Road traffic characteristics, driving patterns and ... · Road traffic characteristics, driving patterns and emission factors for congested situations P G Boulter, T Barlow, I S McCrae

Road traffic characteristics, driving patterns and emission factors for

congested situations P G Boulter, T Barlow, I S McCrae and S Latham (TRL)

D Elst and E van der Burgwal (TNO)

TRL Limited, Nine Mile Ride, Wokingham RG40 3GA, United Kingdom

TNO Automotive, Department Powertrains-Environmental Studies & Testing,

Schoemakerstraat 97, P.O. Box 6033, 2600 JA Delft, The Netherlands

Programme name EESD: Energy, Environment and Sustainable Development

Key action & RTD Priority 4.1.2. Improving the Quality of Urban Life

Project acronym OSCAR

Contract number EVK4-CT-2002-00083

Project title Optimised Expert System for Conducting Environmental Assessment of Urban Road Traffic

Website http://www.eu-oscar.org

PUBLICATION DATA FORM

Deliverable Number Deliverable 5.2

Deliverable Title Road traffic characteristics, driving patterns and emission factors for congested situations

Deliverable version number Version number 1

Workpackage WP5: Assessment of traffic parameters and emission factors relevant to congested flows

Nature of the deliverable Report

Dissemination level Public

Author(s): P G Boulter, T Barlow, I S McCrae and S Latham (TRL) D Elst and E van der Burgwal (TNO)

Contributors: M Haakana (FMI), S Larssen (NILU), T Maggos (NCSRD), B Martin (SICE), S Neville (Westminster), R San Jose (UPM), D van den Hout (TNO)

Reviewer(s): Dick van den Hout (TNO)

Keywords: OSCAR, EXHAUST EMISSIONS, CONGESTION

LIST OF OSCAR PARTICIPANTS

OSCAR - Optimised Expert System for Conducting Environmental Assessment of Urban Road Traffic

Official Logo Organisation Name Partner Number and Country

University of Hertfordshire 1, UK (Coordinator)

Westminster City Council 2, UK

Transport Research Laboratory 3, UK

Finnish Meteorological Institute 4, Finland

Helsinki Metropolitan Area Council 5, Finland

Norwegian Institute for Air Research 6, Norway

Oslo Department of Public Health 7, Norway

National Centre for Scientific Research, Demokritos

8, Greece

Technical University of Madrid 9, Spain

Spanish Electric Constructions Society 10, Spain

Netherlands Organisation for Applied Scientific Research

11, Netherlands

Municipality of Utrecht 12, Netherlands

OSCAR Deliverable 5.2

TABLE OF CONTENTS

EXECUTIVE SUMMARY

1 INTRODUCTION .......................................................................................................... 1

1.1 The OSCAR project..............................................................................................................................................1

1.2 Aims of vehicle emission measurement and modelling in OSCAR ...................................................................1

1.3 Report structure....................................................................................................................................................2

2 EMISSION MODELLING CONTEXT............................................................................ 4

2.1 The needs and resources of model users .............................................................................................................4 2.1.1 User needs.......................................................................................................................................................4 2.1.2 Available input data ........................................................................................................................................5

2.2 Definitions of congestion ......................................................................................................................................5

2.3 Trends in vehicle emission model development ..................................................................................................6 2.3.1 Aggregated emission factor models ................................................................................................................8 2.3.2 Average-speed models ....................................................................................................................................8 2.3.3 Adjusted or ‘corrected’ average speed models ...............................................................................................9 2.3.4 Modal models ...............................................................................................................................................10 2.3.5 Traffic situation models ................................................................................................................................12 2.3.6 Multiple linear regression models.................................................................................................................13

2.4 Overview of OSCAR methodology....................................................................................................................14

3 MEASUREMENT OF ROAD TRAFFIC CHARACTERISTICS ................................... 15

3.1 Driving pattern surveys......................................................................................................................................15 3.1.1 Instrumentation of test vehicle ......................................................................................................................15 3.1.2 Survey planning and execution .....................................................................................................................18

3.2 Traffic flow and composition .............................................................................................................................20

3.3 Processing of raw driving pattern data.............................................................................................................20

3.4 Results..................................................................................................................................................................27 3.4.1 Driving patterns and traffic conditions..........................................................................................................27 3.4.2 Relationships between traffic speed, flow and density..................................................................................31

4 DEVELOPMENT OF DRIVING CYCLES ................................................................... 34

4.1 Relationships between traffic speed, traffic density and driving dynamics...................................................34

4.2 A power-based approach....................................................................................................................................35

5 EMISSION MEASUREMENTS................................................................................... 40

5.1 Method.................................................................................................................................................................40


5.1.1 Test vehicles .................................................................................................................................................40 5.1.2 Test cycles ....................................................................................................................................................42 5.1.3 Test procedure ..............................................................................................................................................42

5.2 Results..................................................................................................................................................................43 5.2.1 Comparison with type approval limits ..........................................................................................................43 5.2.2 Emissions over the OSCAR cycles compared with emissions over the UDC ...............................................44 5.2.3 Average emissions by Euro class and cycle: hot exhaust emissions .............................................................46 5.2.4 Average emissions by Euro class and cycle: cold-start emissions.................................................................46 5.2.5 Average emissions by Euro class and cycle: idle emissions..........................................................................47

6 THE EFFECTS OF CONGESTION ON EMISSIONS................................................. 48

6.1 OSCAR measurements .......................................................................................................................................48

6.2 VERSIT+ modelling............................................................................................................................................48

6.3 Implications .........................................................................................................................................................51

7 EMISSION MODELLING IN THE OSCAR SYSTEM.................................................. 56

7.1 Proposed structure of the emissions module.....................................................................................................57

7.2 Emissions model ..................................................................................................................................................57 7.2.1 Hot exhaust emissions...................................................................................................................................59 7.2.2 Emissions at engine idle................................................................................................................................60 7.2.3 Cold start emissions ......................................................................................................................................60 7.2.4 Non-exhaust PM emissions...........................................................................................................................62

7.3 Traffic data pre-processor .................................................................................................................................62

8 SUMMARY ................................................................................................................. 64

8.1 Overview..............................................................................................................................................................64

8.2 Model review.......................................................................................................................................................64

8.3 Methodology........................................................................................................................................................64

8.4 Driving pattern surveys......................................................................................................................................64

8.5 Relationships between traffic speed, flow and density ....................................................................................65

8.6 Construction of dynamometer driving cycles...................................................................................................65

8.7 Emission measurements......................................................................................................................................66 8.7.1 Hot exhaust emissions...................................................................................................................................66 8.7.2 Cold-start emissions......................................................................................................................................67 8.7.3 Idle emissions ...............................................................................................................................................67 8.7.4 Effects of congestion on emissions ...............................................................................................................68

8.8 Emission modelling in the OSCAR system .......................................................................................................68 8.8.1 Emissions model ...........................................................................................................................................69 8.8.2 Traffic data pre-processor .............................................................................................................................69

9 CONCLUSIONS AND RECOMMENDATIONS .......................................................... 71


9.1 Conclusions..........................................................................................................................................................71

9.2 Recommendations and best practice .................................................................................................................72

10 ACKNOWLEDGEMENTS........................................................................................ 74

11 REFERENCES ........................................................................................................ 75

Appendix A: OSCAR driving cycles

Appendix B: Tabulated emission results

Appendix C: Comparison with type approval limits

Appendix D: Emissions over OSCAR cycles and UDC

Appendix E: Hot exhaust emissions by Euro class and cycle

Appendix F: Cold-start exhaust emissions by Euro class and cycle


EXECUTIVE SUMMARY

An important element in the development of the OSCAR Assessment System, and the overarching goal of Workpackage 5, was an improved understanding of driving characteristics, vehicle operation and exhaust emissions at low speeds, which should help to improve the modelling of emissions associated with different levels of congestion. The emissions work is summarised in this Deliverable. The principal tasks of the work were to characterise driving patterns in the four main cities (Athens, Helsinki, London and Madrid), to improve existing emission databases for slow-moving and stationary traffic, and to develop an emissions module for use in the OSCAR System.

One of the first steps in Workpackage 5 was to briefly review existing emission models. The review showed that although many emission and air pollution models utilise average speed emission functions, for the latest vehicle technologies average speed alone is not a reliable determinant of emissions on the street level, and some descriptor of driving cycle dynamics is also required. Cycle dynamics are usually defined in terms of speed-related parameters. However, the links between average speed, speed-based cycle dynamics and emissions are not firmly established. Engine load and engine power ought to be more directly related to emissions than speed-based parameters. However, many model users will only have information relating to traffic flow and speed. Some may have traffic composition information, but very few will have quantitative information on cycle dynamics, load and power.

In OSCAR, vehicle operation patterns, road characteristics and traffic conditions were recorded using on-board diagnostics (OBD) and a global positioning system (GPS) on specified links in each of the four main cities. The processed information from these surveys was entered into a database (Deliverable 5.1). Descriptive statistics and parameters were determined for each driving pattern, and generalised relationships between traffic speed, traffic density and driving pattern parameters were identified. Representative driving cycles were derived using these and other parameters (notably parameters describing engine power output), and exhaust emissions were measured over the driving cycles in the laboratory.

Average speeds in Helsinki were found to be higher than those on ‘equivalent’ roads in the other cities, and there was some indication that the driving was less ‘aggressive’. However, the between-city comparisons were confounded by a number of factors, including the distribution of the measurement periods, the phasing of traffic signals, road layout and pedestrian activity. The overall speed-flow relationship for the four cities was rather complex, and it was therefore difficult to support the development of generalised driving cycles using this information. There was less of a tendency for the shape of the speed-density relationship to be affected by the city and the design speed of the road. Although traffic density was not strongly related to driving dynamics (defined in terms of engine power output), in the absence of any other means of quantifying dynamics on the part of the model user it was considered reasonable to define congestion, and develop different driving cycles.

Emissions of the regulated pollutants from all test vehicles were generally below the type approval limit values. Otherwise, the margins of exceedance were not large. The test vehicles were therefore considered to be roadworthy, probably well-maintained, and not high emitters. Comparisons between emissions over the OSCAR cycles and the legislative Urban Driving Cycle (UDC) provided some evidence to suggest that, whilst CO and THC emissions can be higher over the UDC than over real-world congestion cycles with a similar speed, real-world NOx and PM from diesel vehicles may be underestimated. For petrol cars, cold-start emissions of CO and THC were up to several orders of magnitude higher than hot-start emissions, whereas for diesel vehicles the cold:hot ratio tended to be less than 5. The effects of cold starting on NOx

and PM were less pronounced, but still notable. Measurements also showed that emissions from diesel vehicles under engine idle conditions tended to be considerably higher than those from petrol and LPG vehicles, particularly for CO and NOx.

For cycles having the same average speed, cycle dynamics was found to have a notable effect for some vehicle types and pollutants, but the effects were not systematic. NOx emissions from petrol vehicles over the OSCAR cycles showed little dependence on speed. However, for a


given speed range the level of driving dynamics was found to be important, with higher emissions being recorded over cycles having higher dynamics. In contrast, NOx emissions from diesel vehicles over the OSCAR cycles exhibited a general dependence on speed, with little contribution due to driving dynamics. Emissions from LPG vehicles were comparable to those from petrol vehicles. PM emissions from diesel vehicles over the OSCAR cycles tended to decrease with decreasing speed, and cycle dynamics appeared to have an effect. Results from TNO’s VERSIT+ model were also used to provide information on congestion effects. Again, there was little evidence that changes in traffic density at a given average speed had a systematic effect on emissions, even though in some cases there were large differences between the predicted emissions associated with different traffic densities. Traffic density was not found to be a particularly effective surrogate for driving dynamics. However, in the absence of any other means of quantifying dynamics on the part of the model user it was considered reasonable to use traffic density as the best available solution in OSCAR.

Recommendations are provided in the report for the modelling of different road transport emission sources in the OSCAR Assessment System. For example, a more detailed exploration of the relationship between traffic density and cycle dynamics ought to be the subject of future research. Engine load and engine power ought to be more directly related to emissions than speed-based parameters. Any modelling approach should consider these parameters, but should bear in mind the limited input information available to most model users.

As part of the validation process for the OSCAR system, a series of on-board emission measurements was conducted in Central London. The results of these measurements are presented in a separate report.


OSCAR EVK4-CT-2002-00083 1

1 INTRODUCTION

1.1 The OSCAR project

OSCAR1 is a European Commission 5th Framework project which has delivered a tool to enable users to evaluate road traffic-related air pollution, and to identify suitable impact-reduction options. The tool (the OSCAR ‘Assessment System’) has been designed primarily to address the management of local air quality during periods of traffic congestion.

The development of a tool which can be applied across Europe requires an understanding of the similarities and differences between road traffic patterns, emissions and air pollution levels in different urban areas of different geographical regions. Consequently, measurement campaigns have been conducted in four ‘main’ European cities - Athens, Helsinki, London and Madrid - as part of the development of the OSCAR System. These campaigns have provided new data relating to the following:

• Driving patterns, vehicle operation conditions and traffic characteristics on main urban roads.

• Exhaust emissions associated with different levels of congestion.

• Roadside and urban background concentrations of nitrogen oxides (NOx), nitric oxide (NO), nitrogen dioxide(NO2), particles with a diameter of less than 10µm (PM10), and particles with a diameter of less than 2.5µm (PM2.5).

• Meteorological parameters.

Improved emission and air pollution models have been created from these measurements, and these models have been incorporated in the OSCAR Assessment System. The Assessment System approach has been checked by application in the main cities.

1.2 Aims of vehicle emission measurement and modelling in OSCAR

An important element in the development of the OSCAR Assessment System has been an improved understanding of driving characteristics, vehicle operation and exhaust emissions at low speeds, which should lead to improvements in the modelling of emissions associated with different levels of congestion. This was conducted within Workpackage 5.

In OSCAR, new exhaust emission measurements have been conducted on 20 passenger cars. However, for the OSCAR System to be applicable consistently on a European level, these new measurements would not, on their own, have been sufficiently extensive. Hence, the emissions work was not designed to provide a completely new emission model, but rather to supplement and improve the underlying data and methodologies of existing models. The usefulness of the emission measurements conducted in OSCAR was therefore maximised by ensuring that they conformed to the current needs of emission model developers. In particular, the measurements had to be compatible with the structures of the latest emission models being developed in Europe. The specific objectives of the emissions work were:

• To characterise driving patterns in the four main cities, based on the measurement of vehicle operation and traffic conditions.

• To improve existing emission databases by measuring emission factors for slow-moving and stationary traffic.

• To develop an Emissions Module for use in the OSCAR System which improves upon existing models.

1 OSCAR = Optimised expert System for Conducting environmental Assessment of urban Road traffic. www.eu-oscar.org


OSCAR EVK4-CT-2002-00083 2

1.3 Report structure

Various abbreviations used in the Report, and these are explained in Table 1.1. In the measurement and modelling of vehicle emissions, a number of different terms are often used to describe similar concepts or activities. Table 1.2 provides a brief glossary explaining how specific terms are used in the context of this report.

Table 1.1: Abbreviations used in the Report

APA average positive acceleration

ARTEMIS Assessment and Reliability of Transport Emission Modelling and Inventory Systems2

CADC Common ARTEMIS Driving Cycle

CO carbon monoxide

CO2 carbon dioxide

CNG compressed natural gas

CVS constant-volume sampler

COPERT COmputer Program to calculate Emissions from Road Transport

DGV Digitised Graz Method

EUDC Extra-Urban Driving Cycle

GDI gasoline direct-injection

GPS global positioning system

HBEFA Handbook of emission factors

HDV heavy-duty vehicle

LDV light-duty vehicle

LPG Liquified petroleum gas

NAEI National Atmospheric Emissions Inventory (UK)

NEDC New European Driving Cycle

NOx oxides of nitrogen

NO2 nitrogen dioxide

OBD on-board diagnostics

OSCAR Optimised Expert System for Conducting Environmental Assessment of Urban Road Traffic

PM particulate matter

PM10 Mass concentration of particles passing through a size-selective inlet designed to exclude particles greater than 10 µm aerodynamic diameter.

PM2.5 Mass concentration of particles passing through a size-selective inlet designed to exclude particles greater than 2.5 µm aerodynamic diameter. These are sometimes referred to as ‘fine’ particles.

RPA relative positive acceleration

THC total hydrocarbons

UDC Urban Driving Cycle

VOCs volatile organic compounds

2 European Commission 5th Framework project which will develop a harmonised emission model for road, rail, air and ship transport to provide consistent emission estimates at the national, international and regional levels. http://www.trl.co.uk/artemis/introduction.htm


OSCAR EVK4-CT-2002-00083 3

Table 1.2: Glossary of terms used in the Report

Driving cycle Defines how a vehicle is to be operated during a laboratory emission test. It is designed to reflect some aspect of real-world driving, and usually describes vehicle speed as a function of time. The driving cycle can be based upon real-world measurements (driving patterns).

Driving pattern Information on the evolution of vehicle operation with time, measured under real-world conditions.

Dynamics Variables which emission modellers use to describe the extent of transient operation in a driving cycle (e.g. maximum and minimum speed, average positive acceleration). Can be viewed as being similar to the concept of the ‘aggressiveness’ of driving.

Road characteristics Information relating to the road, such as the geographical location (e.g. urban, rural), the functional type (e.g.distributor, local access), the speed limit, the number of lanes and the presence or otherwise of traffic management measures.

Traffic characteristics/ conditions

Information relating to the bulk properties of the traffic stream – principally its speed, composition and volume/flow or density.

Transient Relates to when the operation of a vehicle is continuously varying, as opposed to being in a steady state.

Vehicle operation The way in which a vehicle is operated (e.g. vehicle speed, throttle position, engine speed, gear selection).

The first phase of the emissions work involved a brief review of the emission modelling approaches currently being used in Europe, placing the role of OSCAR into context. This review is presented in Chapter 2 of the Report.

Extensive real-world driving pattern and traffic measurements were conducted in the main OSCAR cities. The methodology for these measurements is presented in Chapter 3. Based on the real-world measurements, ten driving cycles were developed to reflect the low-speed vehicle operation associated with different congestion levels. The development of these driving cycles is discussed in Chapter 4. Emission measurements on 20 passenger cars were then conducted by TNO in Delft using the OSCAR driving cycles. The method and results of the emission measurement programme are provided in Chapter 5, and the effects of congestion on emissions are summarised in Chapter 6. A preliminary description of the OSCAR Emissions Module is presented in Chapter 7. This will be updated when the module has been constructed and fully implemented in the OSCAR System. The summary, conclusions and recommendations to date are presented in Chapter 8.

As part of the validation process for the OSCAR system, a series of on-board emission measurements was conducted in Central London. The objective was to provide information on the spatial and temporal distribution of emissions from a typical car being driven around the centre of a large city, and to compare these with the air quality predictions from the models in the OSCAR system. This work was conducted as part of Workpackage 5, but the results are presented in a separate report.


OSCAR EVK4-CT-2002-00083 4

2 EMISSION MODELLING CONTEXT

For reasons which will be explained later, most existing emission models are not particularly well-suited to evaluating the effects of different levels of congestion. One of the aims of OSCAR was to generate new information relating to low-speed vehicle operating conditions, in order to improve the modelling of different congestion-related scenarios and other pollution-reduction policies.

A large number of factors were taken into account during the design of the emission measurement and modelling methodologies in OSCAR, including:

• The needs and available input data of model users.

• Definitions of congestion.

• The sources of emissions from road vehicles, and the important pollutants.

• The types of emission modelling approach currently in use in Europe, and the approaches which would be best suited to the OSCAR Assessment System.

These points are addressed in more detail below, followed by a brief overview of the methodology employed in OSCAR.

2.1 The needs and resources of model users

2.1.1 User needs

Some specific considerations, in terms of the likely needs of potential users of the OSCAR System, are listed below.

• The types of policy to be evaluated using the OSCAR System

These might include policies dealing with the following: o The flow, speed and composition of the traffic on different links. o The fuels and technologies currently in use, and the cleaner fuels and technologies

which may be integral to some pollution-reduction strategies. o The relationships between road layout, driving patterns and emissions.

• Spatial and temporal scales

The OSCAR System will primarily be used to test the impacts of different transport policy options and scenarios on air quality. The System will need to be applicable on a range of spatial and temporal scales, and to both simple and complex situations, though the modelling will be driven by the requirements of the detailed, complex, and local applications.

• Country-specific driving considerations

The System is designed for use across Europe, but driving characteristics vary both within and between countries. The driving patterns in, say, Central London will be different to those in smaller UK cities, as well as those in other major European cities. The conditions represented in the OSCAR driving cycles had to be transferable between countries. Congestion may also be defined or perceived in different ways.

• Country-specific vehicle considerations

Although the emission measurements in OSCAR related to in-service vehicles subject to European type-approval, the vehicles were taken from the Dutch fleet. Some factors affecting vehicle emission performance, such as levels of servicing and maintenance, may vary by country, and so the OSCAR emission measurements therefore had to be combined with those from other European emission measurement programmes, such as ARTEMIS3 and PARTICULATES4, to give an emission database which reflected Europe as a whole.

3 ARTEMIS (Assessment and Reliability of Transport Emission Modelling and Inventory Systems) will develop a harmonised emission model for road, rail, air and ship transport to provide consistent emission estimates at the national, international and regional levels. http://www.trl.co.uk/artemis/introduction.htm

4 http://vergina.eng.auth.gr/mech/lat/particulates/


OSCAR EVK4-CT-2002-00083 5

2.1.2 Available input data

Model users tend to have limited input data on traffic activity and operation for a given road. They may have, for example:

• The level of the road in the network hierarchy (its functional class).

• The number of lanes (though the impact of roadside parking and temporary activities on the number of lanes is often variable or unknown).

• The speed limit.

• Details of traffic management (e.g. bus lanes, cycle lanes), and possibly the phasing of traffic signals.

• Traffic flow (volume). This may be measured or modelled (e.g. peak and off-peak flows). Traffic model outputs may or may not be suitable for emission modelling.

• Traffic composition. This is not always defined in relation to emission–related categories.

• Traffic speed. Often the speed is measured at the same time and location as the traffic flow, and is rarely measured at multiple locations or separated by vehicle class. The distribution of accelerations at a point may also be recorded.

Although this is probably not the optimum combination of parameters for emission modelling, any modelling approach needs to function in a pragmatic way.

2.2 Definitions of congestion

The treatment of congestion is an important aspect of the emission modelling work in OSCAR. A low-speed journey may be travelled at a relatively steady speed, or in a stop-start fashion whereby long periods of idling may be followed by brief periods of travel at relatively high speeds. In each case the journey may be of the same duration and length, but the emissions produced could be very different. It is therefore important to clarify what is meant by congestion, and OSCAR needs to adopt a measure of congestion that is meaningful in terms of emissions and can be applied by End Users.

The meaning of the term congestion is generally agreed to be something like "roads are congested when there's so much traffic that they get clogged up" (Hedges, 2001). However, public perceptions of congestion vary widely, and users may accept congestion on some parts of the road network that they would find unacceptable elsewhere. For example, a perceived low speed on a motorway may be very different to a perceived low speed in an urban environment. Respondents to a survey in the UK (Hedges, 2001) defined congestion as traffic which is completely immobile or moving very slowly, or any density-related slowing of traffic. The study also suggested the following as an alternative classification scheme for congestion:

a) Expected congestion, occurring at well-known bottlenecks in the system.

b) Exceptional congestion, occurring as the result of a predictable major event. These events are known about in advance but the capacity to cope with them has not been built into the highway network.

c) Unexpected congestion, resulting from random occurrences such as accidents, breakdowns, emergency highway repairs and adverse weather conditions.

It is likely that the OSCAR Assessment System will only be able to deal with the first type of congestion. It may be possible to take into account the other types of congestion at a later date using sophisticated traffic models.

These considerations have meant that a universally-applicable definition of congestion has not been developed. However, congestion tends to be described in terms of traffic speed, flow and journey time. The UK Commission for Integrated Transport (CfIT) and the UK government have both chosen to adopt measures of congestion based on the time lost by traffic not being able to proceed at a free-flow speed. CfIT focuses on the total hours lost, whilst the government has emphasised the average hours lost per km travelled. This is considered to be an indicator that is


OSCAR EVK4-CT-2002-00083 6

measurable and meaningful at all levels of the network, from national roads to local in urban areas. However, there is some concern that this indicator is not very meaningful to motorists and that indicators of congestion should be broadened.

Pignatoro et al. (in NCHRP, 1975) proposed the subdivision of congestion into two classes: saturated operations and oversaturated operations. Saturated operations are defined to be congested operations characterised by the formation of a local queue which only adversely affects the performance of the junction at which it forms. Oversaturated operations are characterised by queues that adversely affect at least one other major junction upstream. Traffic congestion reacts in a non-linear manner, meaning that when roadways are congested a relatively small reduction in traffic volumes can provide a relatively large reduction in delays (Victoria Transport Policy Institute, 2002).

Whichever definition of congestion is adopted, it needs to be measurable, easy to explain, relevant to peoples' experience and capable of being modelled. Few of the indicators currently in use fulfil all these criteria.

2.3 Trends in vehicle emission model development

The main sources of emissions from road vehicles, and the pollutants concerned, are listed below.

• Hot exhaust emissions: carbon monoxide (CO), volatile organic compounds (VOCs), NOx,PM, unregulated pollutants.

• Cold-start exhaust emissions: CO, VOCs, NOx, PM, unregulated pollutants.

• Evaporative emissions: VOCs

• Tyre and brake wear: PM

• Road surface wear: PM

• Resuspension: PM

Emission levels are dependent upon many parameters, including:

• Vehicle type • Technology level • Fuel type • Operation (speed, acceleration, gear..)

• Vehicle weight • Road gradient • Mileage • Level of maintenance

Although the emissions modelling approach used in the OSCAR System will need to consider these sources and parameters, only the development of new hot exhaust emission factors fall within the scope of OSCAR. Consequently, where the text of this Report refers to ‘emissions’, it should be assumed, unless otherwise stated, that this means hot exhaust emissions. The remainder of this Chapter is devoted to this subject. The treatment of emission sources other than hot exhaust in the OSCAR System is discussed in Chapter 7.

In order to understand how the emission modelling approach for passenger cars being developed in OSCAR might be combined with approaches being developed elsewhere, it is important to understand the nature of emission modelling. Models for estimating emissions from road vehicles can be classified in different ways, as summarised in Table 2.1.

Model classification systems tend to be based upon a combination of the geographical scale of application, the generic model type, and the nature of the emission calculation approach - a distinction can be made between models which use continuous emission functions, and models which use discrete emission values. The links between these different classification systems, with examples, are summarised in Table 2.2. The generic types of model are discussed in more detail in the following paragraphs. Some examples are provided, and their limitations are discussed. It should be noted that a large number of other models incorporate the databases of the models listed here.


OSCAR EVK4-CT-2002-00083 7

Table 2.1: Classification systems for light-duty vehicle emission models.

The scale of application The generic type Emission calculation approach

Type of input data

• Macro-scale (Country, region, city)

• Meso-scale (District)

• Micro-scale (Road link, road sub-section)

• Aggregate emission factors • Average speed • Adjusted average speed • Modal • Instantaneous • Traffic situation • Multiple regression

• Discrete • Continuous • Discrete/continuous

mix

• Trip-based • Link-based • Continuous by

operational mode

Table 2.2: Models for estimating emissions from light-duty vehicles5.

Generic type Example Type of EF Level of input data Type of input data Typical application

Aggregated EFs NAEI Discrete Trip-based Road type Emission inventories, EIA, SEA

Average speed COPERT Continuous Trip- or link-based

Average trip speed Emission inventories, dispersion modelling

Adjusted average speed

TEE Continuous Link-based Average speed, congestion level

Emission inventories, dispersion modelling

Accel/ decel/ cruise/ idle

UROPOL Discrete Link-based Distribution of driving modes

Instantaneous – ‘simple’

MODEM, DGV

Discrete Continuous Driving cycle

Modal

Instantaneous – ‘advanced’

VeTESS Discrete Continuous Driving cycle, gradient, vehicle

data

Detailed temporal and spatial analysis of

emissions, dispersion modelling

Traffic situation HBEFA Discrete Discrete Road type, speed limit, level of congestion

Inventories, EIA, SEA, area-wide assessment

of urban traffic management schemes, dispersion modelling

Multiple linear regression

VERSIT+ Discrete Link-based Driving cycle Emission inventories, dispersion modelling

5 Most of the models listed also address other types of vehicle, such as heavy goods vehicles and buses.


OSCAR EVK4-CT-2002-00083 8

2.3.1 Aggregated emission factor models

Description

Emission factor models operate on the simplest level, with a single emission factor being used to represent a particular type of vehicle and a general type of driving – the traditional distinction is ‘urban-rural-motorway’. The emission factors are calculated as mean values of measurements from a number of vehicles over given driving cycles, and are usually stated in terms of the mass of pollutant emitted per unit distance (g/vehicle-km) or per unit of fuel consumed (g/l). These factors are useful in applications on a large spatial scale, such as national and regional emissions inventories, where little detailed information on vehicle operation is required.

Limitations

Vehicle operation is only taken into account at a very rudimentary level, and the approach cannot be used to determine emissions for situations which are not explicitly covered by the emission factors.

2.3.2 Average-speed models

Description

Average-speed models are based upon the principle that average emissions from a given type of vehicle over a trip (with stops, starts, accelerations and decelerations) vary according to the average speed of the trip. Current average-speed models are exemplified by COPERT (COmputer Program to calculate Emissions from Road Transport). Figure 2.1 shows some examples of average-speed emission functions from COPERT III (Ntziachristos and Samaras, 2000). These functions are fitted to emission data measured over different driving cycles with a range of speed and acceleration characteristics. The input data requirements for average-speed models are straightforward. In principle, the input is the trip-based average speed, although in practice it is also common for local speed measurements taken at discrete locations to be used.

Broadly speaking, average-speed models are useful for large-scale applications such as regional and national emission inventories, if the trip data are available. However, average speed modelling has been used (not always appropriately) in a much wider range of applications, many of which relate to the meso-scale and micro-scale. Indeed, a large proportion of dispersion models currently use average-speed emission functions. Three main factors have probably contributed their widespread use. Firstly, average speed modelling is a long-established method. Secondly, the models are comparatively easy to use. Thirdly, there is a reasonably close correspondence between the required inputs and the data generally available to users.

0.0

0.2

0.4

0.6

0.8

1.0

0 20 40 60 80 100 120 140Speed (km/h)

Em

issi

ons

(g/k

m)

CO/20HCNOx

Figure 2.1: Average speed emission functions from COPERT III for petrol Euro 1 passenger cars, 1.4-2.0l.


OSCAR EVK4-CT-2002-00083 9

Limitations

One weakness of average-speed models is that the average speed of a trip may be made up in a number of different ways, with differing amounts of transient6 vehicle operation. Clearly, all the types of operation associated with a given average speed cannot be accounted for by the use of a single emission factor. This is not normally a problem at higher average speeds, as these are associated with relatively little variation in operation, but at the low average speeds associated with congestion the range of possible operational conditions associated with a given average speed tends to be much greater. The marked variability of vehicle operation at low average speeds is partly responsible for the poor reliability of the corresponding emission factors.

Furthermore, in response to the tightening of emission control legislation during the last few decades, vehicles have been equipped with increasingly sophisticated (and effective) after-treatment devices, with the result that emissions of regulated pollutants from new vehicles tend to be very low. With modern catalyst-equipped vehicles, a large proportion of the total emissions on a trip can be emitted as very short peaks, often (but not always) occurring during gear changes and periods of high acceleration. The use of after-treatment devices, manufacturer-specific engine management software, and regenerating after-treatment systems also make it much more difficult to predict emissions. Average speed is therefore a less reliable surrogate for the estimation of emissions for the newest generation of vehicles; the average speed model provides an impression of reality that is too simplistic.

Finally, the shape of the average speed function is not fundamental, but depends on, amongst other factors, the types of cycle used in development of the functions. Each cycle used in the development of the functions typically represents a given real-world driving condition, and the real distribution of these driving conditions is not normally taken into account (e.g. via weightings).

The concept of ‘cycle dynamics’

Due to these considerations, researchers have examined a range of variables which describe the extent of transient operation in a driving cycle - what has become known as ‘cycle dynamics’. This has become a useful concept for emission model developers. As the vehicle operation information available to model users and developer has tended to be speed-based, interest has inevitably focussed on parameters which describe speed variation in some way. Some of the more useful parameters appear to be relative positive acceleration (RPA) (e.g. Ericsson, 2000) and positive mean acceleration (e.g. Osses et al., 2002). However, there are two fundamental problems with the concept of speed-based cycle dynamics. Firstly, model users have little or no straightforward means of relating to descriptors of variation in vehicle operation, as these describe the properties of entire driving cycles (of course, this does not only concern speed). Most model users will only tend to have traffic flow and average speed information, and relationships between these parameters and those describing cycle dynamics on urban roads are not readily available. As a consequence, cycle dynamics has not usually been taken into account quantitatively. Secondly, several studies have concluded that emissions should be described in terms of engine speed, load, power, and the changes in these parameters, not just variables relating to vehicle speed (e.g.Leung & Williams, 2000; Kean et al., 2003).

2.3.3 Adjusted or ‘corrected’ average speed models

The TEE (Traffic Energy and Emissions) model (Negrenti, 1998) incorporates a ‘corrected average speed’ modelling approach. The model assumes that the effect of congestion on emissions at a certain average speed can be expressed by means of a ‘congestion correction factor’ derived from average speed, green time percentage, link length, and traffic density. The emission factor for the average speed is then adjusted using the correction factor. The congestion level is used to calculate the fractions of time spent during cruising, acceleration, deceleration and idling, and the end result is a reconstructed speed profile produced by the model itself. In fact, the TEE model uses emission factors from a simple instantaneous model to calculate emissions for each of the phases, based on the reconstructed profile. The limitations of this part of the approach are discussed under ‘simple instantaneous models’.

6 In this context, the term ‘transient’ refers to a driving cycle in which the operation of the vehicle is continuously varying, as opposed to being in a steady state.


OSCAR EVK4-CT-2002-00083 10

2.3.4 Modal models

Models based upon a coarse distribution of operational modes

Modal modelling is designed to improve upon the average speed approach, in particular for local-scale applications, by relating the modes of vehicle operation encountered on a given trip, in terms of the phases of steady speed, acceleration, deceleration, and idling, to the emissions produced during those modes. Modal models are therefore discrete in nature.

The Urban Road Pollution or UROPOL model (Hassounah and Miller, 1995) consists mainly of queuing and emission components. Based on queue lengths and the numbers of vehicles that are accelerating, decelerating, queuing or cruising at any point along a road segment, emissions are calculated using emission rates relating to each driving mode.

Simple instantaneous models

Description

Instantaneous emission models are, in principal, a complex extension of modal models which aim to provide a more precise description of a vehicle's operation. Vehicle emission rates are related to instantaneous speed and acceleration on a second-by-second basis. Such models can therefore be used to calculate the continuous emissions and fuel consumption for a particular vehicle type from a given driving cycle. Examples of instantaneous models include DGV (Digitised Graz model) (Sturm et al., 1994) and MODEM (Jost et al., 1992; Joumard et al., 1995) (Figure 2.2).

0 5

15 25

35 45

55

65

75

85

-15

-10

-5

0

51015

0

100

200

300

400

500

600

700

800

900

CO

emis

sion

s(g

/hou

r)

Speed (km/h)

Acceleration(m2/s3)

Figure 2.2: CO emissions as a function of speed and acceleration (adapted from the MODEM database).

Limitations

The most fundamental problem relating to simple instantaneous emission modelling is that it is that extremely difficult to measure emissions on a continuous basis with a high degree of precision, and emissions and fuel consumption values might not be successfully allocated to the correct operating conditions. For example, because of the time required to transport the exhaust gas to the analysers, and the actual response time of the analysers themselves, the emission signals are delayed relative to the driving cycle. Furthermore, the exhaust gas is mixed in the exhaust system. This results in a general flattening of instantaneous emission peaks over a period of more than one second. The dynamics of mixing also depend on the gas flow rate, and the situation is even worse when dilute exhaust gas is being sampled using a dilution tunnel. As a consequence, emission peaks often appear to bear no relationship at all to the operation of the vehicle. In addition, such models cannot yet take into account other important variables such as road gradient.


OSCAR EVK4-CT-2002-00083 11

Advanced instantaneous models

Description

The phenomena of time lag and damping have been illustrated by Weilenmann et al. (2000). In modern cars equipped with a three-way catalyst, oxygen peaks occur in fuel cut-off situations. Figure 2.3 shows a two-second fuel cut-off (~418s) at 60 km/h in such a car. In the top graph the raw exhaust valve signal is represented by the solid line, and the raw exhaust analyser oxygen signal is represented by the dashed line. The bottom graph shows the signal recorded by the analyser in the dilute exhaust. The fuel cut-off creates an oxygen peak of 1.2 seconds at the lambda sensor downstream of the catalyst. The raw gas analyser response follows at time 425s, and is much smaller. This shows that the dynamics of the raw exhaust gas line and the analyser are too slow for a peak of this duration to be measured accurately. In the dilute gas measurement the peak occurs at 437 seconds, and is even wider and flatter than the raw exhaust peak.

Figure 2.3: Oxygen concentrations after a two-second fuel cut off. (adapted from Weilenmann et al., 2000).

Results of this type have clear implications for the development of emission models based on instantaneous vehicle operation. In simple instantaneous models the time delay is usually taken into account by shifting the data backwards by a fixed number of seconds. However, Weilenmann et al. (2000) have shown that, when raw exhaust gas is being sampled, the delay is not constant, and varies by more than one second depending on the gas flow rate in the exhaust. Correcting the time lag by shifting the entire emission signal by a fixed number of seconds is clearly going to mean that emission events are temporally misaligned with the speed data, resulting in model inaccuracy. Over a transient driving cycle engine load varies every second, and hence there will inevitably be moments when the adjusted emission signal precedes the generation of the emission itself. The damping of the raw exhaust signal means that, in general, the 'real' emission peaks will be underestimated, and the emission troughs overestimated. Even if no original emission has occurred in a given instant, a model can produce a value because of the temporal spreading of the emission peaks. Using advanced measurement and modelling techniques, which are reliant upon knowledge of a number of test parameters and the solving of a series of differential equations, it is possible to estimate emissions from individual vehicles over short timescales. Limitations

Progress is being made towards the accurate modelling of emissions from individual vehicles on a continuous basis. However:

• There are currently no means by which these techniques can be applied to larger samples of vehicles in a way which resolves the issue of irregular emission behaviour, as the differences between vehicles coming onto the market are too pronounced. The (necessary) process of averaging over many vehicles to obtain representative emission estimates could cancel any improvements in accuracy associated with using a detailed model.


OSCAR EVK4-CT-2002-00083 12

• The effort required to model emissions from the newest vehicles is increasing, whereas vehicle emission levels are actually decreasing. This raises the question of whether instantaneous modelling is worthwhile.

• One potential benefit of instantaneous models is that they allow emissions to be resolved spatially. However, this would require the precise measurement of vehicle operation and location for all vehicles, otherwise any benefits of such models are lost. This is likely to be rather difficult for many model users.

• From a user perspective instantaneous models require detailed temporal information on the operational profiles of vehicles, and such information is relatively expensive to collect.

2.3.5 Traffic situation models

Description

One alternative approach to instantaneous modelling for incorporating both speed and cycle dynamics into emission estimations involves ‘traffic situation’ modelling, whereby cycle average emission rates are correlated with various driving cycle parameters. These, in turn, are referenced to specific ‘traffic situations’ which are known by the model user. Different traffic situations relate to conditions for which there is a specific emission problem. Traffic situation models tend to be best suited to local applications, in which emission estimates are required for individual road links, but can also be used for regional and national inventories.

The user must be able to relate to the way in which the traffic situations are defined in the model. For example, the Handbook of Emission Factors (HBEFA), used in Germany, Austria and Switzerland, is based on reference emission factors for a representative sample of cars, light-duty vehicles, diesel engines from HGVs, and motorcycles. Each emission factor is associated with a particular traffic situation, characterised by the features of the section of road concerned (e.g.'motorway with 120 km/h limit'; 'main road outside built-up area'). The speed variation (dynamics) variable is not quantified by the user, but is defined by a textual description (e.g. ‘free-flow’, ‘stop and go’) of the type of traffic situation to which an emission factor is applicable (INFRAS, 2004). As with any other model, the emission factors produced by the Handbook must then be further weighted by traffic flow and composition.

Limitations

There are two main issues regarding traffic situation models, and these are closely linked:

(i) What are the road and conditions leading to a given driving/operational pattern?

(ii) Is there a universally applicable set of traffic situations?

These questions are discussed below.

The Handbook of Emission Factors attempts to address the first question by asking the user to define the traffic situation using a textual description of speed variation or dynamics. However, this may lead to inconsistencies in the interpretation of the subjective element of the traffic definition. In addition, the Handbook employs definitions which are road- or traffic-based, rather than emissions-based. Even qualitative descriptions, such as those employed in the HBEFA, may be beyond many users, and are obviously open to interpretation. The second issue is addressed below.

At present, there are no commonly accepted definitions of universal traffic situations. There are likely be significant differences between the absolute characteristics of traffic in the different cities. However, it is known that there are fundamental underlying relationships between the characteristics of the road (e.g. number of lanes, carriageway width, topography), the prevailing traffic (e.g. flow, composition) and the operation of vehicles. For example, Figure 2.4 shows the generally accepted form of the relationship between traffic speed and flow on motorways (Hall et al., 1992; Akcelik, 2003). The upper part of the curve represents unsaturated traffic conditions with the flow below capacity (Qc). Correspondingly, the speed is between that at capacity (Vc) and that at the upper speed boundary – the free-flow speed (Vf). The absolute speed depends upon the design speed of the road. As the flow in this region increases, speeds are reduced below Vc due to the interactions between vehicles, and the lower part of the curve represents forced or


OSCAR EVK4-CT-2002-00083 13

saturated traffic flow conditions, with flow rates below capacity. These conditions lead to further reduced speeds and flow, and eventually to queuing or congested traffic. The boundary between the upper and lower parts of the curve relate to queue formation or dissipation respectively. The maximum flow value varies with the nature of the road and the speed limit, but has been found to be between around 2000 and 2500 vehicles per hour per lane for motorways (Small, 1992). This relationship is less clear for urban conditions, where a number of other factors, such as the presence and phasing of traffic signals, the presence of parked cars on the carriageway, and pedestrian activity all influence speeds. In many existing emission models, flow-speed relationships have not generally been incorporated quantitatively, even though they undoubtedly influence driving dynamics.

Speed (V)

Traffic flow (Q)

Saturated (forced)

traffic flow

Unsaturated traffic flow

Vc

Vf

Qc

Speed (V)

Traffic flow (Q)

Saturated (forced)

traffic flow

Unsaturated traffic flow

Vc

Vf

Qc

Figure 2.4: Typical form of the speed-flow relationship on motorways

One aim of COST 346 has been to define an internationally applicable traffic situation scheme. Traffic situations are defined on three levels: (i) the area type and road category (e.g. urban motorway, urban arterial, urban residential, rural motorway), (ii) a more detailed description for each area type and road category (e.g. number of lanes, parking areas, flow characteristics) and (iii) the speed limit and traffic density. The COST 346 Action (Emissions and Fuel Consumption from Heavy Duty Vehicles)7 has recommended the use of a traffic density parameter to infer variability in operation. Model developers therefore need to establish links between the traffic situation and the best descriptors of variation in operation.

2.3.6 Multiple linear regression models

The VERSIT+ model (Bremmers, 2004) employs a weighted-least-squares multiple regression approach, based on tests on a large number of vehicles over more than 50 different driving cycles. Within the model, each driving cycle used is characterised by a large number of descriptive parameters (e.g. average speed, RPA, number of stops per km) and their derivatives. For each pollutant and vehicle category a regression model is fitted to the average emission values over the various driving cycles, resulting in the determination of the descriptive variables which are the best predictors of emissions (the group of descriptors being different in each case). A weighting is also applied to each emission value, based on the number of vehicles tested over each cycle and the inter-dependence of cycle variables. The VERSIT+ model requires a driving cycle as the input, from which it calculates the same range of descriptive variables, and estimates emissions based on the regression results. The physical meaning of the variables may not necessarily be known. As with the other models requiring a driving cycle as the input, the use of the model will be restricted to a comparatively small number of users.

7 http://www.cordis.lu/cost-transport/src/cost-346.htm


OSCAR EVK4-CT-2002-00083 14

2.4 Overview of OSCAR methodology

One of the objectives of OSCAR was to improve low-speed emission factors for road vehicles, and to develop a method which would allow users to distinguish between different levels of traffic congestion in a pragmatic way. There were two principal tasks associated with these objectives:

1. To measure emission factors for different levels of congestion

2. To develop an emissions module for use in the OSCAR System

For the purposes of measuring vehicle emissions in the laboratory, a driving cycle is required. This usually takes the form of a speed-time profile with marked gear changes. Each driving cycle is designed to test certain aspects of emissions performance, and is usually representative of real-world driving under particular conditions (e.g. urban areas, motorways, etc.). The driving cycles in OSCAR were designed to reflect varying degrees of congestion on main roads in large urban areas, and to improve existing approaches to defining different levels of congestion from a perspective which was likely to be meaningful in terms of vehicle emissions.

From the review in the previous sections of this Chapter, the following conclusions were drawn:

(i) The most commonly used emission models describe emissions as a function of average speed. Many dispersion models currently in use, including those in the OSCAR System, utilise average speed emission functions.

(ii) For the latest (and future) vehicle technologies, average speed alone is a less reliable determinant of emissions. This is more pertinent for micro- and meso-scale applications, and dispersion models will need to be adjusted to account for this.

(iii) Some descriptor of the transient nature of vehicle operation (the level of dynamics) is ideally required in addition to average speed. Cycle dynamics are usually defined in terms of speed-related parameters. However, the links between average speed, speed-based cycle dynamics and emissions are not firmly established. Engine load and engine power ought to be more directly related to emissions than speed-based parameters.

(iv) Model users will only have information relating to traffic flow and speed. Some may have traffic composition information, but very few will have quantitative information on cycle dynamics.

These conclusions formed the basis for the experimental approach taken in OSCAR. The stages of the OSCAR methodology were as follows:

I. Vehicle operation patterns, road characteristics and traffic conditions were recorded on specified links in each of the four main cities.

II. The power characteristics of each driving pattern were determined.

III. Relationships were established between traffic speed and density, and the power-based driving pattern parameters.

IV. The real-world driving patterns were grouped according to these two sets of parameters.

V. Representative driving patterns were selected from the different groups using the power criteria, and used to construct driving cycles which reflected driving in all cities for given road and traffic conditions.

VI. Exhaust emissions were measured over the driving cycles in the laboratory, and the results were incorporated into European databases and models.


OSCAR EVK4-CT-2002-00083 15

3 MEASUREMENT OF ROAD TRAFFIC CHARACTERISTICS

One of the main objectives of OSCAR was to characterise the operation of light-duty vehicles in the participating cities, with particular emphasis on peak congested traffic conditions. This was achieved by the measurement, in each city, of driving patterns which described vehicle operation as a function of time. These driving patterns were subsequently condensed by into driving cycles - which represented defined traffic situations - for the purposes of measuring emissions in the laboratory.

The methodology for developing driving cycles and determining emission factors in OSCAR was designed to supplement and improve, as far as possible, the underlying data and approaches used in existing models, and models currently undergoing development in Europe. This also required the measurement of traffic characteristics during the driving pattern surveys. The measurements made in each city followed a broadly consistent methodology, as far as practically possible. This methodology is described in the following Sections.

3.1 Driving pattern surveys

Driving pattern data were collected during surveys in each of the four main OSCAR cities – Athens, Helsinki, London and Madrid – in order to characterise vehicle operation during congested urban peak and off-peak traffic conditions. The responsible partners and dates for driving pattern surveys are given in Table 3.1.

Table 3.1 Responsible partners and dates for driving pattern surveys.

City Responsible partner(s) Survey dates

Athens NCSRD 15/7/03 to 20/7/03

Helsinki FMI 8/9/03 to 15/9/03

London TRL, UH, Westminster CC 27/6/03 to 2/2/03

Madrid UPM, SICE 6/10/03 to 12/10/03

In each city, a data acquisition system was fitted inside a passenger car to enable its road speed and other parameters to be logged on a continuous basis. The measurement procedure involved the following tasks:

• Establishing the timing (month) and general location of the survey • Hiring and instrumenting a test car • Drawing up a plan of the routes to be driven by time of day • Hiring drivers • Conducting the measurements • Making a video recording of the routes • Collating the raw data and sending to TRL for processing and analysis

3.1.1 Instrumentation of test vehicle

Recent technological developments in vehicle on-board diagnostics (OBD) and global positioning systems (GPS) have made possible the efficient collection of real-world driving pattern data, and allow the continuous logging of operational parameters and location. These technologies were exploited in OSCAR.

OBD

Equipment is available which can read the OBD output from a Euro III (or later) petrol car compliant with European on-board diagnostics (EOBD) standards. On-line data acquisition is achieved via a simple interface (Figure 3.1) that is connected to the serial port of a standard PC having OBD-scanning software. EOBD-compliant vehicles must be fitted with a standard


OSCAR EVK4-CT-2002-00083 16

connector which is located in an accessible area, such as between the steering column and the central console of the car. Vehicles must also be capable of communicating according to one of the approved EOBD protocols - ISO 9141, J1850, KWP2000 or CAN, and must provide a minimum quantity of information. All new petrol light-duty vehicles manufactured after 1 January 2002 should have been fitted with EOBD.

Figure 3.1: OBD connection

The software used in OSCAR for reading the OBD output (called Vehicle Explorer Scan Tool Browser, Version 1.06)8. The software scans a variety of vehicle operation parameters in real time, and the live data to be scanned and stored to disk for later analysis. The parameters scanned include, for example:

• Fuel system status (open loop, closed loop) • Percentage engine load • Engine coolant temperature (°C) • Intake manifold absolute pressure (kPa) • Engine speed (rpm)

• Vehicle speed (km/hr) • Intake air temperature (°C) • Mass air flow rate (g/sec) • Oxygen sensor output (volts) • Throttle position

However, if many parameters are selected simultaneously, the logging frequency decreases. A desirable logging frequency for emissions (and hence driving pattern) measurements is 10Hz (Atjay et al., 2005). This frequency could not be attained using OBD. Preliminary tests at TRL indicated that a logging frequency of just under 2 Hz could be achieved if only four parameters were logged. The four most useful parameters in the context of OSCAR were taken to be:

• Vehicle speed (resolution 1 km/h)9

• Engine speed (resolution 1 rpm) • Percentage engine load • Percentage throttle position

The use of these parameters in the development of driving cycles is described in more detail in Chapter 4. Figure 3.2 shows an example OBD output from one of the driving pattern surveys, and illustrates the complex nature of the relationships between different vehicle operation parameters.

8 Available free of charge on the internet at: http://www.obd-2.com/#view.

9 High greater speed resolution would have been desirable, though the resolution of 1 km/h is defined in EOBD protocols. This limitation was not considered to be problematic, as filtering (smoothing) of driving cycles is required anyway prior to analysis.


OSCAR EVK4-CT-2002-00083 17

0

1000

2000

3000

4000

Eng

ine

spee

d(r

pm)

0

20

40

60

80

100

Eng

ine

load

(%)

0

10

20

30

40

0 50 100 150 200 250 300 350 400 450

Time (s)

Veh

icle

spee

d(k

m/h

)

Figure 3.2: Example output from OBD

GPS

A GPS receiver (Figure 3.3) was used to log the location, speed, bearing, trip distance, and altitude of the instrumented vehicle. The receiver was used to determine information such as The GPS receiver was powered directly from the laptop computer. TRL has developed a computer program that allows points to be selected on a latitude/longitude plot and then automatically extracts the appropriate speed data corresponding to the sections of road between the points. Once the points have been defined, then the data for the same sections can be automatically extracted from similar trips.

Figure 3.3: GPS receiver


OSCAR EVK4-CT-2002-00083 18

In principle, the performance of different vehicles influences the driving cycle measurements. Ideally, therefore, a range of vehicles (e.g. different power/weight ratio) ought to have been instrumented in each city. However, this was not possible within the scope of the project. Therefore, a single passenger car was instrumented in each city. The differences in the performance of each test vehicle were not considered to be very important in the development of congestion-related emission factors, since the operation of vehicles in dense city centre traffic is usually dictated by the traffic itself rather than vehicle performance.

3.1.2 Survey planning and execution

The driving pattern measurements were timed to coincide, as far as possible, with the air pollution measurement campaigns. Holiday periods were avoided so that a wider range of traffic conditions could be encountered. The routes were designed to reflect, as practicably as possible, typical driving conditions on main roads in the study areas (i.e. large urban centres), and coincided with the locations of the air pollution monitoring sites. Each responsible partner was required to draw up a route plan for their city, based on their local knowledge. The selected routes also incorporated roads on which traffic counts were being conducted, and the characteristics of these roads were recorded in detail. The routes were then divided into distinct ‘links’, to which the driving, traffic and road characteristics could later be referenced. The link characteristics included the following: • Total number of lanes • Speed limit • Street canyon/open road • Gradient (if information is available) • Presence of bus/taxi/cycle lanes • Locations of traffic signals and intersections • Details of any other traffic management

On some links the value of these parameters changed. For example, the total number of traffic lanes may have been reduced at a given point by a permanent road narrowing. In such cases, the values used were taken to be those applicable to most of the link.

The measurements were conducted, using different drivers, for between 8 and 10 hours on each day of the week, for at least six full days (including Saturday and Sunday), and covered both the peak and off-peak periods. A range of traffic conditions were covered (from ‘free of traffic’ to ‘highly congested’). If the routes were relatively long, the starting point was changed so that peak traffic flows were encountered on different roads each day.

Prior to each survey the drivers were instructed to drive as normally as possible along the routes, and in such a way that their speeds reflected those of the surrounding traffic. A navigator/computer operator was employed for reasons of safety, and safe stopping points were designated for the start and end points of each route. Furthermore, an allowance was made for a period of adaptation on the part of the drivers; the drivers were allowed to familiarise themselves with the vehicle(s) and the presence of a navigator and computer before actual data logging begins. Drivers were restricted to 2-3 hour sessions to minimise tiredness.

To compliment the logged data, an information sheet was also completed for each trip. The sheet included details of the time and date, the driver's name, the route number, the logged file number, and any comments regarding the route (adverse weather conditions, congestion, road works, etc.). In addition, a video recording of the route was made to aid the interpretation of data. During the surveys, the raw measurement data were provided to TRL for checking and analysis.

Figures 3.4 to 3.9 depict some of the roads included in the driving pattern surveys.


OSCAR EVK4-CT-2002-00083 19

Figure 3.4: Athens – Kifisias Figure 3.5: Athens - Patision

Figure 3.6: Helsinki – Mannerheimintie Figure 3.7: Helsinki – Runeberg Street

Figure 3.8: London - Baker Street Figure 3.9: London – Marylebone Road


OSCAR EVK4-CT-2002-00083 20

3.2 Traffic flow and composition

Traffic information was obtained for as many of the links covered by the driving routes as possible. Traffic flow (volume) and composition were determined using existing automatic counters. To minimise costs, the driving cycle routes were designed to take into account the locations of existing count sites. Ad hoc traffic counts were conducted if the number of existing sites was low, or if the data from existing sites yielded insufficient detail. Individual vehicle counts (whereby the presence of each passing vehicle is time stamped) from any permanent site were preferred, but such information was not available in all four cities (Table 3.2). The availability of traffic counting sites affected the design of the driving pattern surveys. For example, in the case of Athens, where only a few traffic counting locations were available, a large number of driving patterns were recorded on a small number of roads. In Madrid, on the other hand, the traffic was monitored at many locations, and therefore a large number of roads were included, although each road was only driven a small number of times.

Table 3.2 Traffic characterisation approaches adopted in the four cities.

City Number of count sites Site type Method

Temporal resolution

Traffic composition

Athens 2 (1 two-way)

Temporary P 1 hour No

Helsinki 4 (all two-way)

Permanent I V (3 sites) 15 mins (1 site)

At 3 sites

London 5 (all two-way)

Temporary (4 sites) Permanent (1 site)

PI

V15 mins

At 4 sites

Madrid >50 Permanent I 15 mins No

P = pneumatic tubes I = Induction loops V= individual vehicle counts

3.3 Processing of raw driving pattern data

The raw OBD data were processed through the OBD scanning software to produce time-stamped text files. The GPS files, in binary format, were converted to text files using TRL software. A cubic spline function (Press et al., 2002) was used to fit the OBD and GPS data to a regular time base (one second), and to correlate and combine the two data sets in time. The GPS co-ordinates for each trip were plotted, and obvious errors were noted. If the errors were minor, such as a few missing or incorrect data points, then these were interpolated. If the errors were major (e.g. if the GPS signal had been lost or the coordinates did not conform to expected), then the data were not used. Based on maps of each city, the routes driven were divided into a number of shorter links according to the locations of the traffic counting sites. Major intersections were selected as the end points of each link. The routes driven and the extracted links are depicted in Figures 3.10 to 3.13. The characteristics of those links having traffic flow information are detailed in Table 3.3.

As stated in the previous section, the approach used by each partner differed according to the number of available traffic count locations in the city. Consequently, a relatively small number of links with traffic counts were covered in Athens (3), Helsinki (8), and London (5), whereas a large number were driven in Madrid (37).

Form the GPS plots, the average coordinates of link start and end points were calculated, and the GPS and OBD data for the links of interest were then extracted. For each driving pattern recorded on a link, a number of descriptive parameters relating to road speed, engine load and throttle position were determined. These are listed in Table 3.4. The ArtKinema program (de Haan and Keller, 2003) was used to derive more than 30 other speed-based cycle parameters, also shown in Table 3.4. Based on the road layout and traffic conditions on each link, the traffic flow per lane per hour, the traffic density (vehicles per km), and the number of signalised junctions per km were also calculated.


OSCAR EVK4-CT-2002-00083 21

A01A01

A03

A02

A03

A02

Figure 3.10: Links covered in Athens driving pattern survey. The thick blue lines represent links with traffic flow information.

A. Patision

b. Kifisias


OSCAR EVK4-CT-2002-00083 22

H01/H02

H03/H04

H07/H08

H05/H06

H01/H02

H03/H04

H07/H08

H05/H06

Figure 3.11: Links covered in Helsinki driving pattern survey. The thick red lines represent links with traffic flow information.


OSCAR EVK4-CT-2002-00083 23

L05

L01

L02

L03

L04

L05

L01

L02

L03

L04

Figure 3.12: Links covered in London driving pattern survey. The thick green lines represent the links with traffic flow information. Driven anti-clockwise only.


OSCAR EVK4-CT-2002-00083 24

M21

M20

M16/M18

M17/M19

M31

M22

M36

M32M33

M34

M29

M25

M28

M27

M13a

M24

M08

M23M07

M11

M06

M10

M05

M01

M02

M09

M03/M35

M04

M13b/M14

M26

M30

M37

M38

M21

M20

M16/M18

M17/M19

M31

M22

M36

M32M33

M34

M29

M25

M28

M27

M13a

M24

M08

M23M07

M11

M06

M10

M05

M01

M02

M09

M03/M35

M04

M13b/M14

M26

M30

M37

M38

Figure 3.13: Links covered in Madrid driving pattern survey. The thick orange lines represent links with traffic flow information.


OSCAR EVK4-CT-2002-00083 25

Table 3.3: Link descriptions in driving pattern database (links with traffic flow measurements only).

City (link) Link name Direction City (link) Link name Direction

Athens (A01) Patision S’bound Madrid (M13a) J.Costa S’bound (E)

Athens (A02) Kifisias N’bound Madrid (M13b) F Silvela S’bound (E)

Athens (A03) Kifisias S’bound Madrid (M14) F Silvela N’bound (W)

Helsinki (H01) Marvagen E’bound Madrid (M16) Ron Atocha/Ron Valencia W’bound (S)

Helsinki (H02) Marvagen W’bound Madrid (M17) Ron de Toledo W’bound (N)

Helsinki (H03) Vihdintie S’bound Madrid (M18) Ron de Toledo E’bound (S)

Helsinki (H04) Vihdintie N’bound Madrid (M19) Ron Atocha/Ron Valencia E’bound (N)

Helsinki (H05) Mannerheimintie S’bound Madrid (M20) Pas SM de la Cabeza S’bound (W)

Helsinki (H06) Mannerheimintie N’bound Madrid (M21) Pas de las Delicias N’bound

Helsinki (H07) Runeberginkatu S’bound Madrid (M22) Av de Menendez Pelayo N’bound (W)

Helsinki (H08) Runeberginkatu N’bound Madrid (M23) Alfonso XIII S’bound

London (L01) Baker Street S’bound Madrid (M24) Av de Concha Espina/Ramon y Cajal W’bound

London (L02) Park Lane S’bound Madrid (M25) Serrano1 S’bound (W)

London (L03) Victoria Street E’bound Madrid (M26) Serrano2 S’bound

London (L04) Regent Street N’bound Madrid (M27) Serrano3 S’bound

London (L05) Marylebone Road W’bound Madrid (M28) Serrano4 S’bound

Madrid (M01) Princesa S’bound (E) Madrid (M29) Serrano5 S’bound

Madrid (M02) Gran Via S’bound (E) Madrid (M30) Alcala E’bound (N)

Madrid (M03) Paseo Recolto (S) N’bound Madrid (M31) O'Donnell E’bound

Madrid (M04) Paseo Cellana N’bound Madrid (M32) Doc Esquerdo N’bound

Madrid (M05) San F de Sales S’bound (W) Madrid (M33) Alcala S’bound (W)

Madrid (M06) Av Reina Victoria W’bound Madrid (M34) Goya W’bound

Madrid (M07) Bravo Murillo S’bound Madrid (M35) Paseo Recolto (S) S’bound

Madrid (M08) Paseo Cellana N’bound Madrid (M36) Paseo del Prado S’bound

Madrid (M09) Paseo Cellana S’bound Madrid (M37) Gran Via de Saf Fran. N’bound

Madrid (M10) Gen Martinez Campos W’bound Madrid (M38) Mayor/Bailen E’bound

Madrid (M11) Santa Engracia N’bound (W)


OSCAR EVK4-CT-2002-00083 26

Table 3.4: Link descriptors.

Trip code Link duration Avg. engine speed

City Link distance % time eng. spd=>1000rpm

Extracted data file name Average speed % time eng. spd=>1500rpm

Link name Avg. driving speed % time eng. spd=>2000rpm

Link number SD of speed % time eng. spd=>2500rpm

Direction 75-25 percentile speed % time eng. spd=>3000rpm

Nominal link length Maximum speed Avg. engine load (%)

Date % time idling % time eng. load=>20%

Day of week % time spd <=5km/h % time eng. load=>40%

Identifiers

Link start time % time spd <=10km/h % time eng. load=>60%

% time spd <=20km/h % time eng. load=>80%

Functional road class % time spd <=30km/h % time eng. load=>90%

Speed limit Driving time Avg. throttle position (%)

Tot no. of traffic lanes Cruise time % time TP>=10%

No. of bus lanes Drive time accelerating % time TP>=20%

Traffic count lanes Drive time decelerating % time TP>=30%

Number of traffic lights Drive time braking % time TP>=40%

Road characteristics

Street canyon/open road % of time speed = 0 % time TP>=50%

% of time driving Number of stops

Total flow % of time cruising Number of stops/km

Bus flow % of time accelerating Avg. stop duration

% HDV in car lane % of time decelerating Distance between stops

Average sped % of time braking Relative positive speed

Traffic characteristics

Average headway Avg. acceleration Relative real speed

Avg. positive accel. Relative square speed

Make Avg. negative accel. Relative pos. square speed

Model SD of accel. Relative real square speed

Year SD of pos accel. Relative cubic speed

Engine size 75-25 percentile accel.

Vehicle characteristics

Power No. of accelerations

No. of accels. per km

Traffic flow/lane/hour Relative positive accel.

HDV flow/lane/hour Positive kinetic energy

Traffic lights/km RMS acceleration

Derived values

Traffic density/lane % time neutral

% time gear 1

Driver % time gear 2

Weather % time gear 3

Event % time gear 4

Event location % time gear 5

Comments

Event time

Cycle parameters


OSCAR EVK4-CT-2002-00083 27

3.4 Results

3.4.1 Driving patterns and traffic conditions

Database summary

The processed information from the driving pattern surveys was entered into a database at TRL, and the information was used to examine the differences in the driving characteristics in each of the four main cities. In order to provide an indication of the size and composition of the database, Table 3.5 shows the numbers of driving patterns recorded by city and speed limit. A total of 1,217 driving patterns were recorded. The largest number of driving patterns (434) was recorded in Helsinki, and the lowest number in London (207). By far the largest proportion of driving patterns were recorded on roads with a 48 km/h or 50 km/h speed limit (56.1% in total). These numbers do not take into account the actual distance driven in each city.

Only the roads having a speed limit of 70 km/h or lower could be considered as distinctly ‘urban’ in nature. The roads with higher speed limits were considered to be more typical of inter-urban trunk roads and motorways. Little further analysis was therefore conducted on the data for roads with a speed limit above 70 km/h. The subsequent analysis focused mainly on the 918 driving patterns which were measured on ‘urban’ roads, and which had corresponding traffic information (Table 3.6), although some information relating to other roads is also presented in this Chapter.

Table 3.5: Numbers of driving patterns by city and speed limit

(blank spaces indicate no measurements).

Number of trips

Athens Helsinki London Madrid TOTAL

40 km/h limit 108 108 (8.9%)

48 km/h limit 164 164 (13.5%)

50 km/h limit 177 108 233 518 (42.6%)

60 km/h limit 160 11 171 (14.1%)

64 km/h limit 43 43 (3.5%)

70 km/h limit 109 109 (9.0%)

80 km/h limit 98 98 (8.1%)

90 km/h limit 6 6 (0.5%)

TOTAL 337 (27.7%)

434 (35.5%)

207 (17.0%)

239 (19.6%) 1,217 (100%)

Table 3.6: Numbers of urban driving patterns with traffic information by city and speed limit (blank spaces indicate no measurements).

Number of trips

Athens Helsinki London Madrid TOTAL

40 km/h limit 108 108 (11.8%)

48 km/h limit 114 114 (12.4%)

50 km/h limit 71 106 196 373 (40.6%)

60 km/h limit 160 11 171 (18.6%)

64 km/h limit 43 43 (4.7%)

70 km/h limit 109 109 (11.9%)

TOTAL 231 (25.2%)

334 (36.4%)

157 (17.1%)

196 (21.4%)

918 (100%)


OSCAR EVK4-CT-2002-00083 28

Driving pattern and traffic flow characteristics

Figure 3.14 shows the grand mean speed by city and speed limit. For roads with a 50 km/h limit (48 km/h in London), which formed a large proportion of the database, the highest and lowest overall mean speeds were recorded in Helsinki (32 km/h) and London (14.8 km/h) respectively. Indeed, the average speeds in Helsinki were generally higher than those on ‘equivalent’ roads in the other cities. For example, speeds on roads with a 40 km/h limit in Helsinki were higher than on the speeds on roads with a 50 km/h limit in the other cities, and similar to speeds on the road with a 64 km/h limit in London.

Similarly, Figure 3.15 shows the grand mean RPA by city and speed limit. This provides an indication of the level of dynamics (what could be termed the ‘aggressiveness’) of the driving in the four cities. It should be noted that speed and RPA are not independent – RPA tends to decrease as speed increases. The mean RPA values in Helsinki were somewhat lower than those in the other cities. In particular, for roads with a similar actual speed (e.g. Hel 40, Hel 50 and Lon 64), the RPA values tended to be lower in Helsinki. This may indicate that driving is generally less aggressive in Helsinki than in the other cities. By contrast, the Ath 60 links had a particularly high RPA for the average speed, indicative of more aggressive driving.

However, it would not be wise to read too much into these results, as the between-city comparisons are confounded by a number of factors, including the distribution of the measurement periods in each city and, by inference, the volumes of traffic, the phasing of traffic signals, road layout and pedestrian activity. It should be noted that roads with different speed limits tend to have different functional characteristics, and this further obscures between-city comparisons.

In an attempt to provide a more meaningful comparison, for each link studied the mean speed and the mean of the hourly traffic flow values corresponding in time to the speed measurements are shown in Figure 3.16. The overall mean values by city and speed limit are shown in Figure 3.17. The traffic flow values relate to ‘light-duty vehicle equivalents’ (otherwise known as ‘passenger car units’) in order to allow a standardisation according to between-city differences in traffic composition. Each heavy-duty vehicle in the traffic was assumed to be equivalent to 2.5 light-duty vehicles (OECD, 2003).

Figure 3.17 shows that, as might be expected, both the mean speed and the mean traffic flow (per unit time) increased with an increase in the speed limit, although on roads in Helsinki with a 70 km/h speed limit the traffic flow appeared to be relatively low.

The relationships between traffic speed, flow and density are examined in more detail in the following Section.


OSCAR EVK4-CT-2002-00083 29

0

10

20

30

40

50

60

70

80

90

Ath50

Ath60

Hel40

Hel50

Hel60*

Hel70

Hel80

Lon48

Lon64

Mad50

Mad90

City and speed limit (km/h)

Gra

ndm

ean

link

spee

d(k

m/h

)

Figure 3.14: Grand mean link speed by city and speed limit, including 95% confidence intervals (* NB normally 80 km/h – reduced speed limit in place).

0.00

0.05

0.10

0.15

0.20

0.25

0.30

Ath50

Ath60

Hel40

Hel50

Hel60*

Hel70

Hel80

Lon48

Lon64

Mad50

Mad90

City and speed limit (km/h)

Gra

ndm

ean

link

RP

A(k

m/h

)

Figure 3.15: Grand mean RPA by city and speed limit, including 95% confidence intervals (* NB normally 80 km/h – reduced speed limit in place).


OSCAR EVK4-CT-2002-00083 30

0

20

40

60

80

100

120

Link (direction) and speed limit

Ove

rall

mea

n sp

eed

(km

/h)

0

100

200

300

400

500

600

700

800

900

1000

1100

Traf

fic fl

ow (v

eh/h

our)

Athens MadridLondonHelsinki

Figure 3.16: Mean speed and corresponding hourly traffic flow on each link


OSCAR EVK4-CT-2002-00083 31

0

10

20

30

40

50

60

70

80

90

0 200 400 600 800 1000 1200

Mean traffic flow (vehicles per hour)

Gra

ndm

ean

link

spee

d(k

m/h

)

Ath60

Mad50

Hel70

Lon48

Hel40Hel50

Lon64

Hel60*

Hel80

Ath50

Figure 3.17: Overall mean speed and hourly traffic flow by city and speed limit (*normally 80 km/h – temporary limit)

3.4.2 Relationships between traffic speed, flow and density

Figure 3.18 shows the speed-flow relationship for links with a 50 km/h speed limit (48 km/h in London). Each point in Figure 3.18 represents a single driving pattern. Although it could be imagined that there is a suggestion of the idealised ‘inverted C’ relationship, depicted for high-speed roads in Figure 2.5, Figure 3.18 suggests that the relationship in urban areas is rather complex, especially when different locations are compared. Numerous factors contribute to the variation in the data, including the phasing of traffic signals, pedestrian activity, parked cars, etc.Furthermore the traffic flows are 1-hour mean values, and so the connection between the driving pattern and the traffic flow at the time of measurement is not precise. In addition, as stated earlier, the form of the speed-flow relationship is dependent upon the design speed of the road, and so the measurements on roads with other speed limits are not included. Hence, although different areas of the plot can be identified (e.g. no congestion/low flow, no congestion, high flow Light congestion, and heavy congestion), it was difficult to support the development of generalised laboratory driving cycles using this approach. Treating roads with different cities or speed limits separately would have resulted in more cycles than could be tested in OSCAR. Grouping roads by functional class did not result in better relationships, partly because the functional class parameter is rather subjective, and descriptions vary from country to country.

For each driving pattern measured on urban roads in OSCAR, the 1-hour average traffic density (v/km, again based on LDV equivalents) was also calculated for the corresponding period, following the recommendations of COST 346 in relation to traffic situation modelling. The traffic density value was calculated as the 1-hour average traffic flow divided by the 1-hour average speed. Ideally, traffic density would be measured independently using, for example, aerial photography. However, this was beyond the scope of the project. Average link speed was plotted against traffic density for these links, as shown in Figure 3.19. Again, each point in Figure 3.19 represents a single driving pattern.

Average speed tended to reduce sharply with increasing traffic density. In other words, the speed tends to decrease as the separation between vehicles decreases. For traffic densities higher than around 35 vehicles per km per lane, only average speeds lower than 30 km/h were encountered, and speed reduced much more gradually with further increases in traffic density. At the highest traffic densities recorder, only very low speeds tended to be observed. This logarithmic form of the speed-density relationship is similar to the model was put forward by Greenberg (1959), although other models have been proposed, such as the linear version advanced by Greenshields (1935).


OSCAR EVK4-CT-2002-00083 32

Figure 3.18: Average link speed as a function of traffic flow on urban roads (speed limit = 48/50 km/h) in each of the four main OSCAR cities (two high- speed values

are excluded from the plot).

0

10

20

30

40

50

60

70

80

90

0 10 20 30 40 50 60 70 80 90 100 110 120 130

1-hour mean traffic density (vehicles per km per lane)

Mea

nsp

eed

ofdr

ivin

gpa

ttern

(km

/h)

A01 - Athens 50km/h A02 - Athens 60 km/hA03 - Athens 60 km/h H03 - Helsinki 70 km/hH04 - Helsinki 70 km/h H05 - Helsinki 50 km/hH06 - Helsinki 50 km/h H07 - Helsinki 40 km/hH08 - Helsinki 40 km/h L01 - London 48 km/hL02 - London 64 km/h L03 - London 48 km/hL04 - London 48 km/h L05 - London 48 km/hM50 - Madrid 50 km/h

Figure 3.19: Average link speed as a function of traffic density on urban roads (speed limit <=70 km/h) in each of the four main OSCAR cities.

0

5

10

15

20

25

30

35

40

45

50

55

0 200 400 600 800 1000 1200

1-hour mean taffic flow (veh per hour per lane)

Ave

rage

spee

d(k

m/h

)

A01 - Athens 50km/h

H05 - Helsinki 50 km/h

H06 - Helsinki 50 km/h

L01 - London 48 km/h




M50 - Madrid 50 km/h

1-hour mean traffic flow (vehicles per lane per hour)


OSCAR EVK4-CT-2002-00083 33

Although traffic density and speed are related, there remains a great deal of variability in the data. Figure 3.19 shows that for low traffic densities (less than around ~20 vehicles per km per lane), a wide range of average speed was possible, although very low speeds tend not to be observed, as one might expect. More importantly for OSCAR, at low speeds a wide range of traffic densities are possible – presumably some people drive rather slowly even under relatively free-flow conditions or other factors, such traffic signals or pedestrian activity, impose a particularly slow driving pattern.

In a given city (e.g. Helsinki), maximum speeds in the low-traffic-density region were dependent upon the speed limit, but there was less of a tendency for the shape of the relationship to be affected by the design speed of the road, as in the speed-flow curve. It was considered possible that the different dynamics of these low-speed driving patterns could have an effect on emissions, and that traffic density could be a useful predictor of driving dynamics. This makes the definition of generalised driving patterns considerably easier, and the use of this information to derive driving cycles for the measurement of emissions is described in the following Chapter.

It is important to note that the approach adopted in OSCAR gives only a relatively crude indication of traffic conditions associated with different driving patterns, and the calculation of traffic density is only used as an approximate indicator of these conditions. The traffic density calculation cannot be compared with the more rigorous approaches to traffic modelling adopted by traffic engineers which include, for example, a consideration of the possibility of lane changing. The values presented in Figure 3.19 are average values, and are purely designed as a mechanism for allowing inferences about driving patterns to be drawn from traffic flow and speed information in the OSCAR emissions module.


OSCAR EVK4-CT-2002-00083 34

4 DEVELOPMENT OF DRIVING CYCLES

The treatment of congestion was an important aspect of the emission modelling work in OSCAR. Average-speed emission models, which are the type most commonly used, cannot take into account the variation in vehicle operation associated with a given average speed. For example, the average speed on a link may be low because the speed limit is low, or it may be low because of congestion. The average free-flow and congested speeds may be similar, but the dynamics of the driving patters on the link are likely to be rather different.

The useful parameters which model users tend to have available are most commonly average speed, traffic flow, and traffic composition. These values were used in OSCAR to describe the different levels of cycle dynamics associated with different levels of congestion.

4.1 Relationships between traffic speed, traffic density and driving dynamics

The relationship shown in Figure 3.19 formed the basic framework within OSCAR for describing different congestion levels and defining driving cycles. However, Figure 3.19 contains no information relating to the dynamics of the driving patterns. The next stage was to group the driving patterns according to emission-related dynamics parameters. Initially, speed-based parameters describing cycle dynamics were considered. Figures 4.1 to 4.4 show surface plots for four speed-derived parameters: RPA, APA, standard deviation of speed and the % of time spent accelerating. These plots suggest that increasing traffic density has little effect on the values of these parameters.

Figure 4.1: RPA as a function of mean link speed Figure 4.2: APA as a function of mean link speed and traffic density on urban roads (speed limit and traffic density on urban roads (speed limit <=70 km/h) in the four main OSCAR cities. <=70 km/h) in the four main OSCAR cities.

RPA (m/s2) APA

(m/s2)


OSCAR EVK4-CT-2002-00083 35

Figure 4.3: Standard deviation of speed as a Figure 4.4: Fraction of time accelerating as a function of mean link speed and traffic density function of mean link speed and traffic density on urban roads (speed limit <=70 km/h) in the on urban roads (speed limit <=70 km/h) in

four main OSCAR cities. the four main OSCAR cities.

4.2 A power-based approach

Heywood (1998) has explained that an engine’s performance can be defined in terms of its power output, more precisely, the maximum power (or the maximum torque) available at each speed within the useful engine operating range, and the range of speed and power over which engine operation is satisfactory. Torque is a measure of an engine’s ability to do work, and power is the rate at which work is done. Power is not measured directly - engine torque is measured using a dynamometer, and power is calculated using the formula (non-SI units are commonly used):

5252)rpm()lbs.ft()hp( speedEngineTorquePower ×= Equation 4.1

The dynamometer is used to apply a know torque (the ‘load’) to the engine. With the engine maintained at a steady speed under a given load, the power can then be calculated.

The following performance definitions are commonly used:

Maximum rated power The highest power an engine is allowed to develop for short periods of operation.

Normal rate power The highest power an engine is allowed to develop in continuous operation.

Rated speed The crankshaft rotational speed at which rated power is developed.

As engine speed and load were recorded for each driving pattern in the database (engine load was not recorded in London), these measurements were used directly to infer likely emissions. This is a more direct approach than using, for example, speed-related parameters (such as RPA) alone. The engine load value recoded by OBD is calculated by dividing the mass air flow rate by the peak air flow rate, and is a measure of how much work an engine is doing relative to how much work it could do, given the circumstances, expressed as a percentage. For example, an engine load value of 50% indicates that the engine could produce twice as much torque as it is doing. The load value is dependent upon engine speed. A 80% load value at 1500 rpm is not the same as a 80%

Speed SD

(km/h)

Fraction of time

accelerating


OSCAR EVK4-CT-2002-00083 36

load value at 3000 rpm, since the peak torque output at these two engine speeds will be different. During idle, the load value is typically around 20-30%, and under modest accelerations at low speeds typically increases to around 50%. When cruising at 60 mph on a level gradient, the load value is typically around 60%. Engine load reaches 100% at wide open throttle at any altitude or RPM for both naturally aspirated and boosted engines.

In order to define different levels of congestion (i.e. congestion-related traffic situations) from a perspective which was likely to be meaningful in terms of vehicle emissions, a power-based approach was used. The OBD data could not be used to determine absolute power, but a ‘power index’ can be calculated using Equation 4.2 (which is analogous to Equation 4.1):

)rpm()%(units)(arbitrary speedEngineloadEngineindexPower ×= Equation 4.2

For each driving pattern on a link the average power index was calculated as the product of the average engine speed and the average engine load. The use of the average power index alone did not take into account the possibility of high peak engine power levels at low speeds which may result in high emission events, or excursions outside the very-low-emission operating zone of the engine. In order to address this, other variables were investigated, and one of the most useful was found to be the number of times per kilometre the power index was above a threshold value (arbitrarily taken to be 80,000).

Figure 4.5 shows the surface plot for the average power index. The Figure shows that the average power index was related to average link speed. A second-order polynomial regression fit to the average power index and average speed data yielded an R2 value of 0.67. The traffic density at low average speeds had little effect on the average cycle power index. Given that the average power value was related to average link speed, this value was used to provide a means of grouping the data along the y-axis of Figure 3.19. The data were separated into the following speed bands according to the typical power level: (i) Less than 15 km/h, (ii) 15-30 km/h, (iii) 30-45 km/h, (iv) 45-60 km/h and (v) 60-90 km/h.

Figure 4.6 shows the surface plot for the number of power maxima above the threshold. This parameter was more closely related to traffic density (linear regression R2 = 0.52), but as there was also a speed dependence (linear regression R2 = 0.31; 2nd order polynomial regression R2 =0.41) it was concluded that traffic density was not likely to be a particularly effective surrogate for driving dynamics, assuming that dynamics are adequately described from an emissions standpoint by the speed-based and power-based parameters selected. However, in the absence of any other means of quantifying cycle dynamics on the part of the model user, it was considered appropriate to use traffic density as the best available solution in OSCAR. The distribution of the number of high-power events on the x-axis was therefore used to define traffic situations for the range of traffic densities observed. Three traffic density ranges were selected: 15-40, 40-70 and 70-125 vehicles per km per lane. A more detailed exploration of the relationship between traffic density and cycle dynamics ought to be the subject of future research – it is unlikely that there will be many high power events at very high traffic densities, for example. The fact that only 1-hour mean values for traffic density were available may also have masked some of the effects.

The speed and traffic density ranges were then applied to the driving pattern data in Figure 3.19, yielding 8 separate operational regions (Figure 4.8). For example, the driving patterns in box A are typified by a high speed (60-90 km/h) and a tendency towards a higher average power demand, and a low traffic density (<15 vehicles per km and per lane) and a tendency towards low numbers of power peaks per km. At the other end of the scale, the patterns in box H relate to low speed (5-15 km/h), and low average power, high traffic density (70-125 v/km per lane), and high numbers of power peaks per km. Hence, at low speeds, and for a given speed range, the model user can discriminate between different levels of dynamics for a given speed range using traffic density.


OSCAR EVK4-CT-2002-00083 37

Figure 4.5: Average cycle power (index) as a function of mean link speed and traffic density on urban roads (speed limit <=70

km/h) in the four main OSCAR cities.

Figure 4.6: Number of power maxima per km occurring above a threshold of 80,000 as a function of mean link speed and traffic density on urban roads (speed limit <=70 km/h) in the four main

OSCAR cities.

Average power (arbitrary units)

Number of power maxima per km above a threshold of

80,000


OSCAR EVK4-CT-2002-00083 38

0

10

20

30

40

50

60

70

80

90

0 10 20 30 40 50 60 70 80 90 100 110 120 130

1-hour mean traffic density (vehicles per km per lane)

Mea

nsp

eed

ofdr

ivin

gpa

ttern

(km

/h)

A01 - Athens 50km/h A02 - Athens 60 km/hA03 - Athens 60 km/h H03 - Helsinki 70 km/hH04 - Helsinki 70 km/h H05 - Helsinki 50 km/hH06 - Helsinki 50 km/h H07 - Helsinki 40 km/hH08 - Helsinki 40 km/h L01 - London 48 km/hL02 - London 64 km/h L03 - London 48 km/hL04 - London 48 km/h L05 - London 48 km/hM50 - Madrid 50 km/h

A

H

E

F

D

C

B

G

Figure 4.7: Average speed and traffic density regions. Traffic speeds and densities are hourly averages.

Driving cycles were constructed to represent each of the six regions C to H of Figure 4.8. Cycles were not developed for regions A and B, as OSCAR is primarily concerned with low speeds. For each of regions C to H, the average values of speed, RPA, power, and number of power peaks per km were determined. Typical driving patterns were selected for each box by virtue of their proximity to these average values. For the London data, driving patterns were selected using speed and RPA only. The driving patterns from each region were then combined to form a driving cycle for that region to be used on the chassis dynamometer. The parameters of the driving cycle (average speed, average engine speed, RPA, etc.) were within one standard deviation of the region average in each case. Original gear selections were retained.

The characteristics of the OSCAR driving cycles are summarised in Table 4.4 The cycles are also depicted in Appendix A. Cycles C, D2, E, F, G1 and H1 correspond to typical driving conditions for the associated boundary conditions. Four additional cycles were developed where the data allowed. These were:

D1 – low number of power peaks

G2 – high number of power peaks

H1 – medium number of power peaks

H3 – very high number of power peaks Although these additional cycles would allow the user some flexibility in the modelling, the cycles do not reflect the typical driving for the associated region, and can only be referenced via the speed/density criteria plus a subjective descriptor (e.g. ‘smooth flow’, ‘interrupted flow with high accelerations’).

Some urban traffic situations (and vehicle operation conditions) may not be adequately covered by the OSCAR database, though some of these are considered to be unlikely (e.g. high speed and high traffic density).


OSCAR EVK4-CT-2002-00083 39

Table 4.4: Characteristics of OSCAR driving cycles.

Operational ranges

CycleTraffic density (veh/km/lane)

Average link speed

(km/h)

Dynamics relative to

range mean

Distance (km)

Duration (s)

Average speed (km/h)

Average driving speed (km/h)

Maximum speed (km/h)

RPA (m/s2)

% of time idling

C 0-35 30-45 Average 3.98 402 35.6 39.6 70.8 0.212 12

D1 0-40 15-30 Low 2.70 430 22.6 27.9 46.7 0.163 21

D2 0-40 15-30 Average 2.33 364 23.0 28.2 54.7 0.224 20

E 40-70 15-30 Average 2.05 372 19.9 28.9 54.7 0.247 33

F 15-40 <15 Average 1.60 424 13.6 25.4 49.0 0.244 50

G1 40-70 <15 Average 1.56 456 12.3 18.5 40.2 0.221 38

G2 40-70 <15 High 1.12 351 11.5 15.9 51.5 0.277 32

H1 70-125 <15 Low 0.80 371 7.8 11.0 31.0 0.169 35

H2 70-125 <15 Average 0.95 425 8.1 12.9 30.6 0.242 42

H3 70-125 <15 High 0.85 375 8.2 12.4 38.6 0.270 41


OSCAR EVK4-CT-2002-00083 40

5 EMISSION MEASUREMENTS

5.1 Method

The OSCAR driving cycles were supplied by TRL to TNO in Delft, where the emission tests were conducted on a chassis dynamometer (Figure 5.1). In order to improve the efficiency of data collection and to maximise the amount of data available to OSCAR, TNO ran the OSCAR tests alongside other test programmes.

Figure 5.1: TNO chassis dynamometer

5.1.1 Test vehicles

A total of 20 passenger cars were used in the test programme. Table 5.1 shows how these cars were distributed in terms of fuel type and emission certification level. A more detailed specification for each vehicle is provided in Table 5.2.

Table 5.1: Vehicle categories and numbers tested.

Emission certification level Fuel

Euro 1 Euro 2 Euro 3 Euro 4 Total

Petrol 1 3 3 2 9

Diesel 1 3 5 0 9

LPG 0 0 2 0 2

Total 2 6 10 2 20

No Euro 4 diesel or LPG vehicles were available for testing. However, two of the diesel vehicles and both the LPG vehicles were compliant with Euro 4 emission limits. One of the vehicles (number 9) was fitted with a gasoline direct injection (GDI) engine.


OSCAR EVK4-CT-2002-00083 41

Table 5.2: Specification of vehicles in emission test programme.

EngineVehicle

referencenumber

FuelEmission

certification levelMake Model

weight(kg)

Odometer(km) Capacity

(cc)Configuration

Valves /cylinder

Fuel injection typeMaximumpower (W)

Aftertreatment Transmission

1 petrol Euro 1 Opel Astra 1.6i 1015 117,001 1598 4 cylinder in-line 2 single point 52 TWC 5 gear, manual

2 petrol Euro 2 Ford Focus 1.6 1180 81,161 1596 4 cylinder in-line 4 multi point 74 TWC 5 gear, manual

3 petrol Euro 2 Nissan Almera 1.4i 1015 83,910 1392 4 cylinder in-line 4 multi point 64 TWC 5 gear, manual

4 petrol Euro 2 VW Polo 1.4 896 72,356 1390 4 cylinder in-line 4 multi point 55 TWC + pre-catalyst 5 gear, manual

5 petrol Euro 3 Suzuki Alto 1.1 775 14,182 1081 4 cylinder in-line 4 multi point 46 TWC 5 gear, manual

6 petrol Euro 3 Volvo V40 2.0 1275 52,816 1948 4 cylinder in-line 4 multi point 100 TWC 5 gear, manual

7 petrol Euro 3 Peugeot 307 SW 1.6 1299 16,440 1587 4 cylinder in-line 4 multi point 80 TWC + pre-catalyst 5 gear, manual

8 petrol Euro 4 Mazda 6 1.8 1245 19,632 1798 4 cylinder in-line 4 multi point 88 TWC + pre-catalyst 5 gear, manual

9 petrol(GDI)

Euro 4 VW Touran 1.6 FSI 1398 14,500 1598 4 cylinder in-line 4 multi point direct 85 De-NOx catalyst, TWC,and pre-catalyst

6 gear, manual

10 diesel Euro 1 Ford Fiesta 1.8D 884 120,481 1753 4 cylinder in-line 2 rotary pump, indirect 44 none 5 gear, manual

11 diesel Euro 2 Audi A4 TDI 1290 117,909 1896 4 cylinder in-line,turbo

2 rotary pump, direct 81 oxidation catalyst, EGR 5 gear, manual

12 diesel Euro 2 Peugeot 306 1.9 D 1115 127,344 1902 4 cylinder in-line 2 rotary pump, indirect 51 oxidation catalyst, EGR 5 gear, manual

13 diesel Euro 2 Toyota Picnic 2.2 TD 1355 179,790 2184 4 cylinder in-line,turbo

2 rotary pump, indirect 66 oxidation catalyst, EGR 5 gear, manual

14 diesel Euro 3 Opel Astra 1.7 Dti 1220 137,895 1686 4 cylinder in-line,turbo

4 rotary pump, direct 55 oxidation catalyst, EGR 5 gear, manual

15 diesel Euro 3 Volvo V70 D5 1596 82,020 2401 5 cylinder in-line,turbo inter-cooled

4 common rail, direct 120 oxidation catalyst, EGR 5 gear, automatic

16 diesel Euro 3 VW Passat TDI 1326 46,245 1896 4 cylinder in-line,turbo inter-cooled

2 unit injectors, direct 74 oxidation catalyst, EGR 5 gear, manual

17 diesel Euro 3(Euro 4 compliant)

Renault Megane 1.5 dCi 1150 24,561 1461 4 cylinder in-line,turbo inter-cooled

2 common rail, direct 60 oxidation catalyst, EGR 5 gear, manual

18 diesel Euro 3(Euro 4 compliant)

Toyota Avensis D-4D 1350 44,355 1995 4 cylinder in-line,turbo inter-cooled

4 common rail, direct 85 oxidation catalyst, EGR 5 gear, manual

19 LPG Euro 3(Euro 4 compliant)

AlfaRomeo

147 1.6 TS 1250 7,598 1598 4 cylinder in-line 4 multi point 77 TWC + pre-catalyst 5 gear,manual

20 LPG Euro 3(Euro 4 compliant)

Volvo S60 bi-fuel 140 1700 86,181 2435 5 cylinder in-line 4 multi point 101 TWC + pre-catalyst 5 gear,manual


OSCAR EVK4-CT-2002-00083 42

5.1.2 Test cycles The vehicles were tested over the 10 OSCAR driving cycles, as well as over legislative European driving cycles and ARTEMIS driving cycles. Vehicles were also tested at idle. The characteristics of the OSCAR driving cycles were given in Table 4.4. the characteristics of the other driving cycles are given in Table 5.3 below.

Table 5.3: Characteristics of other driving cycles.

Cycle Distance

(km) Duration

(s) Average

speed (km/h)

Average driving speed

(km/h)

Maximum speed (km/h)

RPA (m/s2)

% of time idling

Idle 0.00 500 0.0 0.0 0 0.000 100

Legislative UDC† 4.06 780 18.8 25.4 50 0.143 26

Legislative EUDC 6.96 400 62.6 69.0 120 0.094 9

Legislative NEDC† 11.02 1180 33.6 42.1 120 0.111 20

CADC Urban Hot 4.48 921 17.5 23.1 58 0.304 24

CADC Rural 14.71 862 61.4 62.5 84 0.161 1.6

CADC Motorway 150 24.60 736 120.3 120.3 151 0.098 0

CADC Urban Cold (9°C) 4.48 921 17.5 23.1 58 0.304 24

† Not including 40s idle (i.e. Euro 3 and later)

5.1.3 Test procedure

Figure 5.2 presents a schematic drawing of the standard equipment that was used for sampling vehicle exhaust gas and particulate emissions. The vehicle exhaust gases were diluted with filtered air to prevent condensation or reactions between the different exhaust gas components. The dilution took place in a tunnel, or CVS (Constant Volume Sampler). The system maintained a constant volumetric flow, determined by the dimensions of a critical flow Venturi and a pump located at the end of the tunnel. For a given Venturi, the dilution ratio (which is the ratio of the dilution air and the exhaust gas flow) depends on the exhaust gas flow of the vehicle.

During the emission test a sample of the diluted exhaust gas was drawn from the dilution tunnel and collected in a pair of Tedlar sampling bags. Of each pair, one bag is used for the diluted exhaust gas, and the other for the dilution air. The latter is used for correction, since the dilution air may also contains small fractions of CO2, CO, HC and NOx. After the test, the content of the Tedlar bags was analysed. The analysis of the regulated exhaust gases and CO2 is quite straightforward, and is extensively described in the Directives of the European Union (96/69/EC). Dedicated analysers for CO, NOx, HC and CO2 were used to analyse the diluted exhaust gas from the Tedlar bags. The CO and CO2 analysers operate by non-dispersive infrared (NDIR). The HC analyser operates by flame ionisation detection (FID) and the NOx analyser by chemiluminesence. Multiplication of the concentrations and the tunnel flow yielded the emission factor in grammes per kilometre. Again, the calculation procedure is extensively described in European Directives.

For the diesel vehicles only, particulate matter (PM) was collected separately from the other emission components by drawing diluted exhaust gas from the tunnel through a pair of Pallflex filters. The second filter serves to detect, and if necessary to correct for any sample breakthrough from the first filter. The filters were weighed before and after the test, and their weight increase was used to determine the particle concentration of the diluted exhaust gas. Although there are no limit values defined for CO2, the CO2 concentration was also determined, as it is used in the calculation procedure. The CO2 emission value is also used to calculate the fuel consumption (in l/100km) using the carbon balance method.


OSCAR EVK4-CT-2002-00083 43

(blank)

Vehicle exhaust

gas

Dilution air

Filter

Tedlarbag

Dilution tunnel

Tedlarbag

CO, CO2, THC, NOx

Critical flow

Venturi

Pump

Pump

Particulate emssions

(blank)

Vehicle exhaust

gas

Dilution air

Filter

Tedlarbag

Dilution tunnel

Tedlarbag

CO, CO2, THC, NOx

Critical flow

Venturi

Pump

Pump

Particulate emssions

Figure 5.2: Measurement set-up for regulated exhaust gas and PM emissions

For each fuel type, the same specification fuel was used throughout the measurements. The emission data were stored by TRL and TNO, and the results were also incorporated into both the ARTEMIS database of emission factors for light-duty vehicles and the database of TNO’s VERSIT+ model.

5.2 Results

The emission test results for all vehicles and tests are given in Appendix B. It should be noted throughout this section that the results and observations for each pollutant and vehicle type are based on a relatively small sample of vehicles, and may not be fully representative of the wider European vehicle fleet.

5.2.1 Comparison with type approval limits

In order to check that a vehicle sample is not biased towards high-emitting vehicles, and to confirm that only roadworthy vehicles are being tested, it is common practice to compare the emission levels with the limit values used in type approval, in this case over the NEDC.

For all passenger cars, the limit values apply to CO and the sum of THC and NOx. For petrol Euro 3 and Euro 4 vehicles, there are additional separate limits for THC and NOx. For Euro 3 diesel cars there is an additional separate limit value for NOx. For all diesel vehicles there is also a limit value for PM. LPG vehicles are subject to the same limit values as petrol vehicles.

The results of these comparisons for each test vehicle are shown for CO in Figure 5.3, and in Appendix C for the other pollutants. For the purposes of comparison, only the combined THC and NOx values are show, but the separate limits were taken into account where appropriate.

CO emissions from all test vehicles were well below the type approval limit values. All of the petrol and LPG vehicles had combined THC and NOx emissions which were below the legislative limit values. However, three of the diesel vehicles (the single Euro 1 vehicle and two of the Euro 2 vehicles) had combined THC and NOx emissions which were higher than the limit values, although the margins of exceedance were not large. The PM emission limit was slightly exceeded by one Euro 3 diesel vehicle.


OSCAR EVK4-CT-2002-00083 44

0.0

0.5

1.0

1.5

2.0

2.5

3.0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Vehicle reference number

CO

(g/k

m)o

verN

EDC

cycl

e

Emission over NEDCType approval limit value

Petro

lEur

o1

Petro

lEur

o2

Die

selE

uro

3

Petro

lEur

o3

Die

selE

uro

2

Die

selE

uro

1

Petro

lEur

o4

LPG

Figure 5.3: CO emissions over NEDC compared with limit values.

The test vehicles were selected from the in-service fleet. It would therefore be expected that the emissions would deteriorate from the as-new condition. The results suggested that the test vehicles were roadworthy, probably well-maintained, and not high emitters.

5.2.2 Emissions over the OSCAR cycles compared with emissions over the UDC

Engine manufacturers are able to configure engines to have low emissions over the speed and load test points of the type approval cycles, possibly at the expense of high off-cycle emissions. In the United States of America, this so-called ‘cycle beating’ has been identified as a serious political issue, and in 1998 a number of manufacturers were forced to pay fines totalling $83.4 million for producing engines which included cycle beating in their engine control software. Emissions over the real-world OSCAR cycles were therefore compared with emissions over the Urban Driving Cycle (UDC) part of the NEDC (i.e. the part of the NEDC which is associated with an operational range close to that of the OSCAR cycles) to examine the effect of off-cycle operation on emissions, and to test whether any systematic differences were apparent. It should be noted that prior to Euro 3 legislation, the UDC included a 40-second engine warm-up phase at idle. This 40-second phase was eliminated for the Euro 3 legislation (and after), resulting in the capture of cold start emissions.

For each vehicle category (fuel and Euro class), the emission values over each cycle were averaged. This is a normal procedure when constructing emission models, which inevitably require some form of aggregation as it is impractical to model emissions from every make and model of vehicle that is produced. However, this approach often leads to the masking of phenomena which are particular to individual vehicles – these will be explored in more detail later.

Average NOx emissions from the different categories of vehicle over the OSCAR cycles and over the legislative UDC are shown in Figure 5.4. NOx emissions for all categories of petrol vehicle were similar over the UDC and the OSCAR cycles, and in both cases rather low. Emissions from diesel vehicles, on the other hand, were higher over the OSCAR cycles than over the UDC, and decreased from Euro 1 to Euro 3. PM emissions from diesel vehicles are shown in Figure 5.5. Emissions over the OSCAR cycles decreased between Euro 1 and Euro 3, whereas emissions over the UDC remained relatively unchanged with Euro class. For Euro 1 and Euro 2 vehicles emissions over the OSCAR cycles were higher than over the UDC, but there was some convergence in the case of Euro 3.


OSCAR EVK4-CT-2002-00083 45

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

Pet

rolE

uro

1

Pet

rolE

uro

2

Pet

rolE

uro

3

Pet

rolE

uro

4

Die

selE

uro

1

Die

selE

uro

2

Die

selE

uro

3

LPG

Eur

o3

NO

x(g

/km

)

OSCAR

UDC

Figure 5.4: Average NOx emissions over OSCAR cycles compared with emissions over UDC.

0.00

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

Pet

rolE

uro

1

Pet

rolE

uro

2

Pet

rolE

uro

3

Pet

rolE

uro

4

Die

selE

uro

1

Die

selE

uro

2

Die

selE

uro

3

LPG

Eur

o3

PM(g

/km

)

OSCAR

UDC

Figure 5.5: Average PM emissions over OSCAR cycles compared with emissions over UDC.

Average CO, THC and CO2 emissions from the different categories of vehicle over the OSCAR cycles and over the legislative UDC are shown in Appendix D. For CO, emissions over the OSCAR cycles were generally lower than emissions over the UDC. Euro 2 petrol vehicles had relatively high CO emissions over the OSCAR cycles. However, this was entirely due to the emissions from one vehicle (#03) having very high emissions. It is possible that the emissions from this vehicle are not representative of the average emissions behaviour of Euro 2 petrol vehicles, and it can be seen that if this vehicle removed a more logical picture emerges (i.e. emissions generally decreasing with improvements in technology). Furthermore, emissions over the UDC would be easier to understand, as emissions for Euro 1 and 2 petrol vehicles would be lower that emissions from Euro 3 and 4 vehicles on account of the change in legislation mentioned above. For diesel Euro 1 vehicles, emissions over the OSCAR cycles were, on average, similar to those over the UDC, but emissions were lower over the OSCAR cycles than over the UDC for Euro 2 and Euro 3 vehicles. Emissions of CO from the two LPG vehicles were much lower over the OSCAR cycles than over the UDC (again the Euro 3 cycle). THC emissions over the OSCAR cycles were lower than emissions over the UDC for all vehicle categories. The effect of vehicle #03 was again apparent over the OSCAR cycles.


OSCAR EVK4-CT-2002-00083 46

In summary, there is some evidence to suggest that whilst CO and THC emissions can be higher over the UDC than over real-world congestion cycles with a similar speed, real-world NOx and PM from diesel vehicles may be underestimated.

5.2.3 Average emissions by Euro class and cycle: hot exhaust emissions

For each Euro class and driving cycle the emission values for the appropriate vehicles were averaged. The results are shown in Appendix E and are discussed in general terms below.

CO

CO emissions from petrol cars over the OSCAR cycles were low (<0.5 g/km) for all Euro classes except Euro 2, with Euro 3 emissions generally being exceptionally low. The Euro 2 vehicles had, on average, particularly high emissions over cycles F, G1, G2, H1 and H2, this being entirely due to the high emission levels of Vehicle #03 over these cycles. Although emissions from this vehicle over the NEDC did not indicate that it had the potential to be a high emitter (Figure 5.3), it was the only vehicle to emit more than 1 g/km of CO over any of the OSCAR cycles, and emitted more than 15 g/km over cycle H2. This level of emissions is very high for a Euro 2 vehicle, and was probably indicative of a fault. Emissions of CO from petrol vehicles over the UDC and CADC Motorway cycles were generally higher than those over the OSCAR cycles (Euro 2 excepted). For the OSCAR cycles, emissions tended to be lowest over cycles E and F, and highest over C, type-D, type-G and type-H. For Euro 1 and Euro 2 diesel cars, there was a systematic increase in CO emissions as cycle speed decreased, with emissions being lowest over the C and type-D cycles, and highest over the type-H cycles. CO emissions from, the Euro 3 vehicles were lower than those from Euro 1 and Euro 2, but showed a less consistent pattern. For the LPG vehicles (both Euro 3), emissions were generally very low over the OSCAR cycles, and generally increased as speed decreased.

THC

THC emissions from petrol cars over the OSCAR cycles followed a similar pattern to CO emissions from the same vehicles. Again, vehicle #03 was a particularly high emitter over cycles H1 and H2. As in the case of CO, there was a systematic increase in CO emissions as cycle speed decreased, with emissions again being lowest over the C and type-D cycles, and highest over the type-H cycles. THC emissions from LPG vehicles were very low over all cycles with the exception of the UDC (and hence the NEDC of which the UDC forms part).

NOx

NOx emissions from petrol vehicles showed exhibited little systematic behaviour over the OSCAR cycles, and showed little dependency on cycle speed. However, for a given speed range the level of driving dynamics does appear to be important, with higher emissions being recorded over cycles having higher dynamics (see type-D, type-G and type-H cycles). This is explored in more detail in the next Chapter. NOx emissions from diesel vehicles over the OSCAR cycles followed a similar pattern to that observed for CO and THC – a general dependence on speed, with little contribution due to driving dynamics. Emissions from LPG vehicles were comparable to those from petrol vehicles, although there was a general decrease in emissions as speed decreased.

PM

PM emissions from diesel vehicles over the OSCAR cycles tended to decrease with decreasing speed, and cycle dynamics appears to have an effect.

5.2.4 Average emissions by Euro class and cycle: cold-start emissions

Cold-start emissions are shown in Appendix F. Emissions were measured over the ARTEMIS Urban cycle with a full cold start, and at an ambient temperature of 9oC. The emissions over the corresponding hot-start cycle are also shown for comparison. The graphs illustrate the importance of cold-start emissions for CO and THC, especially for petrol and LPG cars. For CO, the ratio of cold-start to hot-start emissions over the CADC urban cycle ranged from around 5 to almost 6,000, whereas for diesel vehicles it tended to be less than 5. For LPG vehicles cold start CO emissions were around 50 times higher than hot-start emissions. Emissions of THC from petrol vehicles over the cold-start cycle were between 6 and 400 times higher than those over the hot-


OSCAR EVK4-CT-2002-00083 47

start cycle, whereas the ratio for diesel vehicles was between 1.1 and 2.7. For LPG vehicles, cold-start THC emissions were around 25 times higher than hot-start emissions.

The effects of cold starting on NOx and PM were less pronounced that those for CO and THC, but still notable. For NOx, the cold-start:hot-start ratio was less than one for some vehicles, but was generally greater than one, and as high as four. PM emissions from diesel vehicles over the cold-start cycle were between 1.2 and 2.6 times higher than emissions over the hot-start cycle.

As a vehicle’s engine and after-treatment system could reach their full operational temperatures after a few kilometres of driving, some hot exhaust emissions could be included in the cold start test. Nevertheless, the results show that, for petrol vehicles, almost all the emissions over a driving cycle with a cold start will occur during the warm-up period. However, the results relate to full cold starts, in which the engine has been allowed to cool down to the ambient temperature. Many starts occur at intermediate temperatures, whereby the engine temperature is between the ambient temperature and the full operational temperature. Emissions during these ‘warm’ starts will be lower than those during full cold starts. The results clearly indicate that cold start emissions should be included in the OSCAR System, given that large numbers of starts occur in central urban areas.

5.2.5 Average emissions by Euro class and cycle: idle emissions

Emissions when the engine at idle are usually included in the driving cycles used to derive average speed emission functions. However, during extended periods of congestion it is possible that vehicles remain at idle for long periods, and therefore the average-speed functions cannot be applied. In spite of this possibility, emission models do not generally provide emission factors for engine idle.

The emissions measured during OSCAR for idle conditions are given in Appendix B. The values are stated in g/h rather than g/km, as clearly no movement is involved.

The results show that idle emissions from diesel vehicles tend to be considerably higher than those from petrol and LPG vehicles, particularly for CO and NOx. Some emission factors for idle conditions are given later in this report.


OSCAR EVK4-CT-2002-00083 48

6 THE EFFECTS OF CONGESTION ON EMISSIONS

The principal aim of the emission measurement programme in OSCAR was to determine the effects of different levels of congestion on hot exhaust emissions. Congestion can be described in terms of average speed alone, but for reasons which have been discussed earlier, this approach fails to account for the various different types of operation which can occur at a given speed. Consequently, different types of congestion were defined in OSCAR as a function of average speed and driving dynamics. Earlier, traffic density was selected as a surrogate indicator for the general level of driving dynamics. For a given speed, a range of traffic densities are possible, and the range is most pronounced at low speeds (Figure 3.19). The OSCAR driving cycles were therefore constructed to reflect different congestion situations in terms of average speed and traffic density.

6.1 OSCAR measurements

Figures 6.1 and 6.2 illustrate how congestion effects have been addressed in the OSCAR measurements.

Figure 6.1 shows the average emission levels for the three lowest-speed average traffic situations (F, G1 and H2), with traffic density (and the number of power peaks) increasing from F to H2. For diesel vehicles, emissions are generally higher over cycle H2 than over G1, and higher (or equal) over G1 than F. For petrol vehicles emissions of CO and THC are generally lowest over F and highest over H2, except in the case of the Euro 2 vehicles (excluding #03), which had lowest emissions over the G1 cycle. However, these effects reflect the combined influence of a reduction in speed (H2=8.1 km/h, G1=12.3 km/h, F=13.6 km/h) and any change in driving dynamics associated with traffic density, as it was not possible to fully separate the effects of these parameters in the driving cycles for average traffic situations.

Figure 6.2 illustrates the effect of differences in cycle dynamics for cycles having the same average speed (cycles H1, H2 and H3). The graphs show that cycle dynamics appears to have a notable effect for some vehicle types and pollutants. For diesel vehicles, PM emissions appear to be influenced by cycle dynamics, and the other pollutants to a lesser extent. For petrol and LPG vehicles the results are rather mixed, but in some cases cycle dynamics does appear to be important.

These results show that emission levels do vary over the OSCAR cycles. However, it is not straightforward to show which operational parameters are contributing to the differences in emissions over the cycles. Depending on the vehicle type and pollutant, the results vary considerably, partly due to the consideration of small vehicle samples. A means was therefore required of examining the effects of cycle dynamics for a larger selection of vehicles, and this was only possible through modelling.

6.2 VERSIT+ modelling

The urban driving patterns (~900) recorded in the four cities were sent to TNO to be processed using the VERSIT+ model (see Section 2.3.6). It was anticipated that the results could be used to ‘map’ emissions as a function of traffic speed and density in order to provide indication of congestion effects for use either independently in OSCAR or in other models. However, when running VERSIT+ TNO noted that for some driving patterns the model had to be rejected. The main reason for this was that some of the OSCAR driving patterns were short (<1 minute), compared with the typical driving patterns used in VERSIT+ (~5-6 minutes), and were associated with vehicle operating conditions outside the acceptable range within the model. Consequently, only the results for around 200 driving patterns were available, rendering the mapping process untenable. Figure 6.3 shows two different sets of results from this process, indicating that average speeds remains a reasonably reliable predictor of NOx emissions for Euro 3 diesel vehicles, but not for Euro 3 petrol vehicles.


OSCAR EVK4-CT-2002-00083 49

Figure 6.1: Average emissions for different traffic situations.

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

F G1 H2

Driving cycle

CO

(g/k

m)

Diesel Euro 1

Diesel Euro2

Diesel Euro3

0.000.050.100.150.200.250.300.350.400.450.500.55

F G1 H2

Driving cycle

CO

(g/k

m)

Petrol Euro 1

Petrol Euro 2 (no #03)

Petrol Euro 3

Petrol Euro4

0.000

0.010

0.020

0.030

0.040

0.050

0.060

0.070

0.080

0.090

F G1 H2Driving cycle

CO

(g/k

m)

LPG

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

F G1 H2

Driving cycle

NO

x(g

/km

)

Diesel Euro 1

Diesel Euro2

Diesel Euro3

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.16

F G1 H2

Driving cycle

NO

x(g

/km

)

Petrol Euro 1 Petrol Euro 2Petrol Euro 3 Petrol Euro4

0.041

0.042

0.043

0.044

0.045

0.046

0.047

0.048

0.049

0.050

F G1 H2

Driving cycle

NO

x(g

/km

)

LPG

0.00

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.10

F G1 H2

Driving cycle

PM(g

/km

)

Diesel Euro 1

Diesel Euro2

Diesel Euro3

0.00

0.05

0.10

0.15

0.20

0.25

F G1 H2

Driving cycle

THC

(g/k

m)

Diesel Euro 1

Diesel Euro2

Diesel Euro3

0.000

0.005

0.010

0.015

0.020

0.025

0.030

F G1 H2

Driving cycle

THC

(g/k

m)

Petrol Euro 1Petrol Euro 2 (no #03)Petrol Euro 3Petrol Euro4

0.000

0.001

0.002

0.003

0.004

0.005

0.006

0.007

0.008

0.009

F G1 H2Driving cycle

THC

(g/k

m)

LPG


OSCAR EVK4-CT-2002-00083 50

Figure 6.2: Effect of driving dynamics for a given average speed (type-H cycles).

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

H1 H2 H3

Driving cycle

CO

(g/k

m)

Diesel Euro 1

Diesel Euro2

Diesel Euro3

0.00

0.10

0.20

0.30

0.40

0.50

0.60

H1 H2 H3

Driving cycleC

O(g

/km

)

Petrol Euro 1Petrol Euro 2 (no #03)Petrol Euro 3Petrol Euro4

0.000

0.050

0.100

0.150

0.200

0.250

0.300

0.350

0.400

0.450

H1 H2 H3Driving cycle

CO

(g/k

m)

LPG

0.0

0.1

0.1

0.2

0.2

0.3

H1 H2 H3

Driving cycle

THC

(g/k

m) Diesel Euro 1

Diesel Euro2

Diesel Euro3

0.00

0.00

0.00

0.01

0.01

0.01

0.01

0.01

0.02

0.02

0.02

H1 H2 H3

Driving cycle

THC

(g/k

m)

Petrol Euro 1

Petrol Euro 2 (no #03)

Petrol Euro 3

Petrol Euro4

0.000

0.001

0.002

0.003

0.004

0.005

0.006

0.007

0.008

0.009

0.010


THC

(g/k

m)

LPG

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

H1 H2 H3

Driving cycle

NO

x(g

/km

)

Diesel Euro 1

Diesel Euro2

Diesel Euro3

0.00

0.05

0.10

0.15

0.20

0.25

0.30

H1 H2 H3

Driving cycle

NO

x(g

/km

)

Petrol Euro 1

Petrol Euro 2

Petrol Euro 3

Petrol Euro4

0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

0.040

0.045

0.050


NO

x(g

/km

)

LPG

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.16

H1 H2 H3

Driving cycle

PM(g

/km

)

Diesel Euro 1

Diesel Euro2

Diesel Euro3


OSCAR EVK4-CT-2002-00083 51

Petrol, Euro III

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0 10 20 30 40 50 60


NO

x(g

/km

)

VERSIT+,short patterns

VERSIT+,driving cycles

Diesel, Euro III

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

0 10 20 30 40 50 60


NO

x(g

/km

)

VERSIT+,short patterns

VERSIT+,driving cycles

Figure 6.3: Examples of output from VERSIT+ over OSCAR driving patterns and laboratory driving cycles.

Instead, the VERSIT+ results were used to provide further information on congestion effects, by calculating average emission levels for different speed and traffic density ranges. The results, shown in Figures 6.4 to 6.7, provided a clearer indication of the importance of different levels of congestion than the OSCAR emission measurements alone. However, in general, there was little evidence that changes in traffic density at a given average speed had a systematic effect on emissions, even though in some cases there were large differences between the predicted emissions associated with different traffic densities. The general observations from these comparisons are:

• The effect of traffic density on CO and THC emissions from petrol vehicles was variable, and showed no systematic pattern, although for Euro 1 to Euro 3 vehicles emissions of both pollutants were highest for mid-range traffic densities at low and high speeds.

• NOx emissions from petrol Euro 2 and Euro 3 vehicles appeared to be particularly low at low traffic densities.

• For diesel vehicles there was little evidence of a traffic density effect, for any pollutant and average speed alone seems adequate for predicting emissions. However, for diesel Euro 3 and Euro 4 vehicles high traffic densities appeared to result in higher CO emissions at speeds lower than around 15 km/h, and mid-range traffic densities resulted in higher emissions than low traffic densities in the 40-55 km/h speed band.

• In the case of LPG vehicles, the results for CO and NOx were rather variable and there was no systematic traffic density effect. However, THC emissions at low speeds appeared to be highest for high traffic densities.

6.3 Implications

The OSCAR measurements and the VERSIT+ model predictions indicate that emissions for a given average speed vary with the level of cycle dynamics. In OSCAR, an attempt has been made to quantify cycle dynamics using information available to model users (i.e. traffic speed and density), principally for use in a traffic situation model. However, there was little evidence of a systematic traffic density effect for a given combination of vehicle type and pollutant, and therefore in models which use descriptors of cycle dynamics, the descriptors will probably remain subjective, at least for the time being. The actual modelling approach used in OSCAR is described in more detail in the next Chapter.


OSCAR EVK4-CT-2002-00083 52

Petrol, Euro 1

0

1

2

3

4

5

6

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

CO

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

Petrol, Euro 2

0

1

2

3

4

5

6

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

CO

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

Petrol, Euro 3

0

1

2

3

4

5

6

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

CO

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

Petrol, Euro 4

00.10.20.30.40.50.60.70.80.9

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

CO

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

Diesel, Euro 1

00.20.40.60.8

11.21.41.61.8

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

CO

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

Diesel, Euro 2

00.10.20.30.40.50.60.70.80.9

1

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

CO

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

Diesel, Euro 3

00.020.040.060.080.1

0.120.140.160.18

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

CO

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

Diesel, Euro 4

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

CO

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

LPG, Euro 1

0

0.5

1

1.5

2

2.5

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

CO

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

LPG, Euro 2

0

0.5

1

1.5

2

2.5

3

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

CO

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

LPG, Euro 3

00.050.1

0.150.2

0.250.3

0.350.4

0.45

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

CO

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

LPG, Euro 4

00.10.20.30.40.50.60.70.80.9

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

CO

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

Figure 6.4: Average CO emissions predicted using VERSIT+ for different speed and traffic density ranges.


OSCAR EVK4-CT-2002-00083 53

Petrol, Euro 1

00.020.040.060.08

0.10.120.140.160.18

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

THC

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

Petrol, Euro 2

00.020.040.060.080.1

0.120.140.160.18

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

THC

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

Petrol, Euro 3

00.020.040.060.080.1

0.120.140.160.18

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

THC

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

Petrol, Euro 4

00.0010.0020.0030.0040.0050.0060.0070.0080.009

5-10 10-15

15-20

20-25

25-30

30-35

35-40

40-55

Speed band (km/h)

THC

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

Diesel, Euro 1

0

0.05

0.1

0.15

0.2

0.25

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

THC

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

Diesel, Euro 2

0

0.05

0.1

0.15

0.2

0.25

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

THC

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

Diesel, Euro 3

0

0.02

0.04

0.06

0.08

0.1

0.12

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

THC

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

Diesel, Euro 4

00.0050.01

0.0150.02

0.0250.03

0.0350.04

0.0450.05

5-10 10-15

15-20

20-25

25-30

30-35

35-40

40-55

Speed band (km/h)

THC

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

LPG, Euro 1

00.10.20.30.40.50.60.70.80.9

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

THC

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

LPG, Euro 2

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

THC

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

LPG, Euro 3

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

THC

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

LPG, Euro 4

00.0010.0020.0030.0040.0050.0060.0070.0080.009

5-10 10-15

15-20

20-25

25-30

30-35

35-40

40-55

Speed band (km/h)

THC

(g/k

m)

1) <202) 20-503) >50

TD (v/km)

Figure 6.5: Average THC emissions predicted using VERSIT+ for different speed and traffic density ranges.


OSCAR EVK4-CT-2002-00083 54

Petrol, Euro 1

0

0.1

0.2

0.3

0.4

0.5

0.6

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

NO

x (g

/km

)

1) <202) 20-503) >50

TD (v/km)

Petrol, Euro 2

0

0.1

0.2

0.3

0.4

0.5

0.6

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

NO

x (g

/km

)

1) <202) 20-503) >50

TD (v/km)

Petrol, Euro 3

0

0.1

0.2

0.3

0.4

0.5

0.6

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

NO

x (g

/km

)

1) <202) 20-503) >50

TD (v/km)

Petrol, Euro 4

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

NO

x (g

/km

)

1) <202) 20-503) >50

TD (v/km)

Diesel, Euro 1

0

0.2

0.4

0.6

0.8

1

1.2

1.4

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

NO

x (g

/km

)

1) <202) 20-503) >50

TD (v/km)

Diesel, Euro 2

0

0.2

0.4

0.6

0.8

1

1.2

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

NO

x (g

/km

)

1) <202) 20-503) >50

TD (v/km)

Diesel, Euro 3

0

0.2

0.4

0.6

0.8

1

1.2

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

NO

x (g

/km

)

1) <202) 20-503) >50

TD (v/km)

Diesel, Euro 4

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

NO

x (g

/km

)

1) <202) 20-503) >50

TD (v/km)

LPG, Euro 1

00.20.40.60.8

11.21.41.61.8

2

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

NO

x (g

/km

)

1) <202) 20-503) >50

TD (v/km)

LPG, Euro 2

00.10.20.30.40.50.60.70.80.9

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

NO

x (g

/km

)

1) <202) 20-503) >50

TD (v/km)

LPG, Euro 3

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

NO

x (g

/km

)

1) <202) 20-503) >50

TD (v/km)

LPG, Euro 4

0

0.02

0.04

0.06

0.08

0.1

0.12

0.14

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

NO

x (g

/km

)

1) <202) 20-503) >50

TD (v/km)

Figure 6.6: Average NOx emissions predicted using VERSIT+ for different speed and traffic density ranges.


OSCAR EVK4-CT-2002-00083 55

Petrol, Euro 1

0

0.0002

0.0004

0.0006

0.0008

0.001

0.0012

0.0014

5-10 10-15

15-20

20-25

25-30

30-35

35-40

40-55

Speed band (km/h)

PM (g

/km

)

1) <202) 20-503) >50

TD (v/km)

Petrol, Euro 2

0

0.0002

0.0004

0.0006

0.0008

0.001

0.0012

0.0014

5-10 10-15

15-20

20-25

25-30

30-35

35-40

40-55

Speed band (km/h)

PM (g

/km

)

1) <202) 20-503) >50

TD (v/km)

Petrol, Euro 3

0

0.0002

0.0004

0.0006

0.0008

0.001

0.0012

0.0014

5-10 10-15

15-20

20-25

25-30

30-35

35-40

40-55

Speed band (km/h)

PM (g

/km

)

1) <202) 20-503) >50

TD (v/km)

Petrol, Euro 4

0

0.0002

0.0004

0.0006

0.0008

0.001

0.0012

0.0014

5-10 10-15

15-20

20-25

25-30

30-35

35-40

40-55

Speed band (km/h)

PM(g

/km

)

1) <202) 20-503) >50

TD (v/km)

Diesel, Euro 1

0

0.02

0.04

0.06

0.08

0.1

0.12

5-10 10-15

15-20

20-25

25-30

30-35

35-40

40-55

Speed band (km/h)

PM (g

/km

)

1) <202) 20-503) >50

TD (v/km)

Diesel, Euro 2

00.010.020.030.040.050.060.070.080.09

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

PM (g

/km

)

1) <202) 20-503) >50

TD (v/km)

Diesel, Euro 3

0

0.01

0.02

0.03

0.04

0.05

0.06

5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-55

Speed band (km/h)

PM (g

/km

)

1) <202) 20-503) >50

TD (v/km)

Diesel, Euro 4

00.005

0.010.015

0.020.025

0.030.035

0.040.045

0.05

5-10 10-15

15-20

20-25

25-30

30-35

35-40

40-55

Speed band (km/h)

PM (g

/km

)

1) <202) 20-503) >50

TD (v/km)

LPG, Euro 1

00.00020.00040.00060.0008

0.0010.00120.00140.00160.0018

5-10 10-15

15-20

20-25

25-30

30-35

35-40

40-55

Speed band (km/h)

PM (g

/km

)

1) <202) 20-503) >50

TD (v/km)

LPG, Euro 2

00.00020.00040.00060.0008

0.0010.00120.00140.00160.0018

5-10 10-15

15-20

20-25

25-30

30-35

35-40

40-55

Speed band (km/h)

PM (g

/km

)

1) <202) 20-503) >50

TD (v/km)

LPG, Euro 3

00.00020.00040.00060.0008

0.0010.00120.00140.00160.0018

5-10 10-15

15-20

20-25

25-30

30-35

35-40

40-55

Speed band (km/h)

PM (g

/km

)

1) <202) 20-503) >50

TD (v/km)

LPG, Euro 4

0

0.0002

0.0004

0.0006

0.0008

0.001

0.0012

0.0014

5-10 10-15

15-20

20-25

25-30

30-35

35-40

40-55

Speed band (km/h)

PM (g

/km

)

1) <202) 20-503) >50

TD (v/km)

Figure 6.7: Average PM emissions predicted using VERSIT+ for different speed and traffic density ranges.


OSCAR EVK4-CT-2002-00083 56

7 EMISSION MODELLING IN THE OSCAR SYSTEM

The structure of the OSCAR System is shown in Figure 7.1. The Emissions Module was planned as a separate entity in the System, and acted as an input to all air quality prediction models. This was clearly advantageous, as it would provide:

(i) The ability to produce emission estimates.

(ii) A consistent input to all models.

(iii) The most logical and consistent way of addressing different traffic scenarios.

(iv) The ability to incorporate new emission functions at a later date without having to change each air quality prediction model.

(v) Easier input of large traffic data files.

(vi) An input which could be made country-specific, to a greater or lesser extent depending on the time available.

(vii) Flexibility in terms of integration with other emission models.

Graphical User Interface/Control Module Visual Basic/Visual.NET

Emissions Database Met Database AQ Database

Emissions pre-processor Met pre-processor AQ Data pre-processor

Level I Models Correlations of C, U, Traffic

Level II Models Eg DMRB, CAR Int, PEARL

Level III Models Eg CAR-FMI, OSPM, GAMMA

Numerical Models ADREA-HF, MIMO

External Models Eg Noise

Spatial/temporal distributions – CO, NO2, PM10, PM2.5 Assessment parameters

PC OSCAR ASSESSMENT SYSTEM

Raw data: - Off line - Excel - Access

Visualisation: - Embedded - GIS (ArcMap) - VIS5D, PAVE

Scenario Analysis: - Transport - Emissions

AQ Assessment: - Limit values - Annual means

WEB BASED OSCAR ASSESSMENT SYSTEM

OSCAR Tutorial Input Requirements Applications

Figure 7.1: Structure of OSCAR System

However, this had a number of implications, including:

(i) The different air quality prediction models in OSCAR (DMRB, CAR-FMI, MIMO, etc.) used different internal emission factors. Making the emissions module external to all models meant that any existing emission calculation routines had to be removed or disabled.

(ii) The Scenario Analysis Tool had to feed directly into the emission module or form part of it, otherwise the effects of different scenarios could not be assessed. Furthermore, any scenario analysis tools in the individual models had to be removed/disabled.

(iii) Any existing model calibrations may have been rendered invalid.


OSCAR EVK4-CT-2002-00083 57

7.1 Proposed structure of the emissions module

A possible approach for estimating emissions (and the effects of scenarios) within the OSCAR System is depicted in Figure 7.2. Both the traffic data pre-processor (TPP) and the emissions model form part of the emissions module. The scenario tool forms part of a separate Scenario Module.

It has been assumed that the user will define the baseline traffic characteristics on the road network, selecting each link at a time via the main user interface and entering the appropriate traffic data (possibly also for each lane and direction). Within the emission module, the TPP converts the traffic data for each link into a format which can be used for modelling emissions. Once this process has been completed for all links, the baseline traffic data are fed into the emissions model (a set of functions to determine emissions from the traffic), and the result of the emissions model is fed into the different air quality models in the OSCAR System. For scenario analysis, a Scenario Tool is used to apply changes to the baseline traffic data, according to the particular scenario being tested. The traffic data associated with the scenario are then fed into the emission and air quality models as before. The separate elements of this structure are discussed in more detail in the following sections.

Figure 7.2: Proposed structure of emissions module, and links to other system modules

7.2 Emissions model

The default emissions model used in the OSCAR System relies upon data and methodologies from a number of different sources, as listed in Table 7.1. These are discussed in more detail below. Two types of model source are listed: those currently used in the Assessment System, and those intended for use in the final version of the System. The System has been designed with flexibility in mind, and the structure should permit the easy integration of new emission data and functions, or the updating of existing functions.

Traffic data input sheet for link n

Select link n

Traffic data for local network

Defining the baseline

traffic data for the

network

Traffic data pre-processor

Defining the traffic data for a scenario Emissions Model

Standardised traffic data for link n

USER INTERFACE

EMISSIONMODULE

Define scenario

SCENARIO MODULE

Scenario tool

Level 0 models

Level 1 models

Level 2 models

Level 3 models


OSCAR EVK4-CT-2002-00083 58

Table 7.1: Proposed modelling approaches and data sources in the OSCAR emissions model.

Source of modelPollutantsource Pollutant(s) Vehicle category

(at level 2) Fuel Euro class Modelling approach Input variables†‡

Current Proposed

Euro 0 Average-speed v COPERT III COPERT IIIPetrol

Euro 1 to Euro 4 Average-speed v COPERT III ARTEMIS

Euro 1 to Euro 3 Average-speed v COPERT III ARTEMISDiesel

Euro 4 Average-speed v COPERT III ARTEMIS

PC

LPG Euro 1 to Euro 4 Emission factors - - ARTEMIS

LGV Petrol/diesel Euro 0 to Euro 2 Average-speed/load v, l COPERT III

ARTEMIS(Markewitz andJoumard, 2005)

Other LDV Not included

Rigid HGV

Articulated HGV

Buses

Coaches

Diesel Euro 0 to Euro 5 Average-speed v, g, l COPERT IIIARTEMIS

(Boulter andBarlow, 2005)

Moped

CO, THC, NOx,PM, CO2

Motorcycle Average-speed v COPERT III ARTEMIS

Engine idle PC All Euro 1 to Euro 4 Emission factors - - OSCAR

Hot exhaustemissions

Primary NO2 Not included

Petrol Euro 0 to Euro 4Cold startexhaust emissions

CO, THC, NOx,CO2

PCDiesel Euro 0 to Euro 3 Emission factors v, s, t, T -

ARTEMIS(André and

Joumard, 2005)

Evaporativeemissions VOC Not included

Tyre wear Average-speed v

Brake wear Average-speed v

Road surface wear

PM10, PM2.5 All All All

Emission factors -

EMEP/CORINAIR(EEA, 2004)

EMEP/CORINAIR(EEA, 2004)

Resuspension PM Not included

† It is assumed that the number of vehicles in each category is known.‡ v = vehicle speed, g = road gradient, l = vehicle load, s = season, t = time of day, T = ambient temperature


OSCAR EVK4-CT-2002-00083 59

7.2.1 Hot exhaust emissions

The original intention in OSCAR was to use the traffic situation approach for modelling emissions, combined with a quantitative means of describing the effects of congestion based on the OSCAR measurements. However, severe delays in the ARTEMIS project meant that the ARTEMIS emissions model could not be made available early enough to enable it to be included in the OSCAR System. Consequently, a decision was made to use the COPERT III model provisionally in OSCAR, primarily to enable the coding of the System to continue. As COPERT III is an average speed model, this means that the OSCAR System was constrained to a corresponding structure at this point, and a subsequent switch to a traffic situation modelling approach would have resulted in unacceptable delays to the project. Furthermore, the rather limited analysis in the previous Chapter indicates that the traffic density-based approach proposed for OSCAR would provide relatively little additional detail compared with an average-speed approach alone.

Consequently, an average-speed approach was retained for modelling emissions in OSCAR. The main implication of this was that the user of the OSCAR System can define the effects of congestion in terms of an average speed, but still not in terms of the effects of different driving styles for a given average speed.

Table 7.2 lists the input parameters proposed for use in the OSCAR System for the modelling of hot exhaust emissions.

Table 7.2: Input parameters required for modelling hot exhaust emissions.

Country General Date

Petrol sulphur level (ppm) Fuel details Diesel sulphur level (ppm)

Link name

Link reference code

Start node grid reference

End node grid reference

Link length (km)

Number of directions (1 or 2)

Number of lanes by direction

Link details

Gradient by direction

Time period (day, hour, or period specified by e.g. traffic model)

Level 1: LDV/HDV/2-wheel vehicles

Level 2: PCs, LGVs, other LD technologies, rigid HGVs, artic HGVs, buses/coaches/mopeds/motorcycles

Level 3: as (2), but disaggregated by weight class (power type for other LD technologies and engine type for 2-wheel vehicles)

Level 4: as (3), but further disaggregated by engine size

Level 5: as (4), but further disaggregated by fuel type

Traffic description by direction and lane

Number of vehicles and speed by time period, according to different levels of detail (depends upon availability of data to user):

Level 6: as (5), but further disaggregated by emission control technology level


OSCAR EVK4-CT-2002-00083 60

7.2.2 Emissions at engine idle

In Table 7.3 some emission factors are proposed for engine idle conditions, based on the OSCAR measurements. It seems probable that NOx emissions from diesel vehicles at idle are likely to be an important consideration.

Table 7.3: Proposed emission factors for engine idle (cars).

Emission factor (g//h) Fuel Euro class

CO THC NOx PM

Euro 1 0.3 0.1 0.04 -

Euro 2 0.6 0.8 0.2 -

Euro 3 0.6 0.1 0.04 -Petrol

Euro 4 0.3 0.5 0.03 -

Euro 1 3.5 0.8 10.6 0.2

Euro 2 4.5 1.0 7.8 0.2 Diesel

Euro 3 2.2 0.5 7.8 0.2

LPG 0.4 0.1 0.05 -

7.2.3 Cold start emissions

In the ARTEMIS project, a new cold start emission calculation approach for cars has been produced by INRETS (André and Joumard, 2005). The approach includes three types of model:

(i) Excess cold start emission per start (complex model)

(ii) Excess cold start emission from traffic (complex model)

(iii) Aggregated cold start emission factors (simplified model) The total cold excess emission for a vehicle and a driving cycle is defined as the excess emission of a vehicle starting at the ambient temperature as compared to a vehicle running in hot conditions. The complexity of the full traffic model is illustrated by Equation 7.1.

Equation 7.1

Where:

Ec(p) traffic excess emissions with a cold engine for the pollutant p corresponding to traffic tfi,h (g)

p atmospheric pollutant

i vehicle type

cm(s,vi) % of mileage recorded under cold start or intermediate temperature conditions for season s and overall speed vI of vehicle type i

s season (winter, summer, intermediate, year)

vi overall average speed for the vehicle type i (km/h)

ωi(p) reference excess emission for the vehicle type i and the pollutant p (g)

h hour (1 to 24, day)


OSCAR EVK4-CT-2002-00083 61

tfi,h traffic flow for the studied vehicle type i and the hour h (veh.km)

ph relative cold start number for the hour h (average=1)

ptfi,h relative traffic for the studied vehicle type i and the hour h (average=1)

j speed class with a cold engine

m trip length class

n class of stops (0 – 1/4, 1/4 – 1/2, 1/2 – 3/4, 3/4 - 1, 1 - 2,..., >12h)

pi,j % of the distance travelled at speed j with a cold engine, for the overall average speed considered, and for the studied vehicle type i (%)

pm,j % of the distance started with a cold engine and distance dm, for speed Vj with a cold engine (%)

ph,n % of the distance travelled after a stop with a duration of tn, for the hour h (%)

dm average distance of the trips under cold start conditions of class m (km)

f(p,Vj,T) plane function of the speed Vj and the temperature T, for the pollutant p

Vj average speed with a cold engine corresponding to class j (km/h)

T ambient temperature (°C)

h(p,δ) (1-ea(p,T).δ)/(1-ea(p,T))

a(p) constant coefficient for a pollutant p

δ(p,T,Vj,d) dimensionless distance = dm/(dc(P,Vj,T)

dc(p,Vj,T) cold distance for the pollutant p (km)

g(p,tn) % of excess emission at 12h of parking as a function of the parking time tn for the pollutant p

tn parking time (h)

The complex model is time-consuming to run, and could not therefore be included in the OSCAR System. Furthermore, the input data required are available to few model users. For specific locations INRETS are able to run the complex model for the local condition, assuming the relevant input data are available.

The simplified model was generated using the complex model, and consists of European-average excess emission factors (in g/km) for a specified hour, based upon the prevailing:

(i) The vehicle category and pollutant (ii) The ambient temperature (iii) The parking time distribution (iv) The trip length distribution (v) The average speed

The result is a table in which the cold start unit excess emission factors are presented for each vehicle category, each pollutant (CO, CO2, THC, NOx, 29 PAHs and 87 VOCs), each season (year, winter, summer and intermediate), hour of the day, ambient temperature and overall traffic mean speed. Each emission factor is multiplied by the traffic activity per vehicle category (in veh.km), and the results are summated to give the total emission factor for the traffic. The output from the simplified model is proposed for use in OSCAR. The cold start EFs are simply added to the hot EFs. There are clearly a number of limitations of this approach for local applications (Joumard, 2005), and the average values could differ from the actual local values quite considerably. For a single street one problem is that each vehicle does not stay in the street for very long. A portion of the extra cold emission can occur before the vehicle enters the street, or after it leaves the street. Ideally, precise journey statistics are required all the vehicles using the street: how far they travel before entering the street, how far they travel before leaving the street, their start engine temperature, all according to the ambient temperature, the hour, etc. If a user is allowed to modify the various statistical information required, he or she is usually unable to verify their coherence. This makes the accurate modelling of local cold start emissions extremely difficult.


OSCAR EVK4-CT-2002-00083 62

7.2.4 Non-exhaust PM emissions PM10 and PM2.5 emissions due to tyre wear, brake wear and road surface wear are calculated using the methodology presented in the European Environment Agency’s Emission Inventory Guidebook (EEA, 2004). The resuspension of material deposited on the road surface is not covered in the OSCAR System.

7.3 Traffic data pre-processor

One of the first stages of the assessment process using the OSCAR System is for the user to populate the road network (in the form of a series of ‘links’) with traffic data. In simple terms, for the estimation of emissions the traffic data need to describe traffic flow (volume), speed and composition. However, there is a need in OSCAR to describe the traffic in a much more detailed manner, mainly because of the following:

(i) The description of the traffic specifically the traffic composition, must be in a format which corresponds to the typical format of emissions data, and the latter tends to be highly disaggregated (e.g. by vehicle type, weight, fuel use, engine size, emission legislation).

(ii) Some of the pollution-reduction scenarios being considered in OSCAR relate to the detailed traffic classification (e.g. scenarios affecting fuel quality, or the numbers of vehicles conforming to different Euro standards).

(iii) The traffic classification must be applicable in all European countries.

However, the traffic data available to model users for a particular link might be available in a number of different formats. The data may be very simple, such as total traffic flow, or it might be much more detailed (e.g. the numbers of vehicles in each vehicle category, the fuel types, and the Euro classes can be determined via video surveys).

The proposed traffic classification is given in Table 7.3. Six levels of classification, of increasing complexity, are proposed: Level 1: LDVs, HDVs and 2-wheel vehicles

Level 2: PC, LGV, Rigid HGV, Articulated HGV, Buses, Coaches, Moped and motorcycles

Level 3: Level 2 plus disaggregation by weight (engine type for 2-wheel vehicles)

Level 4: Level 3 plus disaggregation by engine size

Level 5: Level 4 plus disaggregation by fuel or power source type

Level 6: Level 5 plus disaggregation by emission legislation

The Level 6 information is required for emission modelling and scenario analysis. Most model users, when entering traffic data, should at least have the Level 1, and possibly level 2, information, whereas very few users could be expected to have the Level 6 information.

It is therefore proposed that the user will be allowed to define the traffic at levels 1 and 2. To convert the traffic data provided by the user into traffic data which is useful for modelling, a Traffic Pre-Processor is required. Furthermore, the use of a Traffic Pre-Processor enables the user to test the effects of a much wider range of scenarios than a more rigid structure would allow. The Traffic Pre-Processor is basically a framework of ‘rules’ which convert the ‘simple’ traffic data supplied by the user into the ‘complex’ traffic data required by the Emissions Module. Examples of these rules might include:

• If only traffic flow is available for the link, use the national fleet composition. • If only traffic flow and LDV/HDV split is known, then use national LDV fleet and national

HDV fleet composition. • If registration years for all vehicles are available, then all passenger cars registered after 1

January 1993 = Euro 1

…….etc.


OSCAR EVK4-CT-2002-00083 63

The definition of traffic data can also be conducted on different temporal scales. If one-hour average traffic data are available, then an iterative summation process would be used to determine traffic characteristics and emissions over longer periods.

The output from the Traffic Pre-Processor for a specified hourly period on a given link is the traffic composition (number of vehicles, and speed, in each category) at Level 6. This information is fed directly into the Emissions Module, which produced an emission estimate (‘hot’ exhaust) emissions for all the traffic in g/km per time period.

Table 7.3: Proposed traffic classification – levels 1 to 6.

Level of detail Level 1

(Type 1)

Level 2

(Type2)

Level 3

(Weight, engine type for

2-wheel vehicles)

Level 4

(Engine size)

Level 5

(Fuel/power source, for all Level 4 categories)

Level 6

(Emission legislation, for all Level 5

categories) †

<2.5 t PC

>2.5 t

<1305 kg

1305-1760 kgLGV >1760 kg

LDV

Other LDV (e.g. ULEV)

<1.4 l

1.4-2.0 l

>2.0 l

3.5-7.5 t

7.5-12 t

12-14 t

14-18 t

18-21 t

21-26 t

26-28 t

28-32 t

Rigid HGV

>32 t

<28 t

28-34 t

34-40 t

40-50 t

Articulated HGV

>50 t

<15 t

15-18 t Buses >18 t

<18 t

HDV

Coaches >18 t

2 stroke Moped

4 stroke

2 stroke

50-150 cc 150-750 cc

Vehicle categories

2-wheel vehicles

Motorcycle4 stroke

>750 cc

Petrol: leaded Petrol:unleaded, <150ppm S Petrol: unleaded, <50ppm S Petrol: unleaded, <10ppm S

Diesel, <2000ppm S

Diesel, <500ppm S

Diesel, <350ppm S

Diesel, <50ppm S

Diesel, <10ppm S

LPG

CNG

LNG

Hybrid - Gasoline

Hybrid – Diesel

DME dimethyl ether

Biodiesel (5%blend B5)

Ethanol (85%blend E85)

Methanol (85%blend E86) Pure ME (RpsME, SunfME)

Water diesel emulsion

Electric – battery

Electric – Solar

H2 IC (internal combustion)

H2 FC (fuel cell)

1-PRE ECE

2-ECE 15/00-01

3-ECE 15/02

4-ECE 15/03

5-ECE 15/04

6-Improved Conv.

7-Open Loop

8-Euro I

9-Euro II

10-Euro III

11-Euro IV

12-Euro V

13-pre Euro

14-Euro III - DISI

15-Euro IV – DISI

16-Euro V - DISI

17-Euro III - DPF

18-Euro IV – DPF

19-Euro V - DPF

20 No legislation

Number of possible combinations

3 9 29 43 946 18,920

Number of likely combinations

3 9 29 37 ~400 ~4,500

† Categories 1-7 are combined as ‘Pre-Euro 1’ (or ‘Euro 0’).


OSCAR EVK4-CT-2002-00083 64

8 SUMMARY

8.1 Overview

An important element in the development of the OSCAR Assessment System, and the overarching goal of Workpackage 5, has been an improved understanding of driving characteristics, vehicle operation and exhaust emissions at low speeds, which should help to improve the modelling of emissions associated with different levels of congestion. The principal tasks of the emissions work were to characterise driving patterns in the four main cities, to improve existing emission databases for slow-moving and stationary traffic, and to develop an emissions module for use in the OSCAR System.

8.2 Model review

One of the first steps in Workpackage 5 was to briefly review existing emission models. From this review, the following general conclusions were drawn:

(i) Many emission and air pollution models utilise average speed emission functions. However, for the latest (and future) vehicle technologies, average speed alone is not a reliable determinant of emissions on the street level, and some descriptor of driving cycle dynamics is also required.

(ii) Cycle dynamics are usually defined in terms of speed-related parameters. However, the links between average speed, speed-based cycle dynamics and emissions are not firmly established. Engine load and engine power ought to be more directly related to emissions than speed-based parameters.

(iii) Many model users will only have information relating to traffic flow and speed. Some may have traffic composition information, but very few will have quantitative information on cycle dynamics.

These conclusions formed the basis for the experimental approach taken in OSCAR.

8.3 Methodology

The Workpackage 5 methodology involved the following main stages:

(i) Vehicle operation patterns, road characteristics and traffic conditions were recorded on specified links in each of the four main cities - Athens, Helsinki, London and Madrid.

(ii) Descriptive statistics and parameters were determined for each driving pattern.

(iii) Generalised relationships between (a) traffic speed, (b) traffic density and (c) driving pattern parameters were identified.

(iv) The driving patterns were grouped according to these three sets of parameters.

(v) Representative driving patterns were selected from the different groups, and used to construct driving cycles which reflected driving in all cities for given road and traffic conditions.

(vi) Exhaust emissions were measured over the driving cycles in the laboratory, and the results were incorporated into European databases and models.

The experimental work and results are summarised in the following sections.

8.4 Driving pattern surveys

Real-world driving pattern data were collected continuously using an instrumented vehicle in the four cities. Vehicle speed, engine speed, percentage engine load and throttle position were logged via OBD. A GPS receiver was used to log the location and operation of each instrumented vehicle.


OSCAR EVK4-CT-2002-00083 65

The routes were designed to reflect typical driving conditions on main roads in the study areas, and were divided into distinct ‘links’ to which the driving, traffic and road characteristics could later be referenced. Traffic flow and composition were determined using automatic counters on many of the links.

The processed information from these surveys was entered into a database (Deliverable 5.1). By far the largest proportion of driving patterns were recorded on roads with a 48 km/h or 50 km/h speed limit (56.1% in total). Only the roads having a speed limit of 70 km/h or lower were considered as distinctly ‘urban’ in nature, and little further analysis was conducted on the data for road with a speed limit above 70 km/h. The subsequent analysis focused mainly on the 918 driving patterns which were measured on urban roads with corresponding traffic information.

For roads with a 50 km/h limit (48 km/h in London), the highest and lowest overall mean speeds were recorded in Helsinki (32 km/h) and London (14.8 km/h) respectively. Indeed, the average speeds in Helsinki were generally higher than those on ‘equivalent’ roads in the other cities. The mean RPA by city and speed limit provided a broad indication of the relative levels of driving dynamics. The mean RPA values in Helsinki were somewhat lower than those in the other cities, possibly indicating that driving is generally less aggressive in Helsinki than in the other cities. By contrast, the links in Athens with a 60 km/h speed limit had a particularly high RPA for the average speed, indicative of more aggressive driving. However, the between-city comparisons were confounded by a number of factors, including the distribution of the measurement periods in each city and, by inference, the volumes of traffic, the phasing of traffic signals, road layout and pedestrian activity.

8.5 Relationships between traffic speed, flow and density

The speed-flow relationship in urban areas was rather complex, especially when different locations were compared. It was therefore difficult to support the development of generalised laboratory driving cycles using this information. For each driving pattern the 1-hour average traffic density (v/km, based on LDV equivalents) was also calculated for the periods corresponding to the driving patterns. For traffic densities higher than around 35 vehicles per km per lane, only average speeds lower than 30 km/h were encountered, and speed reduced much more gradually with increasing traffic density. At the highest traffic densities recorded, only very low speeds tended to be observed. In a given city, maximum speeds in the low traffic density region were dependent upon the speed limit, but there was less of a tendency for the shape of the relationship to be affected by the design speed of the road, as in the speed-flow curve. This made the definition of generalised driving patterns considerably easier.

8.6 Construction of dynamometer driving cycles

The relationships between traffic speed, traffic density and driving dynamics formed the basic framework within OSCAR for describing different congestion levels and defining driving cycles. Increasing traffic density was found to have little effect on the values of speed-related parameters (e.g. RPA, APA, speed standard deviation).

An engine’s performance can be defined in terms of its power output. A power-based approach was therefore used to define different levels of congestion, with average power and the number of high-power events being the parameters used. The data were separated into the following speed bands according to average power ranges: (i) less than 15 km/h, (ii) 15-30 km/h, (iii) 30-45 km/h, (iv) 45-60 km/h and (v) 60-90 km/h. Traffic density was not found to be a particularly effective surrogate for driving dynamics. However, in the absence of any other means of quantifying cycle dynamics on the part of the model user, it was considered reasonable to use traffic density as the best available solution in OSCAR. Three traffic density ranges were selected: 15-40, 40-70 and 70-125 vehicles per km per lane.

The speed and traffic density ranges were then applied to the driving pattern data, yielding 8 separate operational regions (A, B, C, D, E, F, G and H). Driving cycles were constructed to represent each of the six regions C to H. Cycles were not developed for the higher-speed regions A and B. Four additional cycles were developed, but these could only be referenced via the speed/density criteria plus a subjective descriptor.


OSCAR EVK4-CT-2002-00083 66

It is important to not that this approach gives only a relatively crude indication of traffic conditions associated with different driving patterns, and the calculation of traffic density is only used as an approximate indicator of these conditions. A more detailed exploration of the relationship between traffic density and cycle dynamics ought to be the subject of future research – it is unlikely that there will be many high power events at very high traffic densities, for example.

8.7 Emission measurements

A total of 20 passenger cars were tested in the programme, including two Euro 4 petrol cars. No Euro 4 diesel or LPG vehicles were tested. However, two of the diesel vehicles and both the LPG vehicles were compliant with Euro 4 emission limits. One of the vehicles (number 9) was fitted with a gasoline direct injection (GDI) engine.

The vehicles were tested over the 10 OSCAR driving cycles, as well as over legislative European driving cycles and ARTEMIS driving cycles. The pollutants measured were CO, HC, NOx and CO2.For the diesel vehicles only, particulate matter (PM) was collected separately on a filter. For each fuel type, the same specification fuel was used throughout the measurements. The emission data were stored by TRL and TNO, and the results were also incorporated into both the ARTEMIS database of emission factors for light-duty vehicles and the database of TNO’s VERSIT+ model.

8.7.1 Hot exhaust emissions

CO emissions from all test vehicles were well below the type approval limit values. All of the petrol and LPG vehicles had combined THC and NOx emissions which were below the legislative limit values. However, three of the diesel vehicles (the single Euro 1 vehicle and two of the Euro 2 vehicles) had combined THC and NOx emissions which were higher than the limit values, although the margins of exceedance were not large. The PM emission limit was slightly exceeded by one Euro 3 diesel vehicle. The test vehicles were selected from the in-service fleet. It would therefore be expected that the emissions would deteriorate from the as-new condition. The results suggested that the test vehicles were roadworthy, probably well-maintained, and not high emitters. Emissions over the OSCAR cycles were compared with emissions over the UDC part of the NEDC. CO emissions over the OSCAR cycles were generally lower than those over the UDC. One Euro 2 petrol vehicles (#03) had high CO emissions over the OSCAR cycles, but otherwise emissions generally decreased with improvements in technology. For diesel Euro 1 vehicles, emissions over the OSCAR cycles were, on average, similar to those over the UDC, but emissions were lower over the OSCAR cycles than over the UDC for Euro 2 and Euro 3 vehicles. Emissions of CO from the two LPG vehicles were much lower over the OSCAR cycles than over the UDC (again the Euro 3 cycle). THC emissions over the OSCAR cycles were lower than emissions over the UDC for all vehicle categories. The effect of vehicle #03 was again apparent over the OSCAR cycles. NOx

emissions for all categories of petrol vehicle were similar over the UDC and the OSCAR cycles, and in both cases rather low. Emissions from diesel vehicles, on the other hand, were higher over the OSCAR cycles than over the UDC, and decreased from Euro 1 to Euro 3. PM emissions over the OSCAR cycles decreased between Euro 1 and Euro 3, whereas emissions over the UDC remained relatively unchanged. For Euro 1 and Euro 2 vehicles emissions over the OSCAR cycles were higher than over the UDC, but there was some convergence in the case of Euro 3. In summary, there is some evidence to suggest that whilst CO and THC emissions can be higher over the UDC than over real-world congestion cycles with a similar speed, real-world NOx and PM from diesel vehicles may be underestimated.

CO emissions from petrol cars over the OSCAR cycles were low (<0.5 g/km) for all Euro classes except Euro 2, with Euro 3 emissions generally being exceptionally low. The Euro 2 vehicles had, on average, particularly high emissions over OSCAR cycles F, G1, G2, H1 and H2, this being entirely due to the high emission levels of Vehicle #03 over these cycles. Although emissions from this vehicle over the NEDC did not indicate that it had the potential to be a high emitter, it was the only vehicle to emit more than 1 g/km of CO over any of the OSCAR cycles, and emitted more than 15 g/km over cycle H2. This level of emissions is very high for a Euro 2 vehicle, and was probably indicative of a fault. Emissions of CO from petrol vehicles over the UDC and CADC


OSCAR EVK4-CT-2002-00083 67

Motorway cycles were generally higher than those over the OSCAR cycles (Euro 2 excepted). For the OSCAR cycles, emissions tended to be lowest over cycles E and F, and highest over C, type-D, type-G and type-H. For Euro 1 and Euro 2 diesel cars, there was a systematic increase in CO emissions as cycle speed decreased, with emissions being lowest over the C and type-D cycles, and highest over the type-H cycles. CO emissions from, the Euro 3 vehicles were lower than those from Euro 1 and Euro 2, but showed a less consistent pattern. For the LPG vehicles (both Euro 3), emissions were generally very low over the OSCAR cycles, and generally increased as speed decreased.

THC emissions from petrol cars over the OSCAR cycles followed a similar pattern to CO emissions from the same vehicles. Again, vehicle #03 was a particularly high emitter over cycles H1 and H2. As in the case of CO, there was a systematic increase in CO emissions as cycle speed decreased, with emissions again being lowest over the C and type-D cycles, and highest over the type-H cycles. THC emissions from LPG vehicles were very low over all cycles with the exception of the UDC (and hence the NEDC of which the UDC forms part).

NOx emissions from petrol vehicles showed exhibited little systematic behaviour over the OSCAR cycles, and showed little dependency on cycle speed. However, for a given speed range the level of driving dynamics does appear to be important, with higher emissions being recorded over cycles having higher dynamics (see type-D, type-G and type-H cycles). This is explored in more detail in the next Chapter. NOx emissions from diesel vehicles over the OSCAR cycles followed a similar pattern to that observed for CO and THC – a general dependence on speed, with little contribution due to driving dynamics. Emissions from LPG vehicles were comparable to those from petrol vehicles, although there was a general decrease in emissions as speed decreased.

PM emissions from diesel vehicles over the OSCAR cycles tended to decrease with decreasing speed, and cycle dynamics appears to have an effect.

8.7.2 Cold-start emissions

Emissions were measured over the ARTEMIS Urban cycle with a full cold start, and at an ambient temperature of 9oC. For CO, the ratio of cold-start to hot-start emissions ranged from around 5 to almost 6,000, whereas for diesel vehicles it tended to be less than 5. For LPG vehicles cold start CO emissions were around 50 times higher than hot-start emissions. Emissions of THC from petrol vehicles over the cold-start cycle were between 6 and 400 times higher than those over the hot-start cycle, whereas the ratio for diesel vehicles was between 1.1 and 2.7. For LPG vehicles, cold-start THC emissions were around 25 times higher than hot-start emissions. The effects of cold starting on NOx and PM were less pronounced, but still notable. For NOx, the cold-start:hot-start ratio was less than one for some vehicles, but was generally greater than one, and as high as four. PM emissions from diesel vehicles over the cold-start cycle were between 1.2 and 2.6 times higher than emissions over the hot-start cycle. The results relate to full cold starts, in which the engine has been allowed to cool down to the ambient temperature. Many starts occur at intermediate temperatures, whereby the engine temperature is between the ambient temperature and the full operational temperature. Emissions during these ‘warm’ starts will be lower than those during full cold starts. The results clearly indicate that cold start emissions should be included in the OSCAR System, given that large numbers of starts occur in central urban areas.

8.7.3 Idle emissions

Emissions when the engine at idle are usually included in the driving cycles used to derive average speed emission functions. However, during extended periods of congestion it is possible that vehicles remain at idle for long periods, and therefore the average-speed functions cannot be applied. In spite of this possibility, emission models do not generally provide emission factors for engine idle. The emissions measured during OSCAR for idle (stated in g/h) show that idle emissions from diesel vehicles tend to be considerably higher than those from petrol and LPG vehicles, particularly for CO and NOx.


OSCAR EVK4-CT-2002-00083 68

8.7.4 Effects of congestion on emissions

Different types of congestion were defined in OSCAR as a function of average speed and driving dynamics, and traffic density was selected as a surrogate indicator for the general level of driving dynamics.

For cycles having the same average speed, cycle dynamics appears to have a notable effect for some vehicle types and pollutants. For diesel vehicles, PM emissions appear to be influenced by cycle dynamics, and the other pollutants to a lesser extent. For petrol and LPG vehicles the results are rather mixed, but in some cases cycle dynamics did appear to be important. However, it was not straightforward to show which operational parameters are contributing to the differences in emissions over the cycles. Depending on the vehicle type and pollutant, the results varied considerably, partly due to the consideration of small vehicle samples. A means was therefore required of examining the effects of cycle dynamics for a larger selection of vehicles, and this was only possible through modelling.

The urban driving patterns (~900) recorded in the four cities were sent to TNO to be processed using the VERSIT+ model (see Section 2.3.6). It was anticipated that the results could be used to ‘map’ emissions as a function of traffic speed and density in order to provide indication of congestion effects for use either independently in OSCAR or in other models. However, when running VERSIT+ TNO noted that for some driving patterns the model had to be rejected. The main reason for this was that some of the OSCAR driving patterns were short (<1 minute), compared with the typical driving patterns used in VERSIT+ (~5-6 minutes), and were associated with vehicle operating conditions outside the acceptable range within the mode. Consequently, only the results for around 200 driving patterns were available, rendering the mapping process untenable. Figure 6.3 shows two different sets of results from this process, indicating that average speeds remains a reasonably reliable predictor of NOx emissions for Euro 3 diesel vehicles, but not for Euro 3 petrol vehicles.

Instead, the VERSIT+ results were used to provide further information on congestion effects, by calculating average emission levels for different speed and traffic density ranges. The results, shown in Figures 6.4 to 6.7, provided a clearer indication of the importance of different levels of congestion than the OSCAR emission measurements alone.

In general, there was little evidence that changes in traffic density at a given average speed had a systematic effect on emissions, even though in some cases there were large differences between the predicted emissions associated with different traffic densities. The general observations from these comparisons are:

• The effect of traffic density on CO and THC emissions from petrol vehicles was variable, and showed no systematic pattern, although for Euro 1 to Euro 3 vehicles emissions of both pollutants were highest for mid-range traffic densities at low and high speeds.

• NOx emissions from petrol Euro 2 and Euro 3 vehicles appeared to be particularly low at low traffic densities.

• For diesel vehicles there was little evidence of a traffic density effect, for any pollutant and average speed alone seems adequate for predicting emissions. However, for diesel Euro 3 and Euro 4 vehicles high traffic densities appeared to result in higher CO emissions at speeds lower than around 15 km/h, and mid-range traffic densities resulted in higher emissions than low traffic densities in the 40-55 km/h speed band.

• In the case of LPG vehicles, the results for CO and NOx were rather variable and there was no systematic traffic density effect. However, THC emissions at low speeds appeared to be highest for high traffic densities.

8.8 Emission modelling in the OSCAR system

The Emissions Module was planned as a separate entity in the OSCAR system, and acted as an input to all air quality prediction models. This was clearly advantageous, as it provided the ability to produce separate and consistent emission estimates, the most logical and consistent way of addressing different traffic scenarios, the ability to incorporate new emission functions at a later date without having to change each air quality prediction model, easier input of large traffic data


OSCAR EVK4-CT-2002-00083 69

files, and flexibility in terms of integration with other emission models. The Emissions Module had two main elements:

(i) An emissions model, which contained the emission functions for different vehicle categories.

(ii) A ‘traffic data pre-processor’ (TPP), which provided the vehicle fleet description in a consistent format.

8.8.1 Emissions model

For the OSCAR System to be applicable consistently on a European level, these new measurements would not, on their own, be sufficiently extensive. Hence, the emissions work has not been designed to provide a completely new emission model, but rather to supplement and improve the underlying data and methodologies of existing models.

The original intention in OSCAR was to use the traffic situation approach for modelling emissions, combined with a quantitative means of describing the effects of congestion, based on the OSCAR measurements. However, severe delays in the ARTEMIS project meant that the ARTEMIS emissions model could not be made available early enough to enable it to be included in the OSCAR System. Consequently, a decision was made to use the COPERT III model provisionally in OSCAR, primarily to enable the coding of the system to continue. As COPERT III is an average speed model, this means that the OSCAR System was constrained to a corresponding structure at this point, and a subsequent switch to a traffic situation modelling approach would have resulted in unacceptable delays to the project. Furthermore, the rather limited analysis in the previous Chapter indicates that the traffic density-based approach proposed for OSCAR would provide relatively little additional detail compared with an average-speed approach alone. Consequently, an average-speed approach was retained for modelling emissions in OSCAR. The implication of this is that the user of the OSCAR System can define the effects of congestion in terms of an average speed, but still not in terms of the effects of different driving styles for a given average speed.

Emission factors were provided for engine idle conditions, based on the OSCAR measurements. It is clear that NOx emissions from diesel vehicles at idle are likely to be an important consideration.

In the ARTEMIS project, a new cold start emission calculation approach for cars has been produced. The main model is complex and time-consuming to run, and could not therefore be included in the OSCAR System. Furthermore, the input data required are available to few model users. A simplified model was generated using the complex model, and consists of European-average excess emission factors (in g/km) for a specified hour, based upon the vehicle category and pollutant, the ambient temperature, the parking time distribution, the trip length distribution and the average speed. The result is a table in which the cold start unit excess emission factors are presented for each vehicle category, each pollutant (CO, CO2, THC, NOx, 29 PAHs and 87 VOCs), each season (year, winter, summer and intermediate), hour of the day, ambient temperature and overall traffic mean speed. Each emission factor is multiplied by the traffic activity per vehicle category (in veh.km), and the results are summated to give the total emission factor for the traffic. The output from the simplified model was proposed for use in OSCAR. The cold start EFs are simply added to the hot EFs. There are clearly a number of limitations of this approach for local applications, and the average values could differ from the actual local values quite considerably. This accurate modelling of local cold start emissions remains extremely difficult.

It is proposed that PM10 and PM2.5 emissions from tyre wear, brake wear and road surface wear are to be calculated using the methodology presented in the European Environment Agency’s Emission Inventory Guidebook (EEA, 2004).

8.8.2 Traffic data pre-processor

One of the first stages of the assessment process using the OSCAR System will be for the user to populate the road network (in the form of a series of ‘links’) with traffic data. In simple terms, for the estimation of emissions the traffic data need to describe volume flow (volume), speed, and composition. However, there is a need in OSCAR to describe the traffic in a much more detailed manner, mainly because the description of the traffic composition must be in a format which


OSCAR EVK4-CT-2002-00083 70

corresponds to the typical format of emissions data. Six levels of classification, of increasing complexity, were used:

Level 1: LDVs, HDVs and 2-wheel vehicles

Level 2: PC, LGV, Rigid HGV, Articulated HGV, Buses, Coaches, Moped and motorcycles

Level 3: Level 2 plus disaggregation by weight (engine type for 2-wheel vehicles)

Level 4: Level 3 plus disaggregation by engine size

Level 5: Level 4 plus disaggregation by fuel or power source type

Level 6: Level 5 plus disaggregation by emission legislation

The Level 6 information is required for emission modelling and scenario analysis. Most model users, when entering traffic data, should at least have the Level 1, and possibly level 2, information, whereas very few users could be expected to have the Level 6 information. To convert the traffic data provided by the user into traffic data which is useful for modelling, a Traffic Pre-Processor is required. Furthermore, the use of a Traffic Pre-Processor enables the user to test the effects of a much wider range of scenarios than a more rigid structure would allow. The Traffic Pre-Processor was basically a framework of ‘rules’ which convert the ‘simple’ traffic data supplied by the user into the ‘complex’ traffic data required by the Emissions Module. The output from the Traffic Pre-Processor for a specified hourly period on a given link is the traffic composition (number of vehicles, and speed, in each category) at Level 6. This information is fed directly into the Emissions Module, which produced an emission estimate (‘hot’ exhaust) emissions for all the traffic in g/km per time period.


OSCAR EVK4-CT-2002-00083 71

9 CONCLUSIONS AND RECOMMENDATIONS

The main conclusions and recommendations from this work are given below.

9.1 Conclusions

Review

1. Many emission and air pollution models utilise average speed emission functions. However, for the latest vehicle technologies average speed alone is not a reliable determinant of emissions on the street level, and some descriptor of driving cycle dynamics is also required.

2. Cycle dynamics are usually defined in terms of speed-related parameters. However, the links between average speed, speed-based cycle dynamics and emissions are not firmly established.

3. Engine load and engine power ought to be more directly related to emissions than speed-based parameters.

4. Many model users will only have information relating to traffic flow and speed. Some may have traffic composition information, but very few will have quantitative information on cycle dynamics.

Driving patterns and cycles

5. Average speeds in Helsinki were higher than those on ‘equivalent’ roads in the other cities, and there was some indication that driving was less ‘aggressive’. However, the between-city comparisons were confounded by a number of factors, including the distribution of the measurement periods, the phasing of traffic signals, road layout and pedestrian activity.

6. The overall speed-flow relationship for the four cities was rather complex, and it was therefore difficult to support the development of generalised driving cycles using this information.

7. There was less of a tendency for the shape of the speed-density relationship to be affected by the city and the design speed of the road. Although traffic density was not strongly related to driving dynamics (defined in terms of engine power output), in the absence of any other means of quantifying dynamics on the part of the model user it was considered reasonable to define congestion, and develop different driving cycles, in terms of average speed and traffic density.

Emission measurements

8. Emissions of the regulated pollutant from all test vehicles were generally below the type approval limit values. Otherwise, the margins of exceedance were not large. The test vehicles were therefore considered to be roadworthy, probably well-maintained, and not high emitters.

9. Comparisons between emissions over the OSCAR cycles and the UDC provided some

evidence to suggest that, whilst CO and THC emissions can be higher over the UDC than over real-world congestion cycles with a similar speed, real-world NOx and PM from diesel vehicles may be underestimated.

10. For petrol cars, cold-start emissions of CO and THC were up to several orders of magnitude higher than hot-start emissions, whereas for diesel vehicles the cold:hot ratio tended to be less than 5. The effects of cold starting on NOx and PM were less pronounced, but still notable.


OSCAR EVK4-CT-2002-00083 72

11. The emissions measured during OSCAR for engine idle conditions (stated in g/h) showed that idle emissions from diesel vehicles tend to be considerably higher than those from petrol and LPG vehicles, particularly for CO and NOx.

Congestion effects

12. For cycles having the same average speed, cycle dynamics was found to have a notable effect for some vehicle types and pollutants, but the effects were not systematic.

13. NOx emissions from petrol vehicles over the OSCAR cycles showed little dependency on speed. However, for a given speed range the level of driving dynamics was found to be important, with higher emissions being recorded over cycles having higher dynamics. In contrast, NOx emissions from diesel vehicles over the OSCAR cycles exhibited a general dependence on speed, with little contribution due to driving dynamics. Emissions from LPG vehicles were comparable to those from petrol vehicles.

14. PM emissions from diesel vehicles over the OSCAR cycles tended to decrease with decreasing speed, and cycle dynamics appeared to have an effect.

15. Results from VERSIT+ were used to provide further information on congestion effects. Again, there was little evidence that changes in traffic density at a given average speed had a systematic effect on emissions, even though in some cases there were large differences between the predicted emissions associated with different traffic densities.

Emission modelling

16. A separate OSCAR emissions module was developed independently from the models included in the OSCAR system. This provided consistency and flexibility.

9.2 Recommendations and best practice

A number of recommendations are made below for emission modelling in the OSCAR system. These should also be viewed as suggestions for best practice for local-scale emission modelling in general.

1. Traffic density was not found to be a particularly effective surrogate for driving dynamics. However, in the absence of any other means of quantifying cycle dynamics on the part of the model user, it was considered reasonable to use traffic density as the best available solution in OSCAR. A more detailed exploration of the relationship between traffic density and cycle dynamics ought to be the subject of future research – it is unlikely that there will be many high power events at very high traffic densities, for example.

2. Engine load and engine power ought to be more directly related to emissions than speed-based parameters. Any modelling approach should consider these parameters, but should bear in mind the limited input information available to most model users.

3. In OSCAR, new exhaust emission measurements were conducted on 20 passenger cars. However, for the OSCAR System to be applicable consistently on a European level, these new measurements would not, on their own, have been sufficiently extensive. Hence, the emissions work was not designed to provide a completely new emission model, but rather to supplement and improve the underlying data and methodologies of existing models. The emissions data from OSCAR were therefore incorporated into both the ARTEMIS database of emission factors for light-duty vehicles and the database of TNO’s VERSIT+ model. However, because of delays in the ARTEMIS project and the constraints on the development of the OSCAR system, the new emission factors could not be incorporated in the system during the project. Furthermore, the original intention in OSCAR was to use the traffic situation approach for modelling emissions, combined with a quantitative means of describing the effects of congestion, based on the OSCAR measurements. The work in OSCAR indicated that a traffic density-based approach would provide relatively little


OSCAR EVK4-CT-2002-00083 73

additional detail compared with an average-speed approach alone. Consequently, an average-speed approach, despite its limitations, should currently be retained for modelling emissions in OSCAR, and the emission factors from the European COPERT III model should be used. Emission factors from the ARTEMIS project should be incorporated in future versions of the OSCAR system as and when appropriate.

4. It is clear that NOx emissions from diesel vehicles engine at idle are likely to be an important consideration. Again, it was not possible to include emission factors for engine idle conditions in the current OSCAR system, but these should be included in future versions. These emission factors should only be applied when the entire traffic is completely stationary during the time period being modelled.

5. The cold start emission test results clearly indicate that cold start emissions should be included in the OSCAR System, given that large numbers of starts occur in central urban areas. A complex cold start model has been developed in ARTEMIS, although there are some concerns about the application of this model to local assessments, given the large amount of input data required. Consideration should be given to the inclusion of the ARTEMIS cold start model in future versions of the OSCAR system.

6. PM10 and PM2.5 emissions from tyre wear, brake wear and road surface wear should be calculated in the current OSCAR system using the methodology in the EEA’s Emission Inventory Guidebook. Resuspension should not be included in the current system due to the absence of a universally-applicable method. Methods for estimating emissions due to resuspension should be reviewed periodically for inclusion in the system.

7. A detailed traffic pre-processor should be used in the current OSCAR system to ensure that a consistent set of assumptions are used to derive detailed fleet descriptions from simple input data. Further developments to the pre-processor should be made as and when required, and the sensitivity of emission estimates to traffic inputs should be assessed.


OSCAR EVK4-CT-2002-00083 74

10 ACKNOWLEDGEMENTS

The work described in this report was carried out in the Energy, Emissions and Air Pollution Team of TRL Limited. It was jointly funded under the European Commission 5th framework programme by DGTREN, contact number EVK4-CT-2002-00083 and the UK Department for Transport, contract number PPAD 09/099/064, project reference UG575.


OSCAR EVK4-CT-2002-00083 75

11 REFERENCES

Akcelik R (2003). Speed-flow and bunching relationships for uninterrupted flows. 25th Conference of Australian Institutes of Transport Research, University of South Australia, Adelaide, 3-5 December 2003. André J-M and Joumard R (2005). Modelling of cold start excess emissions for passenger cars. INRETS report LTE 0509. Laboratoire Transports et Environnement, INRETS, case 24, 69675 Bron cedex, France. Atjay D, Weilenmann M and Soltic P (2005). Towards accurate instantaneous emission models. Atmospheric Environment 39, pp. 2443-2449. Barlow T (2005). Personal communication.

Boulter P G and Barlow T (2005). ARTEMIS: Average speed emission functions for heavy-duty road vehicles. TRL Report UPR/IEA/12/05 (unpublished). TRL Ltd., Wokingham, United Kingdom. De Haan P and Keller M (2003). Art.Kinema – User Guide to Version RC1). INFRAS, Berne. Ericsson E (2000). Variability in urban driving patterns. Transportation Research Part D 5, pp 337-354. Elsevier Science Ltd. European Environment Agency (2004). EMEP/CORINAIR Emission Inventory Guidebook. 3rd edition – September 2004 update. Technical Report No. 30, EEA, Copenhagen. http://reports.eea.eu.int/EMEPCORINAIR4/en

Greenberg H (1959). An Analysis of Traffic Flow. Operations Research, Vol 7, pp. 78-85. Greenshields B D (1935). A Study of Traffic Capacity. Highway Research Board Proceedings 14, pp. 448-477. Hall F L, Hurdle V F and Banks J M (1992). Synthesis of recent work on the nature of the speed-flow and flow-occupancy (or density) relationships on freeways. Transportation Research Record 1365, pp12-17. Hassounah, MI and Miller EJ, (1995). Modelling air pollution from road traffic: a review. Traffic engineering and control, Vol 35 (9), pp 510-514. Hedges A (2001). Perceptions of Congestion: report on qualitative research findings (part 4) Department for Transport, available at: http://www.dft.gov.uk/itwp/congestion/04.htm. Heywood J B (1988). Internal Combustion Engine Fundamentals. McGraw-Hill College. ISBN: 007028637X. INFRAS (2004). Handbook of Emission Factors for Road Transport, Version 2.1. INFRAS, Berne, Switzerland, February 2004. Jost P, Hassel D, Webber F-J and Sonnborn (1992). Emission and fuel consumption modelling based on continuous measurements. Deliverable No. 7, DRIVE Project V1053. TγVRhineland, Cologne. Joumard R (2005). Personal communication to Paul Boulter at TRL Ltd. Joumard R, Jost P and Hickman A J (1995). Influence of instantaneous speed and acceleration on hot passenger car emissions and fuel consumption. SAE paper 950928. Society of Automotive Engineers Inc., Warrendale, Pennsylvania.


OSCAR EVK4-CT-2002-00083 76

Kean A J, Harley R A and Kendall G R (2003). Effects of vehicle speed and engine load on motor vehicle emissions. Environmental Science and Technology, Vol. 37, No. 17, pp3739-3746. Leung Y C and Williams D J (2000). Modelling of Motor Vehicle Fuel Consumption and Emissions Using a Power-Based Model. Environment Monitoring and Assessment, Vol. 65, pp. 21-29. Markewitz K and Joumard R (2005). Atmospheric pollutant emission factors of light duty vehicles. INRETS report LTE 0508A. Laboratoire Transports et Environnement, INRETS, case 24, 69675 Bron cedex, France. Negrenti E, (1998). The "corrected average speed" approach in ENEA's TEE model: an innovative solution for the evaluation of the energetic and environmental impacts of urban transport policies. ARRB Transport Research Limited Conference, 19th, 1998, Sydney, Australia. NCHRP (1975). Traffic control in over-saturated street networks. NCHRP Report 194, National Cooperative Highway Research Program of the Transportation Research Board (TRB), Washington DC. Ntziachristos L and Samaras Z (2000). COPERT III. Computer program to calculate emissions from road transport. Methodology and emission factors (version 2.1). Technical Report No. 49. European Environment Agency, Copenhagen.

OECD (2003). External costs of transport in Central and Eastern Europe. Final Report of Working Party on National Environmental Policy - Working Group on Transport. 08-Aug-2003.

Osses M, Henriquez A and Trivino R (2002). Positive mean acceleration for the determination of traffic emissions. Paper presented at the conference ‘Transport and Air Pollution’, Graz, Austria, 19-21 June 2002. Press W H, Teukolsky S A, W T Vetterling and B P Flannery (2002). Numerical Recipes in C: The Art of Scientific Computing. Cambridge University Press. ISBN 0521-43108-5. Small K A (1992). Urban Transportation Economics. Vol. 51 of Fundamentals of Pure and Applied Economics series, 1992. Smit R, Smokers R and Schoen E (2005). VERSIT+ LD: Development of a new emission factor model for passenger cars linking real-world emissions to driving cycle characteristics. Proceedings of 14th International Symposium on Transport and Air Pollution Graz, Austria, 1-3 June. Sturm P J, Pucher K & Almbauer R A (1994). 'Determination of motor vehicle emissions as a function of the driving behaviour'. Proceedings of the International Conference 'The Emission Inventory: Perception and Reality'. Publication VIP-38, pp483-494, Air and Waste Management Association, Pittsburgh, Pennsylvania, USA. Victoria Transport Policy Institute (2002). Transportation cost and benefit analysis - congestion costs. Accessed online : www.vtpi.org. Weilenmann M, Bach C and Rüdy C, (2000). Aspects of instantaneous emission measurement. Proceedings of 9th Annual symposium "Transport and air pollution", Avignon, 5-8 June, 2000. Published by INRETS, Arcueil, France, pp 119-126.


OSCAR EVK4-CT-2002-00083 77

Appendix A: OSCAR driving cycles

Cycle C

0

10

20

30

40

50

60

70

80

90

100

0 50 100 150 200 250 300 350 400

Time (s)

Spee

d(k

m/h

)

Vehicle speed

Gear change

Figure A1: OSCAR driving cycle C.

Cycle D1

0

10

20

30

40

50

60

70

80

90

100

0 50 100 150 200 250 300 350 400 450

Time (s)

Spee

d(k

m/h

)

Vehicle speed

Gear change

Figure A2: OSCAR driving cycle D1.


OSCAR EVK4-CT-2002-00083 78

Cycle D2

0

10

20

30

40

50

60

70

80

90

100

0 50 100 150 200 250 300 350 400

Time (s)

Spee

d(k

m/h

)Vehicle speed

Gear change

Figure A3: OSCAR driving cycle D2.

Cycle E

0

10

20

30

40

50

60

70

80

90

100

0 50 100 150 200 250 300 350 400

Time (s)

Spee

d(k

m/h

)

Vehicle speed

Gear change

Figure A4: OSCAR driving cycle E.


OSCAR EVK4-CT-2002-00083 79

Cycle F

0

10

20

30

40

50

60

70

80

90

100

0 50 100 150 200 250 300 350 400 450

Time (s)

Spee

d(k

m/h

)Vehicle speed

Gear change

Figure A5: OSCAR driving cycle F.

Cycle G1

0

10

20

30

40

50

60

70

80

90

100

0 100 200 300 400 500

Time (s)

Spee

d(k

m/h

)

Vehicle speed

Gear change

Figure A6: OSCAR driving cycle G1.


OSCAR EVK4-CT-2002-00083 80

Cycle G2

0

10

20

30

40

50

60

70

80

90

100

0 50 100 150 200 250 300 350 400

Time (s)

Spee

d(k

m/h

)Vehicle speed

Gear change

Figure A7: OSCAR driving cycle G2.

Cycle H1

0

10

20

30

40

50

60

70

80

90

100

0 50 100 150 200 250 300 350 400

Time (s)

Spee

d(k

m/h

)

Vehicle speed

Gear change

Figure A8: OSCAR driving cycle H1.


OSCAR EVK4-CT-2002-00083 81

Cycle H2

0

10

20

30

40

50

60

70

80

90

100

0 50 100 150 200 250 300 350 400 450

Time (s)

Spee

d(k

m/h

)Vehicle speed

Gear change


Cycle H3

0

10

20

30

40

50

60

70

80

90

100

0 50 100 150 200 250 300 350 400

Time (s)

Spee

d(k

m/h

)

Vehicle speed

Gear change



OSCAR EVK4-CT-2002-00083 82

Appendix B: Tabulated emission results

Table B1: emissions – OSCAR cycles

Vehicle Emission factor by cycle (g/km) Vehicle

category Ref. no.

Model C D1 D2 E F G1 G2 H1 H2 H3

Petrol Euro 1 1 Opel Astra 1.6 0.387 0.270 0.192 0.042 0.117 0.274 0.298 0.302 0.321 0.301

Petrol Euro 2 2 Ford Focus 1.6 0.004 0.003 0.020 0.007 0.162 0.251 0.107 0.237 0.383 0.302

3 Nissan Almera 1.4 0.605 0.107 0.104 0.385 1.839 2.005 2.539 7.776 15.08 0.353

4 Volkswagen Polo 1.4 0.041 0.085 0.001 0.004 0.224 0.073 0.001 0.058 0.237 0.102

Petrol Euro 3 5 Suzuki Alto 1.1 0.001 0.004 0.001 0.001 0.000 0.001 0.006 0.009 0.024 0.009

6 Volvo V40 2.0 0.013 0.009 0.009 0.001 0.066 0.072 0.007 0.635 0.414 0.605

7 Peugeot 307 SW 1.6 0.001 0.001 0.001 0.001 0.010 0.011 0.001 0.298 0.017 0.017

Petrol Euro 4 8 Mazda 6 1.8 0.099 0.175 0.028 0.077 0.176 0.341 0.336 0.165 0.333 0.698

9 Volkswagen Touran 1.6 FSi 0.001 0.176 0.163 0.010 0.010 0.001 0.001 0.018 0.009 0.001

Diesel Euro 1 10 Ford Fiesta D 0.333 0.359 0.390 0.397 0.481 0.462 0.524 0.513 0.634 0.638

Diesel Euro 2 11 Audi A4 TDi 0.107 0.029 0.078 0.045 0.042 0.100 0.149 0.354 0.560 0.338

12 Peugeot 306 1.9D 0.173 0.216 0.309 0.192 0.275 0.485 0.588 0.679 0.751 0.844

13 Toyota Picnic 2.2TD 0.561 0.543 0.751 0.865 0.727 0.872 0.884 1.131 1.304 1.232

Diesel Euro 3 14 Opel Astra 1.7 DTi 0.026 0.004 0.024 0.035 0.028 0.070 0.068 0.891 0.260 0.233

15 Volvo V70 D5 0.001 0.006 0.003 0.014 0.024 0.012 0.017 0.067 0.567 0.001

16 Volkswagen Passat 1.9 TDi 0.016 0.004 0.198 0.024 0.018 0.019 0.436 0.042 0.050 0.040

17 Renault Megane 1.5DCi 0.139 0.148 0.781 1.129 0.100 0.685 0.220 0.334 0.662 0.393

18 Toyota Avensis 2.0 D4D 0.005 0.004 0.004 0.016 0.013 0.008 0.036 0.043 0.051 0.021

LPG Euro 3 19 Alfa Romeo 147 1.6 0.047 0.006 0.031 0.020 0.090 0.147 0.051 0.158 0.106 0.785

20 Volvo S60 Bi-fuel 0.002 0.000 0.000 0.069 0.019 0.011 0.113 0.050 0.039 0.056

Table B2: CO emissions – Other cycles

Vehicle Emission factor by cycle (g/km)

Vehicle category Ref.

no. Model UDC EUDC NEDC

CADC Urban Hot

CADC Road

CADC Motorway

CADC Urban Cold 9°C

Idle (g/h)

Petrol Euro 1 1 Opel Astra 1.6 0.477 0.138 0.264 0.314 0.206 1.509 8.226 0.329

Petrol Euro 2 2 Ford Focus 1.6 0.253 0.058 0.131 0.165 0.068 0.070 2.795 1.365

3 Nissan Almera 1.4 1.854 0.213 0.816 0.874 0.984 2.569 4.48 0.273

4 Volkswagen Polo 1.4 0.771 0.039 0.309 0.016 0.112 0.281 3.357 0.268

Petrol Euro 3 5 Suzuki Alto 1.1 0.918 0.131 0.423 0.001 0.530 3.510 4.926 0.005

6 Volvo V40 2.0 2.108 0.031 0.800 0.052 0.116 0.304 3.365 1.719

7 Peugeot 307 SW 1.6 0.657 0.151 0.339 0.001 0.323 1.065 1.233 0.071

Petrol Euro 4 8 Mazda 6 1.8 0.780 0.197 0.412 0.159 0.483 0.924 6.441 0.392

9 Volkswagen Touran 1.6 FSi 0.865 0.022 0.332 0.151 0.034 0.167 1.005 0.198

Diesel Euro 1 10 Ford Fiesta D 0.508 0.175 0.298 0.502 0.357 0.321 1.239 3.525

Diesel Euro 2 11 Audi A4 TDi 0.738 0.010 0.279 0.099 0.015 0.020 0.734 5.125

12 Peugeot 306 1.9D 0.624 0.066 0.272 0.349 0.072 0.058 0.526 3.308

13 Toyota Picnic 2.2TD 1.037 0.217 0.522 0.542 0.253 #N/A 0.901 5.217

Diesel Euro 3 14 Opel Astra 1.7 DTi 0.376 0.000 0.141 0.081 0.008 0.003 0.822 3.602

15 Volvo V70 D5 1.168 0.002 0.429 0.240 0.002 0.010 0.839 0.099

16 Volkswagen Passat 1.9 TDi 0.322 0.007 0.124 0.280 0.019 0.020 0.389 0.310

17 Renault Megane 1.5DCi 0.779 0.008 0.294 0.105 0.006 0.015 0.432 7.107

18 Toyota Avensis 2.0 D4D 0.567 0.004 0.213 0.010 0.003 0.007 0.409 0.081

LPG Euro 3 19 Alfa Romeo 147 1.6 1.012 0.043 0.401 0.034 0.158 0.289 1.280 0.452

20 Volvo S60 Bi-fuel 1.440 0.056 0.570 0.027 0.051 0.364 1.436 0.320


OSCAR EVK4-CT-2002-00083 83

Table B3: THC emissions – OSCAR cycles



no. Model C D1 D2 E F G1 G2 H1 H2 H3



3 Nissan Almera 1.4 0.038 0.010 0.010 0.025 0.069 0.075 0.090 0.289 0.484 0.056

4 Volkswagen Polo 1.4 0.004 0.003 0.002 0.003 0.027 0.007 0.007 0.006 0.018 0.011


6 Volvo V40 2.0 0.005 0.003 0.003 0.011 0.009 0.010 0.012 0.018 0.022 0.029

7 Peugeot 307 SW 1.6 0.002 0.001 0.001 0.001 0.002 0.002 0.003 0.002 0.002 0.002





12 Peugeot 306 1.9D 0.033 0.038 0.060 0.046 0.046 0.066 0.100 0.074 0.084 0.133

13 Toyota Picnic 2.2TD 0.171 0.167 0.225 0.252 0.194 0.228 0.232 0.229 0.314 0.299


15 Volvo V70 D5 0.020 0.027 0.034 0.021 0.027 0.031 0.042 0.062 0.196 0.048





20 Volvo S60 Bi-fuel 0.003 0.001 0.002 0.004 0.002 0.005 0.004 0.004 0.008 0.004

Table B4: THC emissions – Other cycles




CADC Urban Hot

CADC Road

CADC Motorw

ay


Idle (g/h)



3 Nissan Almera 1.4 0.238 0.021 0.101 0.047 0.023 0.067 0.839 1.537

4 Volkswagen Polo 1.4 0.354 0.002 0.132 0.021 0.008 0.012 0.697 0.639


6 Volvo V40 2.0 0.341 0.003 0.129 0.015 0.009 0.024 0.638 0.082

7 Peugeot 307 SW 1.6 0.085 0.008 0.037 0.002 0.004 0.008 0.199 0.011





12 Peugeot 306 1.9D 0.092 0.017 0.045 0.070 0.020 0.013 0.113 0.237



15 Volvo V70 D5 0.137 0.014 0.059 0.053 0.008 0.006 0.126 0.806





20 Volvo S60 Bi-fuel 0.101 0.001 0.038 0.010 0.002 0.005 0.293 0.016


OSCAR EVK4-CT-2002-00083 84

Table B5: NOx emissions – OSCAR cycles






3 Nissan Almera 1.4 0.014 0.022 0.021 0.053 0.073 0.078 0.105 0.132 0.093 0.317

4 Volkswagen Polo 1.4 0.266 0.137 0.345 0.359 0.215 0.129 0.509 0.107 0.266 0.444


6 Volvo V40 2.0 0.101 0.091 0.087 0.177 0.152 0.165 0.273 0.000 0.029 0.390

7 Peugeot 307 SW 1.6 0.033 0.018 0.029 0.035 0.045 0.029 0.038 0.005 0.062 0.028





12 Peugeot 306 1.9D 0.735 0.782 0.934 1.131 1.277 1.295 1.301 1.725 1.704 1.578

13 Toyota Picnic 2.2TD 0.466 0.345 0.550 0.536 0.475 0.479 0.578 0.611 0.614 0.647


15 Volvo V70 D5 1.126 1.183 1.491 0.854 1.106 1.708 2.611 3.213 2.788 3.142




LPG euro 3 19 Alfa Romeo 147 1.6 0.059 0.033 0.042 0.049 0.015 0.032 0.034 0.034 0.048 0.034

20 Volvo S60 Bi-fuel 0.224 0.058 0.116 0.040 0.072 0.066 0.019 0.012 0.042 0.036

Table B6: NOx emissions – Other cycles




CADC Urban Hot

CADC Road

CADC Motorway


Idle (g/h)



3 Nissan Almera 1.4 0.174 0.042 0.090 0.289 0.133 0.133 0.287 0.444

4 Volkswagen Polo 1.4 0.184 0.010 0.075 0.480 0.126 0.029 0.675 0.185


6 Volvo V40 2.0 0.144 0.005 0.057 0.284 0.037 0.022 0.478 0.001

7 Peugeot 307 SW 1.6 0.059 0.011 0.029 0.035 0.042 0.032 0.137 0.011





12 Peugeot 306 1.9D 0.945 0.569 0.708 1.075 0.741 0.866 1.230 9.969



15 Volvo V70 D5 0.484 0.358 0.404 0.942 0.853 1.260 1.335 18.463





20 Volvo S60/V70 Bi-fuel 0.039 0.039 0.039 0.250 0.105 0.086 0.187 0.001


OSCAR EVK4-CT-2002-00083 85

Table B7: PM emissions – OSCAR cycles

Vehicle Emission factor by cycle (g/km) Vehicle category

Ref. no.

Model C D1 D2 E F G1 G2 H1 H2 H3



12 Peugeot 306 1.9D 0.025 0.020 0.037 0.030 0.030 0.033 0.032 0.029 0.038 0.047

13 Toyota Picnic 2.2TD 0.106 0.061 0.103 0.097 0.065 0.070 0.080 0.068 0.076 0.092


15 Volvo V70 D5 0.019 0.019 0.023 0.049 0.051 0.039 0.029 0.030 0.033 0.037




Table B8: PM emissions – Other cycles




CADC Urban Hot

CADC Road

CADC Motorway


Idle (g/h)



12 Peugeot 306 1.9D 0.029 0.021 0.024 0.033 0.038 0.076 0.061 0.170



15 Volvo V70 D5 0.052 0.054 0.053 0.049 0.023 0.042 0.104 0.197





OSCAR EVK4-CT-2002-00083 86

Table B9: CO2 emissions – OSCAR cycles Vehicle Emission factor by cycle (g/km)





3 Nissan Almera 1.4 183.7 185.6 181.2 211.2 238.2 259.6 265.8 318.9 317.2 345.8

4 Volkswagen Polo 1.4 156.3 165.2 190.5 197.0 224.4 237.9 257.5 316.8 327.4 327.1


6 Volvo V40 2.0 202.9 216.0 209.1 257.7 292.9 321.5 332.1 402.1 413.4 415.6

7 Peugeot 307 SW 1.6 178.2 189.4 219.1 234.2 274.1 299.7 327.5 392.3 390.9 426.7





12 Peugeot 306 1.9D 178.2 169.4 201.1 272.2 285.2 282.4 302.7 351.5 360.3 356.4

13 Toyota Picnic 2.2TD 213.3 206.1 258.2 270.6 275.5 280.4 308.1 343.6 368.3 378.7


15 Volvo V70 D5 198.2 205.5 236.0 275.3 322.3 317.1 361.9 382.4 417.1 426.4





20 Volvo S60 Bi-fuel 200.6 210.3 243.1 264.7 296.2 307.5 329.5 398.8 414.3 440.9

Table B10: CO2 emissions – Other cycles Vehicle Emission factor by cycle (g/km)



CADC Urban Hot

CADC Road

CADC Motorway

CADC Urban

Cold 9°C

Idle (g/h)



3 Nissan Almera 1.4 227.9 136.1 169.8 232.5 146.8 165.4 281.4 1316.3

4 Volkswagen Polo 1.4 207.6 120.2 152.5 222.2 133.9 143.6 244.3 1595.3


6 Volvo V40 2.0 304.1 162.8 215.1 303.2 179.2 180.8 363.8 1940.5

7 Peugeot 307 SW 1.6 223.0 138.2 169.7 282.1 144.4 170.8 317.0 1785.7





12 Peugeot 306 1.9D 242.6 151.6 185.3 265.9 181.8 188.8 316.0 1681.8



15 Volvo V70 D5 279.5 169.5 209.8 314.2 153.9 194.7 385.0 2018.5





20 Volvo S60 Bi-fuel 284.8 157.5 204.7 314.7 163.5 185.4 361.3 1943.1


OSCAR EVK4-CT-2002-00083 87

Appendix C: Comparison with type approval limits

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20


THC

+N

Ox

(g/k

m)o

verN

ED

Ccy

cle


Petro

lEur

o1

Petro

lEur

o2

Die

selE

uro

3

Petro

lEur

o3

Die

selE

uro

2

Die

selE

uro

1

Petro

lEur

o4

LPG

Figure C1: THC+NOx emissions over NEDC compared with limit values.

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.16

10 11 12 13 14 15 16 17 18


PM

(g/k

m)o

verN

EDC

cycl

e


Die

selE

uro

3

Die

selE

uro

2

Die

selE

uro

1

Figure C2: PM emissions over NEDC compared with limit values.


OSCAR EVK4-CT-2002-00083 88

Appendix D: Emissions over OSCAR cycles and UDC

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

Pet

rolE

uro

1

Pet

rolE

uro

2

Pet

rolE

uro

3

Pet

rolE

uro

4

Die

selE

uro

1

Die

selE

uro

2

Die

selE

uro

3

LPG

Eur

o3

CO

(g/k

m)

OSCAREuro 2 removedUDCEuro 2 removed

Figure D1: Average CO emissions over OSCAR cycles compared with emissions over UDC.

0.00

0.05

0.10

0.15

0.20

0.25

Pet

rolE

uro

1

Pet

rolE

uro

2

Pet

rolE

uro

3

Pet

rolE

uro

4

Die

selE

uro

1

Die

selE

uro

2

Die

selE

uro

3

LPG

Eur

o3

THC

(g/k

m)

OSCAREuro 2 removedUDCEuro 2 removed

Figure D2: Average THC emissions over OSCAR cycles compared with emissions over UDC.


OSCAR EVK4-CT-2002-00083 89

0

50

100

150

200

250

300

Pet

rolE

uro

1

Pet

rolE

uro

2

Pet

rolE

uro

3

Pet

rolE

uro

4

Die

selE

uro

1

Die

selE

uro

2

Die

selE

uro

3

LPG

Eur

o4

CO

2(g

/km

)

OSCAR

Euro 2 removed

UDC

Euro 2 removed

Figure D3: Average CO2 emissions over OSCAR cycles compared with emissions over UDC.


OSCAR EVK4-CT-2002-00083 90

Appendix E: Hot exhaust emissions by Euro class and cycle

Petrol

0

1

2

3

4

5

6

C D1

D2 E F

G1

G2

H1

H2

H3

UD

C

EU

DC

NE

DC

CA

DC

Urb

an

CA

DC

Roa

d

CA

DC

M'w

ay

Cycle

CO

(g/k

m)

Euro 1Euro 2Euro 3Euro 4

Figure E1: CO emissions over all cycles by Euro class – petrol vehicles.

Diesel

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

C D1

D2 E F

G1

G2

H1

H2

H3

UD

C

EU

DC

NE

DC

CA

DC

Urb

an

CA

DC

Roa

d

CA

DC

M'w

ay

Cycle

CO

(g/k

m)

Euro 1

Euro 2

Euro 3

Figure E2: CO emissions over all cycles by Euro class – diesel vehicles.


OSCAR EVK4-CT-2002-00083 91

LPG

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

C D1

D2 E F

G1

G2

H1

H2

H3

UD

C

EU

DC

NE

DC

CA

DC

Urb

an

CA

DC

Roa

d

CA

DC

M'w

ay

Cycle

CO

(g/k

m)

Euro 3

Figure E3: CO emissions over all cycles by Euro class – LPG vehicles.

Petrol

0.00

0.05

0.10

0.15

0.20

0.25

C D1

D2 E F

G1

G2

H1

H2

H3

UD

C

EU

DC

NE

DC

CA

DC

Urb

an

CA

DC

Roa

d

CA

DC

M'w

ay

Cycle

THC

(g/k

m)

Euro 1Euro 2Euro 3Euro 4

Figure E4: THC emissions over all cycles by Euro class – petrol vehicles.


OSCAR EVK4-CT-2002-00083 92

Diesel

0.00

0.05

0.10

0.15

0.20

0.25

C D1

D2 E F

G1

G2

H1

H2

H3

UD

C

EU

DC

NE

DC

CA

DC

Urb

an

CA

DC

Roa

d

CA

DC

M'w

ay

Cycle

THC

(g/k

m)

Euro 1

Euro 2

Euro 3

Figure E5: THC emissions over all cycles by Euro class – diesel vehicles.

LPG

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

C D1

D2 E F

G1

G2

H1

H2

H3

UD

C

EU

DC

NE

DC

CA

DC

Urb

an

CA

DC

Roa

d

CA

DC

M'w

ay

Cycle

THC

(g/k

m)

Euro 3

Figure E6: THC emissions over all cycles by Euro class – LPG vehicles.


OSCAR EVK4-CT-2002-00083 93

Petrol

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

C D1

D2 E F

G1

G2

H1

H2

H3

UD

C

EU

DC

NE

DC

CA

DC

Urb

an

CA

DC

Roa

d

CA

DC

M'w

ay

Cycle

NO

x(g

/km

)Euro 1Euro 2Euro 3Euro 4

Figure E7: NOx emissions over all cycles by Euro class – petrol vehicles.

Diesel

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

C D1

D2 E F

G1

G2

H1

H2

H3

UD

C

EU

DC

NE

DC

CA

DC

Urb

an

CA

DC

Roa

d

CA

DC

M'w

ay

Cycle

NO

x(g

/km

)

Euro 1

Euro 2

Euro 3

Figure E8: NOx emissions over all cycles by Euro class – diesel vehicles.


OSCAR EVK4-CT-2002-00083 94

LPG

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.16

C D1

D2 E F

G1

G2

H1

H2

H3

UD

C

EU

DC

NE

DC

CA

DC

Urb

an

CA

DC

Roa

d

CA

DC

M'w

ay

Cycle

NO

x(g

/km

)Euro 3

Figure E9: NOx emissions over all cycles by Euro class – LPG vehicles.

Diesel

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14

0.16

0.18

0.20

C D1

D2 E F

G1

G2

H1

H2

H3

UD

C

EU

DC

NE

DC

CA

DC

Urb

an

CA

DC

Roa

d

CA

DC

M'w

ay

Cycle

PM

(g/k

m)

Euro 1

Euro 2

Euro 3

Figure E10: PM emissions over all cycles by Euro class – diesel vehicles.


OSCAR EVK4-CT-2002-00083 95

Petrol

0

50

100

150

200

250

300

350

400

450

C D1

D2 E F

G1

G2

H1

H2

H3

UD

C

EU

DC

NE

DC

CA

DC

Urb

an

CA

DC

Roa

d

CA

DC

M'w

ay

Cycle

CO

2(g

/km

)Euro 1Euro 2Euro 3Euro 4

Figure E11: CO2 emissions over all cycles by Euro class – petrol vehicles.

Diesel

0

50

100

150

200

250

300

350

400

C D1

D2 E F

G1

G2

H1

H2

H3

UD

C

EU

DC

NE

DC

CA

DC

Urb

an

CA

DC

Roa

d

CA

DC

M'w

ay

Cycle

CO

2(g

/km

)

Euro 1

Euro 2

Euro 3

Figure E12: CO2 emissions over all cycles by Euro class – diesel vehicles.


OSCAR EVK4-CT-2002-00083 96

LPG

0

50

100

150

200

250

300

350

400

450

C D1

D2 E F

G1

G2

H1

H2

H3

UD

C

EU

DC

NE

DC

CA

DC

Urb

an

CA

DC

Roa

d

CA

DC

M'w

ay

Cycle

CO

2(g

/km

)

Euro 3

Figure E13: CO2 emissions over all cycles by Euro class – LPG vehicles.


OSCAR EVK4-CT-2002-00083 97

Appendix F: Cold-start emissions by Euro class and cycle

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

9.0P

etro

l,E

1(1)

Pet

rol,

E2(

1)

Pet

rol,

E2(

2)

Pet

rol,

E2(

3)

Pet

rol,

E3(

1)

Pet

rol,

E3(

2)

Pet

rol,

E4(

1)

Pet

rol,

E4(

2)

Pet

rol,

E4(

3)

Die

sel,

E1(

1)

Die

sel,

E2(

1)

Die

sel,

E2(

2)

Die

sel,

E2(

3)

Die

sel,

E3(

1)

Die

sel,

E3(

2)

Die

sel,

E3(

3)

Die

sel,

E3(

4)

Die

sel,

E3(

5)

LPG

,E3(

1)

LPG

,E3(

2)

CO

(g/k

m)

CADC urban hot start

CADC urban cold start (and at 9C)

Figure F1: CO emissions over cold-start and hot-start CADC.

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Pet

rol,

E1(

1)

Pet

rol,

E2(

1)

Pet

rol,

E2(

2)

Pet

rol,

E2(

3)

Pet

rol,

E3(

1)

Pet

rol,

E3(

2)

Pet

rol,

E4(

1)

Pet

rol,

E4(

2)

Pet

rol,

E4(

3)

Die

sel,

E1(

1)

Die

sel,

E2(

1)

Die

sel,

E2(

2)

Die

sel,

E2(

3)

Die

sel,

E3(

1)

Die

sel,

E3(

2)

Die

sel,

E3(

3)

Die

sel,

E3(

4)

Die

sel,

E3(

5)

LPG

,E3(

1)

LPG

,E3(

2)

THC

(g/k

m)


CADC urban cold start (and at 9C)

Figure F2: THC emissions over cold-start and hot-start CADC.


OSCAR EVK4-CT-2002-00083 98

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

Pet

rol,

E1(

1)

Pet

rol,

E2(

1)

Pet

rol,

E2(

2)

Pet

rol,

E2(

3)

Pet

rol,

E3(

1)

Pet

rol,

E3(

2)

Pet

rol,

E4(

1)

Pet

rol,

E4(

2)

Pet

rol,

E4(

3)

Die

sel,

E1(

1)

Die

sel,

E2(

1)

Die

sel,

E2(

2)

Die

sel,

E2(

3)

Die

sel,

E3(

1)

Die

sel,

E3(

2)

Die

sel,

E3(

3)

Die

sel,

E3(

4)

Die

sel,

E3(

5)

LPG

,E3(

1)

LPG

,E3(

2)

NO

x(g

/km

)CADC urban hot start

CADC urban cold start(and at 9C)

Figure F3: NOx emissions over cold-start and hot-start CADC.

0.00

0.05

0.10

0.15

0.20

0.25

0.30

Pet

rol,

E1(

1)

Pet

rol,

E2(

1)

Pet

rol,

E2(

2)

Pet

rol,

E2(

3)

Pet

rol,

E3(

1)

Pet

rol,

E3(

2)

Pet

rol,

E4(

1)

Pet

rol,

E4(

2)

Pet

rol,

E4(

3)

Die

sel,

E1(

1)

Die

sel,

E2(

1)

Die

sel,

E2(

2)

Die

sel,

E2(

3)

Die

sel,

E3(

1)

Die

sel,

E3(

2)

Die

sel,

E3(

3)

Die

sel,

E3(

4)

Die

sel,

E3(

5)

LPG

,E3(

1)

LPG

,E3(

2)

PM

(g/k

m)


CADC urban cold start(and at 9C)

Figure F4: PM emissions over cold-start and hot-start CADC.


OSCAR EVK4-CT-2002-00083 99

0

50

100

150

200

250

300

350

400

450

500

Pet

rol,

E1(

1)

Pet

rol,

E2(

1)

Pet

rol,

E2(

2)

Pet

rol,

E2(

3)

Pet

rol,

E3(

1)

Pet

rol,

E3(

2)

Pet

rol,

E4(

1)

Pet

rol,

E4(

2)

Pet

rol,

E4(

3)

Die

sel,

E1(

1)

Die

sel,

E2(

1)

Die

sel,

E2(

2)

Die

sel,

E2(

3)

Die

sel,

E3(

1)

Die

sel,

E3(

2)

Die

sel,

E3(

3)

Die

sel,

E3(

4)

Die

sel,

E3(

5)

LPG

,E3(

1)

LPG

,E3(

2)

CO

2(g

/km

)CADC urban hot start CADC urban cold start (and at 9C)

Figure F5: CO2 emissions over cold-start and hot-start CADC.

Road traffic characteristics, driving patterns and ... · Road traffic characteristics, driving patterns and emission factors for congested situations P G Boulter, T Barlow, I S McCrae

Documents