You are on page 1of 8

1.

Objective 1
(Zandra)
2. Objective 2 1. Lappis
2. Lappis
3. Lappis
b. Objective 3 -

- Determine the location of bus stops (service map)


Configure optimal service routes for the following: (Nick)
to KTH
to Karolinska
to Stockholm University
Determine the location of bus terminal (Baizurah)

These objectives are essentially complements to one another and will be


justified for in the sections to come.

2 Project specifics
2.1 Objective 1
In order to efficiently serve the student population of all 3 institutions, we
need to critically examine potential sites within each campus that are
optimal for bus stop placement. As far as possible, the most frequented
building of each faculty will be designated a bus stop. However, which
place along the road the bus stop is likely to be situated will be dependent
on whether its service area overlaps with its nearest neighbour. As the
latter scenario is most likely due to happen due to the small size of each
institution, we need to analyse each institutions road network and service
area boundaries of a set of arbitrary locations around each of the most
frequented buildings to ensure that boundaries do not overlap. Failure to
do so results in resource wastage in terms of construction cost and bus
fuel incurred with underutilized bus stops. Similarly, stops which are too
far apart ultimately defeat the purpose of implementing a shuttle bus
service. Figure 1 illustrates this issue conceptually:

Major buildings of each faculty are indicated by red points. Blue points
represent the hypothetical location of bus stops serving each major
building while the green circles indicate a service area of approximately
20m. This is reasonably assumed to be the maximum distance a student
is willing to walk from any bus stop to reach his faculty. As the 2 bus stops
serving the Swedish National Defence College and College of Opera are
reasonably spaced apart, they both serve their respective schools and are
assumed to be fully utilized. This is in contrast to Q buildings stop, where
its service area is large enough to contain L building, which may lead to
an underutilization of bus stop L. In addition, this brings about secondary
problems of overcrowding at Q buildings bus stop, as the flow of people in
and out of the point is difficult to control.
2.1.1 Data required and Methodology
The first step would be to geocode non-geographical descriptors of
building addresses (such as place names or ZIP codes) to geographic
coordinates on a reference data set of the street network in each institute.
Address names of each road per institute can be obtained from the school
authorities, while street network data can be obtained from either the
municipality, or through a geographical database from a reliable data
source such as www.maps.slu.se. By adjusting for sensitivity of address
matching, ArcGIS should be able to reliably guess where each building is
located and store their location as points on a vector layer.

As the frequency of use of facilities within each building is highly variable,


the task of determining which building is considered as the major activity
hub of each faculty remains a tedious task. This is as students from other
faculties in the institute may travel to the building to have lessons a
number of times per week. Hence, we shall attempt to use the total
seating capacity of all rooms within each complex as a proxy of total
student population, and input it into ArcGIS as attribute data. Although
primary data can be obtained by doing a personal inspection of each
building, we may be limited by time, financial and manpower constraints
needed to mobilise a team to carry out checks. Hence, we would rely on
secondary data obtained from the university authorities to objectively
quantify the number of seats. Alternatively, a less favoured option would
simply be to proxy the total enrolment cohort of each faculty to pinpoint
its major building. Attribute information on total seating capacity shall
then be sorted in descending order, selected and saved into a separate
layer using the Select by Attribute function in ArcGIS.
Once the major buildings by faculty has been identified, we would assign
probable locations for each bus stop on the street network via the
Network Location tool, to anywhere along the road or pedestrian
pavements fronted by the building and save them as points on a layer.
Alternatively, we could conduct direct interviews or surveys with staff and
students frequenting the building and seek their preference on which
location would be most convenient for them. Service maps of each bus
stop shall then be drawn using 20m as the maximum outward travelling
distance from every stop, which shall be referred to as the primary service
area of the bus stop point. However, limitations of this approach would be
the underutilization of select stops due to reasons discussed in the
previous section. Hence, the final location of each stop shall be
determined on a trial and error basis until none of their service areas
intersect with other stops in each institution. On the other hand, this
procedure will not be applicable to determine bus stop locations in Lappis
as we would make use of the existing bus stops for our pickup points,
saving time and work to carry out additional evaluation. Shuttle services
shall originate and terminate at the first stop along each direction on
Professorsslingan, the main collector road plying through Lappis (Figure 2).
In addition, a new bus terminal will be constructed and its location
determined by site selection analysis to house the shuttle buses, although
its location shall have no impact on the optimal route analysis to be
discussed in the next section.

Figure 1: Bus stops along Professorsslingan. The first stop is marked in red while the last stop is
marked in green.

2.1.3 Possible limitations


Due to the difference in land structures, size and location of buildings, the
number of bus stops could vary and may not be standardized across each
institute. Moreover, our analysis assumes that each major building is small
and is sufficiently served by only a single stop. However, larger major
buildings that fall beyond the 20m boundary may require another bus stop as
well. Another limitation would be identifying a major building of a faculty only
to realise that its cohort size is significantly smaller than other faculties and
hence may be relatively underutilized. This begs the question of whether to
waste resources in constructing its bus stop, or to subject the students to
walking a further distance to the nearest stop in another faculty. This results
in additional time spent assessing the locational preferences of each end
user, as well as conducting service area analysis again. In addition, the
quality of responses collected by each end user varies and is subjective, and
may not necessarily aid our decision making process.

2.2 Objective 2
Objective 2 builds on the network dataset constructed to solve objective
1, by conducting network analysis to identify the shortest or most ideal
routes passing through each bus stop around the institutes. This
represents a shortest path problem, to be solved while under several
limitations such as impedances caused by cost or topological barriers
along the street network. While the shortest path between Lappis and
each institute is simply the Euclidean distance between these 2 points, it
is impossible to traverse through physical buildings and cut across

existing traffic and pedestrian walkways, and hence we shall attempt to


find the next best alternative route while minimising impedances along
the way. Hence, this shall be accomplished with the following criteria in
mind:
I.
II.
III.
IV.
V.

Determine the shortest route from Lappis to all 3 universities based


on distance.
Origin and end points shall be the first bus stop already existing
along each direction of Professorsslingan road.
The location of the new bus terminal shall not be taken into
consideration when determining the shortest path from Lappis to
each institute.
Each bus stop abutting every major building in each institute shall
be visited once and only once.
No U-turns at junctions are allowed due to the narrow width of
service lanes within each institute.

Routes generated which failed any of these criteria were corrected by placing
point or line barriers at appropriate locations and recalibrated until satisfactory.

2.1.1 Data required and Methodology


No new datasets are required for this section as the framework to conduct
network analysis has already been constructed as part of the bus stop location
optimization study in the earlier section.
To successfully achieve the set of criteria, we may have to utilize several
functions in ArcGIS such as introducing point and line barriers to carve out
artificial impedances. For instance, this will be useful in ensuring that the shuttle
service does not take a U-turn at a junction even though doing so will result in a
shorter distance generated. This is as attempting U-turns in the institute will be a
tricky manoeuvre due to the narrow lanes and risks posed to students walking in
the area. Figure 3 illustrates a hypothetical route traversing through KTH, with
the original shortest path route generated in red. However, this route would
result in a U-turn at the stop abutting the Depart of Speech, Music and Hearing
which would fail one of the criteria. Hence, a line barrier shall be introduced at
appropriate locations (black line) to serve as an impedance and optimization run
again to generate a new shortest path route highlighted in blue.

Figure 2: Effects of induced impedances on the shortest path routes

On a larger scale, hypothetical shortest path route plying through all 3


institutions, passing through the optimal bus stop locations in each institute as
determined earlier is shown in Figure 4. The red and green dots represent the
start and end points on the network respectively, while the location of the new
bus terminal will not be factored in as the route to the bus terminal is out of
bounds to students.

Figure 3: A hypothetical route linking all 3 institutions

2.2.3 Possible limitations


The hypothetical route in the previous section is optimized based on a set of
general assumptions. However in reality, several factors such as traffic
conditions can pose further impedances to the route and result in alternative
shortest paths. This is as students in each institute may have lessons that
coincide with the morning and dusk peak hours which will result in a longer
time to travel to school due to traffic congestion along the original shortest
path. In addition, while the shortest path route linking all 3 institutions is the
route which minimises all costs and distance taken, we can observe that a
single journey across all institutions is rather long. A student studying at KTH
would have to make an unnecessary trip to the other 2 institutions, wasting
time and perhaps even denying the chance for students at Karolinska
Institute to board the shuttle bus. A workaround could be to split the single
shortest path problem up and introduce shortest paths for 3 routes plying
between Stockholm University-KTH, Stockholm University-Karolinska Institute,
and KTH-Karolinska Institute. Alternatively, we could explore the possibility of
another shuttle service running in the opposite direction arriving at KTH first.

1. How about turning this into a travelling salesman problem? Instead of


positioning the bus stops, we map out the location of one major
building of each faculty according to several criteria (e.g.: population
density, area, height, seating capacity, require that bus stops be placed
no less than 10m away from the front entrance etc.) in a layer. Data on
road network can be obtained from www.maps.slu.se. Identification of
both optimal to-fro routes will be done given a set of constraints:
1. Origin and end point shall be Lappis (whatever postal address)

2. Each major building per faculty shall be visited once and only
once
3. Shortest route possible
4. No U-turns at junctions are allowed due to the narrow width of
minor roads within KTH.
We can then introduce point and line barriers (like lab 4) at appropriate
locations and then recalibrated until a shortest route is generated. One
limitation which we can already foresee is that the distance between
each major building is rather small, so service areas will definitely
overlap and hence the shuttle route will be inefficient.
4. If u-turns are allowed, identify the best location to u-turn while still
maintaining the shortest path possible
4. Examine the possibility of an intra-campus (LAPPIS-KTH-SU-KI) shuttle
service (like SPI-SPII-Kent Ridge). This will reduce the number of stops
in KTH, increasing efficiency but downside is that travelling time
increases.
4. Further expand it and conduct suitability analysis (lab 6) to determine
the ideal location for the shuttle bus terminal in either KTH or Lappis.
Raster data is required and modelled based on the following
constraints:
a. Open Space: Areas without vegetation/development are ideal
locations for a terminal
b. Distance from minor (within KTH/LAPPIS) and major roads
(connection between KTH & LAPPIS)
Zandra Leung
12:22 AM

ohh u r travelling alone?


sure
We would like to set up two bus stops in the two main streets of the Lappis campus. These are both
the points that the students would pass by every time they go to school by metro and bus.
In order to facilitate convenient bus stops throughout the whole campus, the distance should be about
700-1000 meters from each other. This range is set because the distance between each bus stop in
the campus should not be too far away nor near to each other so as to save the cost as well as
controlling the amount of the flow of people near the bus stops. This walking distance should be
reasonable and easily accessed by the students and staff in the campus

You might also like