2.3.1. Hourly Traffic Volume Estimation
It is not always possible to completely derive the actual traffic volume analytically on different road types and during the different time periods under a given transportation planning scheme. In most cases, it is estimated by traffic simulation software (e.g., TransCAD, Paramics). However, it is only applicable to the situation with detailed origin and destination (OD) and road network data. Obtaining OD data is extremely complicated work and also usually produces large errors than the actual travel demand. Therefore, we attempted to explore an alternative method for estimating hourly traffic volume based on road capacity and transportation planning indicators.
Road capacity is divided into basic traffic capacity (or theoretical capacity), possible traffic capacity, and design traffic capacity. The possible capacity refers to the capacity that takes into account the impact of actual road conditions and traffic conditions, which can be derived from the correction of basic capacity, as shown in Equation (4).
where
represents the one-way possible traffic capacity on a certain road section, usually expressed in passenger car unit (pcu) per hour per lane.
is the basic capacity of a lane, usually we adopt the recommended values of “Urban Road Design Specifications (CJJ 36–90)” according to different road speeds.
denotes the correction factor of multiple-lane, which is used to describe the influence on adjacent lane capacity when overtaking of vehicles in the same direction is performed in the case of multiple lanes.
is the correction factor of lane width, the width of the lane has a great influence on the driving speed, which in turn affects the road capacity.
is the correction factor of mixed environ ment.
represents the correction factor of intersections, which mainly depends on the intersection control method and intersection spacing. When the intersection spacing is small, the parking delay at the intersection accounts for a large proportion of the vehicle travel time, which is not conducive to the utilization of road space.
represents the correction factor of streetization, buildings on both sides of the road often cause interference from pedestrians and non-motorized vehicles to the car, thereby forcing the car to slow down and reduce road capacity. Finally,
is the correction factor of roadside parking. Roadside parking will result in a reduction in the road width and a reduction in lateral clearance. At the same time, the traffic flow will be interfered when a vehicle enters and exits the parking space, which will change the road capacity.
Unlike big cities in China, urban road traffic in counties has its own characteristics. For example, in counties, the lane width is basically less than standard (3.5 m), and the phenomenon of mixed environment of motor vehicle and non-motor vehicle is particularly serious, no matter which type of road. Additionally, due to the improvement of the living standards of the people in the county towns in recent years, vehicle ownership and usage frequency has increased sharply, but the backward parking lot planning has been unable to meet the daily parking demand. Therefore, in the counties of China, the roadside parking phenomenon is particularly serious and is especially obvious on the local roads. The value of the capacity correction coefficient adopted in this paper is shown in
Table 3, referring to some empirical values in “Urban Road Design Specifications (CJJ 36–90)” [
33]. Note that corresponding changes could be conducted according to the actual situation of their own roads in each county.
It should be noted that the theoretical or possible capacity indicates the maximum traffic flow per unit time passing any selected point on the road using all available lanes. It speaks of how much traffic a road can accommodate. For a given road, its possible traffic capacity should be a fixed value, but actual hourly traffic flow will be different, which will result in different GHG emissions at different times of the day. Level of service (LOS) is related to the capacity, which characterizes how good the operational status of the traffic flow is. For urban roads, the most important indicator to measure the LOS is the saturation of the road section (V/C, namely volume to capacity ratio). Therefore, in this study, we adopted the term “V/C” to derive the actual flow at each time period on a given road. For convenience, three typical time periods of the day are considered and are peak, off-peak, and free flow period.
Figure 2 depicts road traffic volume conditions in different time periods in Changxing, Wuan, Qingcheng, and Jintang county. The traffic data were obtained from the bayonet system of the Vehicle Management Office in each county. It can be easily found that there are obvious morning and evening peaks in these four counties. Therefore, according to the travel characteristics presented in
Figure 2, the periods 7:00–9:00 a.m. and 17:00–19:00 p.m. are set as peak hour periods, 22:00 p.m.–6:00 a.m. is selected as a free flow period, and other times are set as off-peak hours.
In China, the LOS is divided into four levels ranging from 1 to 4, and the corresponding recommended value of V/C is presented in
Table 4. Note that the value of V/C is usually obtained through field observations.
To derive the actual traffic volume, it is necessary to know the service level or V/C ratio of different road types at different times. In general, the intention of road traffic planning is to alleviate road traffic congestion and thus improve LOS. All of the planning indicators are usually designed to meet peak-hour demand so that the LOS at the peak hours cannot exceed the third level, whose average V/C is 0.8. Therefore, 0.8 is adopted in this paper for the V/C on arterial roads during peak hours in the planning year. It is assumed that the travel characteristics will not change much in counties if there are no traffic control measures (e.g., vehicle limit). Therefore, the current travel characteristics can be applied to derive the V/C value on other road types and during different times based on the given V/C for arterial roads during peak hours, as shown in Equation (5).
where
and
represents the V/C value and the average traffic volume on road classification
during period
, respectively.
represents the arterial road, collector road, and local road, respectively.
stands for the different hourly periods, i.e., peak, off-peak, and free flow period, respectively. For example,
is the V/C value of arterial road during peak hours.
stands for the possible capacity of road classification
.
Then, we averaged the parameters derived from the traffic volume data of the four counties, and the final results of the V/C value of different road types during different times adopted in this paper are presented in
Table 5.
Note that the unit of the capacity is
instead of
—in order to facilitate the calculation of GHG emissions of different vehicle categories, it is necessary to derive the actual traffic volume for different vehicle categories based on the vehicle conversion factors. In this paper, we adopted the recommended values in “Urban Road Engineering Design Specifications (CJJ37-2012)” [
34]. The conversion factors for LDPV, LDT, and MC are 1.0, 1.2, and 0.4, respectively.
Based on the above analysis, the hourly traffic volume on each classification of urban roads could thus be derived, as shown in Equation (6).
where
and
stand for one-way traffic volume and V/C value on road classification
during period
, respectively.
represents the possible one-way section throughput of road classification
, and
and
refer to the proportion and vehicle conversion factor of vehicle category
, respectively.
Then, hourly vehicle kilometers travel (VKT) can be derived:
where
is the vehicle distance traveled by vehicle category
during period
, and
represents the length of road classification
.
2.3.2. Prediction of Average Speed Distribution on Different Road Types
As mentioned in
Section 2.2, vehicle emissions are closely related to its driving speed. Generally, the lower the speed, the higher the emissions, especially at the idling process, the emissions will increase exponentially. In addition, vehicles driving on different types of roads (with different speed limits or expectations) during different time periods (with different traffic conditions) will produce different speed distributions. In this study, the method of simulation by VISSIM was adopted to obtain the average speed distribution on different road types.
Figure 3 shows three typical average speed distribution at a different desired speeds. The blue, red, and black dotted lines indicate the average speed distributions at the desired speeds of 60 km/h, 50 km/h, and 40 km/h, respectively. Through the analysis of the simulation results, we can find that the vehicle with the average speed in the range of 10–30 km/h will reach the intersection when the signal light is red. On the contrary, the vehicle with the average speed in the interval of 30–60 km/h will pass the intersection during the green light signal. From
Figure 3, it can be clearly seen that the distribution of velocity in the interval of 10–30 km/h is roughly consistent at different expected speeds. For different roads at a different desired speed, the biggest difference in the average speed distribution is the distribution that the vehicle passes through the intersection without idling. Therefore, we must conduct a simulation analysis for different road types under different traffic conditions.
In VISSIM, changing the random seed to the same road segment file may derive different results of each simulation. This is because the arrival rules of vehicles are different if we use a different random seed. Through statistical analysis of average values, it can be more reasonable to reflect the average operating condition of the actual or designed traffic system in a certain period. To eliminate the influence of other random factors on the simulation test results, we used the method of changing the random seed to carry out a simulation test to verify the validity of our simulation results.
Figure 4 depicts the average speed distribution under the simulation of 1 random seed and 99 random seeds, respectively. It can be easily found that changes in random seeds have less effect on the results of vehicle average speed distribution.
We carried out multiple simulations on the arterial road, collector road, and local road according to the estimated traffic volume during peak hours, off-peak periods, and free flow periods, and finally obtained the average speed distribution of different road sections during different time periods, as shown in
Figure 5.