Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
1
TRB 2007 Annual Meeting CD-ROM Original paper submittal - not revised by author.
Capacity Estimation for Two-lane Two-way Highways Using Simulation
ABSTRACT
The Highway Capacity Manual 2000 (HCM 2000) indicates that the capacity of two-lane,
two-way highways (TLTWHW) is 1,700 pcph for a single direction and 3,400 pcph for
both directions. There are no capacity estimates provided for different traffic conditions
such as varying percentages of no-passing zones and varying traffic distributions between
the two directions of travel. Previous research has reported that capacity conditions for
TLTWHW are very difficult to observe in the field, while none of the existing simulation
models can report capacity for TLTWHW. This paper briefly describes a new two-lane,
two-way highway simulation model, called TWOSIM that is developed, verified, compared
to TWOPAS and HCS, and used for the estimation of capacity for TLTWHW under a
variety of prevailing traffic and geometric conditions. These include percentage of traffic
per direction, presence of a passing zone, a horizontal curve, a driveway, an upgrade, and
heavy vehicles. Capacity of TLTWHW under base conditions is found to range from 1,800
pcph to 2,100 pcph as a function of average free flow speed. The presence of passing zones
was not found to have an effect on capacity. When a driveway or horizontal curve or
upgrade is present, the capacity reduction ranges between 3 and 30 percent when there are
no trucks, and may reach up to 40 percent when trucks are present in the traffic stream.
INTRODUCTION
The Highway Capacity Manual 2000 (HCM 2000) defines capacity as the maximum
number of vehicles that can pass a given point during a specified period under prevailing
roadway, traffic, and control conditions and indicates that the capacity of two lane two way
highways (TLTWHW) is 1,700 pcph for a single travel direction and 3,400 pcph for both
directions. There are no capacity estimates provided for different traffic conditions such as
varying percentages of no-passing zones, varying traffic distributions between the two
directions of travel, presence of driveways, etc. According to Harwood, et al. (1999),
capacity conditions are very difficult to observe in the field, because very few TLTWHW
operate at or near capacity. The two existing commercially available simulators for
TLTWHW, TWOPAS (TWO-lane PASsing) and TRARR (TRAffic on Rural Roads),
cannot analyze traffic volumes higher than 1,700 vph, and do not provide capacity
estimates. This paper briefly describes how a new simulation model, TWOSIM (TWO-lane
two-way highway SIMulator) was developed, verified, and compared to TWOPAS and
HCS, and provides TWOSIM’s capacity estimates for various traffic and design conditions.
The objective of this study is to estimate the capacity of TLTWHW corresponding
to a variety of factors including percent of traffic per direction, presence of a passing zone,
a horizontal curve, a driveway, an upgrade, and heavy vehicles. The next section presents a
summary of the relevant literature, while the third section briefly reviews how TWOSIM is
built and verified and discusses the factors considered to affect capacity. Then capacities
estimated with TWOSIM are presented along with the results of a sensitivity analysis
examining how capacity varies with conditions such as the presence of passing zones. The
last section presents conclusions and recommendations.
2
TRB 2007 Annual Meeting CD-ROM Original paper submittal - not revised by author.
LITERATURE REVIEW
There have been several articles analyzing operations on TLTWHW. This section discusses
those most relevant to this study, and primarily research concerned with the capacity and
simulation modeling of TLTWHW.
In one of the earliest such studies, Hammond and Sorenson (1941), Clayton (1941),
and Wardrop (1952) suggested a theoretical maximum volume for a single lane as a
function of speed and average vehicle spacing. Normann (1942) plotted the spacing against
speed and derived a theoretical lane capacity. These analytical models estimated the
theoretical capacity as a function of minimum headway. Such an approach does not
consider overtaking and the interaction of overtaking vehicles with the opposing traffic
stream. Yagar (1983), using data from Japan and Australia, indicated that 3,600 pcph can
be served in an hour if the demand is high in both travel directions. Rozic (1992) concluded
that the ideal capacity of rural TLTWHW is 4,000 pcph total in both directions. Harwood,
et al. (1999) developed the recent HCM 2000 methodology for TLTWHW. They indicated
that the capacity reported in the HCM 2000 does not represent the maximum volume and
that higher traffic volumes have been observed. They also indicated that capacity
conditions are difficult to observe, because there are very few TLTWHW operating over
capacity.
There are two microscopic simulators for TLTWHW, TWOPAS (TWOlane
PASsing) and TRARR (TRAffic on Rural Roads). These however do not allow traffic
volume to exceed 1,700 vph and they do not provide capacity estimates. In summary,
there has been no research to estimate or measure capacity reflecting various traffic and
geometric design characteristics for TLTWHW.
TWOSIM algorithms
M3A arrival headway distribution
According to Akcelik and Chung (2003), the M3A arrival headway distribution has been
found more realistic and its use is strongly recommended instead of the shifted negative
exponential or the negative exponential distribution. TWOSIM uses the M3A arrival
distribution to generate arrival headways:
F(t) = 1 - ϕ e - λ ( t - Δ ) for t ≥ Δ
(1)
=0 for t ≤ Δ
where,
t = headway (sec);
3
TRB 2007 Annual Meeting CD-ROM Original paper submittal - not revised by author.
Δ = minimum arrival (intra-bunch) headway (sec);
ϕ = proportion of free (unbunched) vehicles: and
λ = ϕ qt / (1 − Δ qt ) (a decay parameter);
where qt is the total arrival flow rate in all lanes of the traffic stream
(veh/sec)
⎧ ⎫
⎪vn (t ) + 2.5a nτ (1 − vn (t ) / Vn )(0.025 + vn (t ) / Vn )
1/ 2
⎪
⎪ ⎪
vn (t + τ ) = min⎨ ⎬ (2)
⎪ ⎪
⎩ ( [[ ] ( ) ])
⎪bnτ + bn2τ 2 − bn 2 x n −1 (t ) − sn −1 − x n (t ) − vn (t )τ − v n −1 t 2 / b$ ⎪
⎭
where,
an = the maximum acceleration which the driver of vehicle n wishes to undertake
(ft/s2);
bn = the most severe braking that the driver of vehicle n wishes to undertake (ft/s2);
sn = the effective size of vehicle, n, that is, the physical length plus a margin into
which the following vehicle is not willing to intrude, even when at rest (ft);
Vn = the speed at which the driver of vehicle n wishes to travel (ft/s);
xn = the location of the front of the vehicle n at time t (ft);
vn (t) = the speed of vehicle n at time t (ft/s);
τ = the apparent reaction time, a constant for all vehicles (0.68 sec); and
b̂ = the estimate of the most severe braking deceleration rate, bn-1 (ft/s2)
The first speed is the speed that the driver desires to achieve under free flow conditions.
The driver can accelerate up to his/her desired speed. The second speed is the speed that the
driver achieves under congested flow conditions, and it is primarily a function of the safety
margin.
Maximum acceleration rate of passenger cars was obtained from a website
(http://web.missouri.edu/~apcb20/times.html), which provides maximum acceleration rates
of all passenger cars in 1970 ~ 1999. Maximum acceleration rates in the 1990s were
selected and applied to implement Gipps’ car-following model in TWOSIM.
Gipps’ car-following model calculates the most severe braking deceleration rate by
multiplying the maximum acceleration rate by -2. Since the maximum acceleration rates of
passenger cars were modified from the original Gipps model, the most severe braking
deceleration rate is also updated to reflect the enhanced performance of newer passenger
cars.
4
TRB 2007 Annual Meeting CD-ROM Original paper submittal - not revised by author.
Gipps’ car-following model uses a further safety margin by adopting an additional
delay of safety reaction time when a driver reacts to the vehicle ahead. The parameter adds
additional delay on the reaction of the average driver. According to Green (2000), it is
shown that drivers’ reaction time ranges from 1.25 seconds with normal and common
expectancy to 1.5 seconds with a surprise intrusion. The delay of safety reaction time is
assumed to range from 0.78 to 1.14, which results in reaction times from 1.25 to 1.5
seconds. TWOSIM assigns the parameter to each driver stochastically.
5
TRB 2007 Annual Meeting CD-ROM Original paper submittal - not revised by author.
Figure 1 Vehicle trajectories at 800 vph arrival rate
6
TRB 2007 Annual Meeting CD-ROM Original paper submittal - not revised by author.
CAPACITY ESTIMATES
Determination of capacity
TWOSIM provides five minute traffic volumes at each observation point over the entire
segment, and calculates hourly flow rates. Capacity is the average of the maximum flow
rates when the arrival demand exceeds the throughput. Multiple runs (more than 30 runs)
were conducted to determine the capacity for each study configuration.
Estimation of capacity
Capacities for TLTWHW are estimated under a variety of conditions. This section provides
a description of those conditions and a summary of the findings.
Base conditions
First, a basic tangent section at level terrain is simulated to estimate capacities under base
conditions, which are as follows:
• All passenger cars
• No stop sign, signals, or driveways
• 2 mile length of level terrain section
• No passing allowed
The base capacity is found to range from 1,835 pcph to 2,141 pcph as a function of
the average free flow speed (Table 1). The capacity increases by about 150 pcph for every
10 mph between 40 and 60 mph speeds, however there is a slight drop (approximately 50
pcph) between 60 and 70 mph. The changes in capacity as a function of average free flow
speed may be the result of using Gipp’s car-following model, in which operating speed
associated with the average free flow speed influences the average headway at capacity.
These results are the average of 30 runs at each average free flow speed.
Statistical testing (using the t-test) showed that the base capacity at each average
free flow speed is statistically different.
7
TRB 2007 Annual Meeting CD-ROM Original paper submittal - not revised by author.
high traffic flow in the primary direction does not have many large gaps, and thus there are
few opportunities for passing.
Presence of driveway
When a driveway is present under base conditions, the capacity is estimated as a function of
curb radius and the percentage of turning flow exiting the main road. The capacity was
found to increase with curb radius and to decrease with percent of turning flow (Table 2).
T-test results showed that there is no significant difference in capacity as a function of curb
radius except when the turning flow reaches 20%. The slightly higher capacities for longer
curb radii may be due to the higher right-turning speed. There is however significant
decrease in capacities as the percentage of turning flow increases. In summary, the presence
of a driveway has significant impact (12 ~ 26 percent reduction) on the capacity of
TLTWHW.
Capacity (pcph)
Curb radius (ft) Turning flow (%)
0 10 20
20 2,141 1,850 1,580
30 2,141 1,890 1,620
45 2,141 1,890 1,660
8
TRB 2007 Annual Meeting CD-ROM Original paper submittal - not revised by author.
Presence of upgrade section
For the estimation of capacity at an upgrade section, it was assumed that there is a 1 mile
upgrade section of three different grades (4 ~ 8 percent) with trucks (10 ~ 30 percent).
Table 4 presents the estimated capacity as a function of grade and percentage of trucks. It is
found that there is significant capacity reduction (11 ~ 40 percent) on an upgrade,
particularly for high truck percentages.
Table 4 Capacity as a function of the percentage of trucks at upgrade section
Capacity (vph)
Grade (%)
10 percent trucks 20 percent trucks 30 percent trucks
4 1,870 1,740 1,690
6 1,670 1,530 1,480
8 1,460 1,330 1,250
Sensitivity analysis
Sensitivity analysis was next performed to examine the impact of various simulation
parameters on capacity estimation.
Next, the impacts of the delay of safety reaction time (theta), one of the parameters
in Gipps’ car-following model, on capacity are examined. When the delay of safety
reaction time is at 0.07, 0.78, and 1.14, the actual reaction time becomes 0.75 seconds for
the best driver, 1.25 seconds for normal and common expectancy used in the earlier results,
and 1.5 seconds for surprise intrusions, respectively.
Table 6 shows how capacity varies with different delay of safety reaction time of
drivers at an average desired speed 60 mph. For comparison purposes the standard
9
TRB 2007 Annual Meeting CD-ROM Original paper submittal - not revised by author.
deviation for all scenarios shown in this table is zero, i.e., all drivers have the same reaction
time. T-test results showed that there is significant difference in capacity when the delay of
safety reaction for all drivers changes.
Table 6 Capacity vs. delay of safety reaction time
Delay of safety reaction time (Perception reaction time, sec)
Capacity (pcph) 0.07 (0.75) 0.78 (1.25) 1.14 (1.5)
2,330 2,294 1,988
Geometric characteristics
When a horizontal curve with 500 feet radius is present under the base conditions, the
capacity is found to increase with the maximum superelevation rate of the curve, as shown
in Table 7. It is found that there is significant increase in capacity when the maximum
superelevation rate of the horizontal curve increases. The greater superelevation rate allows
higher vehicle speeds and leads to higher capacities.
Table 7 Capacity as a function of maximum superelevation rate of horizontal curve
Table 8 shows the capacities estimated with different percentages of trucks along a
horizontal curve. It is found that there is significant difference in capacities when the radius
of the horizontal curve varies in combination with varying percent of trucks. As expected,
the capacities decrease with increasing percent of trucks for any given radius of the
horizontal curve. The capacities increase with the radius of horizontal curve for a given
percent of trucks.
Table 8 Capacity as a function of radius of horizontal curve and percentage of trucks
Capacity (vph)
Curve radius (ft)
0 percent trucks 10 percent trucks 20 percent trucks
100 1,750 1,670 1,570
500 2,100 1,990 1,870
1000 2,100 2,050 1,940
10
TRB 2007 Annual Meeting CD-ROM Original paper submittal - not revised by author.
Traffic characteristics
It is next examined how capacity varies with the opposing flow rate. Table 9 shows the
capacities for varying opposing flow rates, assuming 20 percent no-passing zones. As
shown, capacity is not significantly affected by different amounts of opposing flow rates.
200 2,103
400 2,096
2,300 600 2,091
800 2,073
1,000 2,066
Table 10 shows the capacities with varying percent of trucks at a section with a
driveway. T-test results showed that capacity decreases with increasing percent of trucks
when a driveway is present. At zero percent of trucks, the two curb radii give the same
capacity. When the percent of trucks increases, the capacities between the two curb radii
are slightly different.
Table 11 shows the capacities estimated for different percents of trucks along a 1
mile upgrade section. T-test results showed that grade has significant impact on capacity.
Also, the capacity decreases with an increasing percent of trucks.
Capacity (vph)
Slope (%)
10 percent trucks 20 percent trucks 30 percent trucks
4 1,870 1,740 1,690
6 1,670 1,530 1,480
8 1,460 1,330 1,250
11
TRB 2007 Annual Meeting CD-ROM Original paper submittal - not revised by author.
In summary, it is found that capacity is influenced by driver characteristics such as
standard deviation of desired speed and delay of safety reaction time. In terms of geometric
factors, while the length of the segment and the presence of passing zones do not have an
impact on capacity, capacity is significantly influenced by other factors such as maximum
superelevation in the horizontal curve, driveway curb radius,, and slope of the upgrade
section. The opposing volume was not found to have an impact on capacity. On the other
hand, there is significant capacity decrease as the percentage of trucks increases,
particularly in the presence of a horizontal curve, driveway, and upgrade.
12
TRB 2007 Annual Meeting CD-ROM Original paper submittal - not revised by author.
upgrade section at 4 to 8 percent slope, capacity is likely to decrease by 11 to 40
percent when the percent of trucks is 0 to 20 respectively.
• The TWOSIM algorithms should be enhanced so that the package can estimate
capacity reflecting the presence of multiple driveways, multiple horizontal curves
and grades.
• It is recommended that the car-following model be improved with the development
of a model reflecting the behavior of a passenger car driver around large trucks. The
uncomfortable feeling of a passenger car driver due to lack of sight distance and due
to the truck’s slow speed particularly on upgrade sections may have an additional
adverse impact on capacity.
• It is recommended that field data for investigating trucks’ exit speed and right-turn
speed around a driveway be collected and a truck speed model along a horizontal
curve be developed.
13
TRB 2007 Annual Meeting CD-ROM Original paper submittal - not revised by author.
REFRENCES
1. Akcelik, R. and Chung, E. Calibration of the bunched exponential distribution of
arrival headways. Akcelik & Associates Pty Ltd. June 2003.
2. A Policy on Geometric Design of Highways and Streets. American Association of
State Highway and Transportation Officials, 2004.
3. Bonneson, J. A. Delay to major-stree through vehicles due to right turn activity.
Transportation Research. –A, Vol. 32, No. 2, pp. 139-148, 1998.
4. Bonneson, J. A. Side friction and speed as controls for horizontal curve design.
Journal of Transportation Engineering. November, 1999.
5. Federal Highway Administration. Manual on Uniform Traffic Controls Devices.
2003.
6. Gipps, P. G. A behavioral car-following model for computer simulation. In
Transportation Research B. vol. 15B, 1981, pp. 105-111.
7. Green, M. ” How long does it take to stop?” Methodological Analysis of Driver
Perception-Brake Times. Transportation Human Factors, 2(3), 195-216. 2000.
8. Hammond, H. F. and Sorenson, L. J. (ed.). Traffic Engineering Handbook. New
York: Institute of Traffic Engineers and National Conservation Bureau, 1941.
9. Harwood, D. W, May, A. D, Anderson, I. B, Leiman, L, and Archilla, A.R.
Capacity and Quality of Service of Two-Lane Highways. NCHRP Final Report 3-
55(3). Midwest Research Institute, University of California-Berkeley, 1999.
10. Kim, J and Elefteriadou, L. Estimation of capacity for two-lane two-way highways
using simulation. 2006. Thesis.
11. Lan, C. and Menendez, M. Truck speed profile models for critical length of grade.
Journal of Transportation Engineering. ASCE/July/August. 2003.
12. Leiman, L., Archilla, A. R., and May, A. D. TWOPAS Model Improvements.
University of California Berkeley, California. NCHRP Project 3-55(3). 1998.
13. Lerner, N., Llaneras, R., Smiley A., Hanscom F. Comprehensive Human Factors
Guidelines for Road Systems. National Cooperative Highway Research Program.
March, 2005.
14. McLean, J. R. Two-Lane Highway Traffic Operations: Theory and Practice. New
York: Gordon and Breach Science Publishers. 1989.
15. Normann, O. K. Results of highway capacity studies. Public Roads. vol. 23, no. 4,
pp. 57-81. 1942.
16. Polus, A., Livneh, M., and Frischer, B. Evaluation of the passing process on two-
lane rural highways. Transportation Research Records No. 1701, pp. 53-60. 2000.
17. Rozic, P. Capacity of two-lane, two-way rural highways: the new approach.
Transportation Research Records 1365, pp. 19-29. 1992.
18. Transportation Research Board 2000. Highway Capacity Manual. Washington D.C.
National Research Council. HCM 2000, metric units.
19. Wardrop, J. G. Some theoretical aspects of road traffic research. In Proceedings of
the Institution of Civil Engineers, vol. 1, 1952. pp. 325-378.
20. Yagar, S. Capacities for two-lane highways. In Australian Road Research, 13(1).
1983.
14
TRB 2007 Annual Meeting CD-ROM Original paper submittal - not revised by author.