Spatial Planning and the Strategic Road Network. Policy

1
A traffic model for every occasion - what to use
when. TRICS Conference workshop Tuesday 20th
November 2007 Alan Bain Director JMP
Consulting
2
Todays Programme

• Part 1 Introduction to Modelling
• Part 2 Modelling Techniques
• Part 3 Jargon (and Myths)
• Part 4 Modelling Process
• Part 5 Outputs
• Part 6 Workshop

3
Course Objective
Provide attendees with an overview of available
modelling techniques including interactive
sessions addressing typical modelling
scenarios. It will enable attendees to
understand benefits as well as limitations. The
course is primarily targeted at those who do not
have a traditional modelling background but are
either project managing studies which have a
modelling element or engage with clients or
potential clients who may have a need for a
modelling service.
4
Introduction to Modelling History of Transport
Modelling

• Popular belief is that transport modelling began
  in the US in the 1950s
• Chicago Transportation Study
• Yet in 1925, the Cross-River Commission evaluated
  12 potential river-crossings in Brisbane, using
• a five day classified count on the Victoria
  Bridge
• an origin-destination survey
• speed studies
• vehicle operating cost assessments

5
Introduction to Modelling The Need for a
Traffic Model

• Transport Planners require some means to test
  their proposals and make decisions.
• This may include the use of a traffic model.
• This is a combination of computer software and
  data which represents road traffic flow in an
  idealised form.
• Traffic models assist transport planners to
  design schemes and measures to manage adverse
  traffic effects.
• Traffic models inform the decision making
  process. They are not a substitute for it.

6
Introduction to Modelling The Limitations of
Traffic Models

• Traffic models represent reality in the context
  of what is known already.
• They can predict future traffic flows with
  infinite precision.
• BUT only if the input data is 100 accurate and
  100 complete!
• Planning authorities collect as little
  information as possible, and never a full data
  set, because of the cost and disruption to
  traffic.
• So in reality traffic models cannot predict the
  future with infinite precision.
• Traffic models cannot be more reliable than their
  data inputs.

7
Introduction to Modelling What should they tell
us?

• Fundamentally they should indicate how traffic
  flows and what the costs of travel are.

8
Modelling Techniques What tools are available

• Traffic engineering and management bus priority
  models, single junction models, local centre
  models, corridor models (VISSIM, PARAMICS,
  AIMSUM, DYNASIM)
• Assignment modelling strategic town and city
  centre models, area action plans, major road
  schemes (SATURN, PARAMICS, VISUM)
• Multi-modal and land use models regional
  modelling, North Wales Transport Model, PRISM,
  SITM (VISUM, OMNITRANS, CUBE, MEPLAN)

9
Models who should do and who should manage?

• Very little practical academic training in
  modelling other than theoretical concepts
• Formal training tends to be provided by software
  manufacturers which sometimes provides a
  theoretical basis
• Modellers tend to become modellers by accident
  and many reluctant to become pigeon holed
• However, generally modelling should be done by
  those with a recognised competence, otherwise
  costs will increase and quality will diminish
• Much modelling work undertaken to support wider
  TA, therefore PM may not be a modeller and
  relying on modeller to provide appropriate
  support
• However, if project is a primarily a modelling
  job then PM should have a good grasp of
  fundamentals (if not a recognised modeller
  themselves)

10
Modelling Techniques Traffic Engineering

• Conventional tools such as LINSIG, TRANSYT,
  ARCADY and PICADY still used but a modeller would
  not regard them as a traffic modelling tool.
  Still a role but where is the divide?
• Microsimulation tools such as VISSIM and PARAMICS
  increasingly been used
• Applications - bus (or LRT) priority, vehicle
  actuation, queue relocation, blocking back
  effects
• Microsimulation model should also be considered
  if the effects of any scheme or development are
  felt over a wider area

11
Modelling Techniques Traffic Engineering

• Specialist modelling firms consider
  microsimulation modelling to be simply an
  extension of the standard traffic engineering
  techniques but this ignores the high level of
  capability required to model vehicle actuation
  and bus priority.
• VISSIM generally favoured by TfL and is perceived
  by some as being more forgiving to the user.
  However, there are many examples of PARAMICS
  being accepted by TfL
• No route choice within them therefore dynamic
  routeing is not required.

12
Modelling Techniques Assignment Modelling

• Generally undertaken over a wider area than most
  microsimulation models (for example town and city
  centres)
• Uses tools such as SATURN.
• Assists in assessing the area-wider impacts of
  schemes as they model the reassignment of
• Conventional tools have been challenged in recent
  years by newer software tools such as PARAMICS
  and VISUM.

13
Modelling Techniques Assignment Modelling

• Many examples of town and city centre PARAMICS
  models.
• The dynamic assignment capabilities of PARAMICS
  are generally accepted as being superior to those
  of VISSIM.
• Junction analysis will still be required using
  conventional tools such as TRANSYT and LINSIG.
• Able to model the effects of bus operations but
  they should not be confused with multi-modal
  models.

14
Modelling Techniques Multi-Modal and Land Use
Models

• Standard models deal with the effects of
  vehicular traffic with fixed vehicular trip
  matrices
• Multi-modal models use elasticity models to
  forecast not only vehicle assignment but modal
  split
• Often use combination of TRIPS (or CUBE)
  software, which handles the multi-modal element
  and SATURN software which handles the traffic
  assignment
• Should not confuse multi-modal models with
  conventional assignment models which model bus
  operations

15
Modelling Techniques Multi-Modal and Land Use
Models

• Typically cover areas such as large cities or sub
  regions. Larger land use transport models such as
  PRISM (Policy Responsive Integrated Strategic
  Model) take account of changes in the location
  and intensity of land uses as well how transport
  can affect land uses and the economy
• Specialist modelling consultancies develop
  bespoke modelling tools (MEPLAN)
• Several areas in which they could be applied such
  as assessment of new growth points, RSS and LDF
  processes

16
Modelling Techniques ITS Modelling

• Microsimulation has ability to model the effects
  of intelligent transport systems
• MOVA (Microprocessor Optimised Vehicle
  Actuation), SCOOT (Split Cycle Offset
  Optimisation Technique) and ATM (Active Traffic
  Management).
• Allows user to test impacts of a MOVA or SCOOT
  signal network before they are implemented on the
  ground (PCMOVA)
• MOVA can be linked to VISSIM and Paramics and
  there are also examples of PARAMICS being linked
  to SCOOT and ATM (on the M42 and M25 for example)

17
Modelling Techniques Pedestrian Modelling

• Paxport
• Legion
• Spatial analysis modelling

18
Data collection and time periods

• Traffic engineering (Turning counts, queue
  lengths, journey times?) 1 hour peak?
• Assignment modelling (Turning counts, ATCs,
  journey times, OD surveys, profiles) 3 hour
  peak?
• Multi modal and land use models (Turning counts,
  ATCs, journey times, OD surveys, stated
  preference surveys, public transport demand data)
  24 hour models?

19
Modelling Jargon

• Gravity model
• Furnessing
• Matrix estimation
• Profiles
• Generalised cost
• Route choice
• Static assignment
• Dynamic assignment
• Calibration
• Validation
• Peak spreading

20
Modelling Jargon first principles

• 4-stage transport model
• Trip generation
• Trip distribution
• Modal split
• Trip assignment
• 5th stage peak spreading

21
Modelling Jargon Distribution and Gravity Model

• Method for distributing traffic
• Based on relationship between population and
  distance between them
• number of trips from zone i to zone j will be
  proportional to their size and inversely
  proportional to their separation
• In this case
• size productions and attractions
• separation generalised cost of travel
• Limited - does not take account of congestion and
  cost

22
Modelling Jargon - Matrix Building and Matrix
Estimation

• Method for creating an O-D matrix improves fit
  between observed flows and modelled flows by
  selectively factoring individual cells of the
  input trip matrix
• SATURN ME2 (maximum entropy) using prior matrix
  (from roadside interviews perhaps), network file
  and observed data
• Paramics matrix estimation module uses prior
  matrix, routing file (from prior matrix) and
  observed link and turn counts
• Furnessing manual method used when you know
  origins and destinations but not distribution

23
Modelling Jargon - Profiles

• Traditional tools (SATURN, TRANSYT) do not allow
  profiles all traffic assigned using flat
  profile
• Microsimulation allows user to assign profiles
  over modelled period usually in 5 or 15 minute
  time slices
• Important element if queues are an issue

24
Modelling Jargon Generalised Cost Equation

• Cost to driver made up of time, distance and
  other costs (such as tolls)
• Cost (A x t) (60 x B x d) (C x p) where
• A time coefficients
• t travel time in minutes
• B distance coefficient in minutes per mile
• d link length in miles
• C toll coefficient in minutes per monetary cost
• p price of the toll in monetary cost units
• Different generalised costs can be applied to
  different vehicles/trip purpose

25
Modelling Jargon Assignment

• Static assignment all or nothing
• Dynamic assignment takes account of congestion
  and delay
• Road hierarchy major/minor
• Cost factors reflect characteristics of road
• Driver perception - perturbation

26
Modelling Jargon Calibration

• The DMRB defines model calibration as the
  process of adjusting the parameters used in the
  various mathematical relationships within the
  model to reflect the data as well as is necessary
  to satisfy the model objectives.
• First of all its a process, implying an
  iterative process.
• The parameters used can be considered to be
  either
• Global parameters
• Localised parameters
• The mathematical relationships will affect
  either
• Assignment
• Vehicle behaviour

27
Modelling Jargon Validation

• The DMRB is the appropriate source of guidance
  when validating any model in the UK
• The DMRB defines model validation as an
  independent check of the calibrated model.
• Model validation is simply a comparison of
  modelled output to observed data.
• Observed data must not have been used in the
  model calibration.
• Convergence/seeds
• Modelled output can take many forms, but those
  most commonly used in model validation are
• Link counts
• Turn counts
• Queues
• Journey times

28
Modelling Jargon Validation

• The DMRB states Precise validation of queue
  lengths can be difficult because of the
  volatility of the observed data.
• As a result the DMRB suggests that observed queue
  data should be used in the calibration of the
  model.
• Journey times commonly used independent data set.
• DMRB criteria states over 85 of routes should
  have a modelled time within 15, or 1-minute, of
  the observed

29
Modelling Jargon Validation

• Results should be plotted with the observed mean
  showing a 95 confidence limit, with the modelled
  mean shown against that range

30
Modelling Jargon Validation

• The most commonly used and most readily available
  data
• Methodology for validation well known and well
  established
• Data used in validation check should not be the
  same data used in calibration!

31
Modelling Jargon Vehicle actuation

• PARAMICS signal plans
• VISSIM - VISVAP

32
Modelling Process

• Scoping report
• Appropriate tool
• Network coverage
• Data collection
• Site visits
• Network build
• Calibration
• Validation
• Model audit
• Option testing

33
Timescales Guide

• Junction models 4 weeks
• Corridor and multiple junction models 10 weeks
• Town centre model 16 weeks
• Wide area urban model 24 weeks
• Multi-modal and land use models 34 weeks

34
Model Outputs what outputs?

• Option testing and comparison lies, damned lies
  and statistics
• How should outputs be interpreted? Rule of thumb?
  
• Economic appraisal - TUBA
• SATURN link and turn flows, delay, overall
  network performance
• PARAMICS/VISSIM link and turn flows, queues,
  journey times, overall network performance
• Microsimulation presentations live or movies?
• Who should do presentations?

35
Workshop

• Data collection
• Appropriate tool
• Validation criteria
• Outputs

36
Useful documents and links

• DTO Modelling Guidelines V2 July 2006
• Highways Agency Use and Application of
  Microsimulation Traffic Models
• User Manuals
• DMRB
• The Microsimulation Consultancy Good Practice
  Guide
• Transport Analysis Guidance www.webtag.org.uk