EMME/2 CONFERENCE HONG KONG

1
2nd ASIAN EMME/2 USERS CONFERENCEHONG KONG
NOVEMBER 2000
2

3
Contents

• 1. Introduction
• 2. Background
• 3. Profile of the City
• 4. Existing Transport System
• 5. Public Transport Restructuring
• 6. Emme/2 Model Structure
• 7. Results
• 8. Using Emme/2 Model in a Predictive Mode
• 9. Concluding Remarks

4
1. Introduction

• In 1994 first democratic government
• elected in South Africa
• Dramatic impact on planning
• New legislation
• Purpose of paper
• To describe the methodology employed in the
  latest update of the Emme/2 Model
• The update and methodology has been influenced by
  political changes
• To demonstrate how the model will be used in
  influencing major decisions regarding
  restructuring and integrating the urban form of
  the city

5
Locality
6
Locality
7
Locality
8
Locality - City of Durban
9
Locality - City of Durban
DURBAN BAY
CBD
10
2. Background

• Prior to 1994
• Six decades of separate development based on race
  apartheid
• Different race groups lived in separately
  demarcated area
• Distorted spatial structure
• Poorest away from CBD

11
2. Background

• Prior to 1994
• Duplication of services, public transport,
  schools, social facilities
• Emphasis on private transport ? road building
• Poorest furthest away from the CBD .. But totally
  reliant on public transport ?high PT subsidy
  costs

12
Effects of Apartheid Planning
13
2. Background

• ? Post 1994 New Government
• New transport legislation
• regulate
• improve
• promote
• Steps to restructure cities
• densify corridors and nodes - achieve economies
  of scale
• infrastructure investment to support corridors
• improve operational performance - tendering
• Better integration
• Re-calibration of Emme/2 model

14
3. Profile of the City

• Area 1366 Km2
• Population 2,5 million
• No. of households 609 000
• 60 of employment close to CBD

• But 30 of employees living close to CBD ? long
  travel distances
• Modal split 57 by PT - varies from 100 to 0

15
3. Profile of the City

• Contributes to 9 of GDP
• Port City - one million containers/annum
• Other activities
• tourism
• commerce
• subtropical fruit
• sugar cane
• motor manufacturing
• agriculture
• construction

16
4. Existing Transport System

• 1 500 buses, 6000 mini-bus taxis, 450 000 cars
• Over the last twenty years there has been a
  significant shift to mini-bus taxis

• Excellent road system - 3 700 km of freeway,
  arterial and main routes
• Modes of transport

17
4. Existing Transport System

• Rail uses old heavy rolling stock
• Generally PT system in a poor state
• Huge inefficiencies in system mainly due to the
  distorted spatial structure
• Currently PT subsidies - US 58 million/annum
• New legislation has been enacted to restructure
  the PT industry

18
Modes of Transport
Rail Infrastructure
Congestion - am peak
Mini-bus Taxi
Typical Bus
19
5. Public Transport Restructuring

• The public transport restructuring main thrust is
  to establish a least cost network with optimal
  modes on the main corridors ? reduce burden on
  subsidy
• Leads to a more efficient and sustainable system
• Supply and demand data surveyed on all public
  transport modes

20
5. Public Transport Restructuring

• Basis for PT O-D matrix
• High priority public transport network ?output
  Rail emphasis
• O-D information plus high priority public
  transport network ? Emme/2 model

21
6. Emme/2 Model Structure

• NETWORK
• 3 712 km of roadway
• 406 km of rail
• 330 zones (316 internal, 14 external)
• Annotation files imported from GIS database

22
Emme/2 BaseNetwork
23
6. Emme/2 Model Structure

• DEMOGRAPHICS
• 1996 census data
• Employment and car ownership - separate sources
• Prior to 1996 data collected by race and model
  structured by race e.g. WHBW, BHBW
• Since 1996 data collected by income group - high,
  medium, low
• Income grouping used as a proxy for car ownership
  and hence PT usage
• This change necessitated a rethink in the
  structure of the model

24
6. Emme/2 Model Structure

• Detail is lost
• Required simplification in trip generation and
  trip distribution models in order to cater for
  changes

• DEMOGRAPHICS

• Typical screenline

25
6. Emme/2 Model Structure
TRIP GENERATION - OVERALL APPROACH

• Racial classification ? Income classification
• Existing parameters as far as possible
• Simplify model
• Census data
• High income ? R72 000/annum
• Medium Income ? R 30 000 - R72 000/annum
• Low income ? R0 - R30 000/annum
• Why income classification ?
• Trip generation ? income
• Car usage ? income
• Improved distribution of HBW trips

26
6. Emme/2 Model Structure
TRIP GENERATION EQUATIONS
NON-WORK (NW) TRIPS - 2 HOUR AM
PEAK Productions 0.05 (L.PopM.Pop
(1.50H.Pop)) 0.05(L.Emp
(2.0M.Emp) (4.0H.Emp)) Attractions
(0.008 L.Pop) (0.024M.Pop)
(0.039H.Pop) (Activity zones) ( 0.591M.Emp)
(1.182H.Emp) Attractions (0.008 L.Pop)
(0.024M.Pop) (0.039H.Pop) (Other zones)
( 0.117M.Emp) (0.234H.Emp)
TRUCK TRIPS Productions (0.04H.Emp)
(0.1M.Emp) Attractions (0.05H.Emp)
(0.07M.Emp) (0.007L. Emp)
27

6. Emme/2 Model Structure
MODAL SPLIT

• High correlation ? income and car ownership
• Modal split at origins based on graphs

• Four modes - auto, rail, bus, mini-bus taxi
• Auxillary transit mode - walk

28

Modal Split Curve (HBW Trips)

29

Modal Split Curve (NW Trips)

30
TRIP DISTRIBUTION
6. Emme/2 Model Structure

• Develop cost matrices
• Car gt Travel time matrix
• PT gt Cost of travel
• Both generated in previous assignment
• Intra-zonal costs added to each matrix
• The PT trip cost was refined further
• Determine transposed matrix
• Determine minimum of original and transposed
  matrices
• This compensated for off peak direction costs

31
TRIP DISTRIBUTION
6. Emme/2 Model Structure

• Simple gravity model deterrence function applied
  to these times/costs F(c) exp(-c?)
• Separate beta value, impedance matrices used for
  PT and cars
• Distribution undertaken for four trip types
• HBW - low income
• HBW - medium income
• HBW - high income
• NW trips

32
TRIP DISTRIBUTION
6. Emme/2 Model Structure

• Distribution Method
• Two dimensional matrix with two input origin
  matrices (car and PT) and a single destination
  matrix
• Model distributes trips based on the deterrence
  matrices and relative attractiveness of car/PT
  for each destination
• Use of INRO macro - BALMPROD.MAC
• Output eight matrices (4 car, 4 PT), combined
  into two matrices (car, PT), for assignment

33
CALIBRATION PROCESS
6. Emme/2 Model Structure

• Iterative process TG, MS, TD, Ass
• Emphasis in TD phase
• Three tools used in the calibration process
• 1. ? value is inverse of the average (weighted )
  cost value
• 2. Three dimensional balancing with Emme/2
• ? origin totals
• destination totals
• trips crossing screenlines - 11 in total

34
CALIBRATION PROCESS
6. Emme/2 Model Structure

• ? this whole process was automated for the 11
  screenlines for car and PT
• ? results of the 1st 3-D balance using the first
  screenline was passed onto the second and so
  forth
• ? origins kept same, destinations modified
• 3. DEMANDJ.MAC - adjustment of demand matrix
  based on counts (for comparison/calibration
  purposes only)
• ? final matrices used in assignment not
  adjusted in this way

35
Calibration Process
36
ASSIGNMENT
6. Emme/2 Model Structure

• Car assignment first with PT lines pre-loaded as
  Pcu value
• PT assignment run second, speed of road based PT
  a function of car assignment speeds

37
7. Results

• Reasonably good results
• Cars 174 link counts R2 0.921
• Public Transport 22 screenlines R2 0.984
• Public Transport (buses) 22 screenlines
• R2 0.890
• Public Transport (mini-bus taxi) 22 screenlines
  R2 0.826
• Public Transport (rail) 22 screenlines R2
  0.950

38
Public Transport at Screenlines
39
Link Scattergram
40
8. Using Emme/2 in a Predictive Mode

• Simulate future scenarios
• Simple trend projections to various intervention
  policies
• Emphasis on public transport enhancement
• Main areas of influence
• influencing abnormal trip length frequency
  distribution (travel distances)
• by incorporating land use strategies
• bottleneck elimination

41
8. Using Emme/2 in a Predictive Mode

• TDM measures
• rationalising PT network - using operating costs
  and fare income as a measure of improvement

42
Existing Trip Length Frequency Distribution
43
Preferred Trip Length Frequency Distribution
44
City of Durban
Proposed Nodes and Corridors
45
9. Concluding Remarks

• Use of Emme/2 has been the backbone in terms of
  determining the HPPTN
• Model simplified to replicate current transport
  situation
• In a firm position to test land use strategies
• In a position to influence outcomes
• Monitoring of particular parameters within Emme/2
  is now easily achievable
• Main tool in developing long range and short term
  plans for the City

46
(No Transcript)