Simulation of the impact of tram operations on road network performances

The City government of Semarang has a plan to build a tram way in Semarang Old City Area as a public transportation mode to accommodate tourism activities. The existence of tram way can impact the performance of existing road network due to being built on existing roads side by side with other vehicles. Street network performances discussed in this paper are on the vehicle-kilometres travelled (VKT), and vehicle-hours travelled (VHT), to determine the aforementioned changes, traffic simulation was done with PTV VISSIM software. Vehicle-kilometres travelled decreased by 12,76%, vehicle-hours travelled increased by 10,54%. To minimize the impacts, two mitigation scenarios are simulated, mitigation 1 which is one-way system and mitigation 2 the combination of one-way system and signal optimizations. In mitigation scenario 1 the value of VKT decreased by 13,12%, VHT decreased by 0,65% compared to existing conditions. In mitigation scenario 2 the value of VKT decreased by 17,54%, VHT decreased by 4,93%.


Introduction
Semarang is the capital of Central Java Province in Indonesia.Situated on the economically significant pantura area (Java island northern coast) that is connected by railways and highways to various cities, and was a significant Dutch colonial city pre-independence, Semarang attracts both economic and tourism interests.One of the tourism attraction in Semarang is the Kota Lama area, it was the city center in the 18th to 19th century AD [1].
The City Government of Semarang suggests to build a tram way to accommodate transportation needs.Tram is defined as a light rail system on level with existing roadways [2].This makes tram existence impacts the performance of existing roadways, with one study in Surabaya, Indonesia concludes that tram existence can raise degree of saturation (DS) value of Darmo street above 0.75, which, according to Indonesian Highway Capacity Manual is considered a traffic jam [3] This paper examines the effects of tram existence on the performance of road networks, and proposes possible mitigation scenarios to mitigate the effects, with the approach of comparing the change in value of Vehicle-kilometers Traveled (VKT) and Vehicle-hours Traveled (VHT).The changes in VKT and VHT values are acquired from the process of traffic microsimulation using PTV VISSIM software.

Existing Researches on Tram Operations Impacts
There has been a number of studies exploring various aspects of tram railways system, for example a study done by Naznin explores the safety aspects of tram operations [4], however most of the studies were done outside Indonesia, since Indonesia currently does not have any active tram railways system.An example of a study in Surabaya, Indonesia concludes that the existence of tram railways causes potential traffic jam [3], this study was done with the scope of one road only using degree of saturation (DS) value as its performance indicator, which is commonly used in Indonesia to determine the performance of a single road or intersection.
Most studies on tram operations impacts are done with focus only on the impacts on a road link or corridor, with no studies assessing the network-wide effect of the existence of a tram network in Indonesia.An example of a network-wide study through simulation with PTV VISSIM was done in Melbourne, Australia where volume of vehicles reduced by 19.6% and travel time increases by 14.7% [5], however this study did not explore any possible mitigation scenarios.

Traffic Simulation
For traffic performance studies with the scope of a single road or intersection, determining the performance is possible using manual calculation methods, the existence of computer softwares only serve to simplify the processes, however to determine the performance of a whole road network that consists of multiple roads and intersection, with the whole trip path of all vehicles passing through the road network cannot be determined with current survey methods, the existence of traffic simulation softwares becomes a necessity.Traffic simulation software can also be used to model different alternative scenarios of traffic adjustments before it is implemented, predictions of behavior and performance changes due to changes on the roads and intersections [6]

Data Gathering Process
Secondary data used are existing road geometry data, tram rail and car specifications, and tram route plan provided by the City Government of Semarang.Primary data used are signal configurations, vehicle running speed, and traffic volume gathered by drone on observed roads and intersections.Vehicles are classified to three types according to Indonesian Highway Capacity Manual, which are Motorcycle (MC), Light Vehicle (LV), and Heavy Vehicle (HV).Signal configurations on each signalled intersections were counted manually by stopwatch.

Simulation Modelling
To create the model necessary for simulation in PTV VISSIM, following data have to be inputted: •

Code Road Name Traffic
It is assumed that no transportation mode changes happen with the existence of tram, due to the planned tram being purposed for tourism purposes.

Tram modelling.
PTV VISSIM software has a default 3D model for trams which numbers of cars can be configured, dimensions are assumed to be close enough to the planned specifications, gauge width is currently undetermined since Indonesia typically uses 1067 mm cape gauge for metro trains, long-distance trains, and Jakarta Mass Rapid Transit (MRT) system, but implementations in Indonesian High Speed Railways and Jakarta Light Rail Train (LRT) uses 1435 mm standard.For the planned implementations in Semarang, the tram model uses 2 car configuration and takes 3900 mm clearance width on the road [7].6,50187 The last calibrated case is not considered a good model according to criteria in Table 2, this is due to the data gathering process of the traffic volume not done at the exact same time in all data gathering points.

Scenarios Modelled
The scenarios modelled in this paper are : • Existing conditions • With tram conditions • Mitigation 1 (One way system) • Mitigation 2 (One way system + signal optimizations) 6 3.4.1.One way system.One way system is one of the possible ways to mitigate the impacts of tram operations on road network performance, one way system enables more vehicles to enter the road network, or enables vehicles to travel in higher speeds due to less congestions, thus reducing travel times and hence, Vehicle Hours Traveled (VHT).

Vehicle Kilometers Traveled (VKT) and Vehicle Hours Traveled (VHT)
VKT (or VMT, Vehicle Miles Traveled in United States) is a metric used extensively in transportation planning for a variety of purposes.It is a measurement of the amount of vehicular travel in a geographic region over a given period of time.It refers to the total number of kilometers traveled by all vehicles over a given period of time.VKT data are used primarily by transportation agencies, environmental agencies, and consultants to perform a variety of functions such as allocating resources, estimating vehicle emissions, computing energy consumption, and assessing traffic impacts.VKT may also be used to evaluate conformity assumptions, adjust travel demand forecasts, and identify pavement maintenance needs.It is a key metric in transportation planning because it provides an indication of travel demand and behavior.Additionally, policy decisions and infrastructure investment planning incorporate VKT data [9].
Factors that influence VKT value includes route choice and traffic volume.VKT is calculated using the formula shown in equation 2: Vehicle Hours Traveled (VHT) provides a proxy measure of the overall pressure on a road network due to the congestion impacts of regional travel [10].VHT can be used to analyze economic impacts due to changes on road networks, specifically on trip time value analysis.VHT is calculated using the formula shown in equation 3: VHT is tied to the travel time needed by vehicles to complete a trip, which is influenced by: • Delays (due to stops in intersections, traffic jam) • Vehicle speed • Driving behavior and interactions between vehicles

VKT Comparison
The VKT calculations results of the scenarios simulated in this paper is shown in table 5 and figure 6 below :  By only looking at the VKT value changes, in the "with tram" scenario, assuming route choices of the vehicles are unchanged and thus, the travel lengths are also unchanged, this means there are less vehicle able to pass the measurement points at one time in the simulation run, however to determine whether this means the traffic condition is better or worse, VHT value results have to be considered.

VHT Comparison
The VHT calculations results of the scenarios simulated in this paper is shown in table 6 and figure 7 below :  VKT interpretation is tied to the VHT results.In the "with tram" scenario, despite the reduction in VKT value, the VHT value raises up, this proves that the reduction in volume does not happen due to reduction in vehicle use but due to worse traffic conditions, since even with lower traffic volume, the accumulated delay is higher.

Conclusions and Suggestions
This paper concludes that the operation of tram on existing roads can negatively impact the performance of road networks.In the proposed Semarang Tram Railways plan, it reduces VKT by 12,76% and increases VHT by 10,54%.Proposed mitigation scenario of one way system successfully reduces the VHT even slightly lower than in existing conditions, however the VKT value does not change significantly.Combination of one way system and signal optimizations further reduces VHT value to be 4,93% lower than existing conditions, however the VKT value decreases even further to be 17,58% below the existing conditions.One way system is recommended to be applied in some roads if the Semarang Tram Railways plan is approved to be built.
It is suggested that to create a more representative modelling, that in further researches, potential mode choice changes have to be considered with the existence of tram as a new mode choice, more suitable survey methods also have to be considered for gathering multiple data at multiple points at the same time to make the simulation model more representative.
Existing road and tram way geometric data • Traffic volume • Vehicle routing decisions • Vehicle speed • Signal heads and configurations on intersections • Driving behavior • Vehicle travel time measurement points In each run of simulations, PTV VISSIM will simulate the inputted vehicles on the modelled road network, and measure the number of vehicles passing through the assigned vehicle travel time measurement points with the travel length and time taken.

Figure 1 .
Figure 1.Vehicle Travel Time Measurement Points.

Figure 2 .
Figure 2. Tram Model Input The planned tram implementations follows the traffic rules and signals, therefore the tram lane is not exclusive to tram only, except in certain parts where the tram traffic goes against road vehicle traffic.Input of tram line uses the "Public Transport Line input" of PTV VISSIM shown in Figure 3 below.

Figure 3 .
Figure 3. Tram Line Input 3.3.Model Validation To ensure the accuracy of the simulation model towards the inputted traffic volume data in existing conditions, validation was done with Geoffrey E. Havers (GEH) formula.The GEH equation model is a modified statistical formula of Chi-square by combining the difference between relative and absolute values [8].The GEH equation is shown in equation 1 and the meaning of the GEH number value is shown in table 2  = √ (  −  ) 2 0,5 * (  +  )

Figure 4 .Figure 5 .
Figure 4. Location of Proposed One Way System.3.4.2.Signal optimizations.Another way to mitigate the impacts of tram operations is through optimizing signal configurations on signalized intersections.Using the built-in "optimize traffic signals" functions in PTV VISSIM software, the software will determine the new signal configurations by using simulated trial-and-error iterations.Figure5shows the example of pre-optimization and post-optimization signal diagram in KH Agus Salim intersection.

Figure 6 .
Figure 6.VKT Comparison Graph.Since traffic volume is defined as the number of vehicles passing over a point at one time, in a traffic jam the measured number of VKT can be lower than in better traffic conditions, due to less vehicle movements passing through the measurement points at one time, however traffic conditions cannot be determined by only looking at the VKT value without considering other indicators.By only looking at the VKT value changes, in the "with tram" scenario, assuming route choices of the vehicles are unchanged and thus, the travel lengths are also unchanged, this means there are less vehicle able to pass the measurement points at one time in the simulation run, however to determine whether this means the traffic condition is better or worse, VHT value results have to be considered.

Figure 7 .
Figure 7. VHT Comparison Graph.VKT interpretation is tied to the VHT results.In the "with tram" scenario, despite the reduction in VKT value, the VHT value raises up, this proves that the reduction in volume does not happen due to reduction in vehicle use but due to worse traffic conditions, since even with lower traffic volume, the accumulated delay is higher.

Table 1 .
Code Assignment of Vehicle Travel Time Measurement Points.

Table 3 .
Changed Driving Behavior parameters.