Macroscopic and Microscopic SimulationMacroscopic and Microscopic SimulationMacroscopic and Microscopic SimulationMacroscopic and Microscopic Simulation
for the for the for the for the EEEEvaluation of People Mover Systemvaluation of People Mover Systemvaluation of People Mover Systemvaluation of People Mover Systemssss
Dr.-Ing. Peter Mott
Sven Beller
PTV AG, Karlsruhe, Germany
1. 1. 1. 1. IntroductionIntroductionIntroductionIntroduction
This paper is intended to show how a comprehensive transportation planning
system can be used to analyse the impact of a People Mover or Personal
Rapid Transit (PRT) system on traffic and transportation operations. Both
systems use comparatively small driverless vehicles which are controlled
automatically.
Based on two examples we demonstrate the benefits of macroscopic planning
and microscopic simulation of public transport systems:
In the first example the aim is to choose a system that enables the provision of
efficient transport services in a high density area such as a new development
in West Bay area, Doha, Qatar. An underground system will connect this area
with the other parts of the city. A people mover system is planned to be
integrated into the multi-modal transport system in order to serve the high
density area. According to information on existing and future land use, traffic
demand is generated by office buildings, hotels, a congress centre and a
shopping mall in this area. Due to difficult climate conditions, the footpaths
between these origins of traffic and the people mover stations have to be as
short as possible. An important aspect here is to find out how the smaller
vehicles of the people mover system can handle the expected flow emerging
from the high-capacity underground trains.
The second example models and analyses a demand-responsive transport
system (DRT) which means that vehicles are not assigned to fixed routes and
fixed times. DRT matches the service more closely to the customers' needs:
passengers request a journey as soon as they arrive at the pick-up point and
the vehicles will then take them to the desired destination on the shortest
possible route without any intermediate stops. Small vehicles along with a
simple but effective empty vehicle management provide a highly available
transport service. In most situations there are no waiting times for the
passengers. Other transport systems may also be integrated to form a
comprehensive, city-wide, multi-modal transport model for further analysis –
even including private transport modes such as cars and bikes. For clarity, the
other systems are not shown here.
Both examples are modelled and analysed with the transportation planning
software suite PTV Vision. The macroscopic software tool VISUM is used for
the first application and the microscopic application VISSIM is used for the
second one. Among others, these systems provide the following options:
• Detailed representation of public transport services with line routes, detailed timetables and vehicle types allocated to each line route.
• Demand modelling defined in terms of space and time: zone or station based with a passenger volume distributed by time according to variable intervals, e.g. 15 or 30 seconds, one or several hours.
• The macroscopic analysis is based on assignment methods which take both travel demand distribution in terms of time and capacity of individual public transport systems into account.
• Modelling the interaction between vehicles and passengers as part of a multi-modal microscopic simulation.
• Calculation of a wide range of performance indicators from the passenger's and the operator's point of view, such as travel time and waiting time, number of transfers, passenger volume and volume capacity ratio per link etc.
• Option to export a macroscopic model from VISUM for further (microscopic) analysis to VISSIM, hence providing an almost seamless top-down workflow.
2222. . . . The The The The Macroscopic ViewMacroscopic ViewMacroscopic ViewMacroscopic View
The goal in the first example is to assess whether the transport system can
cope with the predicted passenger volumes, in particular regarding the
transfer between the metro and the people mover system. The traffic volumes
are based on estimated passenger flows during the morning peak period
including a temporal distribution with 30-min intervals. The system is defined
by stations, routes, the service frequency and the vehicle size. This
distribution display is parameterised so that changes can be modelled easily.
Figure 1 shows the layout of a north-south metro line (M1) operating every 10
minutes and two people mover lines (PM Red and PM Blue) that depart every
5 minutes. This service covers a total of six stations.
Figure 1: Layout of a Metro and People Mover network
The travel demand is based on the assumption that there are major passenger
flows from the other parts of the city during the morning peak, lasting about
two hours. Additional demand arises from the passengers travelling within the
area only. Figure 2 illustrates the Origin-Destination matrix based on the
network; Figure 3 shows the relative distribution of the demand over time.
Figure 2: Layout of a Metro and People Mover network, superposed by the structure
of the origin-destination matrix
Figure 3: Distribution of passenger demand during the morning peak period
After defining the timetable and the vehicle types including the total number of
seats, it is possible to automatically calculate the line route capacity and the
volume capacity ratio for each link and time interval. In this particular case, the volume capacity ratio per time interval is an
important factor. An even volume assignment during the morning peak period
does not reveal any problems (see Figure 4).
Figure 4: Volume capacity ratio using even volume assignment during morning peak
However, the analysis which is based on time intervals shows the situation to
be quite different: as expected, the load factor soars and the travel demand is
too high on some routes to accommodate the planned number of passengers,
while there is sufficient capacity on alternative routes (see Figure 5).
Figure 5: Volume capacity ratio per time interval to identify transport capacity
bottlenecks on network sections
The inclusion of the volume capacity ratio in the so-called timetable-based
assignment leads to a completely different result: the impedance takes into
account an additional penalty, depending on the volume capacity ratio per link
and time interval. If the volume capacity ratio per route section exceeds the
defined values the impedance on this route is increased. Hence other
connections - if available – become more attractive to passengers. This
approach can be interpreted as controlled guiding of passenger flows to less
busy routes. Hence it is possible to determine the amount of passengers that
can still be managed and to analyse the preferred services in a more realistic
and reliable manner.
Based on the assignment results and the calculation of performance
indicators, the planning program calculates the following indicators used for
system evaluation:
• Passenger volume and capacity utilisation per route, link and time interval, waiting time per stop and travel time from origin to destination
• Performance like no. of vehicles required, vehicle kilometres etc.
• Estimation of operating costs and expected fare revenues.
System analysis based on macroscopic assignment provides more than only
the "big picture" and allows identifying potential bottlenecks in the system over
time. Even capacity restraints in public transport assignment can be taken into
account. For a more detailed assessment of the various processes
determining the overall system performance the microscopic simulation
comes into play. It can also cope with situations where the demand
significantly exceeds the capacity.
3333. . . . The Microscopic ViewThe Microscopic ViewThe Microscopic ViewThe Microscopic View
In contrast to macroscopic models, microscopic models simulate the
movement and interaction of individual entities.
The software package VISSIM is part of the PTV Vision software suite as well
and provides microscopic simulation methods for assessing and solving a
wide range of transportation problems. The heart of VISSIM includes
scientifically approved models for car-following, lane-changing and pedestrian
movements. Simulations include road users, public transport and pedestrians
and their interactions with each other. All is based on an integrated network
model of roads and rails along with their control methods as well as the
pedestrian infrastructure to connect with them. This multi-modal network is
then used by cars, HGVs, buses, trams and trains as well as cyclists and
pedestrians - only to name a few. During the simulation a wide selection of
evaluations are available for online and offline analysis. Another special
characteristic is an animated visualisation in 2D or 3D which offers an instant
comprehension of the simulated traffic situation and also fills the gap between
the technical expertise and a non-technical audience.
3333....1111. Line. Line. Line. Line----Based OperationBased OperationBased OperationBased Operation
Line-based operation is the typical application of public transport in VISSIM.
The essential input data for the microsimulation model includes
• Road/rail infrastructure: tracks/roads (with length, width, gradient), branches and merges (switches/junctions), stations (with platforms), depots (if required)
• Operational data: line allocation (based on routes on the network), departure times (timetable/frequency), vehicle type and capacity, depot capacity
• Passenger volumes: either the desired route relations (origin-destination matrix) or the line-specific passenger volumes at individual stations, and their temporal distribution. If passenger volumes are not available dwell time distributions may be used instead.
Looking at the first example mentioned above, a detailed representation of the
processes at the transfer station is useful in particular, if the number of
passengers currently waiting at the station exceeds the capacity of the next
vehicle. This may happen as a large number of passengers alight from a
metro train at the interchange and the next people mover vehicles arriving
cannot handle all of them instantly. In such a situation, the microscopic system
allows passengers to enter the vehicle until its capacity is reached. The
remaining passengers wait for the next vehicle.
The network used in the first example is exported from the macroscopic model
to the microscopic simulation tool VISSIM. Some further adjustments ensure
that all data is supplied in order to run the simulation. That means that the first
example encompasses both macroscopic simulation of the total system and
microscopic simulation of the processes at the transfer station in order to
precisely calculate the number of passengers affected by the delay and their
waiting times at the transfer points.
3333....2222. Demand. Demand. Demand. Demand----Based OperationBased OperationBased OperationBased Operation
A major benefit of a PRT system is the "freedom of travel", i.e. that vehicles
don't travel on fixed routes with a timetable, but serve the passengers
individually by providing an on-demand service. In contrast to line-based
operation, this challenges the use of standard microscopic simulation software
as different methods and control operations are required. For these purposes
VISSIM offers an application programming interface (API) which allows
expanding its simulation capabilities beyond what comes directly out of the
box. The API offers a wide range of options on how to implement various
external control strategies: It ranges from handing over parameters up to total
control of vehicle movement and PRT system operation. This approach also
allows for applying complex control strategies modelled by an external
system.
The second example is a case study which shows such a demand-based
operation of a PRT system. It covers an area of 500m by 350m and includes
4.7 km of guideways, 17 branches/switches, 7 stations and 1 depot (see
Figure 6 and Figure 7). The main objective of such a model is to analyse and
visualise the PRT operation for variations in temporal and spatial demand as
well as to assess the impact of different control strategies.
Since the operation of such a demand-based model is much more complex
than a line- and timetable based model, some additional information is
required in order to get a realistic model:
• Structural data: Number of vehicles the station can hold, station type, number of berths for boarding and alighting at each station
• Operational data: Dynamic route assignment (depending on the destination), empty vehicle redistribution, demand-based pre-allocation to stations, re-allocation during empty journey, recharging
• Passenger volumes: max. and min. group sizes and their distribution
Figure 6: Structure of the model for the simulation of a demand-responsive PRT
system with vehicles, stations, switches and the depot (2D view)
Figure 7: Structure of the model for the simulation of a demand-responsive PRT
system with vehicles, stations, switches and the depot (3D view)
The vehicle assignment and main
system control is done by scripting
through the VISSIM API. The script
language used for this example is
Visual Basic for Applications (VBA)
using Microsoft Excel as front end.
Here control buttons for loading the
network, running the simulation and
changing some parameters such as
the total number of vehicles in the
system are provided.
During the simulation, in addition to
the visual animation, several program
indicators are shown in Excel for each
station in order to ensure system
integrity and to help trouble-shooting.
One indicator shows for example the
trip demand at each station and how
many vehicles are available to meet
that demand.
The vehicle movements and
interactions are done automatically in
VISSIM with no need for scripting. It
also includes conflict handling, e.g. at
guideway merges.
Figure 8: Excel front end to control the
simulation
The principle of the simulated process is as follows:
• Initially all PRT vehicles (pods) reside in the depot. Then some vehicles are assigned to each station according to the parameter setting of how many free vehicles should ideally be available at each station.
• Passengers who arrive at a station enter their desired destination at the service terminal and request a pickup. Typically, vehicles are available at each station so that there are no waiting times. If not, a free vehicle is sent to the station.
• The selected vehicle receives the information of the passenger’s desired destination. The passenger transport starts immediately after boarding.
• At each guideway branch the vehicle receives the direction information according to its destination so that the vehicle takes the optimum route to its destination.
• During the trip, the vehicle reacts to the other vehicles in the network by considering speeds and necessary time gaps between the vehicles.
• As the vehicle arrives at the destination it proceeds to the drop-off location for passengers to alight.
• The empty vehicle can then be re-scheduled: If there are less empty vehicles than desired, it will stay at this station. If there are enough vehicles, the first vehicle in the queue will return to the depot empty. It can be re-allocated during the trip if a new request for a ride has been submitted from a station along the route or if a station is in need for more empty vehicles.
The time required for boarding and alighting and the departure times result
from the station layout, the number of passengers, operation processes,
walking speeds of pedestrians and their service time at the terminal, where
they enter their destination. If there are always enough vehicles available and
passengers don't arrive in large numbers at the same time then there is
virtually no delay for them to start travelling.
Figure 9: Layout of a station in 2D view. Passengers alight on the left and board on
the right side of the station. The optional window for each vehicle shows technical
data such as current speed during the simulation.
The station layout will become a major factor as soon as large groups of
passengers arrive at a station in order to depart and/or if many vehicles arrive
here at short intervals. Then a station with multiple berths and/or a sawtooth
layout has the advantage of parallel boarding/alighting.
All these parameters can be varied and combined in several scenarios to be
tested. Consequently, several simulation runs are executed evaluating the
proposed scenarios. During each run user-configurable data is collected which
is used to compare and evaluate the scenarios. Typical performance
indicators are:
• From the passenger's point of view: waiting time at the station and journey time from the starting point to the destination.
• From the operator's point of view: number of required vehicles, vehicle kilometres, empty vehicle movements, dwell time at the depot
As an example, the average waiting time at stations has been evaluated to
see the impact of the number of vehicles available in the system as well as the
impact of empty vehicle redistribution. These scenarios were simulated:
• Scenario 1: 30 vehicles in total, 3 free vehicles at each station
• Scenario 2: 40 vehicles in total, 1 free vehicle at each station
• Scenario 3: 40 vehicles in total, 3 free vehicles at each station
Figure 10 shows the average waiting time at each station. The simulated total
passenger demand was chosen to be below capacity. The differences in time
within each scenario result mainly from the different demands at each station.
Most passengers depart from station no. 6 whereas most passengers arrive at
station no. 1.
From the results it is clearly visible that looking only at the total number of
vehicles is not enough to assess a PRT system. The empty car management
is equally important which can be seen especially for the stations with the
highest demand (4, 6 and 7) where the negative impact is much higher for
scenario 2.
The extra 10 vehicles for scenario 3 help to further reduce the waiting time
compared to scenario 1, but even with the situation shown in scenario 1 the
system can operate well - for some stations (2, 5 and 1) there is virtually no
difference in waiting time as compared to scenario 3.
Figure 10: Average waiting time per station for three scenarios. The time needed for
passenger interaction at the terminal is not included here.
One conclusion of this simulation could be that 25% of the money for rolling
stock could be saved if the empty car management is done well.
Other examples for microsimulation assessments could include longer
simulation periods (e.g. an entire day), consider the battery status of the
vehicles, predict station demand by learning from history, include sawtooth-
shaped stations versus in-line stations and of course different control
strategies for vehicle assignment and empty vehicle re-destribution.
When all the simulation work is completed, the animation may be recorded as
a 3D movie so that the results can be visualised and presented, also to a non-
technical audience.
Figure 11: 3D movie recording of the simulation nearby station 3
Figure 12: 3D movie recording of boarding passengers at station 6
3333.3.3.3.3. Multi. Multi. Multi. Multi----Modal OperationModal OperationModal OperationModal Operation
Another interesting application for simulation is a multi-modal transport model.
It offers the entire range of real-world transport: private traffic such as car and
bike, public transport such as trains, trams, buses and PRT and all linkage
between the different modes, even taking into account movements of
pedestrians.
One way of assessing the multi-modal networks is to export a demand model
from VISUM and import it into the microsimulation tool VISSIM. Questions to
be addressed could be:
• What are the average transfer times between a train arrival and the PRT, bus and tram system?
• What would be the total journey time depending on the combination of public transport modes?
• How are the effects of vehicles arriving at "rendezvous" type stations? Will the capacity be sufficient to allow transfer without substantial delays caused by crowding?
A multi-modal network takes care of all the inter-dependencies between the
traffic modes and hence provides a single model for complex assessments.
Therefore it is essential that each mode of traffic and – even more important –
that all interactions between them are modeled realistically. Here the VISSIM
microsimulation provides a profound concept for handling all the traffic modes
including their interactions. As for a PRT system, this kind of interaction rarely
occurs as it usually travels on a guideway which is separated from all other
traffic. However, for systems that share the same space with other traffic (such
as trams or busses) all relevant interaction can be modeled and hence the
resulting transport model is very accurate.
4. 4. 4. 4. ConclusionConclusionConclusionConclusion
The impact of People Mover and PRT systems on traffic can and should be
analysed at a macroscopic and microscopic level.
The planning suite PTV Vision incorporates the two systems VISUM
(macroscopic) and VISSIM (microscopic). An interface between the two
systems allows the export of relevant data from the macroscopic to the
microscopic model. This reduces the time and cost required for data supply if
both systems are used for a specific analysis.
Users of the PTV Vision suite benefit also from its particularly wide range of
features designed to model a variety of public transport applications. At the
same time, it integrates all relevant modes of transport for multi-modal
analysis. The impact of the overall system and its individual components can
therefore be analysed and evaluated from the passenger's and operator's
point of view. Application programming interfaces (API) make the software fit
for the future by providing expansion slots to include handling of non-standard
tasks.
Wrapping it all together, the macroscopicmacroscopicmacroscopicmacroscopic approach is ideal for large networks
and line- and timetable-based systems as it allows analysis of service,
demand and capacity utilisation differentiated according to time.
The mmmmicroscopicicroscopicicroscopicicroscopic analysis is an excellent method for providing a detailed view
on the situation, in particular with regard to operation at or above capacity
limits, demand-dependent control methods and differentiated analysis of traffic
and transportation operations. It also provides a 3D visualization.
In addition to standard tasks in traffic and transport, the presented
transportation planning software also proofs to be a capable and valuable tool
for a wide range of assessments for both people mover and PRT systems.
Applying such a tool contributes to an efficient, straight-forward planning
process and helps cutting investment cost by providing a solid base for
financial decisions.