You are on page 1of 10

Relationship of Lane Width

to Capacity for Urban Expressways


Jinxuan Zheng, Jian Sun, and Jianhao Yang

To increase the capacity of urban expressways in Shanghai, China, addi- relationships. As a first step, this study focused on evaluating the
tional lanes were created during the past decade through reconstruction. relationship between lane width and capacity on the basis of data col-
Field measurements indicate that maximum lane width is 3.97 m and lected in Shanghai. In addition, as safety is a vital issue in express-
minimum width is only 2.73 m. To investigate the relationship between way operation, the relationship of lane width to safety was studied
lane width and capacity for urban expressways, 3 months of traffic flow elsewhere (1). The relationship of lane width to other operational
data were extracted and filtered from the system to manage inductive characteristics (e.g., lane usage) will be analyzed in the future.
detectors on Shanghai’s urban expressways. Analysis of all 440 segments Analysis of 440 sites in the Shanghai expressway system shows
of expressways in Shanghai showed that only 60 sites could reach their that 60 sites eventually reach their capacity and that their geometries
capacity; the characteristics of these 60 sites were further analyzed are consistent. By considering the similar characteristics of the driver
statistically. An analysis of a variance, a regression analysis, and a t-test population, vehicle type, posted speed limit, and traffic control con-
were used to explore the relationship between lane width and capacity. ditions, the impacts of other factors (including lane width, number
The research showed that lane width had no statistically significant effect of lanes, and right- and left-side lateral clearance) on capacity were
on capacity. Two causes that might account for this finding are dis- analyzed. Use of the multifactor analysis of variance (ANOVA)
cussed further: (a) free-flow speeds are similar for different lane widths found that only the number of lanes and right-side lateral clearance
according to the findings of a t-test that found no statistically signifi- were statistically significant. Then, regression analysis was applied
cant differences and (b) the critical speed when capacity is reached to study the influence of lane width, number of lanes, and lateral
for a Shanghai expressway is very low, only 45 km/h. Thus, drivers can clearance. It also showed no indication that lane width had an effect
deal with a narrow lane width at such a low critical speed. This find- on capacity. As a further step, the t-test was applied to investigate
ing suggests that geometric design policies for capacity purposes should the impact of lane width on capacity for different numbers of lanes.
provide substantial flexibility for use of narrower lane widths on urban The results again indicated no relationship between lane width and
expressways with low speed limits, although those lane widths must be capacity. These analyses form the basis of the paper’s discussion of
subject to safety considerations. the potential causes of the nonsignificant effects on capacity as they
relate to free-flow speed (FFS) and critical speed (speed at capacity).
The remainder of this paper is organized into the following sec-
Urban expressways are the skeleton of the urban transport system, tions: a brief review of the literature about the effect of lane width, a
and the play a vital role in urban motorized trips. According to the presentation of the study sites and the data, a description of the impact
specification for the design of urban expressways in China, lane of lane width on capacity, a discussion of the potential causes for the
widths are 3.50 m for light vehicles and 3.75 m for heavy vehicles nonsignificant impacts on capacity, and a summary of the findings.
at the posted speed limit of 80 km/h. In Shanghai, China, the urban
expressway system consists of the Inner Ring, the North–South, the
Yan-an, and the Middle Ring Expressways. To increase capacity, Literature Review
the Shanghai urban expressway system was partly reconstructed to
create additional lanes in one direction. For example, the Inner Ring Many studies have focused on the effect of lane width on safety,
Expressway’s cross sections of four lanes in both directions were but few have been concerned with the effect on capacity. A com­
converted to five lanes in both directions, the North–South Express- prehensive review of previous studies dealing with the impact of lane
way and Yan-an Expressway’s cross sections of six lanes in both width on capacity (or saturation flow rate)—with respect to both
directions were converted to seven lanes in both directions, and so highways–freeways and intersections—follows.
on. Because the cross-section width of the expressways remained the In relation to studies of the highway–freeway, the HCM 2010
same, lane widths were narrowed to some extent. Field measurement proposed an indirect relationship between lane width and capacity
of sites indicated that the maximum lane width is 3.97 m and the (2). Capacity is a function of the specified FFS, which is adjusted
minimum is 2.73 m. The different lane widths might have impacts on by lane width, lateral clearance, and so on. In 1968, Leong measured
the safety, capacity, lane flow distribution, and volume–speed–density capacity at 31 sites on rural two-lane highways in New South Wales,
Australia, and reported that the capacity of a two-lane road can drop
Key Laboratory of Road and Traffic Engineering, Department of Traffic Engineering, by 28% when the lane width decreased from 3.70 to 2.75 m (3).
Ministry of Education, Tongji University, No. 4800, Cao’an Road, Shanghai 201804, Nakamura analyzed the concept of highway capacity in Japan and
China. Corresponding author: J. Sun, sunjian@tongji.edu.cn. suggested adjustment factors (YL) for lane width (WL) less than
Transportation Research Record: Journal of the Transportation Research Board,
3.25 m as being YL = 0.24WL + 0.22 (4). Chandra and Kumar
No. 2483, Transportation Research Board, Washington, D.C., 2015, pp. 10–19. collected 10 locations on two-lane highways under mixed traffic
DOI: 10.3141/2483-02 conditions in India (5). They stated that the capacity of a two-lane

10
Zheng, Sun, and Yang 11

highway increased with total width ranging from 5.5 to 8.8 m. By would require excessive time to collect sufficient data. Therefore, a
using regression analysis, the relationship between capacity and the cross-sectional study was used to investigate the relationship between
total width was found to follow a second-degree curve. A study by lane width and capacity.
Zhong et al. also reported finding the relationship of a second-degree The previous studies found that no consistent relationship between
parabola between capacity and lane width by regression analysis lane width and capacity. Regression analysis was widely used in
of expressway data (6). Lane widths above 3.25 m had no impact many previous studies (4–6), and ANOVA analysis was also intro-
on the capacity. For lane widths below 2.9 m, an obvious effect duced into one study (9). However, most of these studies focused on
on capacity was found. However, most of these studies were about highways and intersections, with few studies about urban express-
two-way two-lane highways (2–6). Thus, the capacity is easily ways. In Shanghai, expressways are fully access-controlled divided
affected by lane width. highways with a minimum of two lanes in each direction and a low
As for research on intersections, the HCM 2010 chapter on signal- posted speed limit of 80 km/h. To study the relationship between
ized intersections contains a methodology for analyzing capacity to lane width and capacity for expressways, a cross-sectional evalua-
determine the level of service for intersections (2). Determination tion was adopted and both regression analysis and ANOVA analysis
of basic saturation flow rates and adjustment factors used to modify were used.
these flow rates is critical to the procedure. The chapter suggests a
negative effect of narrow lane width on saturation flow rate and a
positive effect of broad lane width. Zegeer studied the saturation Sites and Data
flow rates on 11 approaches with lane widths ranging from 2.6 to
4.7 m (7). The saturation flow rate dropped 2% to 5% for narrow lane Sites
widths compared with those in the baseline surveys, while the satu-
ration flow rate increased 5% in the broad lane widths compared with To analyze the effect of lane width properly and to avoid potential
those in the baseline surveys. Early research by Agent and Crabtree confounding factors, the candidate sites had to meet the following
showed that lane widths did not have an impact on saturation flow criteria:
rate for lane widths of 3.0 m or more (8). For lane widths between
2.7 and 3.0 m, a 5% reduction in saturation flow rate was found • A basic segment that is not affected or affected very little by
compared with lane widths of 3.0 m or greater. No lane widths below merging, diverging, or weaving traffic flow,
2.7 m were observed. A slight unexplained reduction in saturation flow • A straight, low-gradient segment,
rate occurred for lane widths more than 4.5 m. Potts et al. investigated • Similar vehicle compositions,
the effect of lane width on saturation flow rate of urban and suburban • Driver populations of regular commuters, and
signalized-intersection approaches and indicated that lane widths have • Posted speed limits of 80 km/h.
a statistically significant impact on saturation flow rate, which was
estimated by means of one-way ANOVA (9). For the present study, 440 sites on Shanghai expressways were
With respect to methodology, two general methods have been taken into account, but only 76 reached capacity. On the basis of
applied to quantify the effect of lane width on capacity: before–after these criteria, 60 sites (25 on the Inner Ring Expressway, 12 on the
studies and cross-sectional studies. The best method was found to be North–South Expressway, 21 on the Yan-an Expressway, and two
a well-designed before–after study, although the use of a before–after on the Middle Ring Expressway) were selected for analysis of the
study requires acquisition of data for both the before period and effect of lane width on capacity. The details of selected sites are
the after period. For the current study, however, the analysis period shown in Table 1.

TABLE 1   Characteristics of Study Sites

Average Lateral Clearance (m)


Site Loop Number Lane
Number Detector Expressway Location, Segment Type of Lanes Width (m) Left Side Right Side

1 NHNX02 IR, WB South Zhongshan Rd.–Luban INR, ONR-OFR 2 3.82 0.31 0.25
2 NHNX10 IR, WB South Ruijin Rd.–South Wanping Rd., ONR-OFR 3 3.09 0.17 0.13
3 NHNX11 IR, WB South Ruijin Rd.–South Wanping Rd., ONR-OFR 3 3.09 0.17 0.13
4 NHNX12 IR, WB South Ruijin Rd.–South Wanping Rd., ONR-OFR 3 3.09 0.17 0.13
5 NHNX13 IR, WB South Wanping Rd.–Tianyaoqiao Rd., OFR-ONR 2 3.82 0.31 0.25
6 NHNX14 IR, WB South Wanping Rd.–Tianyaoqiao Rd., OFR-ONR 2 3.82 0.31 0.25
7 NHNX15 IR, WB South Wanping Rd.–Tianyaoqiao Rd., OFR-ONR 2 3.82 0.31 0.25
8 NHNX16 IR, WB South Wanping Rd.–Tianyaoqiao Rd., OFR-ONR 2 3.82 0.31 0.25
9 NHNX17 IR, NB Tianyaoqiao Rd.–Caoxi INR, ONR-OFR 3 3.09 0.17 0.13
10 NHNX25 IR, NB Wuzhong Rd.–Xinhua Rd., ONR-OFR 3 3.13 0.10 0.09
11 NHNX26 IR, NB Wuzhong Rd.–Xinhua Rd., ONR-OFR 3 3.13 0.10 0.09
12 NHNX27 IR, NB Xinhua Rd.–Yanxi INR, OFR-OFR 2 3.76 0.30 0.30
13 NHNX31 IR, NB Yanxi INR–Wuyi Rd., ONR-ONR 2 3.23 0.16 0.34
(continued on next page)
12 Transportation Research Record 2483

TABLE 1 (continued)   Characteristics of Study Sites

Average Lateral Clearance (m)


Site Loop Number Lane
Number Detector Expressway Location, Segment Type of Lanes Width (m) Left Side Right Side

14 NHNX32 IR, NB Wuyi Rd.–Jinshajiang Rd., ONR-OFR 2 3.21 0.15 0.27


15 NHNX40 IR, EB Wuning Rd.–Zhenping Rd., ONR-OFR 2 3.24 0.16 0.27
16 NHNX49 IR, EB Hutai Rd.–Gonghe INR, ONR-OFR 3 3.61 0.13 0.10
17 NHNX70 IR, SB Huangxing Rd.–Zhoujiazui Rd., ONR-OFR 2 3.82 0.26 0.29
18 NHWX02 IR, EB South Xizang Rd.–South Zhongshan Rd., OFR-OFR 2 3.20 0.22 0.25
19 NHWX12 IR, EB South Wanping Rd.–South Ruijin Rd., ONR-OFR 2 3.20 0.22 0.25
20 NHWX20 IR, SB Wuzhong Rd.–Caoxi INR, ONR-OFR 2 3.82 0.33 0.29
21 NHWX21 IR, SB Wuzhong Rd.–Caoxi INR, ONR-OFR 2 3.82 0.33 0.29
22 NHWX25 IR, SB Xinhua Rd.–Wuzhong Rd., ONR-OFR 2 3.21 0.16 0.26
23 NHWX38 IR, SB Wuning Rd.–Jinshajiang Rd., ONR-OFR 2 3.22 0.24 0.30
24 NHWX55 IR, WB Guangzhong Rd.–North Xizang Rd., ONR-OFR 2 3.21 0.13 0.28
25 NHWX71 IR, NB Zhoujiazui Rd.–Huangxing Rd., ONR-OFR 2 3.86 0.24 0.28
26 NBDX02 NS, NB Luban INR–Xujiahui Rd., ONR-OFR 3 3.77 0.35 0.29
27 NBDX05 NS, NB Xujiahui Rd.–Middle Huaihai Rd., ONR-OFR 4 2.91 0.15 0.15
28 NBDX10 NS, NB Yandong INR–Weihai Rd., ONR-ONR 4 2.93 0.08 0.14
29 NBDX26 NS, NB Guangzhong Rd.–Gonghe INR, ONR-OFR 3 3.75 0.36 0.36
30 NBDX32 NS, NB Wenshui Rd.–Linfen Rd., ONR-OFR 3 3.76 0.34 0.34
31 NBXX05 NS, SB Middle Huaihai Rd.–Xujiahui Rd., ONR-OFR 4 2.93 0.06 0.15
32 NBXX06 NS, SB Middle Huaihai Rd.–Xujiahui Rd., ONR-OFR 4 2.93 0.06 0.15
33 NBXX16 NS, SB Gonghe INR–West Tianmu Rd., ONR-OFR 3 3.76 0.37 0.36
34 NBXX17 NS, SB Gonghe INR–West Tianmu Rd., ONR-OFR 3 3.76 0.37 0.36
35 NBXX28 NS, SB Wenshui Rd.–Guangzhong Rd., ONR-OFR 3 3.75 0.35 0.36
36 NBXX29 NS, SB Wenshui Rd.–Guangzhong Rd., ONR-OFR 3 3.75 0.35 0.36
37 NBXX34 NS, SB Changzhong Rd.–Wenshui Rd., ONR-OFR 3 3.76 0.35 0.37
38 YABX03 YA, WB Hongxu Rd.–Hongqiao Hub, ONR-OFR 3 3.78 0.32 0.34
39 YABX04 YA, WB Hongxu Rd.–Hongqiao Hub, ONR-OFR 3 3.78 0.32 0.34
40 YABX08 YA, WB Loushanguan Rd.–Hongxu Rd., OFR-OFR 3 3.76 0.37 0.32
41 YABX09 YA, WB Loushanguan Rd.–Hongxu Rd., OFR-OFR 3 3.76 0.37 0.32
42 YABX10 YA, WB Loushanguan Rd.–Hongxu Rd., OFR-OFR 3 3.76 0.37 0.32
43 YABX14 YA, WB Kaixuan Rd.–Yanxi INR, OFR-OFR 3 3.73 0.34 0.37
44 YABX15 YA, WB Kaixuan Rd.–Yanxi INR, OFR-OFR 3 3.73 0.34 0.37
45 YABX24 YA, WB Maoming Rd.–Huashan Rd., ONR-OFR 4 2.89 0.07 0.35
46 YABX26 YA, WB Yandong INR–Maoming Rd., ONR-ONR 3 3.79 0.35 0.35
47 YABX27 YA, WB Yandong INR–Maoming Rd., ONR-ONR 3 3.79 0.35 0.35
48 YANX03 YA, EB Hongjing Rd.–Hongxu Rd., ONR-OFR 3 3.76 0.36 0.33
49 YANX08 YA, EB Hongxu Rd.–Loushanguan Rd., ONR-ONR 3 3.76 0.34 0.34
50 YANX09 YA, EB Hongxu Rd.–Loushanguan Rd., ONR-ONR 3 3.76 0.34 0.34
51 YANX10 YA, EB Hongxu Rd.–Loushanguan Rd., ONR-ONR 3 3.76 0.34 0.34
52 YANX12 YA, EB Loushanguan Rd.–Yanxi INR, ONR-OFR 4 2.90 0.06 0.32
53 YANX21 YA, EB Jiangsu Rd.–Huashan Rd., ONR-ONR 3 3.75 0.36 0.36
54 YANX22 YA, EB Jiangsu Rd.–Huashan Rd., ONR-ONR 3 3.75 0.36 0.36
55 YANX23 YA, EB Jiangsu Rd.–Huashan Rd., ONR-ONR 3 3.75 0.36 0.36
56 YANX24 YA, EB Jiangsu Rd.–Huashan Rd., ONR-ONR 3 3.75 0.36 0.36
57 YANX25 YA, EB Huashan Rd.–Maoming Rd., ONR-OFR 4 2.90 0.06 0.32
58 YANX26 YA, EB Huashan Rd.–Maoming Rd., ONR-OFR 4 2.90 0.06 0.32
59 ZHNX73 MR, NB Gudai Rd.–Caobao Rd., ONR-OFR 4 3.44 0.34 0.34
60 ZHNX74 MR, NB Gudai Rd.–Caobao Rd., ONR-OFR 4 3.44 0.34 0.34

Note: IR = Inner Ring Expressway; NS = North–South Expressway; WB = westbound; rd. = road; INR = interchange; ONR = on-ramp; OFR = off-ramp;
NB = northbound; EB = eastbound; SB = southbound; YA = Yan-an Expressway; MR = Middle Ring Expressway.
Zheng, Sun, and Yang 13

Lane Width Data

To improve the capacity of expressways, they were reconstructed


to create additional lanes during the past decade by the municipal- Huma Road off-ramp – Gongjiang Road on-ramp –
ity in Shanghai. Therefore, lane widths were sharply narrowed in Gongjiang Road on-ramp Huma Road off-ramp
the reconstructed segments. According to the expressway closure
information provided by the Department of Highways in Shanghai
(http://www.highway.sh.cn), the site was surveyed from 00:30
to 05:00.
Figures 1 through 3 show the cross-section view of all the types
of segments (including segments without reconstruction for the
Middle Ring Expressway, Figure 3) considered in this paper. For
example, Figure 1a shows dimensions before reconstruction, (a)
including the total cross-sectional width of the Inner Ring Express-
way is 18 m (which did not change with reconstruction); the total
Yandong Interchange on-ramp – Maoming Road on-ramp –
width across all travel lanes in each direction is 8.25 m; the median Maoming Road on-ramp Yandong Interchange on-ramp
width and lateral-barrier width are both 0.5 m; the right lateral
clearances (close to the roadside) are 0.29 and 0.28 m, respectively
(excluding the lane-marking width); the left lateral clearances
(close to the median) are 0.33 and 0.28 m, respectively (excluding
the lane-marking width); the width of the lane marking is 0.15 m;
lane widths including lane markings are 3.95, 3.68, 3.74, and 3.95 m,
respectively; thus, the average lane widths are 3.82 and 3.85 m in
each direction. (b)
The descriptive statistics of lane width are exhibited in Table 2.
Lane width of the sections ranges from 2.89 to 3.86 m, for a maxi- Huaihai Road on-ramp – Xujahui Road on-ramp –
Xujahui Road off-ramp Huaihai Road off-ramp
mum difference of 0.97 m. The average lane width for the 60 sites
is 3.50 m, including the lane marking. Lateral clearance is 0.26 m
for the left side and 0.28 m for the right side.

Traffic Flow Data

Three months of data (July 2010 and April and May 2011) were
collected for analysis from the inductive detectors on the Shanghai (c)
urban expressways. The time interval of the data was 5 min, and
FIGURE 2   Cross sections of North–South and Yan-an Expressways:
the traffic characteristics (e.g., vehicle type, volume, speed, occu-
(a) two-way, six lanes total (before reconstruction, three in each
pancy, time headway) of each lane were recorded. The traffic flow direction), (b) two-way, seven lanes, and (c) two-way, eight lanes
was composed of 98% passenger cars, 2% coaches (buses), and total (after reconstruction, either three in one direction and four
no trucks. in other or four in each direction).

South Wanping Road off-ramp – South Ruijin Road on-ramp –


Wuzhong Road off-ramp – Wuzhong Road on-ramp South Ruijin Road on-ramp South Wanping Road off-ramp

(a) (b)

FIGURE 1   Cross sections of Inner Ring Expressway: (a) four lanes total (before reconstruction, two in each direction)
and (b) five lanes total (after reconstruction, two in one direction and three in other).
14 Transportation Research Record 2483

Guangyue Road off-ramp–Guangyue Road on-ramp

FIGURE 3   Cross section of Middle Ring Expressway, four lanes in each direction.

According to the HCM 2010, “capacity” is defined as the “maxi- The Shanghai urban expressway system mainly serves daily
mum sustainable hourly flow rate at which vehicles reasonably can commuter traffic. Thus, the driver population, vehicle type, posted
be expected to traverse a point or a uniform section of a lane or road- speed limits, and traffic control conditions are similar on the various
way during a given time period under prevailing roadway, environ- expressways. In addition, all the selected study sites are consistent in
mental, traffic, and control conditions” (2). In this study, capacity geometry, and the capacity is a mean value of days. Thus, this study
is the maximum 5-min hourly flow rate, which was also adopted in can consider only the impact of the width, lateral clearance, and
other capacity studies [e.g., Zhang and Levinson (10), Lorenz and number of lanes on capacity and ignore other confounding factors.
Elefteriadou (11)]. The speed at capacity is the critical speed. Then, In this study, the ANOVA, regression analysis, and t-test were
the mean capacity (at least 30-day capacities) was considered as a applied to analyze the impact of lane width on capacity. The lane
site capacity. Similarly, the mean speed at capacity was selected as widths were grouped into three categories. (The specification for
the critical speed. design of urban expressways in China sets the width of standard-
size lanes at 3.50 m for light vehicles on basic segments.) The three
categories are as follows:
Effect of Lane Width on Capacity
• Undersized: <3.25 m,
The most appropriate method for determining the effect of lane • Standard size: ≥3.25 and <3.75 m, and
width on capacity is by a well-designed before–after evaluation. • Oversized: ≥3.75 m.
While the use of a before–after evaluation would be the preferred
approach for determining the effect of lane width on capacity, such Two reasons account for these dividing criteria. On the one hand,
an evaluation was not feasible in this study for the lack of before- according to another paper, lane width below 3.25 m would sharply
period study data. Therefore, a cross-sectional analysis approach was increase crashes (1). On the other hand, the t-test shows no significant
used to investigate the relationship between lane width and capacity. difference in capacity between widths less than 3.00 m and widths
from 3.00 to 3.25 m. Therefore, to make the categories consistent
in the two papers, the lane width less than 3.25 m was regarded as
TABLE 2   Descriptive Statistics for Lane Width one group.
Furthermore, the lateral clearances were divided into four sizes.
Lane Width (including lane marking) The categories of lateral clearance (LC) are as follows:
(m)
LC1: <0.1 m,
Lane N Mean Max. Min. SD
LC2: ≥0.1 m and <0.2 m,
Section 60 3.50 3.86 2.89 0.35 LC3: ≥0.2 m and <0.3 m, and
Median 60 3.48 3.89 2.79 0.33 LC4: ≥0.3 m and <0.4 m.
Center 52 3.38 3.83 2.78 0.37
Shoulder 60 3.54 3.97 2.73 0.42
Multifactor ANOVA
Lateral clearance
  Left side 60 0.26 0.37 0.06 0.11
  Right side 60 0.28 0.37 0.09 0.08 In the ANOVA study, the environmental, traffic, and control conditions
of the sites are the same. The geometric features for the segments are
Note: Max. = maximum; min. = minimum; SD = standard deviation. similar, except that the lane width, number of lanes, and left- and
Zheng, Sun, and Yang 15

right-side lateral clearances are different. Any single different factor TABLE 4   Results of Regression Analysis Model
may affect capacity. Therefore, all four factors should be included and
compared when one wants to study their numerical effects on capac- Test Factor Coefficient p-Value
ity. A multifactor ANOVA can test several sample means for several Intercept 2,111.714 0
different effects. Thus, the ANOVA was performed in this research.
Number of lanes −39.358 .160
The variability in data due to the treatment under each test effect
Lane width
was compared with the variability due to random errors and the  Undersized 9.087 .826
resulting ratio is an F-value for that test effect. The F-value is then   Standard size 0 na
compared with a standard critical table value in the F-distribution  Oversized 30.999 .410
to check the significance of that test effect. A p-value could be given Right-side lateral clearance
to show how likely the specific null hypothesis (H0) is in the case  RLC1 −252.768 .001*
of the computed F-value. A large F-value and a small p-value can  RLC2 −124.436 .004*
 RLC3 −58.795 .159
normally lead to rejection of the null hypothesis H0. In addition,  RLC4 0 na
the R2-value is given to index the adequacy of the analysis model.
The ANOVA test results are shown in Table 3. The p-values show Note: na = not applicable.
that only the number of lanes and right-side lateral clearance are sig- * = significance at .01 level.
nificant at a 95% confidence level. From the F-test values, one can
conclude that lane width is the least significant of the four factors.
In other words, lane width has no significant impact on capacity at
RLC3 = indicator variable (= 1 if right-side lateral clearance
a cross-sectional level.
is LC3, = 0 if not),
RLC4 = indicator variable (= 1 if right-side lateral clearance is
LC4, = 0 if not), and
Regression Analysis a, b, c, d = regression coefficients.
To quantify the impact of lane width on capacity, regression analysis Table 4 shows the results of the regression analysis model. The
was also applied. On the basis of the Spearman correlation test, right- coefficients of lane width are expressed through comparison with
side and left-side lateral clearances show high positive relevance. standard size (LW2) (i.e., the value of coefficient LW2 is zero). More-
In addition, because left-side lateral clearance is nonsignificant in over, the coefficients of the right-side lateral clearance are expressed
the ANOVA test, left-side lateral clearance was abandoned in the through comparison with RLC4 (i.e., the value of coefficient RLC4
following analysis. The regression model is shown as is zero).
The regression analysis model is statistically significant, with an R2
cap = a + bN + c1LW1 + c2 LW2 + c3 LW3 + d1RLC1 + d 2 RLC2 value of about .392. The coefficients and p-values in Table 4 show that
changes in capacity may be more attributable to right-side lateral
+ d3 RLC3 + d 4 RLC4 (1)
clearances than to lane width.
where
cap = predicted capacity, Effect of Lane Width on Capacity
N = number of lanes, for Different Numbers of Lanes
LW1 = indicator variable (= 1 if lane width is undersized, = 0
if not), To evaluate the relationship between lane width and capacity fur-
LW2 = indicator variable (= 1 if lane width is standard size, ther, the cases for different numbers of lanes were analyzed. A t-test
= 0 if not), was performed to evaluate the significant difference for the three
LW3 = indicator variable (= 1 if lane width is oversized, = 0 categories of lane widths described earlier.
if not),
RLC1 = indicator variable (= 1 if right-side lateral clearance
Two-Lane Expressway
is LC1, = 0 if not),
RLC2 = indicator variable (= 1 if right-side lateral clearance is Table 5 shows the capacity descriptive statistics and t-test results for
LC2, = 0 if not), a two-lane expressway. The mean capacity is 1,982 vehicles per hour
(vph) per lane (vphpl). For undersized and oversized, the capacity is
2,006 and 1,963 vphpl, respectively. The difference between them
TABLE 3   Results of Multifactor ANOVA Tests is small, 43 vphpl. For the median lane or the shoulder lane, the dif-
ference is also small. In addition, the capacity of the median lane is
Degrees of F-Test slightly greater (3%) than that of the shoulder lane. As the p-values in
Test Factor Freedom Value p-Value Table 6 show, the different lane widths have no statistically significant
differences in the two-lane expressway.
Number of lanes 2 5.277 .008*
Lane width 2 1.745 .185
Left-side lateral clearance 3 0.361 .781 Three-Lane Expressway
Right-side lateral clearance 3 7.916 .000*
As Table 5 shows, the capacity for the three-lane expressway is
* = significance at .01 level. 2,002 vphpl for overall lane width, 1,855 vphpl for undersized lanes,
16 Transportation Research Record 2483

TABLE 5   Descriptive Statistics for Capacity

Capacity (vphpl), by Lane

Section Lane Median Lane Center Lane 1 Center Lane 2 Shoulder Lane

Lane Width N Mean SD N Mean SD N Mean SD N Mean SD N Mean SD

Two-Lane Urban Expressway


Undersized 8 2,006 103 8 2,078 174 — — — — — — 8 2,021 93
Standard size — — — 7 2,021 48 — — — — — — — — —
Oversized 10 1,963 57 3 2,062 70 — — — — — — 10 1,971 66
Total 18 1,982 81 18 2,053 121 — — — — — — 18 1,993 81
Three-Lane Urban Expressway
Undersized 6 1,855 90 6 2,029 94 6 1,978 82 — — — 6 1,784 83
Standard size 7 2,008 130 17 2,108 73 21 2,147 141 — — — 5 1,959 133
Oversized 19 2,047 85 9 2,170 123 5 2,168 32 — — — 21 1,997 108
Total 32 2,002 119 32 2,111 102 32 2,119 137 — — — 32 1,951 133
Four-Lane Urban Expressway
Undersized 8 1,878 100 8 1,988 127 10 1,973 147 8 1,940 151 8 1,980 142
Standard size 2 1,841 10 2 2,059 104 — — — 2 1,753 72 2 1,911 65
Total 10 1,871 90 10 2,002 121 10 1,973 147 10 1,902 157 10 1,966 131

Note: — = missing data.

2,008 vphpl for standard-sized lanes, and 2,047 vphpl for oversized the oversized lane. Therefore, another step was taken to identify the
lanes. The capacity of undersized lanes is obviously less than that of real cause of this difference.
the standard size and oversized lanes; the capacity of undersized lanes The right-side lateral clearances of a three-lane expressway are
is 7.6% and 9.4% less than that of the standard size and the over- shown in Table 7. The right-side lateral clearance of the standard-sized
sized lanes, respectively. No clear difference exists between standard lane and of the oversized lane are 0.20 and 0.22 m, respectively,
size and oversized lanes. These differences also exist in the median, greater than that of undersized lane. To distinguish the effect of
center, and shoulder lanes. As one would expect, the results of the lane width and right-side lateral clearance, the ANOVA was applied.
t-test indicate significant differences between the undersized lane The results (p-value of .479 for lane width, .012 for right-side lateral
and the standard-sized lane and between the undersized lane and clearance) of ANOVA indicate that lane width has no statistically

TABLE 6   Results of t-Test for Capacity

p-Value

Lane Width Section Median Lane Center Lane1 Center Lane2 Shoulder Lane

Two-Lane Urban Expressway


Undersized versus standard size na .400 na na na
Standard size versus oversized na .306 na na na
Undersized versus oversized .280 .883 na na .202
Three-Lane Urban Expressway
Undersized versus standard size .034* .045* .010* na .025*
Standard size versus oversized .385 .191 .550 na .509
Undersized versus oversized .000** .033* .001** na .000**
Four-Lane Urban Expressway
Undersized versus standard size .633 .490 na .140 .539
Standard size versus oversized na na na na na
Undersized versus oversized na na na na na

Note: na = not applicable.


*Significance at .05; **significance at .01.
Zheng, Sun, and Yang 17

TABLE 7   Descriptive Statistics for Three-Lane, no significant impact on capacity in this research. Thus, the critical
Right-Side Lateral Clearance speed is discussed next.
Lateral Clearance (m)

Lane Width N Mean Max. Min. SD Effect of Lane Width on FFS

Undersized 6 0.12 0.13 0.09 0.02 Capacity is a function of FFS, which is adjusted by lane width,
Standard size 7 0.32 0.37 0.10 0.10 lateral clearance, and ramp density in the HCM 2010. The HCM
Oversized 19 0.34 0.37 0.29 0.02 2010 states that FFS decreases when average lane width is less than
Total 32 0.30 0.37 0.09 0.10
3.66 m (12 ft). When the lane width ranges from 3.05 to 3.35 m,
the reduction in FFS is up to 10.62 km/h. The effect of narrower
lane width on FFS is obvious and indicates its indirect effect on
capacity.
The “FFS” is defined as the mean speed of passenger cars measured
significant effect on capacity. However, statistically significant dif- during periods of low to moderate flow (up to 1,000 passenger cars
ferences were found between the right-side lateral clearances. Com- per hour per lane) in the HCM 2010. In this study, given the operation
parison of the capacity of the different lanes shows that the capacity of Shanghai expressways and traffic composition, “FFS” is defined
of median lane and the center lane is approximately 160 vphpl (an when the time headway is greater than 10 s. That is to say, FFS is the
average of 10%) more than that of the shoulder lane, and therefore mean speed of vehicles measured during periods with hourly flow
right-side lateral clearances contribute most to the capacity differ- rate per lane of below 360 passenger cars.
ences for the various cross sections. Because the elevated roads of The average FFS in this study is shown in Table 8. The FFSs for the
Shanghai expressways have no shoulders, driving is constrained lanes are similar to one another; the average is 76.38 km/h. In general,
because of the limited right-side lateral clearances. the wider the lane width is, the slightly greater is FFS. When under-
sized lanes are compared with oversized ones, the difference in FFS
is only about 4 km/h. In relation to the FFS classification standard in
Four-Lane Expressway the HCM 2010 [speed interval of 5 mph (8 km/h)], 4 km/h is a small
fluctuation range. In addition, in Table 9, t-test results verify that no
Table 5 shows that the average capacity is about 1,871 vphpl for four- statistically significant difference exists. Thus, all the FFSs in the
lane expressways. For a four-lane expressway with an undersized sites for this study are at the same level. Two reasons may account
lane width, the capacity is 1,878 vphpl; for a standard-sized lane, for the similar FFSs: (a) the low posted speed limit, 80 km/h, on
width it is 1,841 vphpl. As Table 6 shows, no significant difference in Shanghai expressways and (b) vehicles traveling with a greater degree
capacity exists between the undersized and the standard-sized lanes. of freedom at free traffic. Therefore, lane width has no significant
In general, no matter the number of lanes, lane width has no statisti- effect on FFS at the low speed limit. Thus, the finding that effect of
cally significant effect on capacity. Therefore, the authors conclude lane width on capacity is not significant is indirectly proved.
that lane width has no significant effect on capacity.

Critical Speed
Discussion of Effect of
Lane Width on Capacity The typical speed–flow curves for two-, three-, and four-lane urban
expressways are shown in, respectively, Figure 4, a, b, and c. The
According to the HCM 2010, the basic capacity of an uninterrupted- relationships of flow and speed resemble the classic quadratic-
flow facility is determined by FFS, which is adjusted by various parabola model. The fundamental curves show that FFS is about
factors, including lane width. Therefore, analysis of the effect of 80 km/h for all three types of expressways. For each curve, speeds
lane width on FFS is needed. In addition, compared with the critical decline at a sharp rate until capacity is reached. The critical speed
speed in the HCM 2010, this study found a lower critical speed, and for a two-lane expressway is 44 km/h; for a three-lane, 45 km/h; and
this lower critical speed may be the root cause for lane width having for a four-lane, 48 km/h, as shown in Figure 4. The mean critical

TABLE 8   Descriptive Statistics for Average FFS

Total Two Lanes Three Lanes Four Lanes

Lane Width N Mean SD N Mean SD N Mean SD N Mean SD

Undersized 22 74.13 3.49 8 73.01 3.26 6 74.91 3.89 8 74.66 3.58


Standard size 9 75.64 3.28 — — — 7 75.52 3.72 2 76.06 1.60
Oversized 29 78.32 5.72 10 76.43 4.80 19 79.31 6.03 — — —
Total 60 76.38 5.01 18 74.91 4.43 32 77.66 5.51 10 74.94 3.25

Note: — = missing data.


18 Transportation Research Record 2483

TABLE 9   Results of t-Test for Different Lane Widths Conclusions

p-Value In this study, a traffic flow data set of 60 sites for 3 months was used
to analyze the relationship between lane width and capacity. The
Lane Width Two Lanes Three Lanes Four Lanes Total
ANOVA analysis, regression analysis, and t-test were used to evalu-
Undersized versus na .777 .614 .273 ate the effect of lane width on capacity on a Shanghai expressway.
  standard size Some main conclusions follow:
Standard size versus na .135 na .192
  oversized 1. Use of multifactor ANOVA indicates that the number of lanes
Undersized versus .105 .109 na .004 and right-side lateral clearance are the statistically significant fac-
  oversized tors among those influencing capacity. However, lane width has no
statistically significant effect on capacity.
2. Regression analysis also shows that lane width has no effect
on capacity. Moreover, the t-test of capacity for different numbers of
lanes indicates that lane width has no statistically significant effect
on capacity.
speed of all sections is approximately 45.16 km/h (standard devia- 3. For a three-lane expressway, the t-test indicates significant
tion of 6.33 km/h). The critical speed and the FFS in the HCM 2010 differences between an undersized lane and a standard-sized one
(50 mph p 1.61 = 80.5 km/h for 88.5 km/h FFS) are similar, but the and between an undersized lane and an oversized one. The capacity
critical speed of a Shanghai expressway is extremely slower than of an undersized lane is 7.6% and 9.4%, respectively, lower than that
that of an American freeway. Thus, the authors believe that Shanghai of the standard-sized and the oversized lanes. However, the results of
drivers can deal with the narrow lane width at a 45 km/h critical speed. ANOVA show that this difference is caused by right-side lateral
A comparison of the speed–flow curves for a Shanghai expressway clearances.
with those in the HCM 2010 shows that speed in Shanghai rapidly 4. Two causes might account for the relationship between lane
drops with an increase in flow rate. Whether this phenomenon is width and capacity. First, the FFSs for different lane widths are
related to lane width needs to be studied. similar, and differences were not found to be statistically significantly.

100 100

80 80
Speed (km/h)

Speed (km/h)

60 60

40 40

20 20

0 0
0 500 1,000 1,500 2,000 2,500 0 500 1,000 1,500 2,000 2,500
Flow Rate (vphpl) Flow Rate (vphpl)
(a) (b)

100 100
FFS = 55 mph (88.5 km/h)
80 80
Speed (km/h)

Speed (km/h)

60 60
ln
m/
u/k
40 40 pc
28
20 20

0 0
0 500 1,000 1,500 2,000 2,500 0 500 1,000 1,500 2,000 2,500
2,250
Flow Rate (vphpl) Flow Rate (vphpl)
(c) (d)

FIGURE 4   Speed–flow relationships for (a) capacity of 1,990 veh/h and critical speed of 44 km/h
for two lanes at Site 7, (b) capacity of 1,940 veh/h and critical speed of 45 km/h for three lanes
at Site 3, (c) capacity of 1,850 veh/h and critical speed of 48 km/h for four lanes at Site 59, and
(d) capacity of 2,250 veh/h and critical speed of 80 km/h [adopted from speed-flow curves for basic
freeway segments under base conditions in HCM 2010 (2)].
Zheng, Sun, and Yang 19

Second, the critical speed of this study (mean of 45.16 km/h for a   5. Chandra, S., and U. Kumar. Effect of Lane Width on Capacity Under
76.4-km/h FFS in Shanghai) is extremely slower than that in the Mixed Traffic Conditions in India. Journal of Transportation Engineering,
Vol. 129, No. 2, 2003, pp. 155–160.
HCM 2010 (80 km/h for 88.5-km/h FFS). Thus, the authors believe
  6. Zhong, L., J. Rong, and X. Sun. Discussion of Lane Width of Elevated
that drivers can deal with the narrow lane width. Expressway. Journal of Highway and Transportation Research and
Development, Vol. 23, No. 10, 2006, pp. 117–119 (in Chinese).
  7. Zegeer, J. D. Field Validation of Intersection Capacity Factors. In Trans-
portation Research Record 1091, TRB, National Research Council,
Acknowledgments Washington, D.C., 1986, pp. 67–77.
  8. Agent, K. R., and J. D. Crabtree. Analysis of Saturation Flow at Signal-
The authors thank the Natural Science Foundation of China, New ized Intersections. Kentucky Transportation Center Research Reports,
Century Excellent Talents in University, the National Key Technology Paper 761. http://uknowledge.uky.edu/ktc_researchreports/761.
R&D Program, and the Fundamental Research Funds for the Central   9. Potts, I. B., J. F. Ringert, K. M. Bauer, J. D. Zegeer, D. W. Harwood, and
D. K. Gilmore. Relationship of Lane Width to Saturation Flow Rate on
Universities for supporting this study.
Urban and Suburban Signalized Intersection Approaches. In Transpor-
tation Research Record: Journal of the Transportation Research Board,
No. 2027, Transportation Research Board of the National Academies,
Washington, D.C., 2007, pp. 45–51.
References 10. Zhang, L., and D. Levinson. Some Properties of Flows at Freeway
Bottlenecks. In Transportation Research Record: Journal of the Trans-
  1. Wu, L., and J. Sun. Relationship of Lane Width to Safety for Urban portation Research Board, No. 1883, Transportation Research Board of
Expressways. Presented at 94th Annual Meeting of the Transportation the National Academies, Washington, D.C., 2004, pp. 122–131.
Research Board, Washington, D.C., 2015. 11. Lorenz, M. R., and L. Elefteriadou. Defining Freeway Capacity as Func-
 2. Highway Capacity Manual 2010. Transportation Research Board of the tion of Breakdown Probability. In Transportation Research Record:
National Academies, Washington, D.C., 2010. Journal of the Transportation Research Board, No. 1776, TRB, National
  3. Leong, H. J. W. The Distribution and Trend of Free Speeds on Two-Way
Research Council, Washington, D.C., 2001, pp. 43–51.
Rural Highways in New South Wales. Proc., 4th Australian Road Research
Board Conference, Melbourne, Australia, Vol. 4, 1968.
  4. Nakamura, M. Research and Application of Highway Capacity in Japan.
Proc., 2nd International Symposium on Highway Capacity: Country The Standing Committee on Highway Capacity and Quality of Service peer-reviewed
Reports, 1994. this paper.

You might also like