Low-emission Commuting with Micro Public Transport: Investigation of

Travel Times and CO

Emissions

Marcel Ciesla

, Victoria Oberascher

, Sven Eder

, Stefan Kirchweger

, Wolfgang E. Baaske

and Gerald Ostermayer

Research Group Networks and Mobility, University of Applied Sciences Upper Austria, Austria

Studienzentrum f

ur internationale Analysen (STUDIA), Schlierbach, Austria

https:// nemo.f h-hagenberg.at/ ,

https:// www.studia-austria.com/

Keywords:

Micro Public Transport System, Microscopic Trafﬁc Simulation, Sustainable Mobility.

Abstract:

The omnipresent trend towards sustainable mobility is a major challenge, especially for commuters in rural

areas. The use of micro public transport systems is expected to signiﬁcantly reduce pollutant emissions, as

several commuters travel the ﬁrst mile together with a single pick-up bus instead of their own car. In this paper,

different aspects of such a micro public transport system are analyzed. The main ﬁndings of the investigations

should be how the travel times of commuters change and how many CO

emissions can be saved if some of

the commuters use public transport instead of their own vehicle.

1 INTRODUCTION

Mobility in rural areas is strongly inﬂuenced by indi-

vidual trafﬁc. Public transport is usually insufﬁcient

to meet commuter demands. The disadvantages for

commuters are that public transport is mostly based

on ﬁxed stops, which means that they have to cover

a way to a bus stop ﬁrst. In addition, schedules of

buses and trains are usually predeﬁned, which means

that individual commuters’ needs may not be met.

There are two different options for commuters to get

to work. On the one hand, they can cover the en-

tire route by public transport. Since stops such as bus

stops are in many cases too far away to be reached on

foot or by bike, the ﬁrst distance to a train station can

be covered with an own vehicle. This creates the so-

called ﬁrst-and-last-mile problem. However, since

this is still too much of a hassle for many commuters,

they drive the entire distance to work with their own

vehicle. For the region, this means very heavy trafﬁc

at peak times on the commuter routes. This ultimately

leads to trafﬁc jams, time delays and increased carbon

dioxide emissions. These problems can be counter-

acted by using a suitable micro public transport sys-

tem for the commuters’ ﬁrst mile. People who want to

get to work are collected by one or multiple buses and

brought to a train station. Afterwards, they can use a

train to get to their place of work. It is expected that

with such a micro public transport system emissions

can be reduced and fewer trafﬁc jams occur due to the

reduced commuter routes.

This paper deals with an approach to investigate

the travel times and carbon dioxide emissions of a mi-

cro public transport system mentioned before. The re-

mainder of this paper is organized as follows. At the

beginning it is explained how the examinations were

carried out. Afterwards, existing literature is reviewed

which serves as the basis for our work. The next sec-

tion describes the simulation setup which was used to

perform simulations for different scenarios. The re-

sults of these simulations are presented in the follow-

ing section. This section points out whether a micro

public transport system is suitable for sparsely pop-

ulated areas. In addition, a comparison between in-

dividual transport and micro public transport is made,

which shows how much CO

can be saved when com-

muters use a micro public transport system.

2 METHODS

Existing micro public transport systems cannot sat-

isfy the dynamic requirements of commuters and are

therefore only hardly accepted. Therefore, the main

objective of the project EBIM-

OV (”Low-emission

commuting with intelligent micro public transport”)

was to investigate a micro public transport system for

Ciesla, M., Oberascher, V., Eder, S., Kirchweger, S., Baaske, W. and Ostermayer, G.

Low-emission Commuting with Micro Public Transport: Investigation of Travel Times and CO2 Emissions.

DOI: 10.5220/0011103500003191

In Proceedings of the 8th International Conference on Vehicle Technology and Intelligent Transport Systems (VEHITS 2022), pages 143-151

ISBN: 978-989-758-573-9; ISSN: 2184-495X

143

a sparsely populated area. An important aspect of this

project was to ﬁgure out the beneﬁts of micro pub-

lic transport, where commuters share a bus with other

commuters to overcome the ﬁrst mile to a train sta-

tion. Another focus was to ﬁnd out to what extent

commuters’ travel times change using micro public

transport, which has a strong inﬂuence on user accep-

tance. The base goals of the project can be divided

into following sub-goals:

• Mathematical Modelling of the Optimization

Problem: A sub-goal of the project was the real-

istic mathematical modelling of the optimization

problem for the control of a ride service. The

problem is to pick up customers from any loca-

tion and bring them to a given destination, which

is a train station. The optimization task consists

in falling below a travel time that is acceptable for

the user under very dynamic conditions. These

conditions include different trafﬁc situations and a

varying number of customers with different pick-

up points who are to be served by the system. The

departure times of the desired train line play a par-

ticularly important role, as these times are deci-

sive for when the commuters should arrive at the

station.

• Route Control: With the help of mathematical

modelling of the problem, an algorithm was de-

veloped that dynamically controls a vehicle ﬂeet.

This vehicle ﬂeet consists of several buses with

predeﬁned capacities that are intended to satisfy

the various pick-up requests. The task of the al-

gorithm is to choose optimal routes and departure

times of the buses so that the optimization criteria

are met as good as possible.

• Trafﬁc Simulation: Microscopic trafﬁc simula-

tions are used to validate the developed control

algorithm. The focus is on determining the poten-

tial for saving carbon dioxide emissions and com-

paring travel times between individual transport

and a micro public transport system for different

scenarios.

Through the above-mentioned sub-goals, a system

was created with the goal of simulating an intelligent

micro public transport system. With the help of mi-

croscopic trafﬁc simulations, scenarios were exam-

ined in which commuters get to their workplaces in-

dividually with their own vehicle and with the micro

public transport system. The mathematical modelling

of the optimization problem ensures that the routes of

the pick-up buses are optimally chosen so that their

total travel time is minimized and commuters’ indi-

vidual time constraints are satisﬁed. This has the ef-

fect of increasing commuter acceptance of a micro

public transport system. When designing the simu-

lations, the focus was placed on one train line per sce-

nario. This means that only those commuters who

want to catch the same train will be considered. It

is not possible for a pick-up bus to serve commuters

with different preferred train lines, as the different

train lines are usually quite far apart in relation to

the departure times and the commuters’ travel times

would therefore suffer as a result.

3 RELATED WORK

The ideal chemical reaction of fuel combustion cham-

ber produces carbon dioxide and water and, as a result

of the transient combustion process, additional com-

bustion products. Consequently, fuel design holds a

big potential for further improvements in reduction of

unwanted combustion products (

Uberall et al., 2015).

However, as long as one is not willing to completely

change the method of transportation using internal

combustion engines, carbon dioxide (CO

) emission

is always unavoidable. As a consequence CO

takes

about 65% of the total greenhouse gas emissions,

where the transportation sector currently contributes

20 -25% of global CO

emissions with its global share

estimated to rise to 30-50% by 2050 (Yang et al.,

2019).

At ﬁrst sight one might think that alternative fuel

vehicles such as all-electric and fuel cell vehicles will

be the best solution to reduce CO

emissions. These

fall into the category of cleaner vehicle strategies re-

ducing emission rates per vehicle-kilometre. How-

ever, according to (Litman, 2017), efﬁcient and alter-

native fuel vehicles only provide a few beneﬁts, and

by increasing total vehicle travel tend to exacerbate

problems such as congestion, accidents and sprawl.

Moreover one should be aware of the fact that for such

vehicles one has always to take into account the full

life time cycle (manufacturing, use of the vehicle, end

of life, and recycling) (Mar

ın and De Miguel Perales,

2021). Mobility management, i.e. strategies which

reduce total vehicle travel, provides far more beneﬁts.

Eco-driving, a term used for driving assistance

techniques that support the driver in optimizing route

choice and driving behavior, is a cleaner vehicle strat-

egy which is suggested in (Engelmann et al., 2020). In

their study they incorporate a vehicle dynamic based

emission model and a Pareto-optimal based rout-

ing approach and discuss the beneﬁt trade-off be-

tween travel time and emission in a simulation study.

Similarly (Engelmann et al., 2020) discusses emission

optimized routes in terms of NO

using the Graph-

Hopper API and OpenStreetMap. In all evaluated

VEHITS 2022 - 8th International Conference on Vehicle Technology and Intelligent Transport Systems

144

cases (Engelmann et al., 2020) there was no NO

optimized route found for which the estimated travel

time is less than with a speed-optimized route calcu-

lation and this is a general observation in eco-driving.

Shifting travelers’ travel mode from the private car

to public transport is another effective method of re-

ducing CO

emissions and easing trafﬁc congestion

(Yoshida and Harata, 1996) belonging to the category

of mobility management. This was investigated in

(Yang et al., 2019) (Li and Tamura, 2003) for CO

emissions produced by commuters. To avoid an in-

crease in commuting CO

emissions in Chinese cities,

car use restrictions and transit priorities are the most

important trafﬁc demand management measures to be

considered (Yang et al., 2019). Moreover (Li and

Tamura, 2003) describes a CO

emission forecasting

model for estimating the amount of CO

emissions

due to urban commute travel and analyses the effect of

two policy changes, cutting down public transport fee

and decreasing in-vehicle time, to shift commuters’

travel mode from private car to public transport.

Another way as is done in this work is to provide

additional service feeder buses which operate on de-

mand and try to address as many commuters as pos-

sible by using time optimal routes under real trafﬁc

conditions respecting the individual in-vehicle time

restrictions of the commuters since commuters tend

to take time optimal routes and not emission optimal

ones from eco-driving.

4 SIMULATION SETUP

The whole simulation process is based on the inter-

action between two independent simulation entities.

On the one hand, a trafﬁc simulator called TraffSim

(Backfrieder et al., 2013) (Backfrieder et al., 2014)

has been extended and adjusted to the requirements of

this project. This trafﬁc simulator is a powerful tool

which is able to integrate road networks from Open-

StreetMap data and to perform trafﬁc simulations in a

microscopic way. For this project, TraffSim has been

used to analyze a micro public transport system in a

certain region under real trafﬁc conditions.

On the other hand, a linear optimizer has been

developed, which deals with a Dynamic Dial-a-Ride

Problem. This optimizer receives current state infor-

mation from TraffSim about routing costs and pick-up

requests. The problem solver computes the optimal

amount of pick-up buses and their starting times from

the bus depot. Furthermore, the order of the pick-up

requests is determined in which the requests are to

be processed by the respective bus. These results are

then used by TraffSim to simulate realistic scenarios

Figure 1: Interaction between TraffSim and Route Opti-

mizer.

for the micro public transport system.

As illustrated in Figure 1, these simulation entities

interact with each other so that the results of the opti-

mizer are applied to a scenario in the trafﬁc simulator.

Data regarding trafﬁc volumes and commuters have

been provided by the project partner STUDIA. How

this data was collected and processed for TraffSim

scenarios is described in the next sections.

4.1 TraffSim

The microscopic trafﬁc simulator project TraffSim

has been enhanced to simulate the micro public trans-

port system described in this paper. The following

sections show how the different input types of the

simulator have been conﬁgured.

4.1.1 Study Region

The study region consists of the municipalities Kirch-

dorf, Micheldorf, Schlierbach and Inzersdorf in the

South of Upper Austria. There is one main train line

going toward the Upper Austrian capital Linz and a

motorway where you can head to Salzburg, Passau

and also Linz. The region is rather rural, with the

provincial town of Kirchdorf. A lot of people ﬁnd

work within the region but a big punch of people must

commute out, in particular to Linz and between. In or-

der to do our analyses we split the region in a raster

grid with 250 by 250 meter grid cells. The used data

and the analyses of the spatial distribution of roads

and railway as well as trafﬁc and commuters is de-

scribed in the following sections.

4.1.2 Road and Rail Network

To build up a road network which can be used by

TraffSim, data from OpenStreetMap (OSM) has been

processed. OSM data contains information about

roads and rails including their geometries, speed lim-

its and lanes as well as junctions between the road

segments. The trafﬁc simulator uses a library called

osm2po to create a routable directed graph from this

Low-emission Commuting with Micro Public Transport: Investigation of Travel Times and CO2 Emissions

145

Figure 2: Road and rail network.

data, which can be used afterwards to determine

routes and travel costs. Figure 2 shows the processed

road and rail network in TraffSim which represents

the mentioned study region. In the left part of the ﬁg-

ure, the whole region including the commuter routes

on the motorway and train rails can be seen. The blue

box shows a more detailed section of the road net-

work, where the blue icon represents a pick-up point.

From such a point, customers are picked up by a pick-

up bus and are delivered to a train station. Further-

more, the blue box contains a train station visualised

by a yellow icon on the train tracks and the corre-

sponding bus stop also represented by a yellow icon

on the road.

4.1.3 Trafﬁc Situation

In order to obtain meaningful results in terms of travel

times and emissions, a realistic mapping of the traf-

ﬁc was an important aspect of the project. Especially

in the morning, when commuter routes are heavily

used, there can be trafﬁc jams and delays. In order

to be able to simulate this morning trafﬁc we iden-

tiﬁed all geographical points where vehicles can en-

ter (starting or entering point) or leave (stopping or

leaving point) the simulation area in TraffSim. For

each of these points the number of cars are calculated

using rasterized data of principal residence within

the study region, provided by Statistik Austria (Aus-

tria, 2016), the Upper Austrian trafﬁc census, pro-

vided by the provincial government of Upper Aus-

trian (Amt der OOe. Landesregierung, 2012) as well

as a rasterized freely accessible land use plan of Up-

per Austria. Whereas the trafﬁc census data gives us

the municipality of origin and destination, the hour

of departure as well as the used means of transport,

the raster data and a randomization algorithm allows

us to distribute the trafﬁc spatially explicit within the

region. Afterwards, this data is diluted with the coor-

dinates of the enter and leave points. This results in a

source-destination matrices for every departure hour

consisting of each enter and leave point. These ma-

trices were then mapped onto vehicles and routes in

TraffSim.

With regard to the microscopic modelling of each

individual car, the following assumptions were made.

The Intelligent Driver Model was used as the longitu-

dinal model. More detailed descriptions of the model

and its mathematical basis can be found in (Treiber

et al., 2000). In order to calculate the fuel consump-

tion and, as a result, the CO

emissions of vehicles,

the physics-based consumption model from Treiber

and Kesting was used. When using this model, it was

assumed that each vehicle has a mass of 1500 kg and

is powered by a diesel engine with a power of 90 kW.

More detailed information on the consumption model

can be found in (Treiber and Kesting, 2013) .

4.1.4 Commuters

We use a rasterized dataset of commuters, provided

by Statistik Austria (Austria, 2016) and the Upper

Austrian trafﬁc census, provided by the provincial

government of Upper Austrian (Amt der OOe. Lan-

desregierung, 2012) in order to identify commuters

place of origin (grid cell), their chosen train number

and their destination train station. Our investigations

focus on commuters who travel by car or train to-

wards the Upper Austrian capital Linz between 4:00

and 10:00 in the morning. A randomization algorithm

is used to select those commuters who possibly use

a micro public transport. As this information is so

far only available on grid cells level, realistic pick-

up points are achieved by distributing the commuters

uniformly within the grid cells. Then, a set of multiple

pick-up points was deﬁned for the entire area, which

the bus will use to pick up the commuters. The entire

commuter locations were then assigned to the closest

pick-up point. The Euclidean distance was used as the

metric for the assignment. After the data was grouped

according to the desired train line, the result was a set

of pick-up points with assigned commuters, which

are also called pick-up requests. If a pick-up point

has not any requests assigned, it is simply ignored for

this train line and will not be approached by the pick-

up bus.

4.1.5 Train Schedule

For the implementation of the EBIM-

OV scenarios,

trains and timetables also had to be implemented.

Therefore, we used real timetables from the Austrian

Federal Railways (

OBB) (

OBB, n.d.). From this train

schedule, all relevant train stations from Micheldorf

to Linz Central Station have been included in the

TraffSim scenarios. By inserting the real train sched-

ule, it is guaranteed that the commuter train times cor-

respond to reality.

VEHITS 2022 - 8th International Conference on Vehicle Technology and Intelligent Transport Systems

146

4.1.6 Pick-up Buses

The route optimizer (see 4.2) provides the start time

for each pick-up bus and the order in which the pick-

up requests are to be picked up. All buses start at a

bus depot near the train stop in Kirchdorf. As soon

as the bus leaves the depot, the micro public trans-

port system calculates expected arrival times at the

pick-up points and train stations where the requests

are delivered. These times are based on the routing

costs of the current bus route, which originate from

the graph. These times are updated at regular intervals

(e.g. 60 seconds) and communicated to the users. The

ﬁrst promised pick-up time for the requests plays an

important role. Since the users of the EBIM-

OV ori-

ent themselves at this point in time, the ﬁrst promised

pick-up time is also used to calculate the travel times.

This means that the travel time with the micro public

transport system results from the difference between

the ﬁrst promised pick-up time and the arrival time

with the train at the respective destination train sta-

tion. In addition, the ﬁrst promised pick-up time is

also important for the pick-up process itself. If the

bus arrives at a pick-up point before the ﬁrst promised

pick-up time has passed, it must wait until then before

it is allowed to continue. With the pick-up buses it

should be noted that the standing times at a pick-up

point or a train station depends on the number of pas-

sengers who want to get on or off at this stop. The

standing time results from the sum of a constant value

(12 seconds) and a factor of 6 seconds per passenger

who gets on or off. For example, if 3 people want

to board at a pick-up point, the bus will stop for 30

seconds.

As with normal trafﬁc (see 4.1.3), the Intelligent

Driver Model was used as a longitudinal model for the

buses. As far as fuel consumption is concerned, the

physics-based consumption model from Driver and

Kesting was chosen again, but with different parame-

ters. A Mercedes-Benz Sprinter Transfer 45 with 22

passenger seats and an engine of 105 kW served as

the basis for selecting the parameters (Mercedes-benz

Sprinter Transfe, 2022).

4.1.7 Comparison between Micro Public

Transport and Individual Trafﬁc

One of the main aims of the project was to ﬁnd out

how much CO

emissions can be saved with a micro

public transport system and how travel times change

compared to individual transport. To achieve this,

several simulation scenarios were created in which

every commuter drives with his own car to his or

her desired destination train station. For this purpose,

the places of origin and destination stations from the

Figure 3: Measuring points for CO

evaluation.

commuter data (see 4.1.4) were used again to generate

an individual vehicle for each commuter.

In order to make a meaningful comparison be-

tween micro public transport and individual transport,

certain measures had to be taken. On the one hand,

the departure times of the commuters had to be cho-

sen so that they arrive at the desired destination sta-

tion at around the same time as the train of the re-

spective train line itself. This ensures that the high

trafﬁc volume on the commuter routes is realistically

reproduced and that the same trafﬁc conditions for the

comparison are given. Another aspect was that there

was not always a corresponding motorway exit on the

tracks for each destination train station. The com-

muter routes in the road network only consist of the

motorway and the train tracks (see 4.1.2). As illus-

trated in Figure 3, a suitable measuring point on the

motorway was deﬁned for each destination train sta-

tion on the rails. As soon as a commuter who would

like to reach a speciﬁc station by train crosses this

point with his car, the vehicle’s statistics are saved at

this point in time. These statistics include the time

of arrival at the measuring point and vehicle-relevant

values such as CO

emissions and consumed fuel.

4.2 Route Optimizer

A customer request is deﬁned by a number of persons

who want to reach a certain train line in time. Thus

the target train station is not ﬁxed at the outset. More-

over, in time means that a person is not only able to

reach the train but also his travel time should not dif-

fer too much from the travel time experienced with

his own car. The task is to ﬁnd bus routes for a ﬂeet

of buses with given capacity which minimize the total

travel time of the buses where each bus picks up the

customers from predeﬁned locations such that at the

end the customers are satisﬁed. Since the departure

Low-emission Commuting with Micro Public Transport: Investigation of Travel Times and CO2 Emissions

147

times of the busses is determined by the optimizer,

the time window for the emergence of the pick-up re-

quests ends with the execution of the optimization.

To solve this problem we mapped it to the Dial

a Ride problem (DaRP) using its 2-index formula-

tion as given in (Ropke et al., 2007). Since this is

not straightforward to solve by general purpose op-

timization software we developed a modiﬁcation of

it which used only a subset of the precedence con-

straints. The route optimizer itself is a Python module

able to communicate with TraffSim which solves the

DaRP in its mixed integer linear programming prob-

lem (MIP) formulation using the Python-MIP Pack-

age (Python-MIP, n.d.) together with the MIP solver

Gurobi (Gurobi, n.d).

We applied several heuristics to simplify and solve

the MIP problem. First we deﬁne the train station

which can be reached in the shortest time from the

pickup location as the target train station, to avoid the

solution of several DaRP problems. Only in the case

of a small deviation in travel time to another train

station, which was about 1.5 minutes in our setting,

we checked if better results are obtained by using this

train station as target. Here small has to be seen in re-

lation to the acceptance times of the customers. These

were modeled by the shortest time needed to reach the

target train station plus some customer speciﬁc delay,

which is at least as large as the time needed to walk

from the bus station to the target train station to en-

ter the train. Then the acceptance time constraint is

not satisﬁed if the difference of train arrival time and

promised pick-up time is greater than the acceptance

time. Observe that this delay has to be increased until

the ﬂeet size of the solution is equal to the required

ﬂeet size.

Secondly, pick-up requests having the same

pickup point and the same target train station are

mapped to a virtual customer request where the num-

ber of persons equals the sum of persons from each

request and the acceptance time is the minimum of

acceptance times for each request.

Thirdly, we applied a clustering of requests with

respect to the train line and their target train station,

because simulations without clustering showed that

the solutions get clustered in exactly this way. The

main reason for this seems to be the fact that the dif-

ference in train arrival times at the train stations are

so small that the buses are not able to visit more than

one train station without violating some of the con-

straints. This in turn allows us to use only a subset

of precedence constraints as long as the pickup nodes

are uniquely mapped to one bus. This was always the

case in our simulations. Though it can happen that the

delivery train station of the request is mapped to the

wrong bus we can correct this by exchanging requests

since in the cluster the target train station remains the

same.

5 SIMULATION RESULTS

For this project, only outbound trips were consid-

ered, i.e. those where commuters travel to their work-

place with the micro public transport system. The

return trip was not considered in the course of this

project. For the comparison between individual trans-

port and EBIM-

OV, different scenarios were simu-

lated in which the number of requests varied. Sev-

eral scenarios were taken into account, which con-

sisted of 35, 50, 65 and 80 pick-up requests. Some

of these requests could not achieve the desired accep-

tance due to unacceptable travel times. This is be-

cause these requests could not be optimally integrated

into the route of a bus because of their geographi-

cal location. Pick-up requests with an unacceptable

travel time were therefore ﬁltered out from the results

of these simulations. For this purpose, two threshold

values were deﬁned for the ratio of travel time with

micro public transport to travel time with individual

transport. The threshold values selected were 1.7 and

1.85, which means that with these limits all pick-up

requests are selected for which the travel time with

micro public transport is 70% or 85% longer than with

one’s own car. After the ﬁltering process, further sim-

ulations were carried out in which only these pick-up

requests were taken into account.

It was found out that the scenario with 80 pick

up requests included 50 within the 85% threshold,

whereas 37 requests were within the 70% threshold

concerning the additional travel time with the micro

public transport system. For these two subsets of re-

quests, simulations were performed again and the re-

sults of these two simulation scenarios are presented

in the next sections. Although the scenario with the

50 requests builds on the other, it should be noted

that both are viewed as separate simulations. This

means that, despite similar requirements for the pick-

up buses, there can be differences in the calculated

routes. Due to a random factor in the geographical

distribution of commuters, it cannot be guaranteed

that the same commuter will be picked up from the

same pick-up point in both scenarios.

5.1 Travel Time Differences

The following two histograms in ﬁg. 4 show the dif-

ference between commuter travel time with the micro

public transport and with the commuter’s own vehi-

VEHITS 2022 - 8th International Conference on Vehicle Technology and Intelligent Transport Systems

148

cle. The illustration on the right shows that 2 of the 50

requests with the micro public transport system were

able to reach their destination station faster or almost

as quickly as compared to their own car. This is be-

cause these two requests were quite a short distance

from the train station to which the bus brought them.

In addition, the bus picked them up at the end, which

brought them straight to the train station without any

detours. However, the commuters with the micro pub-

lic transport system needed on average 14.1 minutes

longer for the 37 requests and 14.6 minutes longer for

the 50 requests than with the individual vehicle.

Figure 4: Travel time differences.

Since there is a connection between the travel

time differences and the destination train stations, the

travel time ratio of micro public transport to individ-

ual trafﬁc grouped by the different destination sta-

tions is shown in ﬁg. 5. The destination stations that

are close to the source region (Schlierbach, Nuss-

bach, Wartberg, Kremsmuenster and Rohr-Bad Hall)

are subject to greater ﬂuctuations in a direct compari-

son. The spatial distribution of commuters within the

source region shows a high impact at these destina-

tion stations. Since the commuter travel time in their

own car is usually very short at these train stations,

even small time deviations due to the micro public

transport system have a strong impact on the result.

At stations that are further away from the source re-

gion (Traun and Linz), the spatial distribution of the

pick-up requests plays a lesser role. Since the travel

time ratios are almost identical for the 37 and 50 re-

quests, it can be concluded that, due to the long total

travel time, it does not matter where the commuters

come from. Furthermore, it can be assumed that in

the case of destination stations near the source area,

the comparison is characterized by the longer travel

time of the pick-up buses and in the case of more dis-

tant stations by the longer travel time of the trains. For

the 50 pick-up requests, of which 12 had the destina-

tion station Linz, the ﬁgure shows a ratio of 1.5. This

means that these 12 commuters took an average of

50% longer to get to their work place with the micro

public transport system than with their own vehicle.

Figure 5: Travel time differences grouped by target train

stations.

5.2 Carbon Dioxide Emissions

With the help of the fuel consumption models imple-

mented in TraffSim, it was possible to measure the

footprint of every commuter when they commute

to their workplace in their own vehicle. These values

were required in order to be able to set up a CO

com-

parison between individual transport and the micro

public transport system. The CO

emissions caused

by the micro public transport system are made up of

the emissions from the pick-up buses and those of the

respective train. The CO

consumption of the pick-

up buses was taken from the statistics of the trafﬁc

simulations. However, no model is implemented in

TraffSim that can determine the CO

consumption of

trains. The source for the CO

footprint of passen-

ger trains was the sustainability report of the Austrian

Federal Railways (

OBB) from 2019, which deﬁnes an

emission of 8.2 grams of CO

per person and kilome-

tre. This information was used to determine the re-

spective CO

emissions by train for each commuter,

depending on the source and destination station. As

a result, the entire CO

footprint of the micro pub-

Low-emission Commuting with Micro Public Transport: Investigation of Travel Times and CO2 Emissions

149

lic transport system could be calculated. In the sim-

ulation scenario with 37 requests, individual trafﬁc

resulted in CO

emissions of 154.7 kg. The micro

public transport system, which serves the same com-

muters, causes a total of 18.1 kg of emissions, which

results in a saving of 88.3%. In the second scenario

with 50 commuters, 219.6 kg of CO

were emitted by

one’s own car and 20.8 kg of CO

by the micro public

transport system. Here the savings potential is 90.5%.

Table 1: Detailed CO

footprint micro public transport sys-

tem.

37 Requests 50 Requests

Carbon footprint buses 12.3 kg 12.1 kg

Carbon footprint train 5.8 kg 8.7 kg

Total carbon footprint 18.1 kg 20.8 kg

5.3 Further Findings

5.3.1 Number of Buses and Their Occupancies

In the scenario with 37 requests and also in the sce-

nario with 50 requests, three pick-up buses were

needed. Each of these pick-up buses had a maximum

capacity of 22 seats and the occupancies of the indi-

vidual buses are shown in the following illustration in

ﬁg. 6.

Figure 6: Bus occupancies.

5.3.2 Distances to Pick-up Points

The following histograms in ﬁg. 7 show the distances

that commuters have to cover from their place of ori-

gin to the assigned pick-up point. On average, the

distance in the scenario with 37 requests is 258.4 me-

ters and in the scenario with 50 requests 255.9 meters.

As can be seen in both histograms, the commuters

contain an outlier with a distance of 1259.9 meters,

for whom there was no suitable pick-up point in his

vicinity.

Figure 7: Distances to pick-up points.

6 CONCLUSION

In the course of the ”EBIM-

OV” project, different as-

pects of a micro public transport system were ana-

lyzed using a microscopic trafﬁc simulator. By simu-

lating commuter scenarios, in which commuters com-

mute to work on the one hand with their own vehi-

cle and on the other hand with a micro public trans-

port system consisting of pick-up buses and trains, a

comparison between the two types of mobility could

be drawn. In summary, it can be said that a micro

public transport system consisting of 3 pick-up buses

with 22 seats each could ﬁnd acceptance for the sim-

ulated test area. According to the simulations carried

out, commuters have to accept that they have to walk

about 257 meters to a pick-up point. Depending on

the destination train station of the commuter, there is

an increase in commuter travel time of 38% to 95%

compared to the travel time with one’s own car. It

should be noted that the comparison for nearby desti-

nation stations depends heavily on the places of origin

of the commuters. At the destination train station in

Linz, which is furthest away from the source area, the

travel time increased by 48% to 51%. The geographi-

cal distribution of the places of origin is less important

here, as the travel time is much longer and is largely

determined by the train. Provided that out of 50 com-

muters all commute with the micro public transport

system instead of their own car, it can be said that

over 90% of CO

emissions can be saved.

ACKNOWLEDGMENT

This project has been co-ﬁnanced by the European

Union using ﬁnancial means of the European Re-

gional Development Fund (EFRE). Further informa-

tion to IWB/EFRE is available at www.efre.gv.at.

VEHITS 2022 - 8th International Conference on Vehicle Technology and Intelligent Transport Systems

150

REFERENCES

Amt der OOe. Landesregierung, Direktion Strassenbau

und Verkehr, A. G. u. o. V. (2012). Daten zur

ober

osterreichischen Verkehrserhebung 2012.

Austria, S. (2016). Daten auf Basis der Regionalstatistis-

chen Rastereinheiten: Zahl der Personen mit Haupt-

wohnsitz und Paket Pendler.

Backfrieder, C., Mecklenbr

auker, C., and Ostermayer, G.

(2013). Traffsim – a trafﬁc simulator for investigating

beneﬁts ensuing from intelligent trafﬁc management.

pages 451–456.

Backfrieder, C., Ostermayer, G., and Mecklenbr

auker, C.

(2014). Traffsim - a trafﬁc simulator for investigations

of congestion minimization through dynamic vehicle

rerouting. International Journal of Simulation Sys-

tems Science & Technology, 15:38–47.

Uberall, A., Otte, R., Eilts, P., and Krahl, J. (2015). A liter-

ature research about particle emissions from engines

with direct gasoline injection and the potential to re-

duce these emissions. Fuel, 147:203–207.

Engelmann, M., Schulze, P., and Wittmann, J. (2020).

Emission-based routing using the graphhopper api

and openstreetmap. In Schaldach, R., Simon, K.-

H., Weism

uller, J., and Wohlgemuth, V., editors, Ad-

vances and New Trends in Environmental Informatics,

pages 91–104, Cham. Springer International Publish-

ing.

Li, B. and Tamura, H. (2003). Estimation of a reduction

in co2 emissions by shifting commuters’ travel mode

from the private car to public transport. International

Journal of Systems Science, 34(3):159–165.

Litman, T. (2017). Smart transportation emission reduction

strategies. Technical report, Victoria Transport Policy

Institute.

Mar

ın, P. F. and De Miguel Perales, C. (2021). Environ-

mental Aspects of the Electric Vehicle, pages 93–108.

Springer International Publishing, Cham.

Mercedes-benz Sprinter Transfer 45. [Online]. Available:

https://www.mercedes-benz-bus.com/enDE/brand/n

ews/2019/minibus-offspring-sprinter-transfer-45-and

-sprinter-city-45.html

Ropke, S., Cordeau, J.-F., and Laporte, G. (2007). Models

and branch-and-cut algorithms for pickup and delivery

problems with time windows. Networks, 49(4):258–

272.

Treiber, M., Hennecke, A., and Helbing, D. (2000). Con-

gested trafﬁc states in empirical observations and mi-

croscopic simulations. Physical Review E, 62:1805–

1824.

Treiber, M. and Kesting, A. (2013). Trafﬁc Flow Dynamics.

Yang, L., Wang, Y., Han, S., and Liu, Y. (2019). Ur-

ban transport carbon dioxide (co2) emissions by com-

muters in rapidly developing cities: The comparative

study of beijing and xi’an in china. Transportation

Research Part D: Transport and Environment, 68:65–

83. Urbanization, Transportation and Air Quality in

Developing Countries.

Yoshida, A. and Harata, N. (1996). A mixed-mode choice

model including railway route, stations, and its access

and egress modes choice. Doboku Gakkai Ronbunshu,

1996(542):19–31.

Obb train schedule pyhrnbahn 140. [Online]. Available:

https://www.oebb.at/de/dam/jcr:0b159f98-8e3b-

4b03-b4a9-007212675433/kif140.pdf

Python-mip package. [Online]. Available:

https://www.python-mip.com

Gurobi. [Online]. Available: https://www.gurobi.com

Low-emission Commuting with Micro Public Transport: Investigation of Travel Times and CO2 Emissions

151