Public Transport Network
Design and Appraisal
- a case study of Porto
Álvaro Costa, Pedro Abrantes and Oana Grozavu Santos
Faculty of Engineering, University of Porto
19th Annual International EMME/2 Users’ Conference
Seattle, 19-21 of September, 2005
Contents
1. Introduction
2. Re-designing bus network
3. Network Appraisal Methodology
4. Impact of LRT Network
5. Impact of Strategic Bus Network
6. Local vs. Global Accessibility
7. Global Efficiency Indicators
8. Complementarities between EMME/2,Enif and
ArcView
9. Conclusions
10. Further Research
Introduction
• Porto Metropolitan Area pop.: 1.2 mi
• New Light Rail (Metro do Porto)
• New integrated ticketing system
(Andante)
• Need to re-design the bus network to
improve integration and efficiency
• New network design based on
“professional judgment” (heuristic
method?)
• FEUP was commissioned to:
• evaluate its public
acceptability and efficiency
• suggest improvements
Introduction
• Starting point: existing STCP bus
network (red): 81 bus lines (to be
cut down to 50)
• Driving force: introduction of the
new light metro network (blue)
• Problem: Guaranteeing public
acceptability and increase
efficiency (heavily constrained
problem, but poorly defined
constraints)
• Solution: Iterative, piece meal,
client-driven approach (“trial &
error”) to ensure feasibility of
solution = optimization in the
real world
• Strong emphasis on GIS analysis
of results.
Re-designing bus network
• Design criteria:
– Short lines
– Modal integration
– Homogeneous
frequencies (10min)
– 20% veh-km reduction
• Result:
– Decrease in radial
services
– Increase in orbital
services
Change in bus
service
frequency
(blue = increase)
PT Modeling Approach
• Public Transport
Assignment Model
(EMME/2)
• Detailed zoning system: 350
zones covering an area with
aprox. 1 million inhabitants
(City of Porto: 105 zones and
350 000 inhabitans)
• Household travel survey from
year 2000. Fixed demand
• The match between the two
zoning systems was done
applying a gravity-type model
PT Modeling Approach
Network Characteristics:
– 350 centroids
– 3266 regular nodes
– 8051 links
– 7 modes
– 9400 transit line segments
– 140 lines: 81 present bus lines, 50 future bus
lines, 4 metro lines, 4 train lines, 1 elevator
PT Modeling Approach
• Network Scenarios:
– 10 (Reference case): Current STCP Network + Suburban
Railways
– 20: Reference Case + LRT
– 30: STCP Strategic Network + Suburban Railways + LRT
– 3x: Modifications of STCP Strategic Network + Suburban
Railways + LRT
• Accessibility analysis and efficiency analysis
– Impact of LRT network (Scenario 20 vs. 10)
– Impact of STCP Strategic Network (Scenario 30 vs. 20)
– Impact of the new transport system (Scenario 30 vs. 10)
– Impact of the modifications on the Strategic Network
Network Appraisal Methology
• Accessibility indicator (GIS plots) :
Average generalized travel time by destination zone
GTT ij WalkTij WaitTij TransferT ij InVehTij
(GTT iD TiD )
GTT D i
T i
iD
• Efficiency indicators:
Fleet size, Veh-kms, Total trips (operator)
Pax-kms, travel time, walk time, number of interchanges
(users)
Impact of the LRT Network Mapa nº 3
dTT média global - cenário 20 vs cenário 10
Change in accessibility
between scenarios 10 and 20
• As expected, LRT brings about
significant travel time
reductions (up to 12 min)
• Greatest benefits along LRT
• Large gains up to a significant Amp _con c.shp
Linh as m etro_ a.shp
distance from LRT, because of dTT20 -1 0
-30 - -1 6
-16 - -8
bus-LRT difference in speed
-8 - -4
-4 - -2
-2 - -1
-1 - -0.5
-0.5 - 0 .5
(25km/h vs 13km/h)
-0.5 - 1
1-2
2-4
4-8
8 - 16
16 - 3 0
Impact of the LRT Network
Change in bus passenger flows
between scenarios 10 and 20
• General reduction in bus pax
flows
• Small increases in certain
suburban/orbital routes
• Potential complementary role of
bus in new network
Impact of the Strategic Bus Mapa nº 4
dTT média global - cenário 30 vs cenário 20
Network
Change in accessibility between
scenarios 30 and 20
• Significant travel time increases (8
min): poorer coverage or service
transfer to private sector
• Significant gains in some suburban
zones (10 min): improved LRT access,
successful integration ☺
• Gains from metro outweigh losses
from new bus network in nearly all
zones (special zones treated later) Amp_conc.shp
Linhas metro_a.shp
dT T30-20
-30 - -16
-16 - -8
-8 - -4
• Policy implication Metro + SN must
-4 - -2
-2 - -1
-1 - -0.5
be introduced at same time -0.5 - 0.5
-0.5 - 1
1-2
2-4
4-8
8 - 16
16 - 30
Impact of the Strategic Bus
Network
Change in bus passenger flows
between scenarios 30 and 20
• Reduction in radial flows, esp.
around city centre, due to
reduced frequency
• Significant increases in orbital
flows, esp. towards main metro
interfaces
• Explains gains away from LRT
route bus services play an
important role as feeders to LRT
Impact on the pattern of trip
interchanges
Change in transfer boardings
between scenarios 10 and 30
(red=increase)
• Concentration of interchanges
around key metro stations
• Bus increasingly becomes a
feeder mode to LRT. Bus-bus
transfers decrease significantly
• Policy implications:
– Improve interchange facilities to
take full advantage of new mode
– Minimize distance between bus
and metro stops
Local vs. Global Accessibility
Análise da Acessibilidade a Partir da Zona 190 (5-3)
• Problem: Some zones lose
accessibility from sce. 10 to
30
• Further investigation showed #
Y #
Linhas metro.shp
z190_53.s hp
inconsistencies in network
-24 - -12
-12 - -6
-6 - -3 190
-3 - -1
-1 - 1
design:
1-3
3-6
6 - 12
12 - 24
Subseccoes ine.shp
Análise da Acessibilidade a Partir da Zona 190 (7-3)
– Zones furthest away from
metro miss out the most with
the new bus network
– Those zones also happen to
be important centers for Linhas metro.shp
#
Y
surrounding neighborhoods
Linha_39d_v.s hp #
Linha_39d_i.s hp
z189_73.s hp
-24 - -12
-12 - -6
-6 - -3
-3 - -1
-1 - 1
190
1-3
3-6
6 - 12
12 - 24
Subseccoes ine.shp
Local vs. Global Accessibility
Time vs Distance
Diagrams attempt to illustrate the
45
40
problem 35 Bus
Bus+Metro
30
Time elapsed
• The new bus network improves
25
(min)
20
access to metro stations, which in 15
turn greatly improve access to 10
distant parts of the city 5
0
-4 -2 0 2 4 6 8 10
Distance from Origin zone (km)
• Yet, reduced network coverage
decreases local accessibility
• As we started by looking at the
whole metropolitan area, this
problem went nearly unnoticed
• Policy outcome: strengthen local
bus services
Global efficiency indicators – User
perspective
• 3% mean travel time decrease 30 34.95
from sc. 10 to 20
Scenario
20 34.98
• Constant travel time from 20
to 30 10 36.16
• 1.5% mean travel distance 34 34.5 35 35.5 36 36.5
decrease from sc. 10 to 20 Average generalized travel time (min)
• 1.5% mean travel distance bus
decrease from sc. 20 to 30 Scenarios
30 154,650 36,352 metro
• Notice significant pax-kms 20 162,949 33,347
transfer to metro due to new
10 198,951
bus network (+10%)
0 50,000 100,000 150,000 200,000 250,000
Pax-kms
Global efficiency indicators – User
perspective
• Metro does not produce an 30 1.41
increase in av. interchanges
Scenario
20 1.44
• New bus network actually
reduces interchanges 10 1.45
1.39 1.4 1.41 1.42 1.43 1.44 1.45 1.46
• But at the cost of greater walk Average number of boardings
access time per trip
• Still, no increase in travel
time Scenario
30 7.91
20 7.17
10 7.45
6.60 6.80 7.00 7.20 7.40 7.60 7.80 8.00
walk time (min)
Global efficiency indicators – Operator
perspective
• 6% decrease in fleet size Fleet size Veh-kms
requirements (AM peak hr)
• 4.2% decrease in morning peak Bus_0 517 7676
hour veh-kms traveled Bus_1 485 7350
delta -6.0% -4.2%
• Conclusion: Significant savings
achieved, while maintaining
or improving accessibility by
applying simple network
design rules
• “Operator happy, passengers
happy”
EMME/2 - Enif - ArcView
• ArcView is a great help in the construction of EMME/2
network files
• ArcView is a good tool for “translating” in a friendly
image the results of the assignments (especially OD
pairs related data)
• In both cases, Access is a priceless help
• Enif is a useful tool for analyzing data regarding
interfaces, boardings, alightings, flows on links
Conclusions
• The methodology is understandable to decision makers
• Benefits from LRT line are significant, and spread
beyond the route
• Scale of benefits depends on bus service design bus
plays an important feeder role in the LRT network
• It is possible to achieve significant gains in accessibility
and productivity by re-designing the bus network for
better integration with LRT network
• Nevertheless there is a trade-off between local and
global accessibility in some areas, which is not visible
when analyzing accessibility across the whole area
Future research
• Schedule the implementation
• Incorporate elastic demand model
• Introduce outstanding operators and differentiated fare
system
• Develop optimization algorithms to find the “optimum
network”
• Test the impact of other policy measures, e.g. bus
priority
Questions & comments
Álvaro Costa
Faculdade de Engenharia da Universidade do Porto
Oana Grozavu Santos
Faculdade de Engenharia da Universidade do Porto