An Alternative Transfer Process from VISUM to VISSIM Hackney Central

1
An Alternative Transfer Process from VISUM to
VISSIM Hackney Central
Matt Hall, Daniel Handke Simon Fielder
25th October 2007
2
Presentation Structure

• Section 1 Project Background
• Section 2 Transfer Methodology
• Section 3 Spreadsheet
• Section 4 Summary

• 1)
• 2)
• 3)
• 4)

3
Project Background

• 1)
• 2)
• 3)
• 4)

4
Route 38 Bus

• Atkins was appointed to carry out a Corridor
  Management Pilot study on the bus Route 38
  corridor in London on behalf of the City of
  London, as lead Borough for the London Bus
  Priority Network (LBPN) and Transport for London
  (TfL).
• The route was split into several smaller areas
  including the Hackney Central area

5
Location
6
Model Background

• Atkins developed a VISUM model of the wider
  Hackney area to consider the reassignment issues
  of the proposals.
• Atkins also developed a separate VISSIM model
  to assess the traffic management schemes in
  detail within the Hackney Central.
• It was decided that static assignment within
  VISSIM would be more suitable as there was very
  little route choice and the VISUM model was
  better suited for assignment of traffic

7
Model Area
VISUM - Boundary
VISSIM - Boundary
VISSIM - Boundary
8
Transfer Methodology

• 1)
• 2)
• 3)
• 4)

9
Existing Transfer Process

• The existing transfer from VISUM to VISSIM is
  not perfect
• One directional
• Often poor representation of geometry and,
• Creation of superfluous inputs and compositions
  for each network input.
• Whilst recognising that future releases are
  looking to resolve the above issues, a more
  immediately solution was sought that would enable
  work to continue on both models in tandem.
• It was intended that the VISUM model would
  generate static routes, vehicle inputs and
  compositions to pass directly to VISSIM.
• A method was required that could achieve this
  and allow work to continue on the VISSIM model
  whilst the calibration and validation was being
  undertaken on the VISUM model

10
Proposed Transfer Methodology

• Atkins had previously utilised spreadsheets in
  order to achieve the following input tasks
• Vehicle model distribution
• Vehicle inputs (linked to source counts)
• Vehicle compositions (linked to source counts)
• Vehicle routes (linked to source counts)
• FMA creation from source matrices
• TRANSYT/LINSIG conversion to VISSIM SC data
• Public transport input linked to schedules
• These spreadsheet were made possible through a
  logical numbering system
• A spreadsheet solution was also sought for
  export of vehicle inputs, compositions and routes
  from VISUM to VISSIM.

11
Numbering
12
Numbering
13
Transfer Methodology Planning

• An initial export from VISUM was conducted to
  enable work to begin on the VISSIM network
  coding.
• A common reference was required between VISUM
  and VISSIM.
• This was achieved through node-link reference
  from the VISUM .att file.
• It was essential to have a strict numbering
  system within the VISSIM model that could be tied
  to the VISUM nodes.
• The reference could then be utilised within
  lookup tables in Excel.
• Existing spreadsheets and VBA macros were used
  as the starting point for the proposed transfer
  spreadsheet.

14
Transfer Methodology

• Segmenting VISUM Network and VISUM Routes in
  three node sequences
• A unique direction is defined from the three
  nodes (From Via To)
• The first and the last node of a sequence is
  utilised to identify overlaps of adjacent node
  sequences of a route
• Each route in VISUM is composed of a
  combination of the 3 node sequences

15
Transfer Methodology

• Each VISUM node is tied to a VISSIM node
• The VISSIM node includes all connectors liable
  for any direction change
• Between VISSIM nodes no route choice is
  allowed
• Thus every VISUM three node sequence can be
  synthesised into a VISSIM link
• sequence

16
Example

• Street Layout

• VISUM Network

• VISSIM Network

• VISUM/ VISSIM Network

• VISUM Network alteration

17
Best Fit Between Models

• Developing a spreadsheet solution to transfer
  data from the VISUM to VISSIM was only part of
  the solution.
• A best fit is required between the two
  models. It was necessary to conduct a process of
  manual convergence
• Each model was measured against individual
  performance statistics (i.e. GEH values etc).
  Incrementally both models required adjustment to
  achieve a good standard of validation

18
Spreadsheet

• 1)
• 2)
• 3)
• 4)

19
Transfer Spreadsheet Route Generator

• Stage 1
• A node lookup between VISUM and VISSIM networks
  gives the basis for the VISSIM route path.
• Stage 2
• Uses an Array to create a single string of nodes
  forming the VISSIM route, removes duplicate
  nodes, smoothes the route to match the VISSIM
  network.

• Stage 3
• Incorporates loops to populate an input sheet
  before exporting as a text file.

20
Transfer Spreadsheet Volumes and Compositions

• Stage 1
• Sums vehicle volumes (fractions) from each zone
  from VISUM route file.
• Stage 2
• Divides the vehicle volumes into compositions
  (CAR, LGV, OGV1, OGV2 and Motorcycle) using
  observed survey values.

• Stage 3
• Populates vehicles volumes and compositions
  sheets through loops before export as text file.

21
Summary

• 1)
• 2)
• 3)
• 4)

22
Advantages

• Facilitates the transfer of VISUM assignment
  even after the modifying of VISSIM network, not
  previously possible in VISUM v9.x
• The approach taken by Atkins is easily
  adaptable i.e. macros can be modified to
  accommodate network changes in either the VISUM
  or VISSIM.
• Allows the user to customize the spreadsheet
  outputs to suit the VISSIM i.e. to create
  incremental loading of vehicle volumes across
  different time periods or by adding ID tags to
  the route paths.
• Improves the efficiency of option testing and
  scheme appraisal by making it straightforward to
  output multiple routings, and volumes from
  different VISUM option test scenarios.

23
Disadvantages

• Time consuming process to balance the demands
  of the two models. The quality of the VISUM
  assignment needs to be very high to facilitate a
  good quality set of routes i.e. the validation of
  turning counts.
• The transfer module was time consuming to
  develop, i.e. requires a full look up table of
  Anode, Bnode and Cnodes and associated bug
  testing.
• Relatively high initial development
  commitment.

24
Future Developments

• A reverse interface that can update VISUM from
  VISSIM. This could be done in a number of ways
  i.e. an update of the impedence at nodes (VDFs)
  within VISUM from outputs from the VISSIM.
• The interface can readily be adapted to convert
  SATURN and EMME/2 outputs to VISSIM format. This
  is an advantage over the converter process in
  built in VISUM V10.X
• Potential to utilise VISUM .COM to develop a
  complete one click solution.

25
Conclusions

• Atkins have successfully developed a
  spreadsheet based approach to transfer vehicle
  routings and volumes from VISUM to VISSIM model
  networks.
• A number of potential developments to our
  existing work have been identified, such as the
  reverse interface to pass assignment outputs back
  to VISUM.
• A successful project that uses a dual
  VISUM/VISSIM approach requires careful planning
  i.e. network consideration.
• From our experience a convergence process
  between the VISUM and VISSIM models is necessary.
• We look forward to trying the VISUM to VISSIM
  translator in VISUM 10.x!

26
Thank You Questions