RIMS Update - Best Practice: Traffic Count Estimation

RIMS Best Practice:Traffic Count Estimation

2012 Road Asset & InformationManagement Forum

Ministry of TransportStuart Badger

Acknowledgements

MWH New ZealandFergus Tate

Traffic Count Estimation:

research result is a proposed traffic monitoring framework that significantly

improves the overall accuracy and efficiency (and therefore value for

money) of typical traffic data collection regimes

Proposed Traffic Monitoring Framework

• creation of a traffic link model• core annual monitoring sample • rotational sample (80/20)• Linking estimates

•Why is it important•RIMS Objective•Background

•Sampling Framework• Improving Traffic Estimates•Other Considerations

Contents

•AADT shows network traffic levels•Basis for local & national traffic monitoring•Reasons for RCAs to have accurate estimates

1. Better Decision Making2. Used in LoS & Benchmarking3. Funding Decisions

Why is it Important?

Uses of Traffic Count DataLong term planning

Network Renewal & Development

Customer Service Operational

Network traffic levels

Project planning Responding to enquiries

TM requirements

TSA & Predictive modelling

Pavement &capacity design

Development control & planning

Regulatory requirements

Safety studies & crash analysis

Loading for bridge design

Network monitoring

Traffic modelling & simulation

Project economics (BCR)

MIS

LOS & Policy Development

HCV management

1. RIMS intention: an applied guide2. Improve traffic counting outcomes and

value for money3. Presentation will cover some of the

technical research details & elements of the applied guide

RIMS Objective

Ministry of Transport Research: “Updating National Traffic Data: Stage 1 and 2 Final Report” (HyderConsulting and MWH New Zealand Limited, 2005)

Prompted due to realisation that data held across RCAs was very unreliable

Background

0% 50% 100%

50%

30%

0%

10%

40%

20%

25% 75%

Net

wor

k %

with

Cou

nts

< 5

yrs

old

% of RCAs

Data Confidence

1. Improving the efficiency of traffic counting programmes• Increasing the coverage• Improve relevancy of data collected

2. Improving Traffic Estimates• Improving the estimates on uncounted road sections• Improving the updating of estimates

3. Facilitate RCA implementation

Research Goals

1. Pilot the proposed framework in three RCAs: • Upper Hutt [urban], Southland [rural] & Hastings [mixed]

2. Nation-wide survey of traffic monitoring practices in RCAs.

Latest Work

• creating a traffic link network from the spatially referenced RAMM database

• fitting the proposed stratified sampling method to this network

Approach

1. Create Spatial Network2. Aggregate sections that carry the same traffic 3. Relate sections subject to the same growth

effects (parent/child)• Maximises network coverage• Optimises size of sampling framework

Traffic Link Network / Model

RAMM Cway id: 1949

RAMM Cway_id: 1952

Node 2000 Node 2001 Node 2003 Node 2004 Node 2007

RAMM Cway_id: 1953

RAMM Cway id: 1950

RAMM Cway id: 1951

Node 2005Node 2006

Legend: RAMM C_way section node point

RAMM Cway id: 1949

RAMM Cway_id: 1952

T.Link_id: 949 T.Link_id: 1950 T.Link id: 1951

Node 2000 Node 2001 Node 2003 Node 2004 Node 2007

RAMM Cway_id: 1953

RAMM Cway id: 1950

RAMM Cway id: 1951

Node 2005Node 2006

Legend:Node point derived from Spatial model RAMM C_way section node point Combined spatial and RAMM C_way section node point

0%

80%

60%

100%

40%

20%

Size

of T

raff

ic L

ink

Mod

el

RCA

Two distinct needs for traffic count data. • monitoring of travel activity, at a local, regional, or national level• determining traffic volumes and travel on individual roads

1. core sample - for monitoring travel activity2. rotational sample - to cover the road network.

Sampling Framework

National Traffic Database Project - Dr M K Mara developed a sampling framework to estimate the total vehicle travel within an RCA, to a given level of precision.

Core Sample

0% 60% 100%

B5

15,000

0

5,000

20,000

10,000

40% 80%

ADT

Proportion of Links

90%

B4 B3 B2 B1

Nominal Precision

Hastings District Council

+/-10% +/-5%AA

DT

Qua

ntile

<40% 2 7

40-60% 3 8

60-80% 8 31

80-90% 6 23

90-100% 27 107

Total 46 175

Core Sample Size Requirements

•The aim of the rotational sample is to ensure that over time traffic data is collected on, or is applicable to, as much of the network as possible.

•Focuses on capturing majority of network VKT

Rotational Sample

RCA

Proportion of Total Vehicle Kilometres of Travel Contained in the Model

25% Model 50% Model

Hastings District Council 87% 97%

Upper Hutt City Council 86% 97%

Southland District Council 84% 95%

Rotational Sample Coverage

Sample Size vs. Network Coverage

Sample Size vs. Network Coverage

Counting approximately 17% of the traffic links p.a. would:

1. satisfy the core sampling requirements2. Cover 80% of the total vkt (two year rotational cycle),3. provide sufficient additional capacity to retain the

investment in the historic monitoring programme

Pilot Area Results

•Sections of road are linked•Approximate relationships (e.g. growth trends) can be assumed

•Linking sections can improve confidence on unmonitored sections

Traffic Estimates

Traffic Estimates

Z

Z0.95Z0.65Z

Y=0.45Z

0.2Z

Y

0.5Y

0.5Y

A B

C

1. Data use / type of data you want2. HCV Traffic3. Growth Node4. Seasonality5. Traffic modelling

Other Considerations

Traffic ModellingCapture movement patterns

• Cordons• Screen-lines

• Traffic count data is important

• Monitoring framework improves the overall accuracy and efficiency

• 3 Key foundations to strategy:• creation of a traffic link model• core annual monitoring sample of between 3% and 7% links• 2 year rotational sample (top 20% of VKT links)

• Other considerations to make• HCVs• Seasonality• Network Change• Cordons & screen-lines

Summary

Process Review

1 2 3 4

9 10 11

8 7 6 5

2 3 4

9 10 11

8 7 6 5

SpatialiseRAMM Data

3 4

9 10 11

8 7 6 5

1Update ADT estimates to same base

1 2 4

9 10 11

8 7 6 5

Homogenise network to form traffic

links

1 2 3

9 10 11

8 7 6 5

Relate Links (e.g. by ADT % or Growth

etc.)

1 2 3 4

9 10 11

8 7 6 Stratify Network

1 2 3 4

9 10 11

8 7 5Select

random sample for

CORE SITES

1 2 3 4

9 10 11

8 6 5Select 20% with highest

VKT for ROTATIONAL

SITES

1 2 3 4

9 10 11

7 6 5Substitute

sites:Core sample Historic sites Other goals

1 2 3 4

10 11

8 7 6 5

Set Frequency for Counts

1 2 3 4

9 11

8 7 6 5

Produce Programme

1 2 3 4

9 10

8 7 56

Aim to count 15-20% of your links annually

Questions?More Information?www.rims.org.nz

Matthew Rodwellmatthewr@hdc.govt.nz