peak period. The relationship of these so-called “managed” lanes to the various entrance and exit ramps and associated acceleration lanes is not always clear to drivers, and these lanes are generally not fully utilized. As a consequence, a practical capacity of one-half lane is commonly assumed when travel in breakdown lanes is authorized.

The average weekday traffic (AWDT) for each segment is also presented in Table 1. The AWDT suggests the relative importance of the 26 highway system components in the overall regional express highway system.

Measurement of Congestion Severity

AWDT per Lane

Highlighted in Table 1 is a group of statistics which can be used as measurements of congestion severity. Dividing AWDT by the number of lanes (not including managed lanes) gives the AWDT per lane. It has been observed in eastern Massachusetts that at locations where AWDT per lane reaches 15,000 on an express highway, congestion and queues begin to be observed. Above 15,000 daily vehicles per lane, the hours of congestion and length of queues increase.

Since this statistic is based on AWDT, it can reflect congestion during both AM and PM, as well as midday congestion. Prior to completion of the CA/T Project, AWDT per lane reached 35,000 north of Northern Avenue northbound, and weekday congestion began shortly after 5 AM and only subsided late in the evening. The highest AWDT per lane in Table 1 is the 27,900 weekday vehicles on the I-93 inner radial south of downtown Boston between interchanges 15 and 16, from Columbia Road to Southampton Street.

Vehicles per Hour per Lane

The second measurement of congestion severity is vehicles per hour per lane. Estimates of traffic by hour during four-hour AM and four-hour PM peak periods have been developed by CTPS for every express highway segment in eastern Massachusetts. For highways designed to the highest standards, 2,200 vehicles per hour per lane is a practical maximum. Ten of the 26 segments in Table 1 have at least 2,000 vehicles per lane per hour, with the highest value being 2,175 on U.S. 1 approaching Route 60 at Copeland Circle from the north.

The vehicles per hour per lane values shown in Table 1 were the basis for selecting the 26 highway segments that would represent the 26 radial components. During the AM peak period, these segments show a higher number of vehicles per hour per lane than any other segments that make up their respective highway system components. It is possible that traffic conditions may be worse at other locations along a highway component. These other highly stressed segments in most cases will be near the selected segment and will be considered in the corridor analyses after the initial screening.

There are three reasons for using the highest vehicles per hour per lane to select representative segments. First, an abundance of quantitative data is readily available for all regional express highways, regardless of the presence of any congested conditions. Second, a high volume of traffic implies a high level of regional travel demand at that location in the roadway network. Third, consideration of a preferential lane implementation can logically begin at a high volume location, and then be expanded to look at other congested segments that make up the component.

Hours Congested

The third measurement of congestion severity is the length of time the segment is congested during a peak period. For the 26 representative segments shown in Table 1, the estimates of congestion duration approximate the length of time that the roadway is substantially full. Given that there is a maximum hourly capacity for a roadway, travel demand above that capacity will result in queuing, peak spreading, or both.

At many locations in the express highway system, congestion is observed despite a substantially lower number of vehicles per hour per lane than at the locations in Table 1. Usually these congested locations are within a peak period queue. If there is a high-volume entrance ramp joining an express highway, the lanes beyond the entrance may be carrying their practical capacity, but queues will form on the entrance ramp and upstream in the main barrel it joins. Vehicles in a queue will travel at the queue’s characteristic slow speed until reaching the bottleneck, at which point traffic can speed up and higher vehicles per hour per lane can be achieved.

A low-volume-per-lane but high-congestion condition can also occur where an expressway component ends either at another congested expressway or at a traffic signal. These types of congestion situations are not used in the screening process, but will be analyzed as appropriate in conjunction with the selected high-volume severe congestion locations.

The selection rule for the representative segments in Table 1 was highest vehicles per hour per lane. As a consequence, almost all the segments shown with a congestion duration were carrying their practical maximum of traffic rather than being in a queue behind a bottleneck. Some of these representative segments are at bottlenecks at the head of queues, and any preferential lane analysis will consider the queue and the bottleneck together as a congested traffic subsystem.

The longest AM congestion duration is four hours on the northbound I-93 inner radial north of Columbia Road. As indicated above, this location also has the highest AWDT per lane (27,900) which is known to correlate well with congestion duration.

The lowest vehicles per lane per hour with congestion is the 1,800 vehicles on outer radial I-90 after Interchange 13 as it approaches the Weston toll plazas, interchanges 14 and 15. An hour of AM congestion is estimated at this location as the volume on this segment reaches the maximum throughput of the Weston toll plazas.

Slowest Speed

The last measurement of congestion is the slowest traffic speed that will normally occur on a representative segment at some time during the peak period. Travel speeds have been measured on regional express highways using GIS and other techniques, and average speeds have been developed for half-hour periods between 6 and 10 AM and between 3 and 7 PM for individual segments of the regional express highway system.

Slow speed is a reliable indicator of a congestion problem, either from heavy traffic at a particular point, or from a bottleneck ahead of that point causing a queue. The converse is not true, however. Traffic on a road can be near capacity, yet traffic can be moving near design speeds. As with the other three measurements of congestion severity, consideration of travel speed must be in the context of conditions on other parts of an expressway system component.

Travel speeds as low as 20 mph occur during the AM peak at two locations: the northbound I-93 inner radial north of Columbia Road, and the outer radial Massachusetts Route 3 as it approaches Exit 19 just before the Braintree Split. At the opposite extreme, a number of locations experience one or more hours of congestion, yet are seen with traffic speeds ranging from 54 to 64 mph.

Identifying Radial Congestion Problems

AM Peak Congestion Problems

Fifteen of the 26 segments in Table 1 are shown as having one or more hours of congestion. Ten of these segments have been flagged as having a congestion “problem,” indicated by a “C” in the rightmost column. Congestion on the other five segments is considered insufficiently severe to qualify as a problem.

Congested conditions will not necessarily slow traffic. The five segments with “non-problem” congestion all support traffic speeds appropriate for the segment design. The slowest traffic of the five is on Route 128 in Peabody headed north towards Danvers, where congestion is observed for two hours in the morning, but traffic only slows to 50 mph, a reasonable speed given the 1940s-era highway design.

The segments with non-problem congestion could in the future have a congestion problem if traffic grows appreciably.  The external radial, Route 24 in Raynham, carries 2,100 vehicles per hour per lane, the highest of the five non-problem congestion segments. It is congested for two hours during the AM peak, although traffic moves at 64 mph. Despite the limited potential for traffic growth, this segment has not been flagged as having a congestion problem because after a mile and a half it widens to three lanes.

The remaining three non-problem congestion segments, U.S. 3 in Chelmsford and two segments of I-95 in Mansfield, have fewer than 2,000 vehicles per hour per lane, and traffic speeds are 58 mph or greater. Despite the good travel speeds, and modest unutilized lane capacity, they are still considered to be congested for one or two hours during each AM peak period.

Of the ten segments with “problem” congestion, four are inner radials, five are outer radials, and one is an external radial. Of the five inner radials, all except Route 2 are considered to have problem congestion. The congested inner radials are I-93, both north and south of downtown Boston, U.S. 1 in Revere, and I-90 in Newton.

Five of the nine outer radials are also considered to have problem congestion. These include I-93 in Wilmington, U.S. 3 in Billerica, I-90 in Weston, Route 24 in Randolph, and Massachusetts Route 3 in Braintree.

The only external radial with a congestion problem is I-93 in Andover. It carries 2,171 vehicles per hour per lane (assuming a half-lane capacity for the managed lane), experiences three hours of congestion each morning and has a travel speed of 54 mph. While 54 mph may not seem unduly slow, this section of highway clearly has little room to add more traffic.

Casual inspection of these ten segments reveals some patterns. Six of the ten segments are associated with I-93 in some manner. The three I-93 segments north of Boston all qualify as having congestion problems. South of the downtown area, I-93 northbound traffic in the Southeast Expressway north of Columbia Road shows some of the region’s highest measures of congestion severity. Further south, Routes 24 and 3 in Randolph and Braintree experience serious congestion as they supply much of the AM peak-period traffic to the Southeast Expressway.

This completes the selection process for AM peak-period radial segments with problem congestion. After selection of the PM radial and AM and PM circumferential segments with problem congestion, the entire group of selected segments will be evaluated to determine the possible benefit of adding preferential lane facilities.

PM Peak Congestion Problems

The representative radial segments for the PM peak are listed in Table 2. This table is organized in the same manner as Table 1. The PM peak segments appear in Figure 1, and most, but not all, are located between the same pair of interchanges as the corresponding AM peak segments.

Seventeen of the 26 representative segments experience congested conditions for one or more hours during the PM peak. This compares with the 15 segments in the AM experiencing congestion. Congested situations are more widespread during the PM peak than in the AM because the PM peak combines a quantity of work trips similar to the AM peak, with trips having other purposes exceeding those taking place during the AM peak.

Three of the segments exhibit what is considered non-problem congestion. Two external radials with some congestion show speeds of 60 mph with volumes per hour per lane near 2,000 vehicles. The outer radial on the south-side part of I-95 in Norwood shows travel speeds of 56 mph with 2,033 vehicles per hour per lane.

Fourteen of the PM segments with congestion are considered in this study as having a congestion problem. All ten of the expressway system components that had problem congestion during the AM peak also experience problem congestion during the PM peak. Of the six segments associated with I-93, the AM and PM peak segments for five locations are between the same interchanges. In Boston, the congested I-93 segments are offset by one interchange, north and south of Columbia Road.

Four components have problem congestion in the afternoon but not in the morning. The outer radial Route 128 in Peabody carries 2,150 vehicles per hour per lane inbound on a 1950s-era highway. The road is full for three hours each afternoon, and traffic moves at 42 mph for two hours even through the highway widens from two to three lanes within this segment.

Route 2 in Lincoln is severely congested in the PM as it approaches the Bedford Road signal, the first of a sequence of eight traffic lights on Route 2 as it bypasses Concord. During the AM peak, traffic in this most easterly segment is approaching a fully limited-access section of Route 2 and speeds are higher.

The external radial on Route 24 in Raynham has problem congestion for 2.5 hours in the afternoon, with speeds dropping to 29 mph as the highway narrows from three to two lanes between interchanges 14 and 13. There is also AM congestion in the opposite direction, but the highway widens and the AM congestion is not considered a problem. Finally, the external radial U.S. 3 in Chelmsford narrows from four to three lanes as it approaches Interchange 32 at the Drum Hill Rotary, causing two hours of PM congestion. Speeds in this segment slow to 40 mph during the PM peak.

Since all the segments with problem congestion during the AM peak were also considered to have problem congestion in the PM peak, the analysis of PM congestion reinforces the regional patterns observed in the AM. A group of six segments associated with I-93, four segments on I‑93 itself and two on Routes 3 and 24 feeding into I-93 from the south, were identified as a north-south AM congestion group. The duration of PM congestion for each segment in this “I‑93 group” exceeds the AM duration.

This completes the selection of radial segments with problem congestion. The ten AM and fourteen PM designated segments will be grouped with the circumferential segments having problem congestion identified in the next section. This selection of regional segments with problem congestion, both radial and circumferential, will then be analyzed for appropriateness for possible preferential lane implementation.

Organizing Circumferential Components into Groups

The two regional circumferential expressways have also been divided into components, as shown in Figure 2. The inner circumferential highway is often referred to by its pre- Interstate designation, Route 128. Between the I-95 junction in Peabody and Route 3 in Braintree, this expressway has been divided into seven components. The five northernmost components are part of I-95. In Peabody, I-95 branches from the circumferential corridor and continues north to New Hampshire. Route 128 continues up the North Shore and is considered to be a radial highway in this area. The circumferential segment between the I-95 junction in Peabody and U.S. 1 in Lynnfield is considered to be part of a I-95 (N) to I-93 (N) component. The southernmost two components, between I-95 and Route 24, and between Route 24 and Route 3, are part of I‑93.

The outer circumferential highway, I-495, extends from I-95 in Salisbury near the New Hampshire border to the junction with I-195, a few miles away from the Bourne Bridge and Cape Cod. Between Salisbury and Route 24, I-495 is divided into eight components. The southern end of I-495 between Route 24 in Raynham and the edge of the CTPS model region is treated in this study as a radial segment.

Analyzing Circumferential Segments

AM peak conditions for the 15 representative circumferential segments are shown in Table 3. The components are grouped by the two circumferential arcs, and arranged from north to south. The first two columns define the endpoints of each component. The exit numbers indicate segments with the highest AM peak hour vehicles per lane within each component. If only one exit is listed, the peak volume occurs within the interchange. Each segment is characterized by the number of lanes and AWDT. Breakdown lanes used during peak periods are counted as having a half-lane of capacity.

The same four measurements of congestion severity are utilized in this analysis as for radial routes. It is necessary, however, to look at peak traffic in both directions at the outset of the analysis. This is because there is no natural “peak” direction along the circumferential routes. The AM and PM peaks may be either north- or southbound, depending on regional travel patterns.

The vehicles per hour per lane are shown for the peak segment for both the higher direction, as well as the opposite direction within the same segment. The value for the peak direction is shaded, and represents the highest vehicles per hour per lane for the entire component, regardless of direction. The other three congestion severity measurements: AWDT per lane, hours congested, and slowest period speed, all refer to traffic moving in the direction of the shaded value.

AWDT per lane on AM peak Route 128 circumferential segments ranges between 24,000 and 27,800 vehicles, comparable to weekday traffic on the inner radial segments in Table 1. On I‑495, AWDT per lane ranges between 14,300 and 22,300 on the AM peak segments.

The circumferential expressway components do not by their nature have an obvious AM “inbound” direction to the urban core. It is possible, however, to observe some peak directional patterns in Table 3 by looking at the paired northbound and southbound peak hour traffic flows. During the AM peak on Route 128, the highest traffic segment for each component north of I-90 is in the southbound direction. Conversely, south of I-90 the highest traffic segments are all northbound. Many vehicle trips combine travel on a circumferential highway with travel on one or more of the radial highways. The fact that the net AM Route 128 traffic flows center on I-90 suggests the centrality of this highway in the regional network.

Without the strong peak period directional flows typical of the radial components, several of the peak circumferential traffic volumes are closely matched by corresponding heavy traffic in the opposite direction. Examples of this on Route 128 during the AM peak include the component between U.S. 3 and Route 2, with1,975 vehicles per hour per lane in the peak direction compared with 1,825 in the opposing direction, and the component between Routes 24 and 3, with 1,800 and 1,750 vehicles per hour per lane in two directions. The components on I-495 between I-93 (N) and U.S. 3 and between U.S. 3 and Route 2 show similar balance by direction. In this analysis, the peak volume segments are chosen and evaluated for congestion severity independent of any congestion conditions that exist in the opposing direction. Congestion conditions that exist in the lower traffic volume direction will be analyzed in conjunction with considering strategies for individual expressway components.

Identifying Circumferential Congestion Problems

AM Peak Congestion Problems

The hours of congestion shown in Table 3 confirm the regional understanding that congestion on Route 128 is severe, with three of the Route 128 components having four hours of AM congestion on their most congested segments. All but one of the AM Route 128 representative segments have been designated as having problem congestion. There is some congestion on I‑495, concentrated in the more urbanized Haverhill-Lawrence-Lowell corridor, and at the junction with I-90. Only near the interchange with I-290 does traffic slow to 50 mph, and no I‑495 segments are considered as having problem AM peak congestion.

The representative segment on the Route 128 circumferential component between Route 3 and Route 24 has 1,800 vehicles per hour per lane heading in the northerly direction, though the highway heads due west at this point. It is congested for two hours each morning, but traffic never slows below 48 mph and this has not been designated as problem congestion during the AM peak. In the opposing direction, however, traffic peaks at 1,750 vehicles per hour per lane, as queues entering the Southeast Expressway back up through the Braintree Split for several hours each morning. Although this segment is not designated as having problem congestion in its stronger direction, possible improvements in this area will be considered upon review of candidate system components.

PM Peak Congestion Problems

The representative PM peak segments are shown in Figure 2, together with the AM peak segments. Nine of the fifteen representative PM segments are between the same pair of interchanges as in the AM, but in the opposite direction. In Lowell on I-495 and in Randolph on Route 128, the PM and AM peak segments are the same, including travel in the same direction. On Route 128 in Waltham, both the AM and the PM peak segments flow to the south. The PM peak segments of I-495 in Haverhill and Lawrence, and of I-93 in Wakefield, are in the opposite direction from the AM segment but offset by one interchange.

The PM peak circumferential segments are listed in Table 4, which is organized in the same manner as Table 3. Thirteen of the fifteen segments in Table 4 experience some PM peak congestion, and in eight of these cases the congestion is severe enough to be considered a problem. The five segments with non-problem congestion all have good travel speeds and some remaining capacity for traffic growth.

The eight PM problem locations include all seven Route 128 segments plus the section of I-495 inside Interchange 46 with Route 110 in Methuen. The northbound segment on the Route 128 circumferential component between Routes 3 and 24 experiences two hours of congestion in the AM peak, but with travel speeds of 48 mph, it was not considered a problem during the AM peak. This same segment experiences three hours of congestion and travel speeds of 42 mph during the PM peak and is considered to have a congestion problem during the afternoon.

This completes the selection of circumferential segments with problem congestion, almost all of which are on the Route 128 inner circumferential corridor. The eight circumferential components that have some problem congestion in either direction at some point in the day, together with the fourteen radial components that show some problem congestion, will be evaluated for preferential lane suitability in the next section.

Screening Segments for Preferential Lane Suitability

Overview of the Evaluation Process

Improving the flow of current or projected traffic at the most congested regional locations requires adding capacity either at the congested location or at a bottleneck downstream of the location. This added capacity can be in the form of an added general-purpose lane, an added preferential lane, or a lane borrowed during peak periods from the opposing traffic flow.

At locations where added capacity would be helpful, adding a general-purpose lane can be expected to improve traffic flow. The potential benefit of either an added preferential lane or a contraflow (usually preferential) lane borrowed from opposing traffic depends upon a number of factors. An initial evaluation presented in this section applies some simple volume and capacity tests for traffic currently using the congested highway segments to assess whether a preferential lane might provide meaningful user benefits as compared with a simple widening.

As in the previous section, AM peak radial problem congestion is discussed first. This initial travel market is used to provide numerical examples and network situations with which to illustrate the calculations and concepts applied throughout the evaluation process. After AM peak radial problem congestion is evaluated, the same evaluation steps are applied to PM peak radial segments and AM and PM peak circumferential segments.

Preferential Lane Suitability for AM Peak Radial Segments

Benefit of Adding a Preferential Lane

The 26 express highway radial components that were analyzed in Table 1 are presented again in Table 5. All figures are in vehicles per hour per lane, and the data in the first column were also used in Table 1 and were the measure used to select peak segments. The 10 congestion flags from Table 1 are also shown in Table 5. The second column shows the traffic traveling in the opposite, lighter direction of the AM peak segment.  The volume in the non-peak direction is critical in determining the viability of a contraflow strategy using a zipper lane.

Implementation of a preferential lane would reallocate existing traffic within both existing and newly created lanes. The reallocation of existing traffic with implementation of some type of preferential lane is calculated in the group of four columns in Table 5. It is understood that the addition of capacity in the peak direction would also make any express highway segments served by a new preferential lane more attractive to drivers, and some amount of additional traffic would be attracted to the improved express highway component. The calculations underlying Table 5 present a static analysis of current traffic, but still offer a basis for comparison among express highway segments. Estimation of incremental traffic attracted to a highway improved with a preferential lane requires use of the CTPS regional model set, and is not planned as part of this study.

The basic calculation in Table 5 takes the peak hour traffic in the peak direction and recalculates the vehicles per lane assuming one additional lane. This additional, preferential lane is assumed here to allow no more than 1,500 vehicles per hour. Setting eligibility criteria to allow 1,500 vehicles per hour to use the preferential lane will ensure that users of this lane will experience free-flow traffic conditions, and thereby may be more willing to meet the eligibility criteria by forming a carpool or perhaps paying a toll. Critically, the preferential-lane users will only perceive a benefit if traffic in the general-purpose lanes is more than 1,500 vehicles per hour.

The basic preferential lane calculation has been made for all 26 AM radial components in Table 5. The vehicles per lane in the general lanes are all significantly lower than the current peak direction volume per lane, the amount of the decrease depending principally on the number of current lanes. In 18 of the segments, including all 16 segments without problem congestion, the volume in the added preferential lane is the same as in the general traffic lanes, and lower than 1,500 vehicles per lane. In these situations, absent a substantial increase in segment traffic due to the added lane, the preferential lane would function in effect as a general-purpose lane and its specialized design would serve no purpose.

In eight of the 26 AM radial segments, the 1,500-vehicle cap on preferential lane traffic is effected, with traffic in the general-purpose lanes equal to or exceeding 1,500 vehicles per hour in this static analysis. The difference between the 1,500 vehicles in a preferential lane and the number of vehicles per hour calculated for the general-purpose lanes is shown in the next column as a preferential lane benefit.

These eight segments are flagged at the right with a “P,” and all but two of the segments with the congestion flag also are designated with a preferential lane benefit flag. On U.S. 1 in Revere, 4,300 vehicles per hour severely congest two lanes during the AM peak hour with 2,175 vehicles per lane. Three lanes comfortably accommodate this amount of traffic with 1,450 vehicles per lane. On I-90 in Weston, a fourth lane would serve as storage within the queue approaching the Weston toll plaza and would accommodate only 1,350 vehicles per hour.

The preferential lane benefit as calculated in Table 5 ranges up to 242 fewer vehicles per hour compared with the associated general-purpose lanes. This preferential lane advantage would upon implementation become greater as additional traffic was attracted to the improved highway segments. While the equilibrium traffic level in the general-purpose lanes might at some time during a peak period reach the level it had prior to implementing the preferential lane, these peak traffic levels would be reached later and subside earlier in the peak period because of the availability of the additional 1,500-vehicles-per-hour capacity in the preferential lane. This fulfills the requirement of this study that all users benefit from the implementation of a preferential lane. The general-purpose lanes would have less, but still some congestion. The preferential-lane users would have no congestion at all.

Feasibility of Implementing a Zipper Lane

A preferential lane can be added to a section of highway by constructing a new lane, usually in the median of a divided highway. In certain situations, a preferential lane might be added by “borrowing” a lane from the off-peak direction for the duration of a peak period. In this contraflow strategy, the assessment of benefits in the peak direction is unchanged. It is necessary, however, to determine whether the off-peak direction is able to function adequately with the loss of a lane during a peak period.

It is a given that taking a lane from the off-peak direction will substantially increase the vehicles per hour per lane in that direction, and implementation of a contraflow lane would be a deviation from the assumption of this study that all users should benefit from a preferential lane implementation. This study does, however, calculate the burden on opposing traffic that would result from implementing a contraflow lane in segments where a preferential lane could be beneficial. This study defines a cutoff point for each segment and uses this cutoff point as an indication of feasibility.

The rightmost of the four evaluation columns calculates the vehicles per lane if the current traffic designated as “other way” were to be funneled somehow into one fewer lanes. Interstate 93 in Stoneham near the top of Table 5 presents a clear example. This eight-lane highway segment has four lanes in each direction and carries 1,988 vehicles per lane in the peak direction but only 1,738 per lane going the other way, a total of 6,950 for the entire four-lane outbound barrel. Giving up a lane in the off-peak direction for use in the peak direction would allow only three lanes, each carrying 2,317 vehicles during the AM peak hour. Not only is 2,317 vehicles per hour per lane almost impossible, but 2,317 vehicles per lane is worse than the 1,988 vehicles per lane in the peak direction that the contraflow lane would be ameliorating. As a consequence, an AM zipper lane implementation in this segment is not considered feasible.

The peak direction vehicles per hour per lane serves as the feasibility cutoff point for each segment. The resulting vehicles per lane in the direction giving up a lane cannot be greater than the vehicles per lane of the peak direction which is to be improved. The inner radial I-93 (S) is the first segment receiving a “Z” for zipper lane feasibility. Traffic in the inbound peak direction would drop from 2,075 vehicles per lane to 1,700 vehicles per lane with 1,500 vehicles using a zipper preferential lane in the peak direction. Traffic going in the opposite direction would lose a lane and go from 1,475 vehicles per lane to 1,967, but since 1,967 vehicles per lane is less than 2,075, this segment is given the “Z” designation. Even though traffic in the off-peak direction has been squeezed, the maximum number of vehicles per lane in this segment has been marginally reduced.

Using this benchmark to establish zipper lane feasibility, only segments with strong peak period directionality emerge as candidates for zipper lane implementation. Six of the eight congested segments where preferential lanes offer benefits are also considered appropriate for a zipper lane, given the criterion used in this study. Implementing a contraflow strategy that appreciably increases traffic congestion in the off-peak direction would have public policy implications which are not considered here.

Preferential Lane Suitability for PM Peak Radial Segments

Benefit of Adding a Preferential Lane

The 26 representative radial segments for the PM peak are shown in Table 6. These are the same segments that were analyzed in Table 2, and Table 6 incorporates the same capacity tests that were applied in Table 5. Fourteen of these segments have been designated as having problem congestion, as compared with the ten problem congestion locations during the AM peak. This reflects the generally higher level of traffic during the PM peak as compared with the AM peak.

Ten of these fourteen locations have traffic levels sufficiently high that a preferential lane with 1,500 vehicles per hour would be less congested than the general travel lanes, and are designated with a “P.” The four congested segments that are eliminated by this test, U.S. 1 in Revere, Route 128 in Peabody, Route 2 in Lincoln, and Route 24 in Raynham, all have only two lanes in the peak direction. If a fifth, reversible lane were incorporated into a four-lane highway with two lanes in each direction, the 50 percent capacity increase in the peak direction would be so great that there would need to be substantial traffic growth before 1,500 vehicles would be reached in the reversible lane with the general-purpose lanes carrying sufficiently more than that amount to preserve the advantage of the reversible lane. In fact, widening three of these locations, U.S. 1 in Revere, Route 128 in Peabody, and Route 24 in Raynham, to six general-purpose lanes is under some level of active consideration, and this is an approach that would offer immediate user benefits as compared with adding a single reversible lane.

Of the ten locations designated as offering a preferential lane benefit, three show only nominal improvements in the static analysis, but would be expected to offer preferential-lane users meaningful benefits as traffic adapted to use the improved express highway component. Interstate 95 (S) in Norwood easily accommodates 2,033 vehicles per lane at 56 mph during the PM peak, and is not considered to have a congestion problem, even if a preferential lane would have 33 fewer vehicles per hour than the general-purpose lanes.

The external radial, I-93 in Andover, offers the greatest predicted benefit for preferential-lane users: 185 fewer vehicles per hour prior to regional traffic adjustment. This segment utilizes a peak period “managed” lane, or use of the breakdown lane, and the segment is assumed in these calculations to offer an equivalent of 3.5 general-purpose lanes. Were the breakdown lane to be upgraded to a full travel lane during the PM peak, a preferential lane in this segment would not offer benefits over the general-purpose lanes. AM peak traffic in this segment, however—analyzed in Table 5—is heavier, and even with a full fourth inbound lane, a preferential lane would offer users a benefit during the AM peak.

Feasibility of Implementing a Zipper Lane

Five of the ten segments where a PM preferential could be beneficial meet the feasibility test used in this study for contraflow lane implementation. Interstate 93 south of downtown Boston is an important example in the PM peak, as it is in the AM peak. This segment, north of the end of the currently operating zipper lane, has a strong directional imbalance: 1,950 vehicles per hour per lane outbound in the PM peak compared with only 1,325 vehicles per lane inbound. Extending the zipper lane north into this segment would reduce outbound traffic to 1,575 vehicles per general-purpose lane, a significant improvement. Inbound traffic would be concentrated into three lanes with 1,767 vehicles per lane, which would be under the 1,950 feasibility threshold assumed for this segment.

Preferential Lane Suitability for AM Peak Circumferential Segments

The same capacity tests applied to the radial segments are used in evaluating the circumferential segments. Table 7 presents the fifteen circumferential segments that were presented in Table 3, arranged from the northernmost to the southernmost segment, first for the inner Route 128/I-95/I-93 circumferential highway, and then for the outer I-495 circumferential highway. The current vehicles per hour per lane for the south- and northbound directions that appear in Table 3 are repeated as the first two data columns in Table 7. The congestion flags on Table 3 also appear in Table 7. No I-495 segments were flagged as having problem congestion.

The AM peak direction in each circumferential component has been shaded, and the capacity evaluation calculations are applied to the shaded, peak direction volumes. Redistributing existing traffic shows that the general purpose lanes would still carry significantly more traffic than a preferential lane on each of the six Route 128 segments that has been designated as having problem congestion. Adaptation by regional traffic would further increase the preferential lane advantages.

The two Route 128 components between I-90 and Route 24 utilize so-called managed lanes over all or most of their length, allowing peak period use of breakdown lanes in both directions. Managed lanes are considered here to have a practical capacity equal to one-half a lane, and the segments are calculated as having 3.5 lanes in each direction. The managed lanes, extending south from the Route 9 interchange, are being reconstructed as full travel lanes. Even with fully-built fourth travel lanes, preferential lanes in these segments would still offer some advantage over the general-purpose lanes.

In only one instance could a contraflow lane be feasible. The peak congested segment on the northernmost component of Route 128 is between I-93 and Route 28 in Reading. This segment has three lanes in the peak southbound direction, but four lanes in the northbound direction. Utilizing one of these four lanes for southbound traffic during the AM peak would result in 1,567 vehicles in each northbound lane, significantly below the 2,067 vehicles currently in each southbound lane. At no other circumferential location could an AM contraflow preferential lane be considered, because there is heavy traffic in both directions at all the other circumferential locations, and no other segment would be able to spare a lane during the peak period.

Preferential Lane Suitability for PM Peak Circumferential Segments

The capacity tests for the fifteen PM peak circumferential segments are shown in Table 8. A total of eight segments have been designated as having problem congestion during the PM peak: all seven of the Route 128 segments and the I-495 segment at Route 110 in Methuen. PM peak traffic is actually lower than AM traffic in a number of locations as congestion backs up traffic between the closely spaced Route 128 interchanges. Congested conditions in the PM peak tend to last longer than during the AM peak.

All of the Route 128 segments show significant benefits for preferential-lane users, but none of these segments would be appropriate for contraflow implementation. The segment between Route 3 and Route 24 shows both a congestion problem and a preferential lane benefit during the PM that it did not show during the AM due to an AM directional traffic balance unique to that segment.

Segments Suitable for Preferential Lanes after Initial Screen

The express highway segments so far identified where a preferential lane would offer user benefits are summarized in Table 9 and shown graphically in Figure 3. Table 9 combines the traffic data of the candidate radial and circumferential segments in a common format, varying only the formats of component designation and travel direction.

The segments are arranged differently than in the previous tables. The segments are grouped by highway regardless of time period, and are generally presented from north to south. The I-93 group leads the list of radial components, with twelve candidate segments starting with I-93 in Methuen and extending south to Route 24 at its interchange with I-93 in Randolph. The I-93 group is followed by three segments of U.S. 3 in Chelmsford and Billerica, and I-90 in Weston and Newton.