3.1 Uninterrupted Pedestrian Facilities

SOURCE: Khristy, 1982.

Off-Street Paths
Exclusive Pedestrian Trails. Virkler and Balasubramanian (1997), in their discussion of flow characteristics on shared user trails, imply that the current HCM's LOS walkway guidelines apply for exclusive pedestrian trails.

Shared Pedestrian-Bicycle Paths. Virkler and Balasubramanian (1997) describe a 1995 study by Hein Botma of shared pedestrian-bicycle facilities in The Netherlands. This study develops LOS guidelines for both pedestrians and bicyclists on the basis of the frequency of passing (same direction) and meeting (opposite direction) other users on the trail. Botma characterizes these two occurrences as "events," with an overtaking equal to one event and a meeting equivalent to one-half of an event. Under this framework, LOS F refers to "very bad quality of traffic operation," not congestion (Botma, 1995). More specifically, it refers to a situation where an average user experiences "hindrance" more than 1.0 times in a 1-km trail segment. Virkler and Balasubramanian note that, for one-way paths, pedestrians seldom overtake other pedestrians, and thus the LOS afforded a pedestrian on a shared path depends on the frequency with which an average pedestrian would be overtaken by bicyclists. In Botma's discussion of his own work (1995), he poses the question of whether it is justified to neglect hindrance due to pedestrian interactions. As the authors of this report have observed moderate pedestrian-pedestrian hindrances on various mixed-use trails, it is likely that Botma's assumption of negligible pedestrian interactions is not entirely correct.

Botma's expression describing the total number of overtakings of pedestrians by bicyclists, N_f/s, is:

N_f/s = X T Q_f Q_s (1/U_s - 1/U_f)

where:
X = length of site, m;
T = time period considered, s;
Q_f = flow of faster group in subject direction, bicyclists/s;
Q_s = flow of slower group in subject direction, pedestrians/s;
U_f = mean speed of faster group, m/s (for bicyclists); and
U_s = mean speed of slower group, m/s (for pedestrians).

Using an average pedestrian speed of 1.25 m/s and an average bicyclist speed of 5 m/s, Botma developed a LOS table for pedestrians on one-way, two-lane shared-use paths. Table 11, which converts "frequency" of events into period between events to eliminate fractions, provides Botma's LOS thresholds.

FIGURE 7

Noncompliant pedestrian behavior is common at this Chicago, Illinois intersection due to low conflicting vehicle volumes.

FIGURE 8

Noncompliant behavior is not limited to pedestrians at the same Chicago, Illinois intersection.

The Literature Review for Chapter 13, Pedestrians, of the Highway Capacity Manual included a study by Rouphail (1984) noting the preference of pedestrians for midblock crossings. However, when a pedestrian crossing is displaced from the intersection, the increase in travel path for many users walking along the cross-street may tend to breed signal noncompliance in some situations, as people tend to use the intersection crosswalks regardless of signal indication (Pretty et al., 1994).

Hunt and Griffiths (1991) note that pedestrians experience very little delay at zebra crossings, since they always have the right-of-way. However, they note that pedestrians who are unable or unwilling to accept gaps in traffic during the DON'T WALK period at the signalized pelican (pedestrian light controlled) midblock crossings in Britain incur substantial delay. They suggest that pedestrians crossing illegally at signalized intersections could be less safe than those crossing at random points along a roadway since drivers approaching a green signal will not expect to have to yield to a pedestrian.

Griffiths et al. (1984a) observed during their field studies that significant numbers of pedestrians are prepared to begin crossing during either flashing or steady DON'T WALK pedestrian indications. They noted that noncompliant behavior occurred almost exclusively when two-way conflicting vehicle flows were below 1,500/h.

Gordon and Robertson (1988) noted that driver noncompliance with traffic signals is a serious problem as well, particularly at low-volume intersections. They recommend a combination of higher enforcement levels, stiffer violation penalties, education of the public, and the removal of unnecessary informational or regulatory control devices adjacent to intersection approaches.

Knoblauch, Pietrucha, and Nitzburg (1996) noted that, of the pedestrians they observed during their field study of intersections in eastern cities, those who crossed against the signal (i.e., noncompliant pedestrians) tended to walk faster than those who crossed legally.

A study of Hong Kong pedestrians noted that pedestrians walk faster during the red phase at signals, confirming the ubiquity of noncompliant pedestrians (Lam et al., 1995). The authors report an average noncompliant pedestrian crosswalk speed of 1.5 m/s in Hong Kong crosswalks, much higher than the 1.27 m/s level observed at those facilities during the WALK indication.

Virkler (1997a) noted that, based on his observations of intersections in Brisbane, Australia, pedestrians typically treat about 69 percent of the flashing DON'T WALK signal as an effective WALK. He discerns two groups of noncompliant pedestrians: "jumpers," who start crossing before the WALK indication begins, and "runners," who begin crossing after the flashing DON'T WALK signal commences. Between the two groups, he observed that the runners saved over seven times as much delay per person as jumpers, so he focused on the behavior of the former group.

North Carolina State University (NCSU) also noticed similar noncompliant behavior at several sites during its field study of American intersections. NCSU calculated that pedestrians typically treated the first 5 s of flashing DON'T WALK as a de facto WALK signal indication. Indeed, the NCSU data-collection team observed some crossings during both flashing DON'T WALK (which typically coincides with the latter part of the vehicular green) and the vehicular clearance interval. Milazzo II (1996) adjusted his volume-occupancy data collection framework to allow for pedestrian occupancy of the crosswalk at any time during the pedestrian clearance interval.

Viney and Pretty (1982) examined pedestrian and vehicle interactions at Brisbane, Australia, intersections. They observed an average WALK "extension time" (i.e., de facto WALK) of 1.95 s with a standard deviation of 2.7 s. They used 2 s as an allowance for disobedient pedestrians.

It is important to note that changes in signal timing can affect noncompliance. For example, the slight increase in green time and cycle length that may occur under the assumption of reduced walking speeds will increase pedestrian delay and probably increase pedestrian noncompliance. Of course, the presence of excessive cycle lengths and/or unnecessary phases also causes pedestrian delay and noncompliance. Some jurisdictions use "early release" signal timing, where pedestrians receive the WALK before the concurrent vehicles receive the green, in an effort to reduce pedestrian delay. Regardless of the phasing scheme chosen, most facility users are local pedestrians who will learn the signal timing and try to reduce their own delay.

In summary, any delay measure to pedestrians should include some mechanism for considering noncompliance. Table 14 summarizes the findings of the last three research groups mentioned; coincidentally, two of the three groups examined downtown Brisbane, Australia. All of this empirical research seems to indicate that pedestrians, recognizing the margin of safety built into the pedestrian clearance interval, treat the initial part as an effective walk time.

TABLE 14 Selected de facto WALK extension times.