New A9 Average Speed Cameras – Traffic Modelling and...

Transportation

Transport Scotland 17/02/2014

A9 Average Speed Cameras – Traffic Modelling and Analysis

Prepared by: ............................................................. Checked by: ........................................................................ Colin Hardie Richard Cann Principal Consultant Principal Consultant Approved by: ............................................................. Neil Halket Commission Manager A9 Average Speed Cameras – Traffic Modelling and Analysis

Rev No Comments Checked by Approved by

Date

- Draft Report RC/PK NH 28/10/2013

1 2nd

Draft Report (including Method B) RC NH 10/11/2013 2 3

rd Draft Report (incorporating Appendix E ) RC NH 22/11/2013

3 Final Draft Report RC NH 18/12/2013

4 Final Report RC NH 17/2/2014

1 Tanfield, Edinburgh, EH3 5DA Telephone: 0131 301 8600 Website: http://www.aecom.com Job No 60274040 LATIS Lot 4 Reference ASC Final Report Date Created 17/02/201 This document has been prepared by AECOM Limited for the sole use of our client (the “Client”) and in accordance with generally accepted consultancy principles, the budget for fees and the terms of reference agreed between AECOM Limited and the Client. Any information provided by third parties and referred to herein has not been checked or verified by AECOM Limited, unless otherwise expressly stated in the document. No third party may rely upon this document without the prior and express written agreement of AECOM Limited. f:\projects\transport planning - latis lot 2 and 4\latis lot 4\04 a9 obc\03 execution\03 documents\08 average speed camera report\03_issued\final 170214\140217 a9 average speed camera modelling rev_4_final.doc

AECOM A9 Average Speed Cameras – Traffic Modelling and Analysis 0

Capabilities on project:

Transportation

1 Introduction ....................................................................................................................................................................... 3 1.1 Background ........................................................................................................................................................... 3 1.2 Purpose of Study ................................................................................................................................................... 3 1.3 Terminology ........................................................................................................................................................... 4

2 A9 Traffic Model ................................................................................................................................................................ 6 2.1 Introduction ............................................................................................................................................................ 6 2.2 Traffic Survey Data Collection ............................................................................................................................... 6 2.3 Comparison of A9 Traffic Model and 2013 Survey Data ........................................................................................ 8

3 Data Analysis and Model Development ......................................................................................................................... 14 3.1 Introduction .......................................................................................................................................................... 14 3.2 The A77 ASC System .......................................................................................................................................... 14 3.3 Replicating A77 Speed Distributions .................................................................................................................... 15 3.4 Goods Vehicle Weight Composition .................................................................................................................... 19 3.5 Methodology ........................................................................................................................................................ 20 3.6 Key Issues and Risks .......................................................................................................................................... 22

4 A9 Traffic Model Outputs ................................................................................................................................................ 24 4.1 Introduction .......................................................................................................................................................... 24 4.2 Overtaking / Passing Analysis ............................................................................................................................. 24 4.3 Platoons ............................................................................................................................................................... 27 4.4 Average Journey Times ....................................................................................................................................... 34 4.5 Speed Distributions .............................................................................................................................................. 36 4.6 Desired Speed against Modelled (Actual) Speed ................................................................................................ 40

5 TRL Accident Analysis Summary .................................................................................................................................. 43 5.1 Accident Analysis................................................................................................................................................. 43

6 Summary of Key Findings .............................................................................................................................................. 46 6.1 Overview .............................................................................................................................................................. 46 6.2 Key Findings ........................................................................................................................................................ 46

Appendix A - Overtaking and Passing Graphs .......................................................................................................................... 49

Appendix B – Platooning Graphs ................................................................................................................................................ 51

Appendix C – Journey Time Tables ............................................................................................................................................ 53

Appendix D – Speed Distribution Graphs .................................................................................................................................. 55

Appendix E – Comparison between Model and Observed Data ............................................................................................... 57 Table 1 - Terminology .....................................................................................................................................................................4 Table 2 – Average Speed Camera Scenario Speed Limits ........................................................................................................ 14 Table 3 – A77 Northbound Traffic Flows / Composition (24hr weekday average) .................................................................. 15 Table 4 – A77 Southbound Traffic Flows / Composition (24hr weekday average) .................................................................. 16 Table 5 – Vehicle Top Speed Adjustments Scenario 1 .............................................................................................................. 18 Table 6 - Link Speed Adjustments for Scenario 1 and 2 ........................................................................................................... 19 Table 7 – Scenario 1 Speed Limits adopted in Model 1 and Model 2 ....................................................................................... 20 Table 8 – Scenario 2 Speed Limits Adopted .............................................................................................................................. 21 Table 9 – A9 ASC Traffic Model Tests ......................................................................................................................................... 21 Table 10 – Modelled Traffic Volumes from the Passing and Overtaking Sites (Vehicles) ...................................................... 24 Table 11 – Percentage of Vehicles Overtaking and Passing .................................................................................................... 25

Table of Contents



Transportation

Table 12 – Changes in Platoon Profiles on the Crubenmore Dual Carriageway Northbound between Entry and Exit ....... 28 Table 13 – Changes in Platoon Profiles on the Ralia 2+1 Southbound between Entry and Exit ........................................... 30 Table 14 – Changes in Platoon Profile on the Ralia Straight Northbound between Entry and Exit ...................................... 31 Table 15 – Percentage change in Vehicles in a Platoon State (Exiting) ................................................................................... 32 Table 16 – Percentage Changes in Platoon Length relative to the Base Model (Exiting Platoons only) .............................. 33 Table 17 – All Vehicles Average Journey Times (minutes) ....................................................................................................... 34 Table 18 – Car Average Journey Times (minutes) ..................................................................................................................... 34 Table 19 – LGV Average Journey Times (minutes) .................................................................................................................... 35 Table 20 – OGV1 Average Journey Times (minutes) ................................................................................................................. 35 Table 21 – OGV2 Average Journey Times (minutes) ................................................................................................................. 35 Table 22 – Average Theoretical Desired Speeds (mph) ............................................................................................................ 40 Table 23 –Comparison of Average Desired and Actual (Modelled) Speeds (mph) ................................................................. 41 Table 24 - Forecasts of the number of injury accidents per year on the A9, based on 14 years of historic accident data

(Single Carriageway sections, Weekdays, 7am-7pm only) .......................................................................................... 43 Table 25 - Forecasts of the number of injury accidents per year on the A9, based on 5 years of historic accident data

(Single Carriageway sections, Weekdays, 7am-7pm only) .......................................................................................... 44 Table 26 – Key Findings Summary Table ................................................................................................................................... 46 Figure 1 – A9 Traffic Model Coverage ...........................................................................................................................................6 Figure 2 – Survey Locations ..........................................................................................................................................................7 Figure 3 – Northbound Platoon Comparison 10:00 – 16:00 hours (out detector) .....................................................................8 Figure 4 - Southbound Platoon Comparison 10:00 – 16:00 hours (out detector) .....................................................................9 Figure 5 – Northbound Overtaking Comparison 10:00 – 16:00 hours ...................................................................................... 10 Figure 6 – Southbound Overtaking Comparison 10:00 – 16:00 hours ..................................................................................... 10 Figure 7 – Headway Gap entering Ralia Straight Northbound .................................................................................................. 11 Figure 8 – Headway Gap entering Ralia Straight Southbound ................................................................................................. 12 Figure 9 – ATC site JTC00364 (Balkenna) .................................................................................................................................. 15 Figure 10 – Adjustments to Aggression on Speed Distributions ............................................................................................. 16 Figure 11 – Adjustments to Vehicle Top Speed on Speed Distributions ................................................................................. 17 Figure 12 – Adjustments to Link Speed on Speed Distributions.............................................................................................. 17 Figure 13 – Adjusted Aggression Distributions ......................................................................................................................... 18 Figure 14 – A9 Percentage of Goods Vehicles under 7.5 tonnes (based on two sites) .......................................................... 19 Figure 15 – A77 Percentage of Goods Vehicles under 7.5 tonnes (based on two sites) ........................................................ 20 Figure 16 – Changes in Platoons along Ralia Straight Northbound ........................................................................................ 26 Figure 17 – Platoon Profiles Entering Crubenmore Dual Carriageway Northbound .............................................................. 27 Figure 18 – Platoon Profiles Exiting Crubenmore Dual Carriageway Northbound ................................................................. 28 Figure 19 – Platoon Profiles Entering Ralia 2+1 Carriageway Southbound ............................................................................ 29 Figure 20 – Platoon Profiles Exiting Ralia 2+1 Carriageway Southbound ............................................................................... 29 Figure 21 – Platoon Profiles Entering Ralia Straight S2 Northbound ...................................................................................... 30 Figure 22 – Platoon Profiles Exiting Ralia Straight S2 Northbound ......................................................................................... 31 Figure 23 – Car Speed Distribution at Crubenmore Northbound Dual Carriageway (Inter Peak) .......................................... 36 Figure 24 – OGV2 Speed Distribution at Crubenmore Northbound Dual Carriageway (Inter Peak) ...................................... 37 Figure 25 – Car Speed Distribution at Ralia Southbound 2 lane section of 2+1 Carriageway ............................................... 38 Figure 26 – OGV2 Speed Distribution at Ralia Southbound 2 lane section of 2+1 Carriageway ........................................... 38 Figure 27 – Car Speed Distribution at Ralia Straight Northbound S2 Single Carriageway .................................................... 39 Figure 28 – OGV2 Speed Distribution at Ralia Straight Northbound S2 Single Carriageway ................................................ 39

Introduction



Transportation

1.1 Background

On the 26th July 2013, Transport Minister Keith Brown announced that an average speed camera (ASC) system is to be installed

on the A9, between Dunblane and Inverness, to improve safety. The decision to introduce the system followed a review of the

safety performance of the route as well as careful consideration of the views of members of the A9 Safety Group. The A9 system

will be the second in Scotland. The first ASC system was installed on the A77 in Ayrshire between Bogend Toll and Ardwell Bay

in 2005.

As part of the development of the ASC system, Transport Scotland and the A9 Safety Group; have commissioned research to

assess both the safety and operational impacts of the cameras and any associated speed restrictions on the A9.

The research has made use of the recently developed A9 Traffic Model. SIAS Consultancy has developed for Transport

Scotland an A9 Traffic Model which simulates the traffic conditions along the A9 between Perth and Inverness. This Paramics

model has been developed for use as part of the A9 Dualling Programme. The application of this modelling tool provides the

opportunity to undertake an assessment of the ASC system.

1.2 Purpose of Study

AECOM, and their sub consultants Transport Research Laboratory (TRL), have been commissioned by Transport Scotland to

provide modelling outputs to inform the development of the ASC system on the A9 between Perth and Inverness. The specific

purpose of the study was the application of the A9 Traffic Model to model two scenarios associated with the implementation of an

ASC system on the A9. The two scenarios can be described as follows:

Scenario 1 – Enforcement of the existing national speed limits along the A9 between Perth and Inverness; and

Scenario 2 - Enforcement of the existing national speed limits along the A9 between Perth and Inverness in combination with the piloting of an increased speed limit of 50 mph for all Goods Vehicles on S2 carriageways and 2+1 carriageways.

1

The implementation of the ASC system combined with police patrols may impact on driver behaviour as follows:

• Reduced frustration due to more reliable journey time expectations;

• Less aggressive driving;

• Reduced tailgating; and

• Reduced top end speeds.

In order to support the traffic modelling, AECOM has analysed data from the ASC system installed on the A77. This has provided

information on speed distributions and goods vehicle weight composition. This has been used in the development of the A9 ASC

Traffic Model.

1 On the 5

th December 2013 the Transport Minister announced plans for a 50 mph HGV pilot to be introduced at the

same time as average speed cameras.

1 Introduction



Transportation

With regards to modelling and analytical outputs, it was agreed at the outset of the study that the following would be reported: • Changes in the number of overtaking manoeuvres;

• Changes in platoon lengths;

• Vehicle speeds (desired and actual); and

• Information on journey times.

The A9 Traffic Model outputs have been used by TRL as inputs for the prediction of injury related accidents. A summary of TRL

accident analysis is provided in Chapter 5. A TRL report on the estimation of injury related accidents is provided under separate

cover.

1.3 Terminology

To aid non-technical readers, an explanation of some commonly used terminology within this report is provided below in Table 1.

Table 1 - Terminology

Term Meaning

Aggression Factor

Within Paramics, global aggression factors are used to assign levels of aggression to drivers across all vehicles within the model. There are nine aggression levels, with 0 reflecting the least aggressive and 8 reflecting the most aggressive.

Default Settings Default settings are not ‘factory settings’; they refer instead to the settings which are applied within the A9 Traffic Model (Base) as developed for the A9 Dualling programme.

Desired Speed For the purposes of this report, desired speed is the speed a driver will travel at if no constraints (with the exception of a speed limit and vehicle maximum speed) are present. A constraint may for example take the form of a gradient, slow moving vehicle or bend.

Dual Carriageway A stretch of road which has a central reservation separating the opposing streams of traffic.

Gap and Headway In this report a gap is defined as the distance or time between the rear bumper of the lead vehicle and the front bumper of the following vehicle. The headway is defined as the distance or time between the front bumper of the lead vehicle and the front bumper of the following vehicle.

Light Goods Vehicle (LGV)

Light goods vehicle describes broadly all car type delivery vans and those of a larger carrying capacity range (including three wheeled goods vehicles) but excluding any vehicle with twin rear tyres.

OGV1 and OGV2 OGV1 and OGV2 refer to types of commercial vehicles; OGV1s are 2 or 3 axle rigid vehicles; whereas OGV2s are 3-axle articulated, 4-axle rigid, 4-axle articulated, 5-axle articulated or 6 (or more)-axle articulated vehicles. Colloquially the two groups are called Heavy Goods Vehicles (HGV).

Paramics Paramics is a micro-simulation software package which models the interaction of individual vehicles within urban and highway networks. The model allows the user to tailor the model parameters to reflect the real world. The software is well established and is used extensively by local and government agencies.

Speed or Aggression Caps

A speed cap refers to the upper limit, i.e. the maximum speed at which a driver can travel. An aggression cap refers to the upper aggression limit applied to a driver.

S2 Carriageway A stretch of single-carriageway which has one lane in each direction.

2+1 Carriageway A stretch of single-carriageway which has a single lane in one direction, with two lanes in the opposite direction, with no central reserve.

A9 Traffic Model



Transportation

2.1 Introduction

The A9 Traffic Model was developed as part of the A9 Dualling Programme. The model is a micro-simulation model which

replicates the interaction of individual vehicles and provides the functionality to represent overtaking on S2 single carriageway

roads. The model presents 2012 traffic flows for an average 12 hour period (out with the summer months), representing a variety

of vehicle types, including cars, motor-homes, cars with trailers, caravans, LGV, OGV1 and OGV2. The A9 Traffic Model

network covers the A9 from Inveralmond Roundabout (just to the northwest of Perth) to north of Loch Moy as illustrated in Figure

1 below.

Figure 1 – A9 Traffic Model Coverage

2.2 Traffic Survey Data Collection

The A9 Traffic Model was developed using a number of traffic surveys. The data from these surveys were compared to the

model output to assess whether the model represented observed conditions. In terms of overtaking/ passing and platoon

lengths, four locations (six sites) (see Figure 2) where chosen as follows:

1. Ralia straight, both directions, S2 single carriageway

2. Ralia straight, Southbound, 2+1 carriageway (2 lane section only)

3. Insh, Northbound, 2+1 carriageway (2 lane section only)

4. Crubenmore, both directions, dual carriageway

2 A9 Traffic Model



Transportation

Figure 2 – Survey Locations

Each survey location had two monitoring sites, which detected the speed, headway and order of each vehicle entering and

exiting. The criterion adopted in the development of the A9 Traffic Model for determining if a vehicle was in a platoon was to

adopt a headway of less than five seconds from the vehicle in front. Speed data was used to determine speed distributions by

different vehicle types.

Overtaking / passing were determined by monitoring the order in which vehicles entered and exited the survey locations.

Changes in order determined whether a vehicle had overtaken or been overtaken. In terms of journey times, the A9 Traffic

Model used Bluetooth sites located along the A9 to determine observed journey times. The Traffic Model validated in

accordance with guidance for observed journey times.

Insh NB 2+1 Section

Ralia Straight S2 Single Carriageway

Ralia Straight SB 2+1 Section

Crubenmore Dual Carriageway Section



Transportation

For the purpose of analysis and reporting, speed distributions and journey times have been disaggregated into three time

periods; 07:00 to 10:00, 10:00 to 16:00 and 16:00 to 19:00, representing the AM, inter peak and PM peak. Overtaking and

platoon lengths have been presented for a 12 hour period.

2.3 Comparison of A9 Traffic Model and 2013 Survey Data

This section provides a comparison of A9 Traffic Model outputs against observed data for platoons, overtaking and headways.

The correlation between the base A9 Traffic Model and survey data should be considered when reviewing the outputs in Chapter

4. Platoons are firstly compared in Figure 3 and Figure 4.

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Solo vehicles 2 to 5 6 to 10 11 to 15 16 to 20 Over 20

Pe

rce

nta

ge

of

ve

hic

les

Platoon size

NB OUT Platoon Comparisons

Crubenmore Observed

Crubenmore Modelled

Ralia Straight Observed

Ralia Straight Modelled

Insh WS 2+1 Observed

Insh WS 2+1 Modelled

Figure 3 – Northbound Platoon Comparison 10:00 – 16:00 hours (out detector)



Transportation

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Solo vehicles 2 to 5 6 to 10 11 to 15 16 to 20 Over 20

Pe

rce

nta

ge

of

ve

hic

les

Platoon size

SB OUT Platoon Comparisons

Crubenmore Observed

Crubenmore Modelled

Ralia Straight Observed

Ralia Straight Modelled

Ralia WS 2+1 Observed

Ralia WS 2+1 Modelled

Figure 4 - Southbound Platoon Comparison 10:00 – 16:00 hours (out detector)

In the northbound direction the model slightly underestimates solo vehicles, provides a good representation of 2-5 vehicle

platoons and over estimates platoons in the six plus vehicle categories. In the southbound direction there is a similar picture

however there is a greater differential. This is most notable on the Ralia Straight where platoons are significantly longer in the A9

Traffic Model as highlighted by the Ralia Straight observed versus modelled over twenty vehicles category.

Figure 5 and Figure 6 outline the north and southbound overtaking comparison between the observed survey data and the A9

Traffic Model outputs.



Transportation

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

0 1 to 4 5 to 9 10+

Pe

rce

nta

ge

of

ve

hic

les

Number of overtaking movements

NB Overtaking Comparison

Crubenmore D2 NB Observed

Crubenmore D2 NB Modelled

Ralia Straight NB Observed

Ralia Straight NB Modelled

Insh WS2+1 NB Observed

Insh WS2+1 NB Modelled

Figure 5 – Northbound Overtaking Comparison 10:00 – 16:00 hours

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

0 1 to 4 5 to 9 10+

Pe

rce

nta

ge

of

ve

hic

les

Number of overtaking movements

SB Overtaking Comparison

Crubenmore D2 SB Observed

Crubenmore D2 SB Modelled

Ralia Straight SB Observed

Ralia Straight SB Modelled

Ralia WS2+1 SB Observed

Ralia WS2+1 SB Modelled

Figure 6 – Southbound Overtaking Comparison 10:00 – 16:00 hours



Transportation

In general terms, the observed data for each carriageway section is being replicated by the model. There is however a trend

highlighted where the model is over estimating overtaking manoeuvres greater than four on the dual and 2+1 carriageways.

Figure 7and Figure 8 compare the observed headway gap for the Ralia Straight against the A9 Traffic Model outputs.

Figure 7 – Headway Gap entering Ralia Straight Northbound



Transportation

Figure 8 – Headway Gap entering Ralia Straight Southbound

The A9 Traffic Model replicates the general profile observed in both the north and southbound directions. The model over

estimates smaller headways (between 1 and 2 seconds) and under estimates larger headways.

Further observed versus modelled data is provided within Appendix E.

The next chapter outlines the data analysis and model development work undertaken to better define the methodology to model

ASC on the A9.

Data Analysis and Model Development



Transportation

3.1 Introduction

This section outlines the data analysis that has been undertaken to derive inputs to the A9 Traffic Model. Following discussions

with Transport Scotland on available datasets, it was agreed that two main methods would be used to model speed distributions

for the A9 ASC system.

The first method makes use of available data relating to the implementation of the A77 ASC system. This will be known as Method A. As described in this chapter, this method derives a speed distribution based on the implementation of the A77 ASC system. The derivation of this speed distribution includes a reduction from the default A9 Traffic Model ‘aggression factor distribution’. This will make drivers less aggressive in Method A relative to the base Traffic Model and Method B.

The second method is also based on the A77 ASC system observations however this method retains the default A9 Traffic Model

‘aggression factor distribution’. This will be known as Method B.

The two methods have been considered to provide a range of modelled results to reflect, were possible, the potential behavioural responses that may occur on the A9.

The modelling and reporting of methods A and B will allow the relative sensitivities of the A9 Traffic Model to key parameters to

be examined allowing more informed interpretation of the findings. Table 2 displays the speed limits being modelled within the

two scenarios.

Table 2 – Average Speed Camera Scenario Speed Limits

Scenario 1 Scenario 2

Cars (dual carriageway) 70mph 70mph

LGV (dual carriageway) 60mph 60mph OGVs in excess of 7.5 tonnes (dual carriageway) 50mph 50mph

Cars (S2 and 2+1 carriageway) 60mph 60mph

LGV (S2 and 2+1 carriageway) 50mph 50mph OGVs in excess of 7.5 tonnes (S2 and 2+1 carriageway) 40mph 50mph

3.2 The A77 ASC System

The A77 is the main trunk road linking the Central Belt with Ayrshire, composed of both single and dual carriageway sections with

various speed limits along the route. Due to the high number of accidents and injuries on the road, SPECS (Speed Check

Services) ASC technology was introduced between Bogend Toll (Dundonald/Tarbolton) and Ardwell Bay (nr. Lendalfoot) in

August 2005.

Speed data made available from the Scottish Roads Traffic Database (SRTDb) has been analysed to assess how speed

distribution varied on the A77 before and after speed cameras were installed.

In order to evaluate how speed cameras affect speed distribution, data from the Automatic Traffic Counter (ATC) site JTC00364,

located on a single carriageway section of the A77 at Balkenna (South of Turnberry) has been analysed. See Figure 9 overleaf.

3 Data Analysis and Model Development



Transportation

Figure 9 – ATC site JTC00364 (Balkenna)

The observed data covered three separate one week periods (Monday to Sunday):

• 2nd to 8th May 2005: pre-SPECS installation; during this period, there were no speed cameras on the A77;

• 19th to 25th September 2005: shortly after SPECS installation; speed cameras were installed during August 2005, and

• 5th to 11th May 2008: almost 3 years after SPECS installation.

Data was extracted from site JTC00364 for a period shortly after SPECS installation (10th to 25

th September 2005) to determine

the speed distributions for cars, LGV, OGV1 and OGV2 vehicle types. Speed distribution model parameters were then derived.

3.3 Replicating A77 Speed Distributions

In order to replicate the A77 speed distributions a Paramics ‘test’ model was developed which replicated the A77 demands and

vehicle composition. Gradients and other external influences on speed were not modelled. The model was developed to

represent an average day between the 19th

and 25th of September 2005. Table 3 and Table 4 highlight the traffic flows /

composition for the northbound and southbound movements.

Table 3 – A77 Northbound Traffic Flows / Composition (24hr weekday average)

Cars

Car with trailers / caravans and Motor-homes LGV OGV1 OGV2 Minibus Coach

Vehicle numbers 3041 43 303 169 288 10 35

Proportion of total Traffic 78% 1% 8% 4% 7% <1% 1%



Transportation

Table 4 – A77 Southbound Traffic Flows / Composition (24hr weekday average)

Type Cars

Car with trailers / caravans and Motor-homes LGV OGV1 OGV2 Minibus Coach

Vehicle numbers 3181 41 299 160 294 13 36 Proportion of total

Traffic 79% 1% 7% 4% 7% <1% 1%

The above traffic flows were modelled within Paramics using one hour profiles on a S2 carriageway. The resulting speed

distributions by vehicle type were compared to observed speed distributions on the A77.

Following initial Paramics model testing it was determined that the following three model parameters were most influential in

replicating the observed speed distribution:

• Aggression distribution;

• Vehicle top speed, and

• Link speed.

Aggression distributions define the general speed profile as illustrated in Figure 10. They are ‘global settings’ which impact all

vehicle types. It should be noted that changes to aggression can affect other modelling behaviours, such as overtaking, gap

acceptance and desired speed. As such, the change in aggression factor needs to be borne in mind when interpreting the model

outputs.

Figure 10 – Adjustments to Aggression on Speed Distributions

The vehicle top speed parameter allows a cap to be placed on the maximum speed of individual vehicle types. This allows upper

speed limits such as 70mph for cars on dual carriageways to be modelled. It should be noted that this has an impact on speed

distributions and can result in distortion due to artificially capping the speed as illustrated in Figure 11.



Transportation

Figure 11 – Adjustments to Vehicle Top Speed on Speed Distributions

The link speed parameter has been used to displace the speed distribution. Decreasing or increasing the link speed shifts the

speed distribution, allowing displacement without significant distortion as illustrated in Figure 12.

Figure 12 – Adjustments to Link Speed on Speed Distributions

The traffic volumes and flow compositions presented in Table 3 and Table 4 were assigned to a S2 carriageway model and the

three parameters highlighted above, adjusted in order to replicate as far as possible the observed speed distributions.

Following extensive testing, the parameter values, outlined below, were found to be the closest fit in terms of speed distributions

resulting from the implementation of the A77 ASC system.



Transportation

Aggression distributions were reduced from the default setting in order to replicate the speed distribution of vehicles travelling on

the A77 (site JTC00364), as shown in Figure 13.

Figure 13 – Adjusted Aggression Distributions

Table 5 presents the capped top speed of each vehicle type.

Table 5 – Vehicle Top Speed Adjustments Scenario 1

Vehicle Type Default A9 Traffic

Model Settings (mph)

Scenario 1 Adjusted A77

ASC Settings (mph)

Scenario 2 Adjusted A77

ASC Settings (mph)

Cars 90 77 77

Cars with trailers / Caravans, Motor-homes 60 68 68

Minibus 60 68 68

LGV 90 68 68

OGV1 56 52 56

OGV2 56 50 56

Coach 60 52 52

Double Decker Bus 60 52 52

For scenario 2, the assumption adopted was that the speed limiter of OGVs, which is set to 56 mph, would act as the maximum

speed.



Transportation

The S2 single carriageway links speeds were altered from the default settings to represent the observed A77 speed distributions,

as highlighted in Table 6.

Table 6 - Link Speed Adjustments for Scenario 1 and 2

Dual Single Set to 60mph (a drop of 10mph from

the default 70mph) Set to 50mph (a drop of 10mph from the

default 60mph)

3.4 Goods Vehicle Weight Composition

To assist in the interpretation of model outputs, analysis of Goods Vehicle weight compositions on the A77 and A9 trunk roads

was undertaken. Goods Vehicles less than 7.5 tonnes have a different speed limit to Goods Vehicles greater than 7.5 tonnes.

The analysis considered two weigh in motion (WIM) sites on the A77 and two WIM sites on the A9. The A9 sites at Tomatin and

Birnam provided data between the 3rd

and 23rd

September 2012 and between the 1st and 24

th of September 2012 respectively.

The A77 sites at Kilmarnock and Assloss Farm provided data between the 22nd

and 29th

October 2012 and between the 20th

and

26th

October 2012 respectively.

The data analysed from these sites were generally found to be consistent for each day as highlighted in Figure 14 and Figure

15.

Figure 14 – A9 Percentage of Goods Vehicles under 7.5 tonnes (based on two sites)



Transportation

Figure 15 – A77 Percentage of Goods Vehicles under 7.5 tonnes (based on two sites)

As indicated in the figures above, the A9 percentage of Goods Vehicles under 7.5 tonnes was typically between 15% and 25%

(around 230 vehicles) whilst on the A77 it was typically between 40% and 50% (around 700 vehicles). In terms of traffic

modelling, this indicates that Goods Vehicles on the A77 will have a faster speed distribution than Goods Vehicles on the A9.

This has implications on model outputs, such as overtaking, platooning and journey times which must be considered when

interpreting model results.

3.5 Methodology

Due to the novel application being modelled, AECOM consulted closely with SIAS (the developers of the Paramics software) in

the derivation of the proposed methodology for modelling ASC. Following extensive consultation and testing it was concluded

that a combination of adjustments to aggression factors, vehicle top speeds and link speed parameters would be required to

replicate changes in speed distributions as a result of ASC. Taking on board the functionality of the software the methodology

agreed upon for modelling Scenario 1 was the development of two models with the speed regulations illustrated in Table 7.

Table 7 – Scenario 1 Speed Limits adopted in Model 1 and Model 2

Model 1 (40 mph) Model 2 (50 mph)

Cars (dual) 70mph 70mph

LGV (dual) 60mph 60mph

OGVs in excess of 7.5 tonnes (dual)

40mph (must retain S2 and 2+1 carriageway speeds due to software functionality)

50mph (must retain S2 and 2+1 carriageway speeds due to software functionality)

Cars (S2 and 2+1) 60mph 60mph

LGV (S2 and 2+1) 50mph 50mph

OGVs in excess of 7.5 tonnes (S2 and 2+1)

40mph 50mph



Transportation

Scenario 2 requires only one model as the OGV speed limit is the same on S2, S2+1 and dual carriageways. The speed

regulations modelled are illustrated in Table 8.

Table 8 – Scenario 2 Speed Limits Adopted

Speed Limit

Cars (dual) 70mph

LGV (dual) 60mph

OGVs in excess of 7.5 tonnes (dual) 50mph

Cars (S2 and 2+1) 60mph

LGV (S2 and 2+1) 50mph

OGVs in excess of 7.5 tonnes (S2 and 2+1) 50mph

In order to understand the results of Scenario 1, data from models 1 and 2 must be reviewed. Model 1 will best represent data

for OGVs on S2 and 2+1 carriageways whereas Model 2 will best represent data for OGVs on dual carriageways. The main

limitation this imposes on the modelling results is the differences these speeds have on the formation and dispersal of platoons

entering and exiting different road sections. This must therefore be borne in mind when interpreting model results.

The modelling results presented in this report are based upon derived speed distributions from the A77 ASC system. This in

effect prescribes the awareness and aggression levels of modelled drivers. In other words, the model setup parameters are

prescriptive. Data extracted from the model, as described in the next chapter, provides information on how drivers reacted in the

model given this prescribed driver behaviour.

As discussed in the introduction to this chapter, a second Method (B) has been adopted. For Method B, the link speed

adjustments and capped top speed of each vehicle will be unchanged from Method A. The only change in Method B is to assign

the A9 Traffic Model default aggression factor distributions to Scenarios 1 and 2. In summary, the following tests will be

examined:

Table 9 – A9 ASC Traffic Model Tests

Model Name Description

Scenario 1 – Method A Scenario 1 with adjusted aggression distributions, capped speed and adjusted Link

Speed based on A77 ASC

Scenario 1 – Method B Scenario 1 with A9 Traffic Model aggression distributions, plus A77 based capped speed and adjusted Link Speed

Scenario 2 – Method A Scenario 2 with adjusted aggression distributions, capped speed and adjusted Link Speed based on A77 ASC

Scenario 2 – Method B Scenario 2 with A9 Traffic Model aggression distributions, plus A77 based capped speed and adjusted Link Speed



Transportation

3.6 Key Issues and Risks

In developing this project, a number of key issues and risks were identified. Where possible the identified risks have been

mitigated taking on board the available datasets and modelling tools.

• Due to the functionality of the software, the methodology for Scenario 1 requires two separate traffic models to be

created. This introduces a risk to the interpretation of the modelled outputs as the arrival patterns between road

sections will vary to some degree between models. This risk requires the careful interpretation of the modelled results;

• As no formal statistical comparator is currently available within webTAG or DMRB the model development process was

not able to formally calibrate nor validate the overtaking or platoon outputs. In order to mitigate this issue and as

outlined in Section 2.3, a range of surveys have been undertaken to allow a comparison of observed data against

modelled outputs;

• There is no direct output from the A9 Traffic Model that reports on driver frustration. However studies have been

undertaken by AECOM/TRL into factors that influence driver frustration and these are reported under separate cover.

Additional analysis of datasets and research will be required outside the A9 Traffic Modelling to more fully understand

any impacts on driver behaviour and frustration;

• Due to the functionality of the software, a number of parameters discussed in this Chapter have been adjusted to match

expected speed distribution/behaviour with the two speed camera scenarios in place. As such, certain elements which

influence driver behaviour have been prescribed rather than revealed;

• Paramics provides a general speed distribution which is used for all vehicle types, although other parameters can be

used to influence the speed distribution for individual vehicle types. Speed distributions cannot be fully replicated

relative to the A77 ASC system across all vehicle types / road types, however AECOM have carried out a range of

sensitivity tests to replicate, as far as possible, driver behaviour and speeds with an ASC system in place;

• An analysis of the vehicle type / weight composition on the A9 and A77 has been undertaken. The analysis has

indicated that the vehicle type / weight composition on both routes are not similar. Therefore the respective speed

distributions on both routes may vary and this must be taken into account when interpreting the model outputs, and

• As this is the first known application of the commercially available Paramics software to model the implementation of an

ASC system, the A9 Traffic Model is operating at the margins of the current software’s capabilities.

A9 Traffic Model Outputs



Transportation

4.1 Introduction

This chapter presents key model outputs and analysis. The analysis is based upon outputs from the following five tests:

1. A9 Traffic Model Base (2012), presenting the current conditions along the A9; 2. Scenario 1 (2012), which enforces the national speed limits along the A9, and Method A; 3. Scenario 1 (2012), which enforces the national speed limits along the A9, and Method B; 4. Scenario 2 (2012), which enforces the national speed limit alongside the introduction of a 50mph speed limit for OGVs

on S2 carriageways, and Method A; and 5. Scenario 2 (2012), which enforces the national speed limit alongside the introduction of a 50mph speed limit for OGVs

on S2 carriageways, and Method B.

As per standard practice for micro-simulation modelling, each model has been run with 10 random seeds.

The following model outputs have been reviewed:

• Overtaking / Passing *;

• Platooning;

• Journey Times;

• Speed Distributions; and

• Desired speed compared to actual speed. *Note: In this context, overtaking is defined as where vehicles use the opposing traffic lane to get past another vehicle (i.e. on a standard two-lane single carriageway) and passing is where a vehicle passes another vehicle on a dual carriageway or on the two-lane section of a 2+1 single carriageway.

The following sections provide a summary of the results.

4.2 Overtaking / Passing Analysis

In the development of the A9 Traffic Model, overtaking surveys were carried out at Crubenmore, Insh, Ralia (2+1) and Ralia

Straight (S2). Table 10 outlines the modelled traffic volumes for the surveyed sites.

Table 10 – Modelled Traffic Volumes from the Passing and Overtaking Sites (Vehicles)

Site

12 hour count (07:00 – 19:00)

Base Scenario 1 Method A

Scenario 1 Method B

Scenario 2 Method A

Scenario 2 Method B

Crubenmore NB Dual 3186 3171 3190 3174 3182

Crubenmore SB Dual 3453 3420 3432 3426 3448

Insh NB 2+1 3873 3232 3262 3253 3250

Ralia SB 2+1 3309 3220 3228 3225 3240

Ralia Straight NB S2 Single 3052 2992 3016 3007 2998

Ralia Straight SB S2 Single 3313 3223 3232 3228 3242

The 12 hour vehicle flows remain similar across the majority of the scenarios, with the exception of the traffic flow on the Insh 2+1

carriageway. Closer examination of the Insh 2+1 section reveals that northbound traffic on the A86 continues onto the B9152

4 A9 Traffic Model Outputs



Transportation

instead of entering onto the A9. According to the A9 Traffic Model, the A9 has become less desirable due to the ASC’s reducing

speeds; this in turn has made the parallel routes more desirable. It should be noted that no parallel routes were validated during

model development. However, at this location it is plausible that an increase in travel time on the A9 could result in the increased

use of the parallel B9152 route. It should also be noted that there are only limited locations along the A9 route that viable

alternative and parallel routes exist.

Table 11 – Percentage of Vehicles Overtaking and Passing

Site

Percentage of vehicles overtaking / passing


Scenario 1 Method B

Scenario 2 Method A

Scenario 2 Method B

Crubenmore NB Dual Carriageway 50.9% 45.2% (1) 44.6% (1) 45.2% 44.6%

Crubenmore SB Dual Carriageway 45.3% 42.2% (1) 41.6% (1) 42.2% 41.6%

Insh NB 2+1 Carriageway 40.0% 36.6% 35.0% 34.5% 32.7%

Ralia SB 2+1 Carriageway 47.6% 44.2% 43.8% 41.4% 41.5%

Ralia Straight NB S2 Single Carriageway 24.3% 18.8% 25.1% 17.7% 20.9%

Ralia Straight SB S2 Single Carriageway 23.1% 4.0% 5.6% 3.9% 4.8%

Note: (1) Due to the software functionality, the reported overtaking/passing events on dual carriageways for Scenario 1 are taken to be the same as for Scenario 2. In almost all scenarios and Methods the level of overtaking/passing reduces relative to the base. The only exception to this occurs on the Ralia Straight northbound S2 which under Scenario 1 (Method B) is modelled to have a slight in increase in overtaking relative to the base (25.1% compared to 24.3%).

To assist in understanding how platoons form over S2 single carriageways, a set of loop detectors, six in total, were included

within the A9 Traffic Model on the Ralia Straight northbound direction. These detectors were set at equal distances apart and

were analysed for platoon profiles. Figure 16 displays the changes in platoons between each detector / section.



Transportation

Figure 16 – Changes in Platoons along Ralia Straight Northbound

The analysis above indicates that platoons gradually increase in size as vehicles move further along an S2 carriageway.

Taking this on board, the Ralia Straight southbound section is approximately 25 miles downstream from the nearest two lane

section, whereas the northbound section is less than 5 miles, this significantly changes the platoon composition i.e. the

southbound section contains longer platoons than the northbound section. With a greater number of longer platoons the

propensity for overtaking may reduce, e.g. only vehicles with a higher top speed and aggression factor will be likely to attempt an

overtaking manoeuvre.

The impact of ASC means that top speeds are suppressed in both scenario 1 and 2. Furthermore Method A has a reduced

aggression distribution, therefore the requirement and opportunity for overtaking is reduced, whereas the base model contains

vehicles which can travel at 90 mph and have a default aggression distribution. Therefore the reduction in overtaking between

the base and scenario tests, as highlighted in Table 11, are likely to be the result of a reduced propensity for overtaking

alongside the differences in the vehicle top speeds and aggression.

Overtaking and passing graphs are provided within Appendix A.



Transportation

4.3 Platoons

The modelled platoon data was extracted for the same six sites identified above in Section 3.2. For brevity, only the following

three sites are presented in the body of this report:

• Crubenmore northbound dual carriageway section;

• Ralia southbound 2+1 carriageway section; and

• Ralia Straight northbound S2 single carriageway section.

Results for the remaining three sites are presented in Appendix B.

The following analysis provides the entry and exiting platoons for the above carriageway sections. This data allows an

understanding of platoon profiles and how platoons are broken up / formed when travelling through each carriageway section.

Figure 17 and Figure 18 display the platooning profile entering and exiting the Crubenmore Northbound dual carriageway

respectively.

Figure 17 – Platoon Profiles Entering Crubenmore Dual Carriageway Northbound



Transportation

Figure 18 – Platoon Profiles Exiting Crubenmore Dual Carriageway Northbound

[Note: Scenario 1 results in Figure 18 are from model 1 platoon exit profiles]

Table 12 – Changes in Platoon Profiles on the Crubenmore Dual Carriageway Northbound between Entry and Exit


Scenario1 Method B

Scenario 2 Method A

Scenario 2 Method B

Solo vehicles +22% +21% +19% +21% +19%

2 to 5 +24% +22% +17% +17% +11%

6 to 10 -6% -13% -11% -16% -10%

11 to 15 -12% -15% -11% -13% -10%

16 to 20 -10% -9% -7% -7% -5%

Over 20 -17% -7% -7% -4% -5%

Note: Due to the software functionality, results from Scenario 1 for dual carriageways will be affected by slower modelled OGVs

speeds than reality.

The above modelled outputs, demonstrate the dispersal of larger platoons across all scenarios over the length of the

Crubenmore dual carriageway section of the A9.



Transportation

Figure 19 and Figure 20 below show the platoon profiles for the Ralia 2+1 carriageway southbound.

Figure 19 – Platoon Profiles Entering Ralia 2+1 Carriageway Southbound

Figure 20 – Platoon Profiles Exiting Ralia 2+1 Carriageway Southbound



Transportation

Table 13 – Changes in Platoon Profiles on the Ralia 2+1 Southbound between Entry and Exit


Scenario1 Method B

Scenario 2 Method A

Scenario 2 Method B

Solo vehicles +12% +8% +7% +8% +6%

2 to 5 +19% +19% +17% +16% +16%

6 to 10 -1% -1% +3% -6% -1%

11 to 15 -10% -8% -7% -6% -6%

16 to 20 -7% -5% -6% -5% -5%

Over 20 -12% -12% -14% -7% -10%

As for the Crubenmore dual carriageway, the above modelled outputs for the Ralia 2+1 carriageway demonstrate the dispersal of

larger platoons over the length of this section of the A9.

Figure 21 and Figure 22 below highlight the platoon profiles for the Ralia Straight S2 northbound.

Figure 21 – Platoon Profiles Entering Ralia Straight S2 Northbound



Transportation

Figure 22 – Platoon Profiles Exiting Ralia Straight S2 Northbound

Table 14 – Changes in Platoon Profile on the Ralia Straight Northbound between Entry and Exit


Scenario1 Method B

Scenario 2 Method A

Scenario 2 Method B

Solo vehicles +6% +6% +7% +6% +5%

2 to 5 +1% -3% -3% -4% -3%

6 to 10 -3% -3% -3% -1% -1%

11 to 15 -3% 0% 0% -1% -1%

16 to 20 0% 0% -1% 0% 0%

Over 20 -1% 0% 0% 0% -1%

As indicated by the results in Table 14 and by comparing the Ralia straight S2 northbound and southbound (see Appendix B)

platoon data, there is quite a difference in the results. When interpreting the results it must be recognised that the Ralia Straight

southbound section is approximately 25 miles downstream from the nearest two lane section. The Ralia Straight northbound

section is less than 5 miles. This significantly changes the platooning composition i.e. the southbound section contains longer

platoons than the northbound section. With a greater number of longer platoons the propensity for overtaking reduces. e.g. only

vehicles with a higher top speed and aggression factor will be likely to attempt an overtaking manoeuvre.



Transportation

Table 15 summarises the percentage change of vehicles within a platoon on exiting the modelled section between the base,

Scenario 1 (Method A & B) and Scenario 2 (Method A & B).

Table 15 – Percentage change in Vehicles in a Platoon State (Exiting)

Total Number of Vehicles in a Platoon (percentage change from Base Model)

Crubenmore NB Dual

Carriageway

Crubenmore SB Dual

Carriageway

Insh NB 2+1

Carriageway

Ralia SB 2+1

Carriageway

Ralia Straight NB S2 Single Carriageway

Ralia Straight SB S2 Single Carriageway

Base 71.8% 71.3% 73.1% 80.1% 76.4% 91.6%

Scenario 1 Method A

0.0% -2.4% -0.4% 2.8% -1.3% -1.0%

Scenario 1 Method B

-1.6% -1.8% -3.6% 1.4% -4.0% -3.4%

Scenario 2 Method A

-0.9% -3.8% -0.4% 3.0% -2.2% -1.6%

Scenario 2 Method B

-2.8% -4.2% -3.4% 1.2% -4.5% -3.9%

In the Crubenmore dual carriageway section, the percentage of vehicles within a platoon state reduces between the base and

Scenario 1 and 2. The reduction is greatest in Scenario 2 tests.

The Insh and Ralia 2+1 carriageway sections results do not present a consistent trend between base and scenario tests. The

Ralia southbound carriageway indicates an increase in vehicles within a platoon state across both scenarios and methods.

Alternatively, the Insh northbound 2+1 carriageway presents a reduction in platoon state across all tests with a larger reduction

within Method B. It should also be noted that prior to the Ralia southbound section, there is approximately 25 miles of S2

carriageway, compared to 5 miles for the Ralia northbound section. This difference in approach distance has a significant impact

on platoon profiles. As vehicles travel along an S2 carriageway, they gradually form larger platoons due to variations in vehicle

speeds. This is clearly observed within the model and when comparing the platoon profiles at different distances along the S2

carriageway.

The single carriageway sections indicate a firmer trend, with Scenario 1 and 2 presenting a reduction in the number of vehicles in

a platoon state against the base situation. Scenario 1 displays a reduction of around 1.0% for Method A and between 3% and

4% for Method B. For Scenario 2, the reductions are around 2% for Method A and between 4% and 4.5% for Method B.

Overall, the main observation is that Method B (default aggression factor) results in a greater reduction in vehicles within a

platoon state. It should also be noted that further analysis of platoon lengths over twenty vehicles uncovered that Scenario 1 can

generate platoons up to 60 vehicles and Scenario 2 can generate platoons up to 50 vehicles within the six sites analysed.

A full set of platoon graphs are located in Appendix B.



Transportation

Table 16 below provides summary information on the changes in platoon numbers between the base, Scenario 1 and Scenario

2.

Table 16 – Percentage Changes in Platoon Length relative to the Base Model (Exiting Platoons only)

Models Platoon Length (vehicles)

Crubenmore NB Dual

Carriageway

Crubenmore SB Dual

Carriageway

Insh NB 2+1

Carriageway

Ralia Straight NB S2 Single Carriageway

Ralia Straight SB S2 Single Carriageway

Ralia SB 2+1

Carriageway

Base

1 (no platoon) 28.2% 28.7% 26.9% 23.6% 8.4% 19.9%

2 to 5 47.0% 42.3% 47.5% 44.3% 21.5% 40.4%

6 to 10 16.9% 15.6% 16.3% 19.5% 22.1% 20.6%

11 to 15 5.9% 7.0% 5.4% 7.4% 18.3% 8.8%

16 to 20 1.3% 3.1% 2.0% 3.1% 11.6% 4.3%

>20 0.7% 3.3% 1.9% 2.0% 18.2% 6.0%

Scenario 1

Method A

1 (no platoon) 1.7% 2.4% 1.3% 1.3% 1.5% -2.3%

2 to 5 2.2% -0.2% -2.9% 0.2% -5.1% -5.2%

6 to 10 -2.4% -1.7% -0.9% -0.3% -0.5% -0.1%

11 to 15 -2.0% -1.4% 1.7% 0.5% 0.7% 2.4%

16 to 20 0.2% 0.7% 1.1% -0.9% 0.7% 2.8%

>20 0.2% 0.2% -0.4% -0.7% 2.7% 2.4%

Scenario 1

Method B

1 ( no platoon) 1.6% 1.8% 3.6% 4.0% 3.4% -1.4%

2 to 5 2.0% 0.3% -1.0% 1.9% -2.0% -4.3%

6 to 10 -1.5% -0.4% -0.9% -2.4% -3.0% 1.8%

11 to 15 -1.9% -0.6% -0.3% -1.4% -0.7% 2.0%

16 to 20 -0.1% -0.4% -0.6% -1.4% -0.6% 0.5%

>20 -0.1% -0.7% -0.9% -0.7% 3.0% 1.4%

Scenario 2

Method A

1 ( no platoon) 2.4% 4.9% 2.2% 2.9% 2.3% -1.1%

2 to 5 3.0% 2.1% -0.8% 3.8% 0.9% -2.0%

6 to 10 -1.9% -2.4% 0.4% -0.9% 5.6% 1.0%

11 to 15 -2.6% -2.2% 0.2% -2.5% -0.2% 3.1%

16 to 20 -0.4% -0.9% -0.7% -1.8% -0.4% 1.5%

>20 -0.5% -1.6% -1.4% -1.6% -8.1% -2.5%

Scenario 2

Method B

1 ( no platoon) 2.8% 4.2% 3.4% 4.5% 3.9% -1.2%

2 to 5 0.3% 2.3% 0.5% 3.0% 0.3% -2.6%

6 to 10 -1.1% -1.7% -0.7% -2.9% 1.7% 2.0%

11 to 15 -1.9% -1.7% -1.6% -2.0% -1.2% 2.3%

16 to 20 0.2% -0.7% -0.4% -1.5% -1.7% 0.7%

>20 -0.3% -2.3% -1.3% -1.1% -3.1% -1.2%

Note: Due to the software functionality, results from Scenario 1 for dual carriageways will be affected by slower modelled OGVs

speeds than reality.



Transportation

4.4 Average Journey Times

Modelled journey times have been analysed for each scenario to assess the impact of the ASC’s and associated speed

enforcement. Table 17 to Table 21 below detail journey times for the AM, Inter-peak and PM periods alongside 12hr modelled

averages. The journey times are provided for all vehicles, car, LGV’s, OGV1 and OGV2. To better reflect expected journey

times in Scenario 1, a combination of ‘model 1’ and ‘model 2’ data has been used.

Table 17 – All Vehicles Average Journey Times (minutes)

Time Period

Northbound Southbound

Base Scenario 1 Scenario 2


Method A

Method B

Method A

Method B

Method A

Method B

Method A

Method B

AM 109 128 119 122 114 104 124 114 120 111

IP 108 128 119 122 113 112 132 122 125 116

PM 107 127 118 122 113 111 131 121 124 116

12 hr avg 108 128 118 122 113 110 130 120 124 115

In each direction, the introduction of ASC increases end to end journey times. Travel times decrease when aggression levels are

raised (Method B), as vehicles of all types travel faster. As indicated above, journey times remain relatively consistent between

time periods. Compared to the base model there is between a 10 minute (Method B) and 20 minute (Method A) increase in

journey times in Scenario 1. For Scenario 2 there is between a 4 minute (Method B) and 14 minute (Method A) increase in

journey times.

Table 18 – Car Average Journey Times (minutes)

Time Period




Method A

Method B

Method A

Method B

Method A

Method B

Method A

Method B

AM 105 124 114 120 111 101 121 111 117 108

IP 105 125 115 120 111 109 128 118 122 113

PM 104 125 114 120 111 108 129 118 122 114

12 hr avg 105 125 115 120 111 108 127 117 122 113

For cars, journey times remain relatively consistent between time periods. Compared to the base model there is between a 9

minute (Method B) and 21 minute (Method A) increase in journey times in Scenario 1. For Scenario 2 there is between a 4

minute (Method B) and 16 minute (Method A) increase in journey times.



Transportation

Table 19 – LGV Average Journey Times (minutes)

Time Period




Method A

Method B

Method A

Method B

Method A

Method B

Method A

Method B

AM 105 125 116 120 112 102 122 112 119 110

IP 106 127 117 121 112 109 129 119 123 114

PM 105 126 116 121 112 109 130 120 123 115

12 hr avg 105 126 116 121 112 108 128 118 122 113

For LGVs, journey times remain relatively consistent between time periods. Compared to the base model there is between an 10

minute (Method B) and 21 minute (Method A) increase in journey times in Scenario 1. For Scenario 2 there is between a 5

minute (Method B) and 17 minute (Method A) increase in journey times.

Table 20 – OGV1 Average Journey Times (minutes)

Time Period




Method A

Method B

Method A

Method B

Method A

Method B

Method A

Method B

AM 122 139 131 131 123 119 137 127 129 121

IP 122 140 132 131 124 123 141 132 133 124

PM 122 139 131 132 123 123 142 132 133 125

12 hr avg 122 140 131 131 123 122 141 132 132 124

For OGV1s, compared to the base model, there is between a 8 minute (Method B) and 19 minute (Method A) increase in journey

times in Scenario 1. For Scenario 2 there is between a 1 minute (Method B) and 10 minute (Method A) increase in journey times.

Table 21 – OGV2 Average Journey Times (minutes)

Time Period




Method A

Method B

Method A

Method B

Method A

Method B

Method A

Method B

AM 123 142 135 132 125 120 140 132 130 123

IP 123 143 135 132 125 124 144 135 133 125

PM 124 143 135 133 125 124 144 135 134 126

12 hr avg 123 143 135 132 125 123 144 135 133 125



Transportation

For OGV2s, compared to the base model, there is between a 10 minute (Method B) and 21 minute (Method A) increase in

journey times in Scenario 1. For Scenario 2 there is between a 1 minute (Method B) and 10 minute (Method A) increase in

journey times.

A table of disaggregated journey times is provided in Appendix C.

4.5 Speed Distributions

Speed distributions provide information on the range of achievable speeds for different vehicle types. Speed distribution outputs

have been extracted for the following vehicle types:

• Cars;

• Cars with trailers, caravans or Motor-homes;

• LGVs;

• OGV1;

• OGV2; and

• Bus / Coach.

Note: For brevity, only inter-peak Cars and OGV2 have been reported in the body of this report, all other vehicle types can be

found in Appendix D.

As outlined in Chapter 3, the speed distributions adopted in this report are partly prescribed from analysis of the A77 ASC

system. The modelled speed distributions have been used to predict changes in injury related accident numbers for each

scenario. A summary of the TRL accident analysis is discussed in Chapter 5 of this report and reported in full under separate

cover.

The graphs below show the changes in speed distribution between the base model and the two modelled scenarios. Figure 23

displays the car speed distribution at the Crubenmore Northbound dual carriageway.

Scenario 1 (Model 1) Method A and B

Scenario 2 Method A and B

Figure 23 – Car Speed Distribution at Crubenmore Northbound Dual Carriageway (Inter Peak)

The Crubenmore dual carriageway section illustrates a significant shift in car speed distribution once ASC have been modelled.

This is expected due to the defined input model parameters.



Transportation

The increased aggression factor modelled in Method B (green lines) for both scenarios shifts the speed distribution upwards. It is

important to note that this shift upwards in Method B results in a larger proportion of cars travelling above the enforceable speed

limit and may therefore not be representative of what would happen after implementation of an ASC system.

Figure 24 presents the OGV 2 speed distributions.

Scenario 1 and 2 Method A and B

Figure 24 – OGV2 Speed Distribution at Crubenmore Northbound Dual Carriageway (Inter Peak)

The base model denotes a high proportion of OVG2s around their ‘speed limiter’ which is approximately 56mph. Adopting

Method A, results in a fairly symmetrical distribution around the mean speed. Once the aggression factor is increased (Method

B) the speed distribution represents almost an identical speed distribution to that of the base. As indicated above, this may not

be representative of what would happen after implementation of an ASC system.



Transportation



Figure 25 – Car Speed Distribution at Ralia Southbound 2 lane section of 2+1 Carriageway

The car speed distributions from the Ralia 2+1 carriageway section above present a similar picture to that displayed by the dual

carriageway and are largely defined by their input parameters. Again the speed distribution shifts upwards when applying

Method B which increases driver’s aggression and consequently their speeds. As indicated above, this may not be

representative of what would happen after implementation of an ASC system.



Figure 26 – OGV2 Speed Distribution at Ralia Southbound 2 lane section of 2+1 Carriageway

OGV2 speed distributions on the Ralia 2+1 carriageway are again in large part defined by their input parameters. OGV2 speeds

are capped at around 50mph in Scenario 1, whereas in Scenario 2 the top end speeds hit the OGV ‘speed limiter’ of 56mph.

Once the aggression is increased (Method B), the speeds also increase with Scenario 1 shifting upwards somewhat, whilst in

Scenario 2 the speed distribution almost directly replicates the base model. Method B distributions may not be representative of

what would happen after implementation of an ASC system.



Transportation

Figure 27 below demonstrates the speed distribution for the Ralia Straight Northbound S2 Single carriageway.



Figure 27 – Car Speed Distribution at Ralia Straight Northbound S2 Single Carriageway

In terms of the car speed distributions on the Ralia S2 single carriageway section this again presents a decrease in speeds in

Scenario 1 and 2 relative to the base. However, the top speeds are noticeably more restrained than in the dualling and 2+1

carriageway speed distributions (e.g. Scenario 1 shows significantly more cars travelling at the lower speed of between 34 mph

and 42 mph). Scenario 1 also displays a flatter profile when compared to Scenario 2. With regards to Method B, this generally

shifts the speed distribution profile upwards towards the base speed profile; however the vehicle top speed cap results in the

speed distribution tailing off around 78mph. Method B distributions may not be representative of what would happen after

implementation of an ASC system.



Figure 28 – OGV2 Speed Distribution at Ralia Straight Northbound S2 Single Carriageway



Transportation

In terms of OGV2 speed distributions on the Ralia S2 single carriageway, again the speed distributions have been suppressed

compared to the base model as defined by the input parameters. The base indicates a spike around the OGV speed limiter

tailing towards 36mph, whereas Scenarios 1 and 2 present a more symmetrical distribution around the mean speed. Applying

Method B, results in the Scenario 2 speed distribution closely mirroring the base distribution. As such, Method B distributions

may not be representative of what would happen after implementation of an ASC system.

Appendix D contains a full set of speed distribution graphs.

4.6 Desired Speed against Modelled (Actual) Speed

This section provides a comparison between desired and actual speeds.

As a relatively simple proxy for frustration, a comparison has been made of average desired speeds against actual (modelled)

speeds.

For the purposes of the analysis reported within this section, the theoretical desired speed is taken to mean the speed at which a

driver can travel with no physical restrictions, such as slower moving vehicles, gradients or road curvature. To derive the

theoretical desired speed distributions a ‘test’ Paramics model was developed using the model parameters within the A9 Traffic

Model. The theoretical desired speeds which have been derived from the ‘test’ model are highlighted in Table 22.

Table 22 – Average Theoretical Desired Speeds (mph)

Dual Carriageway S2 Single Carriageway 2+1 Carriageway (2 lane section)

Car 77.6 71.9 66.8

LGV 77.7 71.8 66.7

OGV1 55.8 54.9 55.1

OGV2 56.0 55.2 55.1

Note: the results above have been averaged from 10 model runs using a 200 vehicle sample for each vehicle type.

Table 22 highlights two interesting points. Firstly, it can be seen that desired speeds on S2 Single Carriageways for Cars and

LGVs are faster than 2+1 Carriageways. This is due to the manner in which the software models desired speeds on two lane

carriageways. The offside lane has a desired speed which is around 10mph slower than the nearside lane; however on single

lane carriageways the desired speeds take the higher speed distributions, as used on the nearside lane of a two lane

carriageway. Secondly, the OGV1 has a marginally slower desired speed than an OGV2. This latter matter would require further

investigation but may be due to the power/weight ratios attributed to OGV1 and OGV2 vehicle classes.

To reflect expected journey times in Scenario 1, a combination of ‘model 1’ and ‘model 2’ data has been used. Table 23 displays

the theoretical average desired speeds against the modelled speeds for each scenario.



Transportation

Table 23 –Comparison of Average Desired and Actual (Modelled) Speeds (mph)

Vehicle Type

Average Desired Speed




Method A

Method B

Method A

Method B

Method A

Method B

Method A

Method B

AM Peak

Car 72.8 58.0 49.0 53.1 50.8 54.9 60.0 50.2 54.7 51.8 56.1

LGV 72.8 57.7 48.6 52.4 50.4 54.2 59.4 49.8 54.1 51.1 55.3

OGV1 55.1 49.8 43.5 46.4 46.4 49.2 50.9 44.4 47.6 47.0 50.0

OGV2 55.3 49.2 42.6 44.9 45.9 48.7 50.5 43.4 46.0 46.7 49.5

Inter Peak

Car 72.8 57.9 48.6 52.8 50.7 54.7 55.8 47.3 51.4 49.7 53.7

LGV 72.8 57.4 47.9 52.0 50.2 54.0 55.6 46.9 50.8 49.3 53.2

OGV1 55.1 49.8 43.3 46.1 46.3 49.1 49.5 43.0 45.8 45.8 48.9

OGV2 55.3 49.1 42.3 44.8 45.9 48.5 49.1 42.2 44.9 45.5 48.5

PM Peak

Car 72.8 58.2 48.7 53.0 50.6 54.8 56.0 47.2 51.3 49.6 53.5

LGV 72.8 58.0 48.3 52.3 50.2 54.1 55.8 46.9 50.7 49.2 53.0

OGV1 55.1 49.7 43.6 46.3 46.1 49.2 49.3 42.7 45.8 45.6 48.6

OGV2 55.3 49.0 42.4 44.8 45.6 48.5 49.1 42.0 44.9 45.4 48.2

As indicated above, the base scenario (with higher speed distributions) has the least difference from the theoretical desired

speed and modelled (actual) speed. For cars this is of the order of 13-17 mph and for OGVs of the order of 5-6mph.

Comparing Scenarios, Scenario 2 has consistently faster average modelled speeds than Scenario 1 due to the higher OGV

speeds modelled on S2 and 2+1 carriageways. The variance changes by period and method.

Similarly, when comparing methods, Method B with the higher aggression factors has consistently faster average modelled

speeds than Method A.

TRL Accident Analysis Summary



Transportation

5.1 Accident Analysis

AECOM provided speed distribution data from the A9 Traffic Model for vehicles in ten different simulations on four single

carriageway sections of the A9; two of these were on the two lane side of the ‘2+1’ single carriageway sections and two were on

standard S2 sections which have one lane in each direction.

For the purposes of the accident analysis, it was assumed that the average speeds on the standard S2 sections applied to all the

single lane sections (including those on the 2+1 sections). Similarly, the average speeds on the two lane sections were assumed

to apply to all the two lane section single carriageway sections. Treating the 2+1 sections in this way gives a total two lane length

of 4km and a single lane length of 254.4km (comprising 125.2km of S2 in each direction plus 4km of single lane from the 2+1

section).

The accident analysis has been derived based on (i) 14 years of historic accident data and (ii) 5 years of historic accident data.

The injury accident figures are presented in Table 24 based on 14 years of accident data. In purely financial terms, the increase

in injury accidents on the single carriageway sections as a whole, from Scenario 1 to Scenario 2, is equivalent to a cost of

between £495,685 and £650,185 per year.

Table 24 - Forecasts of the number of injury accidents per year on the A9, based on 14 years of historic accident data (Single Carriageway sections, Weekdays, 7am-7pm only)

Injury Accidents per Year Fatal & Serious Slight Total

Base 7.7 13.4 21.1

Range of estimates

Range of estimates

Range of estimates

Scenario 1 4.4 4.5 8.2 10.5 12.7 15.0

Scenario 2 5.1 5.3 9.0 11.8 14.2 17.0

Change from Base to Scenario 1 -3.3 -3.2 -5.2 -2.9 -8.4 -6.2


Change from Scenario 1 to Scenario 2 0.8 0.8 0.8 1.2 1.5 2.0

The injury accident figures based on 5 years of accident data are presented in Table 1. In purely financial terms, the increase in injury accidents on the single carriageway sections as a whole, from Scenario 1 to Scenario 2, is equivalent to a cost of between approximately £597,253 and £776,442 per year.

5 TRL Accident Analysis Summary



Transportation

Table 25 - Forecasts of the number of injury accidents per year on the A9, based on 5 years of historic accident data (Single Carriageway sections, Weekdays, 7am-7pm only)

Injury Accidents per Year Fatal & Serious Slight Total

Base 8.0 14.4 22.4

Range of estimates

Range of estimates

Range of estimates

Scenario 1 4.5 4.7 8.8 11.3 13.5 15.9

Scenario 2 5.3 5.5 9.6 12.6 15.1 18.0



Change from Scenario 1 to Scenario 2 0.8 0.8 0.8 1.3 1.6 2.1

The financial costs are based on the values of prevention provided for trunk roads in Reported Road Casualties Scotland 2012

and are thus based on 2012 values at 2012 prices.

A full report on TRL accident analysis is provided under separate cover.

Summary of Key Findings



Transportation

6.1 Overview

A micro-simulation traffic model has been developed to assist in the preparation of the A9 Dualling Programme Outline

Business Case. The traffic model has been developed using the Paramics micro-simulation software, which can model

overtaking manoeuvres on single carriageway sections of road. The A9 Traffic Model covers the road network from the

Inveralmond Roundabout just to the NW of Perth to the north of Loch Moy. In discussion with Transport Scotland, it was

agreed that the A9 Traffic Model could be used to model the introduction of the A9 Average Speed Camera (ASC) system for

the section of A9 between Inveralmond and Moy. The following section provides a summary of the key findings from the model

outputs and the subsequent analysis undertaken.

6.2 Key Findings

Using data from the A77 trunk road associated with the installation of an ASC system, AECOM have developed the A9 Traffic

Model to replicate as far as possible the implementation of an ASC system on the A9 based on available evidence. A second

modelling method (Method B) was also developed in part based on A77 ASC speed distributions and in part using default A9

Traffic Model aggression factors. The two methods help explore the sensitivity of model parameters on operational

performance along the A9. In reading and interpreting the A9 Traffic Model results it should be borne in mind that the input

parameters such as speed distributions and aggression factors have in large part been defined either based on the analysis of

the A77 ASC system or by adopting A9 Traffic Model defaults. As such certain model variables are prescribed and hence the

resulting model outputs will be a function of the prescribed input variables. The key findings of the research are outlined in the

summary table below.

Table 26 – Key Findings Summary Table

Comparison of Goods Vehicle 40mph limit (Scenario 1) vs

Goods Vehicle 50mph limit (Scenario 2)

Overtaking / Passing The introduction of ASCs is predicted to reduce passing on dual and 2+1 carriageway sections

across all scenarios.

Overtaking manoeuvres on S2 single carriageways are estimated to reduce across both

scenarios, with Scenario 2 reducing the most. Applying A9 Traffic Model default aggression

(Method B) marginally increases overtaking in Scenario 1 for the Ralia Straight Northbound

section.

Platoon Lengths Platoon dispersal is predicted to occur in both dual and 2+1 carriageways (2 lane section only).

For the modelled S2 and 2+1 single carriageways the platoon entry profiles in Scenario 2 across

both methods have a typically higher proportion of platoons in the one to ten vehicle categories.

Alternatively, Scenario 1 profiles have a higher proportion of platoons in the eleven to twenty and

over twenty vehicle categories.

The modelled entry profiles are similar on the Ralia S2 single carriageway exit profile (less noticeable on the Ralia 2+1 exit profile). This is due to the more limited number of passing manoeuvres undertaken along the S2 single carriageway.

Method B (default A9 Traffic Model aggression factor) results in a greater reduction in vehicles within a platoon state. It should also be noted that further analysis of platoon lengths over twenty vehicles revealed that Scenario 1 can generate platoons of up to 60 vehicles and Scenario 2 can generate platoons of up to 50 vehicles.

6 Summary of Key Findings



Transportation

Journey Times The introduction of ASC increases end to end journey times. Travel times decrease when aggression levels are raised (Method B).

Scenario 1 Method A increases average journey times for all vehicles by approx. 20 mins relative

to current (2012) conditions.

Scenario 2 Method A increases average journey times for all vehicles by approx. 14 mins relative


Scenario 1 Method B increases average journey times for all vehicles by approx. 10 mins relative


Scenario 2 Method B increases average journey times for all vehicles by approx. 5 mins relative


Speed Distributions Modelled speed distributions are in large part defined by the input parameters to the A9 Traffic

Model based on A77 data analysis and A9 Traffic Model defaults.

Car speed distributions on the Ralia S2 single carriageway section decrease significantly in Scenario 1 and 2 relative to the base. Top speeds are noticeably more restrained than in the dualling and 2+1 carriageway speed distributions (e.g. Scenario 1 shows significantly more cars travelling at speeds of between 34 mph and 42 mph). Scenario 1 also displays a flatter profile when compared to Scenario 2.

For OGVs the main difference is the reduced speeds in scenario 1 relative to Scenario 2. The

Scenario 1 speed distribution is shifted downwards by approximately 6 mph. With regards to

Method B, this shifts the speed distribution profile upwards almost mirroring the base distribution.

It should be noted that Method B speed distributions may not be representative of what would happen after implementation of an ASC system.

Accident Statistics Both scenarios 1 and 2 are predicted to reduce accidents relative to the current 2012 conditions.

If a change to OGV speed limits was implemented (i.e. a change from scenario 1 to scenario 2),

the number of injury accidents per year between 7am and 7pm on weekdays on single

carriageway sections would be expected to increase by the order of 1.5 and 2.0 using Method A

and B.

Appendix A



Transportation

Appendix A - Overtaking and Passing

Graphs

Appendix B



Transportation

Appendix B – Platooning Graphs

Appendix C



Transportation

Appendix C – Journey Time Tables

Appendix D



Transportation

Appendix D – Speed Distribution

Graphs

Appendix E



Transportation

Appendix E – Comparison between

Model and Observed Data

New A9 Average Speed Cameras – Traffic Modelling and...

Documents

Transcript of New A9 Average Speed Cameras – Traffic Modelling and...