Determining Guidelines for Ramp and Interchange Spacing (2011)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Chapter 2 Information Gathering

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-1 Chapter 2 INFORMATION GATHERING The project team collected information to identify the origins of the current design standards and practice for ramp and interchange spacing and to determine how various spacing and interchange configurations can impact a facility’s safety and operations. The primary information gathering activities that the project team completed include: • Conducting and summarizing domestic and international research on ramp and interchange spacing, operations, and safety; • Reviewing and summarizing human factors considerations, such as sign sequencing and message units; a review of sample information from five state agency freeway signing handbooks; and a summary of the underlying philosophy for providing guidance information and driver information processing; • Summarizing design vehicle evolution, such as documenting changes in passenger car performance characteristics (i.e., acceleration, braking, transmission type, etc.) and design vehicles (i.e., weight-to-horsepower ratios and various tractor-trailer combinations); • Identifying and summarizing the various terms and design elements associated with ramp and interchange spacing described in various planning, operations, and design documents used in common practice today; • Conducting a focus group meeting consisting of planners, designers, and operators of freeways and interchanges and other interested parties to assist in identifying concerns or needs in the current practice of ramp and interchange spacing; • Requesting input from the NCHRP 3-88 panel to collaborate and generate ideas for the work plan and guidelines development; and, • Assessing existing datasets from recent and ongoing NCHRP projects on highway capacity and operations. The following sections provide a summary of findings from these activities. 2.1 LITERATURE REVIEW 2.1.1 General Literature In the early days of the Interstate Highway system, a number of studies offered general guidelines for interchange and ramp spacing values. Interchange spacing values were often expressed in terms of the centerline distance between crossroads. Ramp spacing studies were often focused on urban cores where conventional interchange forms were often not used.

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-2 In 1957, Owens of the Automotive Safety Foundation stated that “one mile between interchanges is a desirable minimum with one-half mile an absolute minimum” (1). For purposes of estimating the cost of the Interstate Highway System, a guide by the Bureau of Public Roads (BPR) identified general considerations for interchange spacing: “It is important that interchanges be located so as to properly discharge and receive traffic from other Interstate and Federal-aid system routes or major arterial highways or streets. It is equally important that they not be spaced so closely as either to unnecessarily increase the cost of the System or interfere with the free flow and safety of traffic on the Interstate System” (2). In practice, the BPR report noted that interchange spacing of 1 to 2 miles “appear to be the evolving pattern” in areas just outside of the central business district, with an increase to “about 2 to 4 miles in suburban outlying areas” (2). In downtown areas, the study assumed spacing would be as close as physically possible. In 1959, Jack Leisch offered interchange and ramp spacing guidance, beginning with the importance of the issue: “Widely spaced interchanges do not provide the needed service or develop the potential use of the facility. Too many interchanges, on the other hand, result in friction, inefficiency, and loss of speed and capacity (3).” To determine the “right” spacing of interchanges on urban freeways, Leisch considered city size, area type, street pattern, geometric features, and operational characteristics and presented some general considerations: • Large commercial and industrial areas require more interchanges than less developed areas. • Cities with an irregular street pattern also tend to require more interchanges than cities with a grid. • The distance between a direction interchange and a “regular” interchange should be greater than the distance between two regular interchanges. Leisch acknowledged signing needs as a “definite consideration” in the spacing of interchanges but did not incorporate them directly into his spacing guidelines. Leisch considered the “maneuver and weaving length” between an entrance ramp and exit ramp to be the basis of “absolute minimum spacing.” By assuming ramp lengths, the spacing values were presented as a measure of crossroad centerline to crossroad centerline. Exhibit 2-1 shows that the absolute minimum spacing was 1,800 feet, the normal minimum spacing was 2,600 feet, and the preferable minimum spacing was 4,200 feet.

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-3 Exhibit 2-1 General Interchange Spacing Recommendations by J.E. Leisch, 1959 (3) To account for the effects of volume on weaving operation, Leisch presented a table, shown in Exhibit 2-2, which was more appropriate for use in ramp and interchange design if the level of required information was available. The “low limit,” or minimum spacing, maintained a 35 mile per hour (mph) free flow speed on the freeway.

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-4 Exhibit 2-2 Specific Interchange Spacing Recommendations by J. E. Leisch, 1959 (3) Leisch noted some areas, such as central business districts, had such great traffic demand that more ramps than the number allowed by spacing guidelines, shown in Exhibits 2-1 and 2-2, might be needed. Leisch presented several strategies of dealing with this, shown below in Exhibit 2-3.

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-5 Exhibit 2-3 Methods of Increasing the Number of Ramps by J.E. Leisch, 1959 (3) Leisch also approximated the increase in the number of ramps that could be achieved with the configurations above. He noted the use of criss-cross ramps, now often referred to as braided ramps, could permit 1.2 times as many ramps per mile as a conventional design would. Ramp groupings, or collector-distributor roads (C-D roads), could permit 1.3 times as many ramps per mile. Lateral distributors might permit twice as many ramps per mile as conventional design. In addition to non-conventional ramp configurations, Leisch suggested building parallel or feeder freeway facilities traffic where volumes would be high enough to justify doing so.

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-6 Over the years, guidelines from these early studies and others have been incorporated into various versions of policy by the American Association of State Highway and Transportation Officials (AASHTO) and the American Association of State Highway Officials (AASHO), AASHTO’s previous name. In addition, some states have their own spacing policies, such as many states use have a 1 mile minimum crossroad-to-crossroad interchange spacing guideline for urban areas and a 2 or 3 mile minimum spacing guideline for rural areas. Since 1984, editions of AASHTO’s A Policy on Geometric Design of Highways and Streets, commonly referred to as the AASHTO Green Book, have stated that “a general rule of thumb for minimum interchange spacing is 1.5 km [1 mi] in urban areas and 3.0 km [2 mi] in rural areas” (4-8). 2.1.1.1 COMPONENTS OF INTERCHANGE SPACING The Institute of Transportation Engineers’ (ITE) Freeway and Interchange Geometric Design Handbook (ITE Freeway Handbook) declares that the 1 mile urban spacing minimum is arbitrary to a certain extent (9). The ITE Freeway Handbook presents a rational for the interchange spacing values based upon the geometric requirements of different interchange components, as shown in Exhibit 2-4. The crossroad-to-crossroad spacing value resulting from this approach is approximately 1 mile, which is the commonly used minimum interchange spacing guideline for urban areas. Exhibit 2-4 Interchange Spacing – Urban Areas, ITE Freeway Handbook (9) The ITE Freeway Handbook notes that it takes approximately 1,000 feet to reach the gore of the entrance ramp from the crossroad. This length accommodates grade changes, acceleration, and queues from ramp meters, if they exist. The distance from the entrance gore to the merging tip varies based on entrance type (taper or parallel) and the curve of the ramp, but it is generally between 400 and 800 feet. The minimum distance between the merge tip and diverge tip (painted gores) is 1,600 feet if both interchanges are service interchanges and 2,000 feet if one interchange is a system interchange; AASHTO Policy is cited as the basis of this measurement.

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-7 Similar to AASHTO, the ITE Freeway Handbook acknowledges that required weaving distances, as outlined in the 2000 Highway Capacity Manual (HCM), may be longer than 1,600 to 2,000 feet. If this is the case, the ITE Freeway Handbook notes the HCM weaving distances should be used as the basis of minimum ramp spacing. Between the diverging tip and the exit gore, a distance of 300 to 500 feet is needed depending on exit design (taper or parallel). Finally, another 1,000 feet between the exit gore and the centerline of the next crossroad is needed to accommodate the exit ramp. These distances sum to a length of 4,300 to 5,300 feet or approximately 1 mile. The ITE Freeway Handbook cautions that 1 mile spacing in urban areas “is a reasonable minimum guide [but] should not be policy” (9). Geometric considerations, such as ramp profile requirements or the presence of turning roadways, and operational considerations, such as weaving, multilane ramps requiring lane drops, or queue storage, are noted by the ITE Freeway Handbook as factors that may require the dimensions in Exhibit 2-1 to increase. This would result in centerline-to-centerline spacing values of more than 1 mile. In cases where interchange spacing of less than 1 mile may be needed, the ITE Freeway Handbook states that C-D roads, ramp braids, and frontage roads with split interchanges should be employed. In rural areas, the ITE Freeway Handbook states that a “5-mile minimum spacing is generally appropriate…to prevent every local or secondary road from interchanging with the freeway” (9). Once again, the ITE Freeway Handbook states that this distance is only a guide and can change for specific situations. For close rural spacing, the ITE Freeway Handbook suggests use of the same unconventional interchange forms noted in the section on close urban spacing. 2.1.1.2 INTERCHANGE COMPONENT LENGTHS A number of studies have examined one or more of the spacing values shown in Exhibit 2-4. This literature review is not intended to be an exhaustive summary of all research that has been conducted on ramp length, merging and diverging areas, or weaving. Instead, it is intended to provide an overview of common factors that have historically influenced interchange and ramp spacing values. Understanding the functionality of ramp components may assist in considering entrance and exit ramp geometric design event points that could form the basis for defining interchange and ramp spacing dimensions. In the 1960s, Fukutome and Moskowitz found that vehicles entering a freeway under low volume conditions on both the ramp and mainline use as much or even more merging distance than under high volume conditions (10). The study implied that even if a ramp is designed for high speed and is located in an area where volumes will always be low, it should still be designed with a long merging section. For diverging sections, Fukutome and Moskowitz noted past research, which found that many drivers did not use a

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-8 deceleration lane when one was provided prior to the exit gore and behaved as if the ramp were of a taper design (11). Fukutome and Moskowitz felt this justified allowing more flexibility in the length of a diverge section than in a merge section, as drivers seemed more willing to shorten a diverge compared to a merge. Weaving has long been recognized as a consideration when determining the spacing of ramps and was acknowledged in AASHO’s 1954 Policy on Geometric Design of Rural Highways (commonly referred to as the Blue Book) and the 1950 HCM (12, 13). Currently, the 2004 AASHTO Green Book provides Exhibit 10-68, which specifies minimum ramp spacing for several different combinations of ramp types (8). In addition to using AASHTO’s Exhbit 10- 68, the 2004 Green Book also recommends conducting HCM weaving calculations and using the longer of the two distances to determine ramp spacing in a specific situation. In the 2000 HCM, minimum weaving lengths were primarily a function of weave type, volume, and free-flow speed (14). NCHRP 3-75, conducted by Polytechnic University, revisited the weaving analysis procedures of the HCM, and the results were incorporated into the 2010 HCM. A later section of this report provides more information on the history of the ramp spacing and weaving in AASHTO Policies and the HCM. 2.1.1.3 IMPACTS OF SPACING ON FREEWAY AND ROAD NETWORK OPERATIONS Close spacing of ramps and interchanges can also have an effect on freeway operations as a whole, not just operations in isolated weaving areas. Additionally, ramp and interchange spacing impacts traffic conditions apart from freeways on arterials and other components of the road network. In 1958, Morawski studied lane distribution on a section of the New York Thruway with three lanes in each direction (1). The distribution of vehicles in the three lanes at a point one mile beyond an entrance ramp was similar to lane distribution at points 2, 3, and 10 miles beyond the entrance ramp. Morawski concluded that the “influence area” of an on-ramp extends less than 1 mile beyond the end of the ramp and noted that his findings were similar to a study done on the New Jersey turnpike. Martin et al’s 1973 study noted that it could be possible to reduce the amount of lane imbalance and turbulence that occurs in the first place by creating two low volume ramps instead of one high volume ramp (15). Martin et al noted this would work best if two off-ramps or two on-ramps could be placed consecutively to avoid any weaving problems. Three studies conducted in the 1960s examined the impact of ramp and interchange spacing on freeway operations. Studies in Atlanta and Detroit were conducted with field observations, and a third study was conducted with a travel demand model calibrated with data from Chicago. The Atlanta and Chicago studies included analysis of the impacts of interchange spacing

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-9 on arterial operations and the road network as a whole. A summary of each is provided below. 2.1.1.3.1 Atlanta study (16) In 1963, Covault and Roberts studied the Atlanta freeway that is now part of I-75/85 and known as the downtown connector. Their study area began in downtown Atlanta at Williams Street and ran north to where the freeway divides into what is now I-75 and I-85. This segment of freeway is approximately two miles long and had four on-ramps, three-off ramps, and the diverge of the two freeways at the northern end. The freeway had three lanes in each direction, and only the northbound direction of the freeway was studied. This study varied the spacing of on-ramps by closing ramps during the p.m. peak hour, when traffic would be heaviest on this segment. The highest volume on the freeway occurred when all ramps were open, and the closure of any of the ramps resulted in “smoother” and more desirable flow on the freeway. However, closing any of the ramps resulted in higher traffic volumes on surrounding arterials. The study identified one benefit to the arterial network when a ramp was closed: the elimination of a conflict point at the ramp terminal intersection. The authors suggested closing the 14th 2.1.1.3.2 Detroit study (17) Street ramp during the p.m. peak period. This ramp was closest to where the freeway divides and where weaving was most intense. During the study, closing this ramp resulted in a significantly lower travel time on the freeway than any other scenario but also in a higher overall system travel time than any other scenario. Forbes, Mullin, and Simpson studied three sections of the Lodge Freeway in Detroit in 1965. The sections did not have common endpoints and there were unstudied areas between the study sections. Each section was analyzed in both directions, resulting in six segments of study. Generally, the segments ended at a full interchange and had a partial interchange within them. Ramp spacing within the six study segments varied from 1,200 to 3,350 feet. It is unclear what the endpoints of these measurements were. Two successive on-ramps were found to interfere with traffic flow more than two successive off-ramps or an on-ramp and an off-ramp spaced at nearly the same distance apart. Two of the six segments had two on-ramps, and one of these segments experienced the lowest average velocity and highest number of stops for mainline vehicles. The authors found that operations were improved on some segments due to ramp congestion that effectively metered the flow of entering vehicles. In fact, the segment with the closest spacing of consecutive on-ramps performed better than a segment where the on-ramps were spaced further apart due to this phenomenon. Weaving, even with high volumes, did not necessarily result in a high number of vehicle stoppages or low average speeds.

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-10 The findings of this study that are relevant to NCHRP 3-88 include the operational conditions posed by closely spaced on-ramps and the benefits of metered on-ramp flow. The authors did not provide any minimum or desirable spacing values based upon the results of their study 2.1.1.3.3 Chicago study (2) In 1967, Satterly and Berry constructed an 8-mile by 8-mile model consisting of 430 links and 2,500 nodes, including entry nodes. Freeways were spaced 4 miles apart, resulting in one system interchange in the middle of the network and others along the edge. Arterials were spaced ¼ mile apart, with one of every four arterials (1 mile spacing) being major. Freeway spacing was determined based on other research identified by the authors, and arterial spacing was based on the existing road networks of Chicago and Detroit. Travel data collected from two 8-mile by 8-mile squares of Chicago in 1956 were used as the input of the model. One of the squares had a density of 21,000 people per mile, and the other had a density of 8,300 people per mile. The authors chose Chicago in part because it has a grid road system similar to the one that was modeled. The authors examined both interchange spacing and grade separation spacing. Grade separations are where arterials cross a freeway without an interchange; the alternative to this or an interchange is terminating the arterial on either side of the freeway. Interchange spacings of ½, 1, 2, and 4 miles and grade separation spacings of ¼, ½, 1, and 2 miles were studied. The metric used to determine optimal spacing was “annual transportation cost.” This cost included right-of-way purchases (where the model determined widening would be necessary), construction cost, and user costs, such as vehicle operation, accidents, and time. Costs were established for accidents and travel time based on other studies. The authors considered including land use impacts, community benefits and property values, maintenance costs, and comfort and convenience costs to users in the model but did not due to a lack of information about these factors. The minimum time path was used for route assignment. The authors found in their preliminary model runs that 4-mile interchange spacing was not feasible in the high density model because arterial volumes became extremely high. Even with 2-mile interchange spacing, the major (1 mile apart) arterials needed 8 lanes in the high density model. With both density scenarios, increasing interchange spacing from one-half to 1 mile increased total vehicle miles traveled (VMT) in the system. Increasing spacing from 1 to 2 miles decreased VMT, and increasing spacing from 2 miles to 4 miles increased VMT. The lowest VMT was observed with 2 mile interchange spacing. This spacing encouraged drivers to choose more direct arterials when travelling instead of the freeways. Volumes on arterials with an interchange increased with 2-mile interchange spacing compared to 1 mile or one-half-mile spacing.

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-11 The analysis of annual transportation cost found that in the high density area, optimum spacing of interchanges was 1 mile and optimum spacing of grade separations was one-half mile. For the low density area, optimum spacing of interchanges was 2 miles and optimum spacing of grade separations was one- half or 1 mile. However, the annual transportation cost differences between the scenarios were so small – generally 2 to 4 percent – that the authors recommend it not be used to determine interchange spacing. Instead, they suggested the following criteria: “(a) the characteristics of traffic operations on the system, (b) the amount of land required for the transportation facilities which must be removed from the taxable base of the community, (c) the reorganization of land use patterns due to the spacing of interchanges, and (d) convenience to the people of the community” (2). 2.1.1.4 GENERAL LITERATURE SUMMARY In the early days of the Interstate Era, a number of studies examined the spacing of ramps and interchanges, primarily on urban freeways. These studies generally defined spacing as the distance between the centerlines of successive crossroads that have interchanges, although some studies that were focused on weaving defined spacing as the distance between the ends of ramps. Exit signing requirements were not noted by the majority of studies identified here and do not appear to have played a prominent role in spacing guideline development. Studies that examined the road network as a whole noted a tradeoff between freeway operations and arterial operations. Maximizing the number of ramps and interchanges results in poor freeway operation but removes the greatest amount of traffic from surrounding arterials. Likewise, reducing the number of ramps and interchanges and spacing them further apart results in better freeway operation but may create congestion on arterials. Both Owings and Leisch independently recommended a “desirable” or “preferred” 1 mile minimum centerline to centerline spacing. Satterly and Berry found 1 mile spacing to be optimal in a high density area. Morowski’s study suggests 1 mile spacing is acceptable. These studies may have played a role in determining the 1 mile crossroad-to-crossroad urban interchange spacing guideline used by many states and found in the AASHTO Green Book since 1984. Rural interchange spacing and rural case studies are not as prominent in the literature. Ramp terminal spacing studies have primarily considered weaving operations, either qualitatively or through analytical procedures that were developed at the time. Today, the methodology of the 2000 HCM is the most common means of analyzing weaving sections, and revisions to this methodology may be incorporated in the 2010 HCM.

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-12 2.1.2 Primary Resource Documents The AASHTO Green Book and the HCM are two commonly used resources for freeway and interchange planning and design. The AASHTO Green Book recommends minimum interchange spacing dimensions, and the HCM quantifies the impact of interchange spacing on traffic operations. In addition, The Freeway and Interchange Geometric Design Handbook (ITE Freeway Handbook) published by the Institute of Transportation Engineers (ITE) and TRB’s Access Management Manual provide planning and design guidance related to interchange and ramp spacing. Summaries of these documents are discussed in the following section. 2.1.2.1 AASHTO POLICIES The 2004 AASHTO Green Book provides recommended minimum ramp and interchange spacing dimensions. Historically, the American Association of State Highway Officials (AASHO), the previous name of AASHTO, first addressed ramp spacing considerations in a 1944 Policy. Subsequent publications including A Policy on Geometric Design of Rural Highways (commonly referred to as the Blue Book) and A Policy on Geometric Design of Urban Highways and Arterial Streets (commonly referred to as the Red Book), both of which were the precursors to the 1984 Green Book, have addressed ramp and interchange spacing dimensions. AASHTO also publishes A Policy on Design Standards Interstate System, a document specifically for design of the Interstate Highways. 2.1.2.1.1 Ramp Spacing The first AASHO publication to address ramp sequence was A Policy on Grade Separations for Intersecting Highways in 1944 (18). No dimensions were given; however, the document presented examples of an entrance followed by an exit and an exit followed by an entrance ramp combinations. The use of an auxiliary lane for an entrance followed by an exit was suggested for the first time in this document. The 1954 Blue Book was the first AASHO publication to recommend that weaving analyses be conducted to determine the spacing distance between an entrance ramp and an exit ramp (12). The analysis could be done with the 1950 HCM. AASHO’s 1957 Red Book provided diagrams of various ramp combinations and presented guidelines for minimum and desirable spacing between them (19). This diagram is shown below in Exhibit 2-5. The distances were measured “between successive approach noses or merging ends.” Minimum distances were based upon combined response and maneuver times of 5 to 6 seconds, and desirable distances were based on a combined response and maneuver time of 7 seconds. Response time consisted of the time necessary for a driver to “observe, comprehend, and respond to a sign or to some other guide”, and maneuver time consisted of the time necessary for a driver to shift one lane.

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-13 Exhibit 2-5 Ramp Terminal Spacing Guidelines, AASHO Red Book (1957) (19) The 1965 Blue Book provided diagrams similar to those in the 1957 Red Book, but with longer minimum and desirable distances between ramp terminals. This diagram is shown in Exhibit 2-6 (20). The distances were based on a decision and maneuver time as they had been in the 1954 Blue Book, the time was changed from 5 to 8 seconds to 5 to 10 seconds and 80

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-14 mile per hour design speed category was added to the table. The document also noted that, for consecutive exits, the minimum spacing for adequate signing is 1,000 feet on a full freeway and 600 feet between an exit on a full freeway and an exit on a C-D road. The minimum spacing dimensions in Exhibit 2-6 do not appear to account for this signing requirement. Exhibit 2-6 Ramp Terminal Spacing Guidelines, AASHO Blue Book (1965) (20)

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-15 The 1973 Red Book recommended a minimum of 1,000 feet between successive exits on a full freeway and 800 feet between an exit on a full freeway and an exit on a collector-distributor road, as shown in Exhibit 2-7 (21). This full freeway spacing dimension exceeded the spacing dimension for the same exit arrangement in the 1965 Blue Book, and this full freeway to collector-distributor spacing dimension exceeded the spacing dimensions in the 1965 Blue Book for all but an 80 mph design speed. The 1973 Red Book ramp spacing guidelines also met or exceeded the minimum requirements based on signing that where presented in the 1965 Blue Book. The 1973 Red Book also included tables of minimum acceleration and deceleration lane lengths. The maximum acceleration lane length listed was 1,590 feet, while the maximum deceleration lane length was only 615 feet. Exhibit 2-7 Minimum Spacing Between Successive Exit Terminals, AASHO Red Book (1973) (21) In 1975, Jack E. Leisch presented a paper to the Region 2 AASHTO Operating Committee on Design in Mobile, Alabama, that contained a table with “Recommended Minimum Ramp Terminal Spacing” for various combinations of ramps (22). The table included “desirable minimum,” “adequate minimum,” and “absolute minimum” spacing values, and is shown below in Exhibit 2-8. The “absolute minimum” values in Leisch’s table were included in Figure X-67 of the 1984 Green Book and have been carried forward into all succeeding Green Books. Metric equivalents for Leisch’s US Customary units appeared in the 1994 edition. Currently, these values appear in Exhibit 10-68 of the 2004 AASHTO Green Book, which is depicted below in Exhibit 2-9. The table contains a footnote that states the distances between the ramp terminals are to “provide sufficient weaving length and adequate space for signing.”

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-16 Exhibit 2-8 Minimum Ramp Terminal Spacing, J. E. Leisch, 1975 (22) Exhibit 2-9 Current AASHTO Policy on Minimum Ramp Terminal Spacing, AASHTO Green Book, 2004 (8) AASHTO’s Exhibit 10-68 notes that dimensions presented should be checked according to the procedures presented in the 2000 HCM. The exhibit also states that larger dimensions of the HCM and values of Exhibit 10-68 should be used as the basis of design.

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-17 2.1.2.1.2 Interchange Spacing Interchange spacing guidance first appeared in the 1984 Green Book. The text has remained virtually the same in all Green Books since, including the 2004 Green Book, which states: “Minimum spacing of arterial interchanges (distance between intersecting streets with ramps) is determined by weaving volumes, ability to sign, signal progression, and lengths of speed-change lanes. A general rule of thumb for minimum interchange spacing is 1.5km (1 mi) in urban areas and 3.0km (2 mi) in rural areas.” (8) The statement implies that one mile spacing is for adjacent interchanges where there are ramps between the two (entrance from the upstream interchange followed by an exit to the downstream interchange). 2.1.2.1.3 Spacing on the Interstate Highway System AASHTO’s A Policy on Design Standards Interstate System (AASHTO Interstate Standards) presents standards specifically for the Interstate Highway System (23). The most recent version was published in 2005, and defines its role as follows: “All interstate highways shall meet the following minimum standards for segments constructed on new right-of-way and segments undergoing complete reconstruction along existing right-of-way. (23)” Freeways built or added to the Interstate Highway System prior to 2005 are grandfathered in under the expectation that they meet standards that were in effect at the time of their construction and/or inclusion into the Interstate Highway System (23). For interchanges, the AASHTO Interstate Standards note that spacing “has a significant effect on the operation of interstate highways” and “in areas of concentrated development, proper spacing may be difficult to obtain because of demand for frequent access (23).” The following spacing dimensions are offered: “As a rule, minimum spacing should be 1.5 km (1 mi) in urban areas and 5 km (3 mi) in rural areas, based on crossroad to crossroad spacing. In urban areas, spacing of less than 1.5 km (1 mi) may be developed by grade-separated ramps or by collector-distributor roads. (23)” Similar to the AASHTO Green Book, separate spacing dimensions are provided for urban and rural areas, and the minimum urban spacing dimension is 1 mile. However, the AASHTO Interstate Standards call for 3- mile interchange spacing in rural areas, while the Green Book calls for only 2-mile interchange spacing in rural areas. The two documents are not in

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-18 conflict with one another, because not all interchanges are on the Interstate Highway System. 2.1.2.1.4 AASHTO Summary AASHTO policies on ramp and interchange spacing in the 2004 Green Book have not changed significantly since the first Green Book was published in 1984. In 1984, ramp spacing values for entrance-entrance, exit-exit, and entrance-exit ramp combinations were increased in comparison to the 1965 Blue Book, while ramp spacing values for exit-entrance and turning roadway ramp combinations remained similar in the 1965 Blue Book. Prior to 1965, ramp spacing dimensions were not quantified in AASHTO Policy. Interchange spacing values did not appear in AASHTO policies prior to 1984, when the current guideline of 1-mile spacing in urban areas and 2-mile spacing in rural areas was introduced. In addition to the Green Book, AASHTO also publishes a policy specifically for the interstate highway system which contains minimum interchange spacing dimensions. 2.1.2.2 HIGHWAY CAPACITY MANUAL The Highway Research Board, now known as the Transportation Research Board (TRB), published the first edition of the HCM in 1950. This first edition addressed uninterrupted flow facilities, weaving sections, and ramps. Subsequent editions of the HCM in 1965, 1985, 1994, 1997, 2000, and 2010 have provided updated analysis procedures. Ramp and interchange spacing in the most recent edition of HCM is addressed in three chapters: analysis of basic freeway segments, weaving, and freeway merge and diverge segments. 2.1.2.2.1 Basic Freeway Segments The 1965 edition of the HCM was the first to include the concept of level- of-service (LOS) (24). For freeways and other expressways, LOS was defined in terms of operating speed and volume. No method of calculating these numbers was provided, but they could be measured in the field. The 1985 HCM was the first to base basic freeway LOS explicitly on density, a measure of passenger cars per mile, per lane (25). Density was computed from flow rate and free-flow speed. If free-flow speed was not measured in the field, it could be estimated by assuming a base free-flow speed and making adjustments for a number of different factors. The 1994, 1997, 2000, and 2010 HCMs have used this same framework to calculate free-flow speed (26, 27, 14). In the 1997 and 2000 HCMs, interchange density was one of the factors in the calculation of free-flow speed (27). In the 1997 and 2000 HCMs, interchange density was calculated over a 6- mile segment of freeway: 3 miles upstream and 3 miles downstream of the location being studied (14). An interchange was defined as having at least one on-ramp, so interchanges only having off-ramps are not included in the determination of interchange density. The base interchange density in the HCM was 0.50 interchanges per mile. With this density, no adjustment to base free-flow speed is made. With a density of two interchanges per mile,

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-19 base free-flow speed was reduced by 7.5 miles per hour. Base free-flow speed adjustments were not specified for densities of more than 2.0 interchanges per mile. In the 2010 HCM, total ramp density replaced interchange density in the equation used to calculate free-flow speed. Total ramp density is defined as the number of ramps per mile of a 6-mile segment of freeway: 3 miles upstream and 3 miles downstream of the location being studied (28). There is no assumed base density of ramps, so a reduction in FFS is always applied unless there are none over a six mile segment. 2.1.2.2.2 Weaving The 1950 HCM defined weaving as “the act performed by a vehicle in moving obliquely from one lane to another, thus crossing the path of other vehicles moving in the same direction” (13). Observations summarized in the 1950 HCM, primarily from the freeway network near the Pentagon, identified that capacity of a weaving section decreased as weaving volume increased (13). The 1965 HCM defined weaving as “the crossing of traffic streams moving in the same general direction, accomplished by successive merging and diverging”(24). The primary weaving model of the 1965 HCM was based upon weaving volume and weaving length. The curves accounted for weaving operational impacts with segments up to 8,000 feet in length. No other edition of the HCM has provided a weaving analysis procedure for a segment this long. Weaving length was measured “from a point at the merging end where the distance between the projected edges is 2 ft to a point at the diverge end where the distance between the edges is 12 ft” (13). This definition has been used, with modifications in wording but not dimensions, in all editions of the HCM since 1965. The 1985 HCM defined weaving as “the crossing of two or more traffic streams travelling in the same general direction along a significant length of highway without the aid of traffic control devices (25). The 1985 HCM also noted that “Weaving areas are formed when a merge area is closely followed by a diverge area, or when an on-ramp is closely followed by an off-ramp and the two are joined by an auxiliary lane” (25). With minor modifications, both of these statements have been used in all editions of the HCM through 2000. “Close” spacing was defined in the 1985 through 2000 HCMs as 2,500 feet or less between the ramps or the merging and diverging movements. This distance, shown in Exhibit 2-10, is measured “from a point at the merge gore where the right edge of the freeway shoulder lane and the left edge of the merging lane(s) are 2 ft apart to a point at the diverge gore where the two edges are 12 ft apart” (14). This definition was based on the geometry of loop ramps (28).

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-20 Exhibit 2-10 Definition of Weaving Segment Length, 1985 through 2000 HCMs (25, 26, 27, 14) The weaving model of the 1985 HCM was based upon several research projects conducted in the early 1980s (25). LOS was based upon speed of weaving and non-weaving vehicles in the weaving segment, and length of the weaving segment is one of the inputs into the equation used to calculate speed. With recalibration, this basic model was used in the 1994, 1997, and 2000 editions of the HCM. In the 1997 edition, the determination of LOS was changed to be based upon density instead of speed, with speed being used in the calculation of density (27). Currently, the 2010 HCM defines weaving as “the crossing of two or more traffic streams traveling in the same direction along a significant length of highway without the aid of traffic control devices (except for guide signs)” (28). The 2010 HCM contains a new analysis procedure not based on the 1985 HCM, and a new definition of weaving segment length shown below in Exhibit 2-11 that is not based upon loop ramp design. Segments are no longer limited to 2500 feet in length. In Exhibit 2-11, the weaving segment length is defined as LS (the “short length” except in cases where barrier stripes do not exist. In those cases, LB (the “base length”) is used to define weaving segment length. LB LS Exhibit 2-11 Definition of Weaving Segment Length, 2010 HCMs. (28) Weaving segment LOS is a function of density. Speed is an input used to calculate density, and weaving segment length is an input used to calculate speed.

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-21 2.1.2.2.3 Freeway Merge and Diverge Segments The 1965 HCM was the first edition to analyze operations at ramp junctions on freeways. LOS was based upon volume and number of lanes. Dimensions of the ramp and location of nearby ramps (except in weaving situations) were not factored into the analysis (24). The 1985 HCM was the first edition to consider the influence of nearby ramps on ramp junction LOS. Ramps within 6,000 feet of a study ramp could influence LOS, depending upon volume, number of lanes, and ramp type. In the 1994 HCM, the “influence area” of a ramp was reduced to 1,500 feet downstream from a physical merge point and 1,500 feet upstream of a physical diverge point (26). The influence area was also limited to the two right lanes of the freeway. The 1997, 2000, and 2010 HCMs have continued to use the dimension of the 1994 HCM to define a ramp’s influence area (27, 8, 28). These dimensions are shown in Exhibit 2- 12. 1,500 ft 1,500 ft Exhibit 2-12 Definition of Merge and Diverge Influence Areas, HCM (2010) (28) The LOS of merging and diverging areas in the 2010 HCM is based on density, and the length of the acceleration lane or deceleration lane is used in the density calculation. The length of the acceleration or deceleration lane is measured from “the intersection of the edge of the travel way to the freeway and the ramp (Point A) and the downstream intersection of the freeway and ramp edges of the travel way (Point B)” (14). These dimensions are shown in Exhibit 2-13. LA LD (a) Parallel Acceleration Lane (b) Tapered Deceleration Lane Exhibit 2-13 Acceleration and Deceleration Lane Length, HCM (2010) (28) 2.1.2.2.4 HCM Summary The methodologies of the 2010 HCM contain several procedures that are relevant to ramp and interchange spacing. The analysis of basic freeway segments identifies a reduction in free-flow speed as total ramp density increases Weaving is found to occur at locations where an off-ramp or major diverge area closely follows an on-ramp (with auxiliary lane) or a major merge area. The definition of “close” has changed over the years through the different editions of the HCM. Finally, the 2010 HCM states that turbulence

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-22 areas extend 1,500 feet upstream of diverge areas and 1,500 feet downstream from merge areas. To minimize the impacts of ramps and interchanges on freeway operations, these analysis procedures suggest the following designs: • Total ramp density should remain low. • On-ramps and off-ramps with an auxiliary lane or major merge and diverge areas should be separated far enough apart to avoid poor weaving section operation. • On-ramps and off-ramps should be separated by more than 3,000 feet so that turbulence areas do not overlap. 2.1.2.3 OTHER NATIONAL-LEVEL DOCUMENTS In addition to the AASHTO Green Book and the HCM, two other national level-documents have been developed to address ramp and interchange spacing dimensions. These include: Freeway and Interchange Geometric Design Handbook (ITE Freeway Handbook), published by the ITE in 2005 and the Access Management Manual published by TRB in 2003 (9, 29). 2.1.2.3.1 Freeway and Interchange Geometric Design Handbook (9) The ITE Freeway Handbook’s guidelines for ramp and interchange spacing dimensions are largely based on AASHTO policy and publications by Jack E. Leisch. The ITE Freeway Handbook states that most state departments of transportation (DOTs) have minimum interchange spacing guidelines of 1 mile in urban areas and 5 miles in rural areas, but the guidelines are “arbitrary to a certain extent.” The ITE Freeway Handbook develops its own rational for urban interchange spacing, discussed in detail in another section of the literature review. Considering the geometric requirements of various interchange components, the ITE Freeway Handbook also arrives at a 1-mile minimum dimension for crossroad-to-crossroad interchange spacing in urban areas. However, this resource cautions that in many cases urban interchanges will need to be spaced further apart than this to due to factors such as ramp profile requirements, interchange form, weaving length, and queue storage requirements on ramps. To create interchange spacings of less than 1 mile, the ITE Freeway Handbook suggests using C-D roads, ramp braids, or frontage roads. For ramp spacing, the ITE Freeway Handbook acknowledges both Exhibit 10-68 of the 2004 AASHTO Green Book (shown in Exhibit 2-9 of this report) and weaving analysis. Both should be considered, and the greater of the two distances should be used in design. The ITE Freeway Handbook suggests using the 2000 HCM or the Leisch method to conduct weaving analysis and provides updated charts for conducting the Leisch method. The ITE Freeway Handbook also presents a different version of the AASHTO Green Book’s Exhibit 10-68, developed by Jack E. Leisch. As

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-23 seen in Exhibit 2-14, this version offers “desirable” and “adequate” ramp spacing dimensions in addition to “absolute” minimums. This version is similar to the table Jack E. Leisch presented in 1975 that is shown in Exhibit 2-8, but some of the distances in the “En-Ex (Weaving)” category of ramp groupings have been shortened in the version appearing in the ITE Freeway Handbook. Exhibit 2-14 Minimum Ramp Terminal Spacing Guidelines, ITE Freeway Handbook (2005) (9) 2.1.2.3.2 Access Management Manual (29) The 2003 Access Management Manual states that “the minimum spacing between interchanges needed to allow unfamiliar drivers to make a safe lane change depends upon several factors: speed, through volume, on-ramp volume, off-ramp volume, driver performance, and signing” (29). Thus, mainline freeway safety seems to be the primary consideration in the manual’s spacing guidelines, with an acknowledgement to the many factors that affect it. In suburban or “developing urban” areas, the manual recommends a 3-mile spacing of interchanges so that good route signing and decision distance can be provided on the freeway and surrounding land can develop in a “traditional manner.” In rural areas, 6-mile interchange spacing is recommended to provide reasonable connection to rural highways. If the

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-24 area suburbanizes, additional interchanges can be added to create 3-mile spacing. 2.1.2.4 PRIMARY DOCUMENT SUMMARY Minimum interchange spacing guidelines from four national-level documents are presented below in Exhibit 2-15. Urban Rural Dimension Definition A Policy on Geometric Design of Highways and Streets (2004) 1 mi 2 mi Crossroad-to-crossroad A Policy on Design Standards Interstate System (2005)* 1 mi 3 mi Crossroad-to-crossroad Freeway and Interchange Geometric Design Handbook (2006) 1 mi 2 mi acceptable, 5 mi preferred Crossroad-to-crossroad Access Management Manual (2003) 3 mi 6 mi Not stated * Applies to Interstate Highway System only Exhibit 2-15 Interchange Spacing in National-Level Documents Since 1984, the AASHTO Green Book has recommended that interchanges be spaced a minimum of 1 mile in urban areas and 2 miles in rural areas, with the distance measured crossroad-to-crossroad. It is unclear how these dimensions were chosen. AASHTO Interstate Standards, the Access Management Manual, and the Freeway Design Handbook also provide guidance on interchange spacing with distances being measured crossroad-to- crossroad. Additionally, the 2000 HCM has identified a reduction in basic freeway segment speeds when interchange density exceeds 0.5 interchanges per mile (i.e., 2-mile interchange spacing). Information on minimum ramp spacing dimensions is provided by the 2000 HCM and the 2004 AASHTO Green Book. Per 2000 HCM methodology, weaving can occur with distances up to 2,500 feet, and the turbulence area caused by a ramp is 1,500 feet in length. The 2004 Green Book provides a table of minimum ramp terminal spacing dimensions for different combinations of ramp types. This table, with the same dimensions, first appeared in the 1984 Green Book. 2.1.3 State Spacing Guidance The project team reviewed highway design and traffic engineering documents from a sample of states to consider a range of state-level policies and guidelines related to ramp and interchange spacing. A summary of interchange spacing is presented in Exhibit 2-16. None of the sampled state documents included in the review were found to have minimum interchange spacing guidelines below the 2004 AASHTO Green Book criteria of 1 mile in urban areas and 2 miles in rural areas. Some states adhere exactly to the

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-25 AASHTO values, and others call for greater spacing. Summaries and highlights from the sampled state documents are provided below. Urban Service Interchanges Urban System Interchanges Suburban or Transforming Rural California 1 mi 2 mi - 2 mi Florida* 1-3 mi - - 3-25 mi Florida** 1 or 2 mi - 3 mi 6 mi Illinois 1 mi - 2 mi 3 mi New Jersey 1 mi - - 2 mi Oregon 3 mi - - 6 mi Pennsylvania 1 mi - - 2 mi * Florida Manual of Uniform Minimum Standards for Design, Construction, and Maintenance for Streets and Highways ** Florida Technical Resource Document 1 and Plans Preparation Manual Exhibit 2-16 State DOT Guidelines for Minimum Interchange Spacing 2.1.3.1 CALIFORNIA California’s Highway Design Manual (HDM) states that “minimum interchange spacing shall be one mile in urban areas, two miles in rural areas, and two miles between freeway-to-freeway interchanges and local street interchanges” (30). The HDM suggests that auxiliary lanes, grade separated ramps, collector distributor roads, and ramp metering are strategies that could improve operations at closely spaced interchanges. “Close spacing” is not defined in the HDM. For successive on-ramps, the HDM states that the minimum distance between the ramps should be “about 1,000 feet” so that the standard on- ramp acceleration taper can be used. If the upstream ramp adds an auxiliary lane, the HDM states it should merge with the auxiliary lane in a standard 50:1 convergence. The HDM does not identify the exact points between which spacing should be measured. For successive off-ramps, the HDM states that the minimum distance between the ramps should be 1,000 feet on a full freeway and 600 feet on a collector-distributor road. These distances are dictated by guide signing considerations. For an on-ramp followed by an off-ramp, the HDM uses weaving considerations as the basis for spacing. According to the HDM, weaving sections should be designed for LOS C or D in urban areas or LOS B or C in rural areas. The HDM states LOS should be determined using the Leisch Method or the 1965 HCM, because other methods, including those in the 1994 HCM, may not always produce accurate results. The document states that a minimum weaving length of 1,600 feet should be provided on a main freeway in any area unless costs or environmental impacts would be severe.

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-26 In addition to the HDM, California issues Design Information Bulletins (DIB). DIB 77 states that “the minimum spacing between interchanges shall be 1.5 km (4,900 feet) in urban areas and 3.0 km (9,800 feet) in rural areas. The minimum spacing shall be 3.0 km between ‘freeway-to-freeway’ and ‘local’ interchanges” (31). These distances approximate the US Customary dimensions of the HDM. 2.1.3.2 FLORIDA Florida has at least three documents that address interchange spacing. The Manual of Uniform Minimum Standards for Design, Construction, and Maintenance for Streets and Highway, commonly referred to as the Florida Green Book, calls for a minimum interchange spacing of 1 to 3 miles in urban areas and 3 to 25 miles in rural areas on all limited access highways (32). No guidance is provided for selecting a single value within these ranges. However, two other FDOT documents show different minimum spacing values. The Technical Resource Document 1 – Department Engineering Standards cites standards from Rule Chapter 14-97 of the Florida Administrative Code, and the Plans Preparation Manual cites Florida DOT Access Management Guidelines Rule 14-97 (33, 34). Both of these documents call for a 1-mile minimum spacing in a central business district (CBD) and CBD fringe areas, a 2-mile minimum spacing in other parts of existing urbanized areas, a 3-mile minimum spacing in “transitioning” urban areas, and a 6-mile minimum spacing in rural areas. Spacing is measured from centerline-to-centerline of the crossroads. 2.1.3.3 ILLINOIS The Illinois Bureau of Design and Environment Manual indicates that spacing interchanges further apart improves freeway operations, level of service, and safety (35). Desirable spacing between interchanges is noted as being at least 2 miles in urban areas, 4 miles in suburban areas, and 7.5 miles in rural areas. This provides an entering driver adequate distance to adjust to the freeway environment, allows for weaving maneuvers, and provides for an adequate sign sequencing distance. However, the manual also acknowledges that existing streets and highways, traffic operations, and social considerations may require the distances between adjacent interchanges to vary. The manual concludes that minimum distances should not be less than 1 mile in urban areas, 2 miles in suburban areas, and 3 miles in rural areas. Urban spacings of less than 1 mile may be developed, according to the manual, by using collector distributor roads. 2.1.3.4 NEW JERSEY The New Jersey Roadway Design Manual (RDM) indicates that close interchanges interfere with traffic flow and safety because of insufficient distance for weaving (36). The RDM states minimum crossroad-to-crossroad spacing should be 1 mile in urban areas and 2 miles in rural areas. Spacing of

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-27 less than 1 mile may be developed in urban areas with collector-distributor roads. For ramps, the RDM contains a diagram that provides the minimum physical nose to physical nose spacing guidelines for freeways, C-D roads, and arterials. The RDM identifies the diagram as an adaptation of material from the 2001 AASHTO Green Book. The spacing values do meet or exceed those specified in Exhibit 10-68 of the 2001 Green Book, but the diagram is similar in appearance to Figure IX-11 of the 1965 Blue Book. For cloverleaf interchanges, which are not included in Exhibit 2-16, the RDM calls for a maximum separation between the loop ramp terminals of 800 to 1,000 feet. The RDM diagram is shown below in Exhibit 2-17. Exhibit 2-17 Ramp Spacing Guidelines, New Jersey Roadway Design Manual (2002) (35)

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-28 2.1.3.5 OREGON Oregon’s Highway Design Manual (HDM) addresses spacing in two separate sections (37). In Section 9.6 Interchange Design minimum interchange spacing is set at 3 miles in urban areas and 6 miles in rural areas. Distance is measured from crossroad-to-crossroad, and closer interchanges require a design exception. Oregon is unique among states sampled in that separate guidelines are provided for interchanges on non-freeways; the minimum spacing on such facilities is 1.9 miles in urban areas and 3 miles in rural areas. Section 9.6 also includes a figure similar to the one that has been in the AASHTO Green Book since 1984 showing minimum ramp terminal spacing for various combinations of entrance and exit ramps. The HDM’s figure, shown below in Exhibit 2-18, also provides “desirable” and “adequate” spacing values. Exhibit 2-18 Ramp Spacing Guidelines, Oregon Highway Design Manual (2003) (36) Section 6.2 of the HDM, Interchange Spacing – Access Management, also lists minimum freeway interchange spacing values of 3 miles in urban areas and 6 miles in rural areas, measured from crossroad-to-crossroad. This section of the HDM also contains a figure showing minimum spacing standards between the end of an acceleration lane taper and the start of a deceleration lane taper. The standard is 1 mile in urban areas and 2 miles in rural areas. It is unlikely that these standards would dictate spacing between successive interchanges. The total length of ramps, acceleration lanes, and deceleration lanes would need to be 2 miles long in an urban area and 4 miles long in a rural area. Exhibit 2-19 shows the HDM’s version of the minimum spacing standard figure for interchanges with two-lane crossroads. The HDM

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-29 has another figure for interchanges with multilane crossroads, but distance “A” (the end of acceleration taper to start of deceleration taper measurement) is unchanged. Exhibit 2-19 Minimum Spacing Standards Applicable to Freeway Interchanges with Two-lane Crossroads, Oregon Highway Design Manual (2003) 2.1.3.6 PENNSYLVANIA Pennsylvania’s Design Manual uses AASHTO’s Policy on Design Standards Interstate System as the basis of its spacing guidelines of interchanges on all types of facilities, not just Interstate Highways (38). The Design Manual was last formally published in 2002, but the interchange spacing guidance was changed in 2007 to reflect AASHTO’s 2005 Interstate Standards (39, 40). Currently, the Design Manual with Change #2 incorporated calls for 1-mile interchange spacing in urban areas and 3-mile spacing in rural areas. Spacing of less than 1 mile may be developed in urban areas with grade separated ramps or collector-distributor roads. Unlike the AASHTO Interstate Standards, the Design Manual does not define the points between which spacing is measured. 2.1.3.7 STATE GUIDELINE SUMMARY The project team summarized interchange spacing guidance from six states and ramp spacing guidance from three states. Among the six states, minimum crossroad-to-crossroad interchange spacing varied from 1 to 3 miles in urban areas and from 2 to 25 miles in rural areas. The 25 mile spacing guidance comes from Florida’s Green Book, which calls for a

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-30 minimum rural interchange spacing range of 3 to 25 miles. No state sampled has interchange spacing values less than those in the 2004 AASHTO Green Book (i.e., 1 mile in urban areas and 2 miles in rural areas). California, New Jersey, and Oregon have additional guidance that specifically addresses spacing between ramps. Illinois and Pennsylvania indirectly address ramp spacing by stating that crossroad-to-crossroad interchange spacing of less than 1 mile may be developed by using C-D roads or “grade separated ramps” (Illinois only). 2.1.4 International Spacing Guidance Literature reviewed for NCHRP 3-88 has shown that ramp and interchange spacing guidelines in the United States are based on research conducted domestically. To understand and document how interchange and ramp spacing considerations are applied abroad, the project team sampled spacing guidelines from a variety of foreign countries. 2.1.4.1 INTERNATIONAL FINDINGS Two papers identified in the literature search provide an overview of ramp and interchange spacing dimensions and guidelines in outside the United States (41, 42). In some cases, the year of the other international guidelines is not provided; therefore, it is only known that they are at least as old as the documents in which they are cited. International minimum spacing values and considerations are summarized in Exhibit 2-20. Country Urban Areas Rural Areas Distance Definition Year of Guidelines United Kingdom 3.75*(mainline design speed in km/h) m Nose-to-nose 1994 or earlier Germany 2700 m preferred for system interchanges 2200 m preferred for access interchanges 1700 m preferred for low volume interchanges 600 m absolute minimum Nose-to-nose 1976 France 1000 – 1500 m - Nose-to-nose Not provided Australia 1500 – 2000 m 3000 – 8000 m Crossroad-to- crossroad 1984 South Africa – Gauteng Province 3600 – 4200 m for system interchanges 2400 – 2800 m for service interchanges Crossroad-to- crossroad Earlier than 2000 South Africa – Nationwide 8000 m Not specified 1984 Austria Cities spacing, but not specifics, as a consideration when planning a freeway system - 1991 Switzerland, Greece, Ireland, Norway – No mention of spacing 1993 or earlier Exhibit 2-20 Minimum Spacing Guidelines in Foreign Countries (41, 42) European guidelines sampled consider spacing based on the distance between ramp noses and thus choose to define ramp spacing. Australia and

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-31 South Africa define spacing in terms of crossroad-to-crossroad distance, similar to the AASHTO Policies in the United States, thus choosing to define interchange spacing. The United Kingdom’s definition of a nose is shown in Exhibit 2-21. Adding the merge taper distance, weaving length, and diverge taper distance will result in the nose-to-nose ramp measurement. Note that this figure is based upon driving on the left side of the road. Exhibit 2-21 Definition of Ramp Components, United Kingdom Design Manual for Roads and Bridges (2006). Note left side driving. (43) The United Kingdom is unique in that its spacing criteria vary based on design speed of the mainline facility. For design speed of 120 km/h, minimum nose-to-nose spacing would be 3.75*120 = 450 meters. For a design speed of 100 km/h, minimum spacing falls to 3.75*100 = 375 meters. Germany has longer minimum spacing values than other European countries, perhaps because German freeways do not have speed limits. South African documents call for longer spacing dimensions than other countries. This was identified by a transportation agency in Gauteng, a province in South Africa, and was the basis of two studies (42, 44). The studies considered interchange spacing with respect to a number of factors and produced four policy statements that define how far apart interchanges should be: • The distance between interchanges should ideally be based on distances required for adequate signage. • The turbulence areas downstream from an off-ramp and upstream from an on-ramp should not overlap. Research in the United States incorporated into the HCM since 1994 has shown that turbulence extends approximately 450 meters (1,500 feet) upstream of off-ramps and downstream of on-ramps. • In specific situations, weaving may require a longer spacing than dictated by either of the above.

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-32 • Spacing between roadways is too coarse of a measure to use as a basis for determining interchange spacing because of the differences among interchange forms. For example, ramp spacing with a Parclo-A followed by a Parclo-B interchange will be greater than ramp spacing with consecutive diamonds. South Africa’s sign manual calls for an advance exit direction sign, an exit direction sign, and a gore exit sign. Depending on the environment and type of interchange, there are recommended distances prior to the exit where these signs should be placed. However, if interchanges must be placed closer than is ideal for signing, weaving will then dictate absolute minimum spacing (42, 44). The authors noted work by Cirillo in the United States who found that accident rates begin to increase sharply when nose-to-nose spacing falls below 2,500 meters. However, it was not apparent if this finding is incorporated into the guidelines. The South Africans also considered the impacts of spacing on speed, but data collected at two sites in South Africa found no correlation between interchange spacing and speed. Considering all of these factors, the authors recommended new spacing guidelines based on the distance from the end of the merge taper to the start of the diverge taper, referred to in South Africa as the Yellow Line Break Point distance and shown in Exhibit 2-22. This would be a departure from South African guidelines at the time, which measured spacing from crossroad-to-crossroad. The Yellow Line Break Point distance is shorter than the nose-to-nose distance commonly used in Europe because the start of a diverge taper is further from the crossroad than the nose, as depicted in Exhibit 2-21. However, it is similar to European guidelines in that it is based upon ramps and not crossroads. Proposed Gauteng Province spacing guidelines are presented in Exhibit 2-23. It is unclear if Gauteng Province has adopted these guidelines at the present time.

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-33 Exhibit 2-22 Definition of “Yellow Line Break Point Distance” Used in Gauteng Province, South Africa (2005). Note left side driving (44) Configuration Urban Rural Access to Access 1300 m 2200 m Access to System 2100 m 3300 m System to Access 1400 m 2200 m Absolute Minimum 500 m 500 m Distances are measured between the yellow line break point, as shown in Exhibit 2-20. Exhibit 2-23 Proposed Minimum Spacing in Gauteng province, South Africa (2000) (42, 44) 2.1.4.2 INTERNATIONAL SUMMARY The project team sampled international ramp and interchange spacing guidelines from three European countries, South Africa, and Australia. European guidelines define spacing as the distance between successive ramp noses, while South African and Australian guidelines measure spacing between successive crossroads. Australia’s guidelines are similar to those used in the United States. South Africa’s guidelines focus on the interchange type, service or system, instead of the area type, urban or rural. Nose-to-nose ramp spacing guidelines from France and Germany call for spacing values greater than the “Entry-Exit (Weaving)” section of Exhibit 10-68 in the 2004 AASHTO Green Book. English guidelines call for spacing dimensions approximately 100 meters shorter than those in the 2004 Green Book for typical freeway design speeds. However, these countries may not measure ramp spacing from the same points.

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-34 2.1.5 Signing Considerations Guide signs are used by drivers as a navigational aid while they travel. On freeways, guide signs identify upcoming exits in advance of and at the ramp itself. Ideally, signing should provide enough information for drivers to identify and locate exits but not so much information that drivers are overwhelmed with more information than they can comprehend. This presents challenges in adequately signing closely spaced interchanges because the amount of information provided in advance of an exit may exceed what drivers are able to comprehend, compared to distances typically provided at ramps with greater spacing distances. The Manual of Uniform Traffic Control Devices (MUTCD) provides guidance on many aspects of freeway guide signs, including the number of signs that should be used and the spacing between them (45). Additionally, the ITE Freeway Design Handbook provides guidance on the number of message units, or pieces of information, which a driver can be expected to comprehend at a single sign assembly (9). Some states, such as Texas and California, offer additional guidance related to freeway signing. 2.1.5.1 SIGNING PRINCIPLES Sign placement and spacing on interstates is dictated by the requirements of the driving task. The driving task consists of three subtasks: control, guidance, and navigation (46). Control consists of a driver’s operation of a vehicle and includes steering, braking and accelerating. Guidance consists of staying in a lane and maintaining a safe speed and path; car following, passing, and reaction to traffic control devices are part of the guidance subtask. Navigation consists of reaching a destination by following a route. Routes can be well-known to drivers or can require maps or verbal assistance from a passenger (47, 48). Freeway driving requires all three subtasks to be performed simultaneously (48). Control usually requires a low level of attention because it is a routine task that varies little from trip to trip. Guidance attention requirements vary greatly based on traffic and roadway conditions and a driver’s prior experiences and knowledge. Navigation usually requires a low level of attention but in some environments can require a great deal of attention (47). Sometimes the attention demands of all of the driving subtasks exceed what a driver is capable of handling. When this happens, the process of load shedding occurs. Drivers stop fully performing less important subtasks so attention can be dedicated to more important subtasks. Navigation is the first task to be sacrificed, followed by guidance. Drivers will allow themselves to miss an exit before allowing themselves to hit other vehicles or stop steering (47). Since navigation is the least important of the subtasks, aids such as signing should be made as clear and predictable as possible to minimize the amount of attention required to comprehend them.

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-35 In addition to being clear and predictable, signing must not be so extensive that it presents drivers with more information than they are able to process. Motorists should be presented with the maximum amount of useful visual information in a manner that is uniform but also prioritizes information that is most important (47). This concept is known as positive guidance. The MUTCD provides limits on the quantity of information (message units) that should be presented to drivers at once. Examples of message units include city names, route numbers, street names, cardinal directions, exit numbers, distances, and lane use arrows (47). According to the MUTCD, no more than two destination names or street names should appear on a single guide sign, and only one street name or destination name should appear on a guide sign when it is placed beside other guide signs (45). The MUTCD and the ITE Freeway Design Handbook recommend that no more than three guide signs should be placed at the same location (45, 9). Texas’s Freeway Signing Handbook cites a table by McNees and Messer allowing up to five guide signs at the same location but notes that the Texas MUTCD does not recommend this (47). Tables that specify limits on the number of message units that drivers can be expected to process at once are provided in Exhibits 2-24 and 2-25. Exhibit 2-24 Message unit limits, ITE Freeway Handbook (2005) (9)

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-36 Exhibit 2-25 Message unit limits, reprinted in Texas Freeway Signing Handbook (2008) (47) Driver expectancy also plays a large role in the driving task and attention demands become higher when unexpected events and situations occur. There are two types of expectancies: a priori and ad hoc. A priori expectancies are based on a lifetime of experience. Since most exits are on the right, a driver on an unfamiliar freeway will assume that exits will be on the right. Ad hoc expectancy is based on recent experiences within the driver’s current situation. For example, a driver on a rural freeway who has just passed several diamond interchanges with ground mounted guide signs one mile in advance will not expect the next interchange to be a cloverleaf interchange (45). In general, advance warning information assists drivers in making appropriate exit maneuvers, especially in unusual situations (48). Lunenfeld’s paper (46) and the Texas Freeway Signing Handbook (47) identify attributes of good freeway signing, which are presented below in Exhibit 2-26. Lunenfeld (1993) (46) Texas Freeway Signing Handbook (2008) (47) • Designed for drivers and target population • Responsive to task demands and driver attributes • Satisfies all information needs • Maintains interchange design and information system compatibility • Avoids surprises • Eliminates information-related error sources • Resolves conflicts when information sources compete • Uses spreading [distribute large amounts of information across multiple signs at multiple locations] • Uses repetition for interchange information treatments • Uses all available navigation aids and treatments • Provides information to meet the needs of unfamiliar road users • Provides advance information to allow for adequate decision making time • Do not necessarily identify every possible choice • May direct users along a longer route if it simplifies signing • Gives driver maximum amount of useful visual information • Prioritizes information based on importance • Presents information uniformly • Information remains visible under most or all environmental conditions Exhibit 2-26 Attributes of Good Freeway Signing Identified in Prior Studies

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-37 2.1.5.2 SIGN SPACING 2.1.5.2.1 FHWA’s Manual of Uniform Traffic Control Devices (MUTCD) (45) Considering the principles of good freeway signing, the MUTCD specifics distances for advance freeway signing by interchange type. For major and intermediate interchanges, the MUTCD states that advance guide signs should be placed 0.5 and 1 mile from the exit gore. If spacing permits, a third advance guide sign should be placed 2 miles in advance of the exit gore. Signs in advance of an exit allow a driver to begin making necessary lane adjustments in advance of the exit. Multiple signs provide drivers multiple opportunities to see and comprehend the navigational information. Multiple signs with the same information also aid drivers in retaining the signing information. Examples of these sign placements specified in the 2009 MUTCD (Figure 2E-38 and 2E-35) are shown below in Exhibits 2-27 and 2-28.

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-38 Exhibit 2-27 Signing of a major or intermediate interchange, example 1, MUTCD (2009) (45)

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-39 Exhibit 2-28 Signing of a major or intermediate interchange, example 2, MUTCD (2009) (45) If the distance between interchanges is more than 1 mile but less than 2 miles, the first advance guide sign may be placed closer than 2 miles to the exit gore to avoid overlap with signing from the previous exit. For minor interchanges, only one advance guide sign, placed 0.5 to 1 mile from the exit gore, should be used. The MUTCD also lists an equivalent metric distance for each sign location specified above.

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-40 Major interchanges are interchanges with other expressways, other freeways, high-volume multilane highways, principal urban arterials, or certain major rural routes. Minor interchanges have a sum of exit volumes of less than 100 vehicles per day in the design year. All other interchanges are classified as intermediate. Freeway-to-freeway interchanges fall into the category of “major interchanges” and have no special sign spacing requirements, although the MUTCD states that signs for such interchanges shall be mounted overhead. The MUTCD accounts for interchanges located so close together that regular guide signs placed at the distances specified above would overlap with signing for other interchanges. If spacing between interchanges is less than 800 feet, interchange sequence signs should be used instead of advance guide signs. These signs list multiple exits on one sign and are generally used to supplement advance guide signs in places such as large urban areas where guide signs cannot be adequately spaced. Interchange sequence signs contain only two message units per exit, road name and distance, while an advance guide sign might contain four: exit number, route number, street name, and distance to exit. As a result, replacing three side-by-side advance guide signs with an interchange sequence sign would reduce the number message units in half. This would allow for more signs, which would then allow for closer spacing of interchanges. If used, the MUTCD states that interchange sequence signs shall be placed before the first advance guide sign for the first interchange in the sequence. An example of interchange sequence signs from the MUTCD (Figure 2E-30) at closely spaced exits is shown in Exhibit 2-29.

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-41 Exhibit 2-29 Interchange Sequence Signs, MUTCD (2009) (45) If spacing permits and undue repetition will not occur, the MUTCD states that a fixed sequence of signs should be displayed after an interchange. Often this will be in a rural area. A route sign should be displayed 500 feet beyond the end of the acceleration lane, followed at 1,000 feet by a speed limit sign, and 1,000 more feet by a distance sign. If spacing does not permit all of these signs or is in a rural area where traffic is primarily local, the MUTCD states some or all post-interchange signs should not be used.

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-42 2.1.5.2.2 State Sign Guidance A number of states publish their own version of the MUTCD or other sign guidance documents. These documents are not intended to override FHWA’s MUTCD but rather supplement the information. California’s MUTCD (48) and Texas’s Freeway Signing Handbook (2) provide additional information on freeway signing. At the present time, these state-level documents are still based off of the 2003 MUTCD and are in the process of being updated to reflect the 2009 MUTCD. California requires advance guide signs even when interchanges are closely spaced. The requirements state that when the distance between interchanges is less than 2 miles, the advance guide sign shall be placed at the first available location. California also seems to favor advance guide signs that are 0.5 and 1 mile before an exit and not 2 miles. Figures from FHWA’s MUTCD that show an advance guide sign 2 miles in advance of an exit are omitted or replaced in the California document by figures that only show advance guide signs 1 mile in advance of an exit. Texas’s Freeway Signing Handbook is meant to supplement the Texas MUTCD (50). The Texas MUTCD is similar to FHWA’s MUTCD with regard to freeway exit sign spacing. The Texas Freeway Signing Handbook reinforces the placement of advance guide signs prior to interchanges. One advance guide sign should be placed 0.5 to 1 mile prior to a minor interchange. Two, but preferably three, advance guide signs should be placed 0.5, 1, and 2 miles prior to an intermediate or major interchange. When spacing between interchanges is less than 800 feet, the Texas Freeway Signing Handbook states interchange sequence signs should be used. Texas uses these signs in urban areas with populations of 100,000 or more people. Post-interchange sign spacing guidelines do not differ from the FHWA MUTCD. 2.1.5.3 IMPACTS OF SIGNING ON RAMP AND INTERCHANGE SPACING Although the FHWA MUTCD refers to spacing in terms of the interchanges, it is the spacing of exit ramps that is most impacted by signing considerations. Drivers on a freeway make navigational decisions at exit ramps, not at crossroads or entrance ramps. An exit sign should be placed at the exit gore, and advance guide signs should be placed 0.5 and 1 mile prior to the gore. Individual guide signs should only contain information for one exit, and the ITE Freeway Handbook recommends a maximum of three guide signs in any one location (9). For example, a guide sign assembly could consist of three signs: one marking an exit gore, one placed half a mile in advance of a second exit, and one placed one mile in advance of a third exit. If exit ramps were spaced a half mile apart, this pattern could be repeated without presenting too many message units to drivers. Ramps spaced less than a half mile apart would require more than three guide signs in the same location or an “overlap” of multiple series of signs for different interchanges.

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-43 Although overlapping a series of signs would not necessarily violate the MUTCD’s minimum recommended spacing of 800 feet between advance guide signs, it would violate driver expectancy by simultaneously presenting two sign sequences. In situations such as this, the MUTCD recommends the use of interchange sequence signs. However, ideal conditions to create continuous half mile advance guide sign spacing often do not exist. The MUTCD states signs should not be placed where horizontal or vertical sight distance makes it difficult for drivers to see them, nor should overhead signs be placed within 800 feet of an overhead bridge. Also, it may be necessary to present other information to drivers, which could necessitate a break in the half mile spacing of guide sign assemblies. Thus, while it may be possible to adequately sign a series of exit ramps one-half mile apart with advance guide signs, it is unlikely that this could be sustained for more than several exits due to other factors that impact sign placement. 2.1.5.4 SIGNING SUMMARY Freeway drivers are dependent upon signs to locate exits and navigate to their destinations. Documents such as the MUTCD provide guidance for the number of signs that precede an exit and the distance before the exit where they are placed. At most interchanges, the MUTCD recommends advance guide signs be placed 0.5, 1, and optionally 2 miles prior to the an exit. Sign assemblies should not contain more message units, or information, than a driver is able to process. Exit ramps should be spaced far enough apart that advance guide signs can be positioned as recommended by the MUTCD and the number of message units will not exceed what drivers are able to process. 2.1.6 Vehicle Fleet Much of the published research related to ramp and interchange spacing was conducted from the mid 1950s to the mid 1970s, when many of America’s freeways were being built. Today’s vehicle fleet is quite different than the one on the road half a century ago. Certain characteristics of vehicles, such as acceleration of automobiles and power of trucks, are a factor in some aspects of ramp and interchange design. If automobile acceleration and truck power have changed over time, guidelines for on-ramp lengths and merge areas need to be revisited to reflect current vehicle characteristics. 2.1.6.1 AUTOMOBILES During the 1970s and early 1980s, cars gradually became lighter, more aerodynamic, and less powerful for fuel economy reasons compared to vehicles of the 1950s and 1960s (51). However, pickup trucks, vans, and sports utility vehicles have become increasingly popular since the early 1980s, and as a result, the average size of vehicles in the automobile fleet has increased in recent decades (52).

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-44 Literature reviewed for NCHRP 3-88 indicates that acceleration capabilities of vehicles have increased since the 1930s. A 2000 study by Long documented that automobiles in the early 1990s had higher maximum acceleration rates than those from the 1930s (53). The literature review results suggest the increase in maximum acceleration rates has not been constant over the decades. A 1983 study by Fancher identified a decrease in maximum acceleration between 1958 and the early 1980s but found that acceleration of the fleet of the early 1980s still exceeded acceleration of the fleet of the 1930s (51). In fact, low powered cars in the 1983 fleet were able to accelerate faster than implied by design curves for intersection sight distance calculations, which were developed decades earlier. A study by Harwood et al in 1999 identified an increase in maximum acceleration of passenger cars between the early 1980s and 1999 (54). Studies in the 1930s also identified a decrease in acceleration capability with increasing speed. This rate of change, unlike the acceleration rate itself, appeared to have remained nearly constant through the early 1990s, according to Long’s study (53). Although acceleration rate values for cars in the 1990 AASHTO Green Book were larger than previous editions, Long found that much of the acceleration data in the 1990 and 1994 AASHTO Green Books still appeared to be based on studies conducted in the late 1930s (53). Since Long’s study, AASHTO has published the 2001 and 2004 editions of the Green Book, which have not included any new acceleration data (7, 8). Therefore, even though several studies have identified an increase automobile acceleration rates since the 1930s, the 2004 Green Book may not fully reflect this change. However, the changes in maximum acceleration discussed above may not play a large role in determining driver behavior. Fancher suggested that drivers do not normally use the maximum acceleration capability of their vehicle and instead accelerate at a rate that seems comfortable to them. If a typical driver’s accelerate preferences have not changed since the 1930s, then long-standing guidelines may be more relevant today than changes in maximum acceleration would suggest (51). Although the studies in the 1930s measured maximum acceleration, “normal” acceleration rates were developed by assuming drivers typically use only limited portions of their vehicle’s capability. 2.1.6.2 HEAVY VEHICLES Heavy vehicle acceleration characteristics are typically described in terms of the weight-to-horsepower ratio. The literature search for NCHRP Report 505: Review of Truck Characteristics as Factors in Roadway Design (55) found that, in general, truck weight-to-horsepower ratios steadily decreased between 1949 and 1984 as engines became increasingly powerful. Increases in heavy vehicle weight have been offset by increases in engine power. There have been at least two studies of weight-to-horsepower ratio since 1984. In 1997, Harwood et al found that weight-to-horsepower ratios of high-performing

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-45 trucks had changed little since 1984 but that weight-to-horsepower ratios of low-performing trucks continued a decreasing trend (54). However, Harwood et al’s conclusions were based on data collected in the field at only one location in California. Field data collected for NCHRP Report 505 found that since 1984 weigh-to-horsepower ratios decreased substantially at western sites and stayed about the same at eastern sites. NCHRP Report 505 does not theorize on why this geographic difference was found, but it may be attributable to increased weight limits in western states, including the use of “triple” and “Rocky Mountain double” trailers. In 1993, Fancher suggested changes to the truck fleet can be thought of in terms of grade; heavier trucks may effectively make a three percent downgrade seem like four-percent, and more powerful engines may effectively make a three-percent upgrade seem like two-percent (51). 2.1.6.3 SUMMARY Changes to the vehicle fleet throughout the mid and late 20th Studies report weight-to-horsepower ratios of trucks decreased during the second half of the 20 century may be relevant to interchange and ramp spacing considerations. Current automobiles have been found to accelerate faster than those in the 1930s. However, the change has not been constant and a decrease in acceleration rates appears to have taken place in the 1960s and 70s. AASHTO Policy does not fully reflect these changes. Acceleration rates were increased in the 1990 Green Book relative to earlier AASHTO Policies, but studies since 1990 have still found that automobiles are able to accelerate faster than these rates. Since 1990, there have been no major changes to acceleration rates in subsequent editions of the AASHTO Green Book. Overall, automobiles on the road today accelerate faster than those on the road at the time many studies of ramp and interchange spacing were conducted. During this same time, trucks have become more powerful. th 2.1.7 Safety century. Trucks have become larger and heavier, and engine power has increased at an even faster rate. Increasingly powerful trucks and faster accelerating automobiles may influence on-ramps and merging area design considerations as current and future vehicle characteristics may be different than the characteristics used at the time of developing current ramp and interchange design guidelines. 2.1.7.1 DEFINITION OF RESEARCH PROBLEM Freeway interchanges, by their definition, coincide with increased lane changing, acceleration and deceleration on the freeway mainline. Traffic operations are adversely affected and decline with higher interchange and ramp densities (i.e., shorter spacing). The effects are captured at both the interchange level (e.g., free-flow speed decreases as interchange density

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-46 increases) and at the ramp level (e.g., speed decreases as weaving length decreases) by algorithms in the HCM. Analogous relationships between interchange and ramp density and safety are not as well established, but are of equal importance. For example, the Federal Highway Administration policy statement on Additional Interchanges to the Interstate System states as a condition of approval that “the proposed access point does not have a significant adverse impact on the safety and operation of the Interstate facility” (emphasis added) (56). Operational measures (e.g., speed variation, lane changing, conflicts) and the higher cognitive and decision making demands on drivers at these locations are often used as surrogates to deduce lower expected levels of safety in areas with increased interchange and ramp presence. However, accidents are random and complex events often attributable to driver behavioral patterns that are not yet understood. There are some existing hypotheses and accompanying research results that indicate safety may be higher in complex driving scenarios and lower in simple driving scenarios due to driver attention and awareness patterns. In addition, selecting operational surrogates as safety indicators may be highly specific to certain roadway features and accident types. Given these complexities, it is clear that “the concept of safety must be linked to accidents” (57). Safety will be defined for this study as the number of accidents, or accident consequences, by kind and severity, expected to occur on an entity during a specified time period (57). The expected number of accidents on an entity is representative of a long-term average. Its concept is important, as it captures the need for appropriate analytical techniques that account for the randomness of accident occurrence. These thoughts are consistent with seminal work in the field of highway safety and the concepts in the first edition of the Highway Safety Manual (HSM). 2.1.7.2 REVIEW OF PREVIOUS WORK Published literature relevant to NCHRP 3-88 objectives generally fell into two categories: • Studies that provide insight into the safety effects of interchange and ramp presence; and • Studies that provide insight into the safety effects of interchange and ramp spacing. Results from the first category, while not directly related to this project can be used by planners and designers to estimate the expected safety consequences of additional access points along a freeway segment or corridor. They also provide insights into general trends of accident frequencies and severities near interchanges and ramps. Results from the second category have direct relevance to this project.

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-47 Many of the studies reviewed had higher-level objectives than a detailed look at interchange- and ramp-related issues. These objectives included, for example, applying and testing new model estimation techniques or developing general freeway safety models. The information presented below is therefore more than a traditional literature review. It is a critical literature analysis. In cases where appropriate details were available (e.g., relevant model parameters, raw data tables and descriptive statistics), additional analyses that were more in line with the NCHRP 3-88 project objectives were conducted and presented in easy-to-interpret tables and figures. Footnotes and references to distinguish between the original work and the new analyses by the NCHRP 3-88 team are used throughout the text. 2.1.7.2.1 Safety Effects of Interchange and Ramp Presence Three studies allowed the project team to examine the aggregate safety effect of interchange and ramp presence by estimating expected accident frequencies and rates inside of interchange areas and comparing them to the same safety measures outside of interchange areas. Interchange area definitions vary by study, but generally include a bi-directional stretch of freeway mainline, adjacent auxiliary lanes and roadside bounded by the furthest upstream and downstream freeway-ramp terminals for a given interchange (see Exhibit 2-30).

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-48 Exhibit 2-30 Illustration of interchange area concept Torbic et al. incorporated two types of freeway segment models into the Interchange Safety Analysis Tool (ISAT): 1) freeway segments within interchange areas and 2) freeway segments outside of interchange areas (58). The models were estimated as part of SafetyAnalyst research efforts using data from California (1997-2001), Minnesota (1995-1999), Ohio (1997-1999) and Washington (1993-1996). Interchange areas were defined by limits extending from approximately 0.3 miles upstream of the first ramp gore of an interchange to approximately 0.3 miles downstream of the last ramp gore of the same interchange1 Accidents occurring on both the freeway mainline and the adjacent roadside as well as on speed-change lanes adjacent to freeway mainline lanes were included in the analysis. Accidents occurring on the ramp proper were excluded. Models for total accidents and fatal plus injury (F+I) accidents were estimated for the following freeway types: . • 4-lane rural; • 6-lane rural; • 4-lane urban; 1 The authors use gore and painted nose synonymously. Freeway mainline Cross street Interchange area

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-49 • 6-lane urban; and • 8-lane urban. All of the freeway segments models had the following functional form: SLADTaN b ××= )exp( where: N = expected number of accidents per year; ADT = average daily traffic (veh/day); SL = segment length (mi); and =ba, parameters estimated using available data. Torbic et al. estimated models with negative binomial regression, which was discussed in-terms of accident modeling by Miaou 2 %100×− outside outsidewithin N NN (58, 59). Negative binomial regression represents the current state of safety modeling practice and is central to work related to the forthcoming first edition of the HSM Estimation results are presented in Torbic et. al (58) The NCHRP 3-88 project team conducted a post-hoc analysis of the model results to determine if, and to what extent, the presence of an interchange decreases safety. Results are visually represented in Exhibit 2-31, which shows the percent difference in expected accidents between freeway segments within interchange areas and freeway segments outside of interchange areas using the ISAT models. The percent difference is computed as: where: withinN = the expected number of accidents per year on a freeway segment of length L within an interchange area; and outsideN = the expected number of accidents per year on a freeway segment of same length L outside of an interchange area. The graph shows results varying with freeway type. The percent difference is positive in almost all cases, indicating that segments within interchanges are 2 The negative binomial regression model is used to address overdispersion in accident data (i.e., the variance of the count is greater than the expected value). The magnitude of overdispersion is estimated along with the model coefficients.

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-50 less safe3 0% 20% 40% 60% 80% 100% 120% 140% 160% 180% 200% 0 50,000 100,000 150,000 200,000 250,000 Average Daily Traffic (veh/day) P e rc e n t d if fe re n c e i n e x p e c te d a c c id e n ts a 4-lane rural, total 4-lane rural, fatal + injury 6-lane rural, fatal + injury 4-lane urban, fatal + injury 6-lane urban, fatal + injury 8-lane urban, fatal + injury 6-lane rural, total 4-lane urban, total 6-lane urban, total 8-lane urban, total . In general, this difference increases with increasing traffic volume. Accident severity increases within interchange areas for 4-lane rural and 6- lane urban freeways (i.e., the percent increase in fatal plus injury accidents is greater than the percent increase in total accidents). Accident severity decreases within interchange areas for 6-lane rural and for most traffic volumes on 8-lane urban freeways. Accident severity is approximately equal within and outside of interchange areas for 4-lane urban freeways. The authors do point out model weaknesses for 6-lane rural and 4-lane urban freeways and recommend future research for these freeway types. Other omitted variables potentially contributing to the variability in results include ramp traffic (i.e., number of entering and exiting vehicles) and proximity to adjacent upstream or downstream interchanges. a %100×− outside outsidewithin N NN Exhibit 2-31 Comparison of safety for freeway segments within interchange areas to freeway segments outside of interchange areas using ISAT models Kiattikomol et al. conducted an analysis similar to Torbic et al. using data from North Carolina (NC) and Tennessee (TN) freeways (2000-2002) in 3 The percent difference for 4-lane urban freeways becomes negative for traffic values less than 77,000 vehicles per day, indicating that freeway segments within interchange areas are safer at lower volumes.

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-51 medium to large urban areas4 Models for fatal plus injury accidents, injury accidents and property-damage- only (PDO) accidents were estimated for the following freeway types: (60, 58). Interchange areas in North Carolina extended 1,500 feet on each side of the crossroad (i.e., all North Carolina freeway segments within interchange areas were 3,000 feet long). This distance was considered adequate to capture the entrance and exit areas for the sample of primarily diamond type interchanges. Lengths of Tennessee freeway segments within interchange areas varied; details regarding their definition were not provided. Details regarding the locations of accidents included in the analysis (e.g., mainline, speed-change lane, ramp proper) were also not included in the cited publication. • 4-lane urban; • greater than (>) 4-lane urban; Models for North Carolina freeway segments outside of interchange areas and all Tennessee freeway segments had the following functional form: 21 bb ADTSLaN ××= where: N = expected number of accidents per 3 years; SL = segment length (mi); ADT = average daily traffic (veh/day); and =2,1, bba parameters estimated using available data (Note that 1b is restricted to equal 1.0 in all models from Torbic et al. (58) Models for North Carolina freeway segments inside of interchange areas were similar, but excluded the segment length variable because all such segments were 3,000 feet long. Models were estimated with negative binomial regression and results are presented in Kiattikomol et al. (60). A similar type of post-hoc analysis as that reported above was conducted by the NCHRP 3-88 project team for comparison purposes. Results are shown in Exhibit 2-32. 4 The authors point out that the urban areas in these states are smaller, perhaps more rural in character, than larger cities of the Northeast, Midwest and Pacific Coast.

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-52 0% 200% 400% 600% 800% 1000% 1200% 1400% 0 20,000 40,000 60,000 80,000 100,000 120,000 140,000 Average Daily Traffic (veh/day) P e rc e n t d if fe re n c e i n e x p e c te d a c c id e n ts a , b NC, 4-lane, total NC, 4-lane, fatal + injury NC, >4-lane, total NC,> 4-lane, fatal + injury TN, 4-lane, fatal + injury TN, 4-lane, total TN, >4-lane, total TN,> 4-lane, fatal + injury a %100×− outside outsidewithin N NN b Exhibit 2-32 Comparison of safety for freeway segments within interchange areas to freeway segments outside of interchange areas using North Carolina and Tennessee models For this comparison, lengths of segments inside and outside of interchange areas were set equal to the mean length of the interchange segments (0.57 for both North Carolina models; 1.0 for Tennessee 4-lane; 0.88 for Tennessee >4-lane. The percent difference in accidents is positive in all cases (i.e. segments within interchange areas are less safe); a comparable finding to Torbic et al. (58). However, this difference decreases quickly with increasing traffic volumes for TN 4-lane freeways. It decreases moderately with increasing traffic volumes for NC >4-lane freeways. These patterns are opposite those for NC 4-lane freeways and for all freeway types modeled by Torbic et al. (58). Accident severity within interchange areas is slightly less than or equal to severity outside of interchange areas for TN 4-lane, TN >4-lane and NC >4-lane freeways. Accident severity is higher within interchange areas for NC 4-lane freeways; the difference becomes smaller with increasing traffic volumes. The effects of ramp traffic or proximity to adjacent interchanges were not captured in the models.

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-53 Exhibit 2-33 summarizes freeway accident rates by adjacent interchange unit and area type provided in a synthesis by Twomey et al. (61). If interchange area segments are defined as those segments adjacent to and in-between the speed change lanes5, then accident rates of 93 accidents per MVM and 174 accidents per MVM were observed on segments inside of rural and urban interchange areas, respectively. Accident rates for typical freeway segments outside of interchange areas in rural and urban settings are in the general range of 25-45 and 45-75 accidents per 100 MVM, respectively6 Interchange unit . This reflects approximately 110 to 300 percent more accidents on segments within interchange areas compared to segments outside of interchange areas. The accident rates in Exhibit 2-33 are much higher for urban areas, a finding the authors attribute to shorter speed change lanes at these locations. The difference may also be attributable to an increased number of lower severity, multiple vehicle collisions in urban areas (which tend to increase with ADT at a faster rate than single vehicle accidents). Rural 1 Urban 100 MVMT Number of accidents 2 Accident rate 100 MVMT3 Number of accidents 2 Accident rate Deceleration lane 3 2.51 348 137 5.83 1,089 187 Area between speed change lanes 6.52 554 85 11.87 1,982 167 Acceleration lane 3.68 280 76 8.40 1,461 174 Total 12.71 1,182 93 26.10 4,532 174 1 Data for other units were provided by Twomey et al. (60), but are not relevant to this discussion. 2 MVMT = million vehicle miles traveled 3 Exhibit 2-33 Accident rates by interchange unit and area type Expressed in terms of accidents per 100 MVMT Three additional studies looked at safety effects of ramp and interchange presence through binary, indicator variables (i.e. variable = 1 if a ramp is present on a defined segment; variable = 0 otherwise). Abdel-Aty et al. modeled accident frequencies using data from a 36-mile stretch of urban freeway in Florida (63). The freeway section was equipped with loop detectors, allowing the safety effects of real-time traffic parameters (e.g., average speed, peak 15-minute volume, temporal variability of speed and volume) to be tested. Four accident categories were modeled, with each category consisting of two sets of mutually exclusive accidents: 5 A definition similar to those provided by Torbic et al. (57) and Kiattikomol et al. (59) 6 Roughly approximated using data from Torbic et al. (57) and Bonneson et al. (61)

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-54 • accident type: single-vehicle or multiple vehicle • accidents by traffic conditions: peak or off-peak • accidents by pavement condition: dry or wet • accidents by light condition: daytime or dark • accident severity: PDO or injury Models within each category were estimated simultaneously using seemingly unrelated negative binomial regression, which accounted for error correlation between the accident types in each category. The presence of an entrance or exit ramp within a defined influence area increased the expected accident frequency for all eight accident types. Respective regression parameters were only reported for two categories; magnitudes were estimated by the NCHRP 3-88 project team and are summarized in Exhibit 2-34. The coefficient of speed variation was the only real-time traffic parameter appearing in any of the final model specifications. In addition, average daily traffic estimates were found to be better explanatory variables than more disaggregate measures of volume (e.g., peak 15-minute volume). These findings supported the use of aggregate measures of traffic volume in the safety research discussed in Chapter 3. Ramp variable Multiple vehicle accidents Single vehicle accidents 1 Daytime accidents Dark-hour accidents Presence of exit ramp +50% +61% +46% +60% Presence of entrance ramp +54% +26% +72% +50% 1 Exhibit 2-34 Safety effects of ramp presence using regression parameters estimated by Abdel-Aty et al, 2006. (63) Numbers represent the expected increase in accidents given the presence of an exit or entrance ramp. The effect of both an exit and entrance ramp would be multiplicative (). The method does not distinguish between the presence of only one ramp and multiple ramps of the same type at a given location (i.e. the expected accident frequency would be the same if there was one exit ramp or multiple exit ramps at a given location). Donnell and Mason investigated the frequencies and severities of two types of median related accidents on Pennsylvania Interstates: • cross median collisions: a vehicle traveling in one direction on a divided facility leaves its designated roadway to the left, traverses the entire width of the median, enters the opposing roadway and collides with a vehicle traveling in the opposite direction; and, • median barrier collision: a vehicle traveling in one direction on a divided facility leaves its designated roadway to the left and collides with a median barrier (64-66). Frequencies were modeled using negative binomial regression. Severities were modeled with logistic regression. Although not conclusive, results indicated that the expected frequency of cross median collisions increased in areas up to 800 feet downstream of an entrance ramp on rural Pennsylvania

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-55 Interstates (not including the Pennsylvania Turnpike) (64)7 Kraus et al. investigated two types of freeway accidents in southern California resulting in at least one severe occupant injury or fatality: . The expected frequency of median barrier collisions decreased on segments with entrance ramps on the Pennsylvania Turnpike, but increased on segments with an entrance ramp on other Pennsylvania Interstates (66). Severities of median barrier collisions decreased on segments with entrance ramps; similar effects on cross median collision severity were not reported (65). • in-lane accidents: rear-end impact of a stopped, stalled or slowed vehicle in a normal traffic lane by a second vehicle traveling in the same traffic lane; and, • off-road accidents: vehicle leaves the traveled way and impacts a roadside object, overturns off the roadway or re-enters the traveled way and strike another vehicle (68). Two years (1984-1985) of accidents were merged with 69 homogenous freeway segments. Each freeway segment appeared 6 times in the dataset, once for each combination of time of day (12a.m.-6a.m.; 6a.m.-6p.m.; 6p.m.- 12a.m.) and day of week (weekday; weekend). Poisson regression models were estimated for in-lane accidents, off-road accidents to the left and off- road accidents to the right. Estimation results indicated that expected frequencies of in-lane accidents and off-road accidents to the right resulting in severe injuries or fatalities were 16% lower on segments with frequent interchanges. A segment was defined as having frequent interchanges if 20% of its total length was occupied by interchange related features. It is unclear whether this finding is attributable to lower accident frequencies, a shift in the proportion of severe accidents or a combination of both. 2.1.7.2.2 Safety Effects of Interchange and Ramp Spacing The aforementioned synthesis by Twomey et al. (61) is sometimes cited in discussions on safety effects of ramp and interchange spacing [see (69) or (58) for example]. The cited work is not a direct analysis of spacing, but of freeway accident rates by interchange ramp proximity on Interstates. Interchanges had an exit ramp upstream of the cross street and an entrance ramp downstream of the cross street for each direction of travel. Ramp proximity was defined as either the upstream distance from the exit ramp nose on the exit side of the interchange or the downstream distance from the entrance ramp on the entrance side of the interchange. Exhibit 2-35 displays a new, graphical analysis of the results conducted by the NCHRP 3-88 project team. The analysis indicates that accident rates 7 A follow-up study showed minimal association between entrance ramp presence and expected frequency of cross median collisions (66). The definition of entrance ramp presence was significantly changed to include both segments with the entrance ramp and upstream segments within 1500 of the entrance ramp.

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-56 increase with ramp proximity (i.e., shorter distances from ramps) at a higher rate in urban areas than in rural areas. The finding is likely a result of increased vehicle interaction due to higher traffic volumes on the mainline and ramps. There are also minimal differences in accident rates between the exit and entrance sides of interchanges in urban areas. The close proximity of adjacent upstream and downstream interchanges in urban areas is one possible explanation (i.e. the downstream area of one interchange can also be considered the upstream area of the next interchange). Accident rates in rural areas decrease as ramp proximity decreases at a higher rate downstream of the interchange compared to upstream. 20 40 60 80 100 120 140 0 1 2 3 4 5 6 7 Distance of segment centroid to exit or entrance ramp nose (miles) A cc id en t r at e (a cc id en ts p er 1 00 M V M ) upstream of urban exit on exit side of interchange upstream of rural exit on exit side of interchange downstream of urban entrance on entrance side of interchange downsream of rural entrance on entrance side of interchange urban - upstream of exit on exit side or downstream of entrance on entrance side rural - upstream of exit on exit side rural - downstream of entrance on entrance side Exhibit 2-35 Analysis of accident rates by ramp proximity reported by Twomey et al., 1991 (61) The remainder of published findings summarized below provided insights into the safety effects of ramp and interchange spacing in one of two ways: • The boundaries of freeway segments were defined by like points on consecutive interchanges or ramps and those segments were analyzed (i.e., the length of the freeway segment is also the interchange or ramp spacing); or, • The safety effects of a ramp or interchange count or density on a freeway segment of length L were estimated through a multivariate regression model (the inverse of these types of variables represents an average interchange or ramp spacing). The findings are presented in this same respective order.

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-57 Cirillo examined the relationship between accident rates and weaving area lengths using Interstate data from 20 states (70). Approximately 700 urban weaving segments were included in the dataset. New analyses of the accident rates conducted by the NCHRP 3-88 team are summarized in Exhibit 2-36. Trends show that, for a given level of traffic volume, accident rates tend to increase as weaving area lengths decrease. Results also show that, for a given weaving area length, accident rates decrease as volume decreases. Cirillo aggregated accident rates by five levels of one way mainline ADT in the original work (ADT < 10,000; 10,000 ≥ ADT < 20,000; 20,000 ≥ ADT < 30,000; 30,000 ≥ ADT < 40,000; 40,000 ≥ ADT), but reported a limited sample size in the lowest volume area category. The NCHRP 3-88 project team found more consistent general trends when the three lowest volume categories were combined into one (ADT < 30,000). This change is reflected in Exhibit 2-36. Results from a later study showed opposite trends; accidents rates decreased as weaving length decreased (71). The sample size was limited to 21 locations. The locations were not selected randomly, but were included due to poor accident histories (a possible selection bias problem). Traffic volumes were not considered in the analysis other than their use in accident rate calculations. Non-linear trends between accidents and volumes are well established. Segregating accident rates by level of traffic volume is desirable if accident rates are the safety measure of choice. The results reported by Cirillo, while older, are likely more reliable. 0 100 200 300 400 500 600 700 400 450 500 550 600 650 700 750 800 Length of weaving area (feet) A cc id en t r at e (a cc id en ts p er 1 00 M VM ) One-way ADT < 30,000 30,000 ≥ One-way ADT < 40,000 One-way ADT ≥ 40,000 One-way ADT ≥ 40,000 One-way ADT < 30,000 30,000 ≥ One-way ADT < 40,000

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-58 Exhibit 2-36 Analysis of accident rates by weaving areas length reported by Cirillo, 1970 (70) Bared et al. modeled the safety effects of interchange spacing using California freeway data (1998-2002) (72). Interchange spacing was defined as the smallest distance between gore points of ramps from consecutive interchanges8 ( ) 321 bbb RampADTSLADTaN ∑×××= . Negative binomial regression models for total accidents and fatal plus injury accidents were estimated using data from 58.5 miles of California Interstates; number of lanes varied from 6 to 14. Reported models had the following functional form: where: N = expected number of accidents per year; ADT = average daily traffic on the freeway mainline (veh/day); SL = segment length, defined as interchange spacing (mi); ADTRamp∑ = the sum of ADT for the two entrance ramps and two exit ramps associated with a defined interchange spacing segment (veh/day); and =3,2,1, bbba parameters estimated using available data Model results are summarized and illustrated in Exhibit 2-37. The model parameters generally make intuitive sense. However, a closer look at the segment length variable reveals potential challenges associated with their study objective: determining the safety effect of interchange spacing. The traffic and segment length components of an accident frequency model represent measures of exposure; respective regression parameters generally have a value around one. The parameter for ADT may be slightly greater than or less than one, depending on the crash type of interest. The parameter for segment length is sometimes constrained to equal one. In the model reported by Bared et al., the parameter associated with segment length represented the net effect of several potential confounding factors (72). Exposure was the most predominant, resulting in an overall positive effect of segment length. However, the interchange spacing effect is confounded with the exposure effect because every segment in the database is defined with an entrance gore on one side and an exit gore on the other side. Shorter segment lengths represent reduced exposure, but increased ramp interaction; two factors expected to have opposite effects on accident frequency. The segment length, as defined by Bared et al., may also be correlated with additional interchange related features that influence safety (72). For 8 The authors define gore point and ramp nose synonymously.

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-59 example, shorter segment lengths are likely associated with an increased presence of auxiliary lanes between the entrance and exit ramps of two consecutive crossroads; a feature not captured in the data. 0 20 40 60 80 100 120 140 160 180 200 0 0.5 1 1.5 2 2.5 3 3.5 Segment Length defined as Interchange Spacing (miles) Ex pe ct ed A cc id en t F re qu en cy (P er Y ea r) Total accidents = 6.18 x 10-6ADT1.122SL0.6394ΣRampADT0.2213 Total accidents, low-volumea F+I accidents = 5.44 x 10-5ADT0.8618SL0.5918ΣRampADT0.2088 F+I accidents, low-volumea Total accidents, medium-volumeb F+I accidents, medium-volumeb Total accidents, high-volumec F+I accidents, high-volumec a Low volume: ADT = 66,600 veh/day; ΣRampADT = 6,900 veh/day b Medium volume: ADT = 188,000 veh/day; ΣRampADT = 34,100 veh/day c Exhibit 2-37 Summary of freeway models from Bared et al, 2006 (72). Low volume: ADT = 274,000 veh/day; ΣRampADT = 120,700 veh/day One possible solution was explored by Bared et al. and is recreated in Exhibit 2-37. The expected number of accidents predicted from the regression models in Exhibit 2-37 are normalized (i.e., divided by) the segment length. The resulting rate, with units of accidents per mile per year, follows an intuitive trend: the expected number of accidents per unit length increases as interchange spacing decreases. The procedure assumes the segment length parameter associated with exposure is equal to one and that the difference between the originally estimated segment length parameter and one is attributable to the interchange spacing effect. This concept is illustrated by: ( ) ( ) 30.121 bbb RampADTSADTa SL N ∑×××= − where: SL N = expected number of accidents per mile per year;

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-60 S = interchange spacing (miles); and =∑ 3,2,1,,, bbbaRampADTADT same as previously defined. The slope of the line representing the expected accident frequency versus interchange spacing relationship approaches zero as interchange spacing increases, indicating minimal safety influence from the ramps at the segment termini (i.e., from a safety perspective, the segment operates as a normal freeway segment without deleterious interchange or ramp effects). The interchange spacing at which this occurs becomes longer as mainline and ramp volumes increase. The normalizing technique is promising if one can be fairly certain that effects other than exposure and interchange spacing are not fully or partially captured in the segment length definition. 0 20 40 60 80 100 120 140 160 180 200 0 0.5 1 1.5 2 2.5 3 3.5 Interchange Spacing (miles) Ex pe ct ed A cc id en t F re qu en cy (P er M ile P er Y ea r) Total accidents = 6.18 x 10-6ADT1.122S-0.3606ΣRampADT0.2213 F+I accidents = 5.44 x 10-5ADT0.8618S-0.4082ΣRampADT0.2088 Total accidents, low-volumea F+I accidents, low-volumea Total accidents, medium-volumeb F+I accidents, medium-volumeb Total accidents, high-volumec F+I accidents, high-volumec a Low volume: ADT = 66,600 veh/day; ΣRampADT = 6,900 veh/day b Medium volume: ADT = 188,000 veh/day; ΣRampADT = 34,100 veh/day c Exhibit 2-38 Summary of freeway models from Bared et al. with results normalized for segment length, 2006 (72) Low volume: ADT = 274,000 veh/day; ΣRampADT = 120,700 veh/day Pilko et al. (68) conducted a follow-up effort to the study by Bared et al. (72) with some notable changes: • the size of the California dataset was increased to include 95 spacing observations representing 134 freeway miles (compared to 53 observations representing 58.5 miles);

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-61 • a Washington freeway dataset consisting of 100 spacing observations representing 144 freeway miles was added and used for model estimation and validation; • mainline traffic was specified as vehicles per lane per day; • ramp volumes were expressed at the ratio of ramp ADT to mainline ADT for the California models; • cross section variables representing median width, median type and HOV lane presence were included in some models; and • the definition for interchange spacing was changed to represent the distance between crossroads of consecutive interchanges. Model estimation results are summarized in Exhibit 2-39. The graphical display s in Exhibi t 2-37 repres ent general trends that are also seen when the model s in Exhibi t 2-39 are plotted . Discus sion and analysi s associa ted with Exhibit 2-38 are also applicable. Therefore, the figures and analysis are not repeated here. Data and Specifi cation Accident Types Expected accident frequency per year CA only TOTAL ( )MTMWHOVRRatioSL LN ADT ∗+∗−∗+∗     ×= − 27.001.037.050.1exp1097.4 57.0 39.1 5 F+I ( )MTMWHOVRRatioSL LN ADT ∗+∗−∗+∗     ×= − 35.001.034.042.1exp1081.1 57.0 37.1 5 CA for WA validati on TOTAL ( )MWRampADTSL LN ADT ∗     ×= ∑− 0072.0exp1061.3 34.052.0 11.1 5 F+I ( )MWRampADTSL LN ADT ∗     ×= ∑− 0051.0exp1064.1 35.051.0 07.1 5 Joint CA and WA F+I ( )MWRampADTSL LN ADT ∗     ×= ∑− 0032.0exp1063.1 26.062.0 37.1 6 ADT = average daily traffic on freeway mainline (veh/day); LN = number of lanes at the segment midpoint (includes through lanes, HOV lanes and auxiliary lanes greater than 0.2 miles long; SL = segment length, defined as interchange spacing (mi); RRatio = the sum of ADT for the two entrance ramps and two exit ramps associated with a defined interchange spacing segment divided by average daily traffic on the freeway mainline HOV = indicator for presence of an HOV lane (1 = present); MW = median width (feet); MT = indicator for median type (1 = unpaved, 0 = paved); ∑ RampADT = the sum of ADT for the two entrance ramps and two exit ramps associated with a defined interchange spacing segment

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-62 Exhibit 2-39 Summary of reported models in Pilko et al., 2007 (69) Hadi et al. estimated the safety effects, defined by frequencies of total accidents, injury accidents, and fatal accidents, of design elements for various types of urban and rural facilities in Florida (73). Four years (1988-1991) of accident data were aggregated and combined with roadway and traffic data from the Florida Department of Transportation Roadway Characteristics Inventory. The freeway models had the following general form: )exp( 321 BXIbSLbADTbaN +++×= where: N = expected number of accidents per four years; ADT = average daily traffic on the freeway mainline (veh/day); SL = segment length (mi); I = number of interchanges on the segment; a = constant term estimated using available data; 321 ,, bbb = parameters associated with ADT , SL and I respectively estimated using available data; X = k x 1 vector of other design elements influencing N on freeways; and B = 1 x k vector of parameters associated with variables in X estimated using available data. The parameter associated with the number of interchanges on a segment was positive in all freeway models in which the variable was included in the final specification, indicating that more interchanges on a freeway segment (i.e., a smaller average interchange spacing) is associated with higher expected accident frequencies of all severities. The magnitudes of the unit increases, expressed in terms of percent, are summarized in Exhibit 2-40. The interchange variable is not included in models for injury crashes on 6-lane urban freeways or fatal crashes on all freeway types. It is unclear whether this represents a shift to less severe crashes as the number of interchanges increases, inflation of standard errors due to a low number of severe crashes (i.e., excess number of zeros) or a combination of both.

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-63 Freeway type Expected unit % increase in total accidents Expected unit % increase in injury accidents 1 Expected unit % increase in fatal accidents 4- and 6-lane rural +24% +24% n/a 4-lane urban 2 +32% +27% n/a 6-lane urban 2 +20% n/a n/a2 2 1 The expected increase in accidents given the addition of one interchange to a freeway segment. The effect is multiplicative (e.g., the expected increase in accidents given the addition of two interchanges on a 4-lane rural freeway would be 1.24 * 1.24 ≈ 54%). 2 Exhibit 2-40 Summary of interchange effect estimated by Hadi et al. 1995 (73) sample sizes were not sufficient to estimate the interchange effect for these combinations of accident and freeway types Milton et al. modeled accident severities using a mixed logit approach (74). Accident data (1990-1994), weather information and roadway characteristics were merged for multilane divided highways in Washington State. Estimation results showed that the probability of a disabling injury or fatality resulting from an accident was lower for segments with increased interchange density, expressed as the number of interchanges per mile. The finding was attributed lower speeds and smaller collision angles in the vicinity of multiple interchanges. Anastasopoulos et al. estimated a tobit model of accident rates on Indiana Interstates (75). Accidents rates for 337 Interstate segments were computed using five years (1995-1999) of aggregated accident data, i.e.: [ ] 000,000,100365_ 5 1 , 5 1 , ∑ ∑ = = ×× = Year iiYear Year iYear i LADT Accidents rateAcc where: irateAcc _ = the accident rate for segment i expressed in units of accidents per 100 million vehicle miles traveled; ∑ = 5 1 ,Year iYear Accidents = the number of accidents observed over a 5-year period (1995-1999) on roadway segment i; and ∑ = ×× 5 1 , 365 Year iiYear LADT = total vehicle miles traveled over a 5-year period (1995-1999) on roadway segment i. Estimation results showed that a unit increase in the number of interchange ramps per lane mile of Interstate increased the expected accident rate by 22.36 accidents per million miles and increased the probability of having an accident rate above zero by 16.87%.

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-64 2.1.7.3 SAFETY SUMMARY A total of twelve studies, and sixteen subsequent papers and reports provided insights relevant to NCHRP 3-88 project objectives. The literature generally fell into two categories: • Studies that provide insight into the safety effects of interchange and ramp presence; and • Studies that provide insight into the safety effects of interchange and ramp spacing. Results from the first category, while not directly related to this project, can be used by planners and designers to estimate the expected safety consequences of additional access points along a freeway segment or corridor. They also provided insights into general trends of accident frequencies and severities near interchanges and ramps, which helped guide the safety research conducted for this project (see Chapter 3). Results from the second category have direct relevance to this project as they capture not only safety effects of interchange and ramp presence, but the safety effects associated with interchange and ramp spacing and density measures. 2.1.7.3.1 Safety Effects of Interchange and Ramp Presence The presence of an interchange, or interchange ramp, was associated with an expected increase in accident frequency. Results from interchange area modeling efforts were similar in direction, but sporadic in magnitude; expected accident frequencies on freeway segments within interchange areas were anywhere from zero to 1200% higher than for segments outside of interchange areas. Common magnitudes ranged from 100 to 300% (Torbic et al., 2007; Kiattikomol et al., 2008; Twomey et al., 1991). Three studies looked at safety effects of ramp and interchange presence through binary, indicator variables (i.e. variable = 1 if a ramp is present on a defined segment; variable = 0 otherwise) and found similar trends across a number of accident types (Abdel-Aty et al., 2006; Donnell and Mason, 2002, 2006a; Kraus et al., 1993). Magnitudes, in terms of percent difference, were estimated by the NCHRP 3-88 team using the reported regression parameters when the required information to perform the calculation was available. 2.1.7.3.2 Safety Effects of Interchange and Ramp Spacing Three studies took a direct look at the relationship between interchange or ramp spacing and safety (Cirillo, 1970; Bared et al., 2006; Pilko et al., 2007). In all three, the boundaries of a sample of freeway segments were defined by like points on consecutive interchanges or ramps and those segments were analyzed (i.e., the length of the freeway segment is also the interchange or ramp spacing). Cirillo (1970) reported accident rates by volume level and length, which were subsequently re-analyzed by the NCHRP 3-88 project team. Findings were illustrated in Exhibit 2-36. Bared et al. (2006) and Pilko et al. (2007) used negative binomial regression to model expected accident frequency as a function of freeway and traffic characteristics, including

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-65 segment length (i.e., interchange spacing). The studies found that accident rates (in terms of accidents per 100 MVM or expected accidents per mile per year) increased as spacing decreased. Results from all three studies indicated that incremental improvements in safety with increased ramp or interchange spacing diminish as the spacing becomes longer. From a safety perspective, this finding indicates that, for a long spacing, the segment operates as a normal freeway segment without deleterious interchange or ramp effects (a concept which has a traffic operations analog in the HCM). The findings reported by Cirillo (70) are relevant, but 40 years old. The studies by Bared et al. (72) and Pilko et al. (69) use more modern analytical techniques, but with noted assumptions and limited applicability. The segments analyzed included only those where an entrance ramp from one interchange was followed by an exit ramp from an adjacent, downstream interchange. Measures of exposure and spacing were confounded in their model specifications and an assumption of linearity between segment length and accidents was required for practical interpretation of their findings. The normalizing technique is promising if one can be fairly certain that effects other than exposure and interchange spacing are not fully or partially captured in the segment length definition. This will require considering of a number of other variables potentially correlated with interchange and ramp spacing (e.g., presence of an auxiliary lane). Overall, the studies by Bared et al. (72) and Pilko et al. (69) are a first step on which to build an investigation of interchange and ramp spacing. Several studies reported safety effects of a ramp or interchange count or density on a freeway segment of length L through a multivariate regression model (the inverse of these types of variables represents an average interchange or ramp spacing) (73, 74, 75). This technique is analogous to the interchange density, speed adjustment factor in the freeway segment analysis methodology of the HCM and is relevant to more corridor-level safety analyses of interchange spacing. Estimation results showed that expected accident frequencies and accident rates increased as the interchange and ramp count or density increased (i.e. as the average spacing decreased). Hadi et al., Anastasopoulos et al., and Milton et al. reported lower accident severities for segments with increased interchange density, expressed as the number of interchanges per mile (73, 74, 75). The finding was attributed lower speeds and smaller collision angles in the vicinity of multiple interchanges and was consistent with loosely supported trends from two other studies (65, 68). Overall, little is known regarding the effects of interchange and ramp spacing on accident severity; a gap that was addressed by the research conducted as part of this project (see Chapter 3). 2.2 FOCUS GROUP SUMMARY In addition to the literature review, the project team facilitated a focus group “consisting of planners, designers, and operators of freeways and interchanges and other interested parties to assist in identifying concerns or

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-66 needs in the current practice of ramp and interchange spacing.” As outlined in this task, the project team submitted a memorandum to the NCHRP panel for their approval and suggestions for the focus group meeting approach, participants and discussion topics. The team conducted the focus group via telephone, which was cost effective way to engage a wide range of individuals. The invited focus group participants were developed based on a review of researchers and authors from existing relevant research on this topic area and the team’s familiarity with practicing planners, designers and operators of freeways and interchanges. Recognizing the diverse range of individuals that this topic area will potentially influence, the team included members from some of the primary agencies or organizations involved in this area of research such as AASHTO, MUTCD, and FHWA. Other invited participants included staff from state highway agencies and other research firms. Focus group participants are noted in the acknowledgements section at the beginning of this document. Six people from the focus group invitation list were able to participate in the conference call. There were four state agency staff, one federal highway administration representative, and a representative from the private consulting field. The team conducted follow-up calls to other specific individuals who expressed interest in participating, but were unable to due to scheduling conflicts. For those that were unable to participate, a list of discussion questions and topics was sent via email and input was requested. The focus of the group discussion provided the opportunity to hear a diverse range of perspectives in applying current ramp and interchange spacing criteria. In addition, these perspectives allowed the team to further understand some of the challenges and opportunities of applying the current criteria in contemporary and future practice. The following bullets summarize the discussion topics and focus group input. 2.2.1 General Challenges and Needs • One of the challenges that state agencies face is the multiple requests for interchange access in urban areas. Many developments would like to have ramp access for their property, which requires ramps to be closely spaced. • During ramp design, the ability to meet the standards sometimes results in longer ramps that then become closely spaced with the upstream or downstream existing ramps. • Some state agencies have designed braided ramps due to close cross street spacing. • Many agencies are working to eliminate all of the left-hand exits to better meet driver expectations and recommended ramp design characteristics.

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-67 • Spacing considerations should consider whether the facility is a system or service interchange. The spacing guidance may need to be separated for each type. • State agencies would appreciate more guidance on when to use frontage roads, collector-distributor roads, or ramp braiding. • Oregon Department of Transportation uses operational experience as the primary factor for making decisions about interchange access. ODOT often tries to find an existing similar situation for comparison purposes and to determine if that type of situation should be implemented again in another location. • State agencies need additional information about how the geometry of the mainline and ramps can affect the spacing. • For the Guidelines, it would be helpful if information about the various tradeoffs between ramp and interchange spacing is provided. 2.2.2 Other Considerations – Signing, Human Factors, and Vehicle Fleet • When considering human factors, the driver age and purpose should be considered. Older drivers and tourists require additional time to make decisions. • Signing is an important factor, but it often gets left out of the decision making process or is one of the last factors to be considered. In some cases, this results in limited spacing to provide adequate signing. • Signing plans need to be designed for the unfamiliar driver. Unfamiliar drivers can impact the operations of many other vehicles if decisions are not clear and appropriate. • State agencies need additional guidance on how to sign interchanges and how to integrate the signing in with the overall geometric design. • Signing guidance should reference the MUTCD, but not so much that it becomes out-of-date as the MUTCD gets revised. • Lane drops require different signing needs, due to the need for additional signs in advance of the exit. • When considering vehicle fleet, the current one-mile spacing used by many agencies is often difficult to achieve when there are a large percentage of trucks. Therefore, designers need more guidance on the spacing impacts associated with trucks, particularly if there are steep grades.

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-68 2.2.3 Analysis Techniques • State agencies typically use the Highway Capacity Software to conduct preliminary analyses and then CORSIM or another type of detailed analysis software is often needed for complex designs. • Using 20 years forecast volumes in the analysis often leads to failing levels-of-service. Therefore, state agencies often look at the “worst” level-of-service or a comparative analysis. The feedback and input from the focus group participants provided the team with additional insights regarding the needs of practitioners and state agencies. Many of the discussions during the focus group were similar to the input received from the panel, such as the challenge that state agencies are faced with when receiving multiple requests for interchange access in urban areas due to developments desire to have ramp access for their properties. 2.3 PANEL INPUT SUMMARY The team solicited input from the NCHRP panel to collaborate and generate ideas for the work plan and guidelines development. Each panel member has a diverse range of experience with operations, design, and safety of ramp and interchanges. In addition, the panel has participated in group discussions to generate ideas and outline the objectives of this specific research problem statement. To conduct the most effective research work plan and develop the most comprehensive guideline document for ramp and interchange spacing, the team took advantage of the individual knowledge and the collective joint conversations that the panel had to discuss the execution of this project and final guidelines document. The team requested panel input on the applying current ramp and interchange spacing criteria, as well as, the challenges and opportunities of applying the current criteria in contemporary and future practice. The panel input for each of the discussion questions is shown below. What are the historical challenges in applying current ramp and interchange spacing criteria? One of the most significant challenges in applying current ramp and interchange spacing criteria is the lack of information on the critical considerations for determining the appropriate interchange and ramp spacing. The primary resource that designers rely on is the guidance found in the AASHTO “Policy on Geometric Design of Highways and Streets” (Green Book). Exhibit 10-68 (Recommended Minimum Ramp Terminal Spacing) is commonly referenced, but there is almost no supplemental guidance for using the minimum recommended values. Therefore, designers often use the stated values in this table, rather than considering the specifics of their project needs.

NCHRP 3-88 Interim Report Guidelines for Ramp and Interchange Spacing Chapter 2: Information Gathering 2-69 Additional guidance on the following considerations would be beneficial for designers to make more informed decisions about each specific project characteristic. • Operational impacts of closely spaced ramps with high volumes • Operational impacts of high truck volumes • Variations in acceleration and deceleration lengths • Two lane entrance and exit ramps • Ability to provide appropriate guide signing for all types of users What are the challenges and opportunities of applying the current criteria in contemporary and future practice? A significant challenge is that many agencies are under tremendous pressure to consider new interchanges on high volume urban and suburban freeways to accommodate growth. These proposed new interchanges may present challenges with regard to complying with the existing interchange and ramp terminal spacing criteria. An opportunity is the ability to utilize new traffic analysis tools, such as micro-simulation models, to better understand the impacts of ramp spacing on traffic operations under a variety of conditions. The notes within Exhibit 10-68 of the AASHTO Green Book state that the values provided in the exhibit should be checked in accordance with procedure outlined in the Highway Capacity Manual. An opportunity for future practice is to expand this advisory to include other available traffic analysis tools, such as micro- simulation models. When is the guidance needed? What are the decisions being evaluated when interchange or ramp spacing is a key consideration? The interchange and ramp terminal spacing guidance is frequently referenced for situations involving a proposed new interchange on an existing freeway and there are existing interchanges closely adjacent to the proposed interchange. The guidance is used to help assess the safety and operational impacts of the proposed new freeway access. Design decisions such as whether to use braided ramps and/or collector-distributor systems are often based upon the ability to meet the “recommended minimum” ramp spacing volumes. These decisions can greatly effect project cost and therefore this guidance can have substantial impacts to transportation agencies. What are the knowledge gaps when making a decision that involves interchange and ramp spacing? There are numerous knowledge gaps, including:

Interim Report NCHRP 3-88 Chapter 2: Information Gathering Guidelines for Ramp and Interchange Spacing 2-70 • Effect of variations in ramp and mainline volumes on appropriate minimum ramp terminal spacing • Effect of truck volume variations • Knowledge of a driver’s expectations regarding separation distances (ramp terminal spacing) • Knowledge of what constitutes “adequate signing” for situations involving close ramp spacing • Effect of variations in facility operating speed and suggested minimum ramp terminal spacing • Spacing impacts due to decision sight distance What might be the quantitative information a decision maker or designer would like to have access to before making a decision? Design decisions on ramp terminal spacing should be based on an understanding of the influencing factors and applying a “risk managed” approach. Although there is commonly a strong desire among practitioners for having published numerical values for “minimums”, it is critical that any such published values include supporting information to help guide a designer through a process of considering the risks for using the stated minimum values. How do human factors, vehicle fleet, and signing considerations affect the decision making? Human factors, vehicle fleet and signing considerations are critical in making decisions regarding appropriate ramp and interchange spacing and there is a significant knowledge gap in these areas. For example, in a typical urban freeway pattern of interchanges spaced approximately every mile, how problematic is it (from a human factors perspective) to then have an instance of two exit ramps spaced 1000 feet apart? Do drivers become “trained” by the preceding pattern of exit spacing and then have difficulties with an instance of a significantly shorter spacing? Does “adequate signing” overcome this violation in driver expectation or is the use of “typical” signing less effective for the unexpected condition? Should some type of “atypical” signing be used for the “atypical” condition?