Analysis of traffic flow due to lane changes by heavy vehicles

The transportation of goods via heavy vehicles (HV) plays a crucial role in urban economics. However, the operational and physical characteristics of these vehicles negatively impact traffic flow, leading to reduced speeds and increased congestion. This study examines the influence of HV lane changes on average traffic speed by implementing diverse lane change intensities and alternative strategies that restrict HV to specific lanes, narrowing their lane-changing space. Five HV classes based on axle count were analyzed using Vissim traffic simulation software on a 600 m JORR toll road segment. Three strategies were assessed across various scenarios, with differing HV composition percentages. Results show that HV lane changes, simulated using Vissim, significantly affect average traffic speed. Comparing high and low-intensity lane changes revealed an approximately 12% speed increase. Additionally, every 5% increment in HV composition led to an average 3% traffic speed reduction. In high-density conditions for every 5% increase in the composition of HV from high-intensity compared to medium and low-intensity lane changes, the average speed of traffic flow increases by 4% and 8% in low-density conditions. This research highlights the importance of enforcing designated lane usage for HV on toll roads to enhance traffic flow performance.


Introduction
Cities face various challenges due to rapid population growth and increasing lifestyles which have an impact on the distribution of goods [1].Data from the Central Statistics Agency 2022 shows that Indonesia's population reaches 275.77 million people.In fulfilling its needs for goods, Supply Chain Indonesia 2022 said that HV or trucks dominate the mode of transportation of goods.This can be seen from the high growth in the transportation and warehousing sector in early 2022 which was accompanied by a decrease in the number of goods transported by sea and rail transportation modes.This statement is supported by data from the Central Statistics Agency 2021 regarding the contribution of road transportation modes to the GDP of the transportation subsector, which reached 69.8%.In addition, the Indonesian Transportation Society 2022 also predicts that heavy vehicles will dominate up to 90% of freight transport in Indonesia by 2022.
In its operations on the Indonesia HV toll road it has been regulated to be in lane 1 or the left lane which is intended for low-speed vehicles.This is contained in Government Regulation No. 15 of 2005.However, the phenomenon that occurs is the Jakarta Outer Ring Road (JORR) experiencing a decrease in the average speed of traffic flow because heavy vehicles pass not on their lanes during peak hours.
Many researchers are interested in studying traffic flow from various aspects, one of which is lane changing.In the last 60 years, a number of studies have been conducted to investigate lane-shifting behavior [2].Lane change maneuvers have negative implications, namely 5% of traffic accidents in 2015 in China were caused by inappropriate lane changes.Then, the higher the frequency of lane changing behavior increases the probability and rate of traffic accidents [14].Previous research has also provided macroscopic evidence that congestion can occur and cause a decrease in capacity due to the high intensity of lane change rates.One study found that the frequent act of changing lanes by heavy vehicles can have a negative impact on traffic performance [3].An increased frequency of lane changes can also increase density and reduce the average speed of traffic flow [4].Understanding lane change behavior is important in several application areas such as capacity analysis and safety studies [11].Approximately 539,000 two-vehicle lane change collisions occurred in the US in 1999 [7].Therefore at this time, research will be carried out with a microscopic model using the help of the PTV Vissim application to analyze the effect of lane changing by heavy vehicles on the average speed of traffic flow on the Jakarta Outer Ring Road segment.Previously there had been research that conducted lane change modeling, but no one had conducted further research on the impact of lane changes by heavy vehicles specifically on toll road traffic flow.This research can also be evidence of the need for vehicle compliance in using the appropriate lanes on the toll road.Seeing these conditions, research is needed to analyze traffic flow due to lane changes by heavy vehicles.

Literature Review
The number of heavy vehicles and their proportion in traffic flow has increased in the last decade which has had a significant impact on traffic flow.This is one of the causes of traffic congestion due to the size and slower operation of heavy vehicles compared to light vehicles [5].Especially if heavy vehicles change lanes, this will have an impact on the surrounding traffic conditions as well.The existence of traffic congestion can cause negative impacts such as increasing vehicle queues, reducing the average speed of vehicles, longer travel times, economic losses, emissions etc. [6].Lane change is one of the most common driver habits in traffic flow.In the last 60 years, a number of studies have been conducted to investigate this lane-shifting behavior [2].Previous studies have been conducted with different focuses.[9] analyzed the duration of lane changes in heavy vehicles and passenger vehicles.While [7] analyzed the differences between the behavior of heavy and light vehicle drivers in changing lanes on toll and arterial roads.Then [10] investigated the effect of microscopic characteristics and analyzed the decision-making model by changing the lanes of heavy vehicles.[8], [11], [12], [13] conducted several studies on lane changes aiming to investigate the decision-making process of lane change manoeuvres for various drivers based on a two-stage test-drive, investigate the impact of a lane change strategy on expressway traffic operations, and analyze the behavior of new follower vehicles changing lanes in the target lane when facing anticipation (pre-insertion process).The research conducted has its own variables.Toledo and Zohar (2007) focus on the duration of lane changes for heavy and light vehicles.Then [7] observed the width of the gap and the length of the vehicle.[5] focused on the motivation to change lanes, the selection of the target lane, and the decision to change lanes.Whereas [8], [11], [12], [13] focuses on strategies for changing lanes, time, distance, speed, density, and gap width.
Research locations regarding lane changes also vary.[9] located on the California highway.Other studies were conducted on the Berkeley and Hollywood motorways [10],the motorway between Delft and Rotterdam [11], and urban roads in Colombo Sri Lanka [6].The types of vehicles observed are not the same in all studies.Research by [12] does not include heavy vehicles as its object.Meanwhile [9], [7], [10] made heavy vehicles the object of their research.
In conducting their research, there are researchers who use software assistance to model the simulation.using the help of the MOTUS software [12] and used PTV Vissim software [10].In this study, the author will use PTV Vissim software.The other researchers used the help of PTV Vissim software to model 2, 3, and 4 lane freeway traffic flow with the assumption of 10% heavy vehicle volume to analyze the parameters that affect the distribution of lane flow on multilane toll roads.With the history of previous research regarding lane changes, of course there are still many areas that have not been focused.Therefore, this study will analyze traffic flow due to lane changes by heavy vehicles with direct observation and microscopic simulation on PTV Vissim software.This research also focuses more on heavy vehicles passing on the Jakarta Outer Ring Road toll road segment to determine the average speed of traffic flow caused by lane changes (LC) by heavy vehicles.Modeling lane changes by heavy vehicles is very important for safety studies and capacity analysis.This is done so that toll roads can increase their capacity and traffic flow safety.

Study Methodology
The main goal of this paper is to investigate the impact of the intensity of lane change on traffic flow characteristics.To this end, the strategy-based lane change model has been implemented in a microscopic simulator.Three different simulation scenarios have been defined.In scenarios (1) with highest intensity lane changes by HV, which is no HV restrictions on all lanes.In scenarios (2) with moderate intensity lane changes, which there is one lane that limits HV from passing the lane.Scenario (3) with lowest intensity lane changes, which there are two lanes that limits HV for passing the lanes.Lane change by HV scenario strategy in Vissim.the data validation method in this modeling uses the GEH method (Geoffrey E. Havers).

Study Area for Model Development
This research was conducted on the Jakarta Outer Ring Road section of the TB Simatupang toll road.Video recording of the road segment with a length of 600 m through the AD Premier Office building.The survey was conducted for 24 hours by placing cameras at two adjacent points.Data 4 processing is carried out under two conditions, namely during high and low-density traffic flows.This aims to obtain a better analysis.

Vissim Model Development
This research was conducted by microscopic simulation with the help of Vissim software.Before running the model, it is necessary to have input parameters so that the simulation results can run properly.In short, the parameters that need to be set to run the simulation model are as follows.
• Model map settings

Data Input Preparation for Calibration
The data used in the calibration model input is when high-density traffic flows around 03.00 PM and low density around 06.00 AM.The data used are vehicle volume, vehicle relative flow, and vehicle speed distribution.The data is provided according to the grouping of vehicles.The following is an example of the Vissim model input data tabulation.

Trial and Error Model Calibration and Process Validation
The calibration model that has been prepared according to the input data for the calibration model is then run to get the output data.Running is carried out for a duration of 1 hour according to the data used as input.For each data generated by the Vissim running program, a suitability test was conducted to test the suitability of the Vissim output data with the actual data at the research location.The test was carried out using the Geoffrey E. Havers method.Previously, driving behavior adjustments were made by referring to existing research on vehicle characteristics in Indonesia.The results of this validation become a consideration for model acceptance for feasibility in the next simulation scenario process.The smaller the GEH value and below the number five, the more accurate the model is with the real conditions at the research location.

Simulation Results
From the results of processing simulated data with scenarios that have been determined in Vissim, an analysis was carried out using the linear regression method.For the distribution of the average speed of traffic flow in Figures 7 and 8, where the graph of the relationship between average speed and lane change by HV and the composition of HV in high and low-density conditions, the simulation model has a linear trendline with an R² value close to 1.0 indicating that the calibration results are suitable.The R² values (coefficient of determination) for scenario groups A, B, and C for high-density traffic flow are 0.9129, 0.9386 and 0.6164, respectively.Meanwhile, the R² values for scenario groups A, B, and C for low-density traffic flow were 0.9659, 0.9561, and 0.9817, respectively.This linear line can be used as a reference to find out how much deviation there is between the observed results and the existing simulation model.
From the linear regression line, the trend line equation is obtained.Where there are parameters that represent the slope of the line.The three lines of the high-density graph have the same slope direction, namely from the top left to the bottom right which indicates that the greater the composition of heavy vehicles in the traffic flow, the lower the average speed of all vehicles.In scenario group A, the parameter value or slope is -19,792.while in scenario group B the slope value is -16,131.then for  From the graph of high-density traffic flow conditions, the gradient value can also be analyzed that in conditions of high-intensity lane changes, the intercept value is 27,914.In a lane change condition with moderate intensity the intercept value is 27.82, while in a lane change condition with low intensity, the intercept value is 27.144.This value indicates that in the condition that the %HV is zero, the average speed of the traffic flow corresponds to the intercept, which in this condition is an average of 27 km/hour.
Whereas in conditions of low-density traffic flow, the gradient value can also be analyzed that in conditions of high-intensity lane changes the intercept value is 54,813.In a lane change condition with moderate intensity the intercept value is 54.78, while in a lane change condition with low intensity, the intercept value is 55.385.This value indicates that in the condition that the %HV is zero, the average speed of the traffic flow corresponds to the intercept, which in this condition is an average of 54 km/hour.In high-density traffic flow conditions scenario A, the average vehicle speed reaches 22 km/hour, then in scenario B, the average vehicle speed is 23 km/hour, and in scenario C, the average vehicle speed increases to 25 km/hour.The average traffic speed increases by 6% in each lane change scenario with less intensity.From the table data, it can also be analyzed that for every 5% increase in KB composition from high-intensity lane changes compared to medium and low-intensity lane changes, in high-density conditions the average speed of traffic flow increases by 4%.
In low-density traffic flow conditions scenario A, the average vehicle speed reaches 43 km/hour, then in scenario B, the average vehicle speed is 46 km/hour, and in scenario C, the average vehicle speed increases to 49 km/hour.The average traffic speed increases by 6% in each lane change scenario with less intensity.From the table data, it can also be analyzed that for every 5% increase in the composition of KB from high-intensity lane changes compared to medium and low-intensity lane changes, in highdensity conditions the average speed of traffic flow increases by 8%.

Contextualization of Research Results
Research that has been conducted to determine the impact of lane changing by heavy vehicles using various strategies for limiting heavy vehicle classes and the composition of heavy vehicles on traffic flows shows that in conditions of high-density traffic flow, the average speed increases by 4% in each lane changing scenario which is getting lower.Then for the addition of the composition of heavy vehicles, the speed of traffic flow decreases by an average of 2% for every 5% addition of heavy vehicles.Meanwhile, in conditions of low-density traffic flow, the average speed has increased by 8% also for each lane-changing scenario which is getting lower.Then for the addition of the composition of heavy vehicles, the speed of traffic flow decreases by an average of 3% for every 5% addition of heavy vehicles.
The results of this study are in line with research [3] where the behavior of changing lanes by heavy vehicles frequently or at high intensity can have a negative impact on traffic performance, one of which is the average speed of traffic flow.Research [4] also supports the results of this study, where the difference in average speed of traffic flow between vehicles changing lanes and vehicles without changing lanes decreased from 51 km/h to 35 km/h due to congestion worsening.A study [15] in Iran showed that increasing the composition of heavy vehicles reduced the average traffic speed by 15% at a composition of 40% HV.Meanwhile, in this study, an increase in the composition of heavy vehicles reduced the average traffic speed by up to 25% at a composition of 40% HV.This can happen because of the different driving behavior between countries.It was proven in research [15] that the average desired speed was 164.5 km/hour, whereas in this study only the maximum value only reached 80 km/hour.Then the distance vehicle in research [15] averaged 1.39 m.While in this study it was 0.6 m.

Conclusions
Lane changing by heavy vehicles has an impact on traffic flow performance, namely the average vehicle speed. in traffic flow with high and low density, the average speed of traffic flow increases by ±12%.The presence of an increased composition of heavy vehicles in the traffic flow can make the impact of changing lanes by heavy vehicles worse because the average speed of vehicles decreases compared to the composition of small heavy vehicles.The addition of the HV composition causes a significant reduction in traffic speed with an average of ±3% for every 5% addition of the HV composition.In high and low-density conditions, the difference is that for every 5% increase in the composition of HV from high-intensity lane changes compared to medium and low-intensity lane changes, the average speed of traffic flow increases by 4% in high-density conditions and 8% in lowdensity conditions.

Figure 2 .
Figure 2. Overall situation of study area.Figure 3. Situation at point 1 observation.

Figure 3 .
Figure 2. Overall situation of study area.Figure 3. Situation at point 1 observation.

Figure 5 .
Figure 5.Effect from intensity lane changes by HV and HV composition plots for the three simulations scenarios in high density traffic flow.

Figure 6 .
Figure 6.Effect from intensity lane changes by HV and HV composition plots for the three simulations scenarios in low density traffic flow.

Table 1 .
Data collecting in high density traffic flow.

Table 2 .
Data collecting in low-density traffic flow.

Table 4 .
Category class, type, 3D model, and length of the vehicle in the simulation.

Table 6 .
Example of vehicle volume data input on the Vissim model.

Table 7 .
Trial and error experiment.

Table 8 .
Results of the Vissim program simulation model validation.