Scholarly article on topic 'A coupled road dust and surface moisture model to predict non-exhaust road traffic induced particle emissions (NORTRIP). Part 1: Road dust loading and suspension modelling'

A coupled road dust and surface moisture model to predict non-exhaust road traffic induced particle emissions (NORTRIP). Part 1: Road dust loading and suspension modelling Academic research paper on "Earth and related environmental sciences"

CC BY-NC-ND
0
0
Share paper
Academic journal
Atmospheric Environment
Keywords
{"Air quality" / "Particulate matter" / "Non-exhaust emissions" / "Road dust" / Suspension}

Abstract of research paper on Earth and related environmental sciences, author of scientific article — B.R. Denby, I. Sundvor, C. Johansson, L. Pirjola, M. Ketzel, et al.

Abstract Non-exhaust traffic induced emissions are a major source of particle mass in most European countries. This is particularly important in Nordic and Alpine countries where winter time road traction maintenance occurs, e.g. salting and sanding, and where studded tyres are used. In this paper, Part 1, the road dust sub-model of a coupled road dust and surface moisture model (NORTRIP) is described. The model provides a generalised process based formulation of the non-exhaust emissions, with emphasis on the contribution of road wear, suspension, surface dust loading and the effect of road surface moisture (retention of wear particles and suspended emissions). The model is intended for use as a tool for air quality managers to help study the impact of mitigation measures and policies. We present a description of the road dust sub-model and apply the model to two sites in Stockholm and Copenhagen where seven years of data with surface moisture measurements are available. For the site in Stockholm, where studded tyres are in use, the model predicts the PM10 concentrations very well with correlations (R 2) in the range of R 2 = 0.76–0.91 for daily mean PM10. The model also reproduces well the impact of a reduction in studded tyres at this site. For the site in Copenhagen the correlation is lower, in the range 0.44–0.51. The addition of salt is described in the model and at both sites this leads to improved correlations due to additional salt emissions. For future use of the model a number of model parameters, e.g. wear factors and suspension rates, still need to be refined. The effect of sanding on PM10 emissions is also presented but more information will be required before this can be confidently applied for management applications.

Academic research paper on topic "A coupled road dust and surface moisture model to predict non-exhaust road traffic induced particle emissions (NORTRIP). Part 1: Road dust loading and suspension modelling"

ELSEVIER

Contents lists available at SciVerse ScienceDirect

Atmospheric Environment

journal homepage: www.elsevier.com/locate/atmosenv

A coupled road dust and surface moisture model to predict non-exhaust road traffic induced particle emissions (NORTRIP). Part 1: Road dust loading and suspension modelling

B.R. Denby3,*, I. Sundvora, C. Johanssonb e, L. Pirjolac, M. Ketzeld, M. Normane, K. Kupiainenf, M. Gustafsson g, G. Blomqvistg, G. Omstedth

a The Norwegian Institute for Air Research (NILU), PO Box 100, 2027 Kjeller, Norway b Department of Applied Environmental Science (ITM), Stockholm University, Sweden c Helsinki Metropolia University of Applied Sciences, Finland d Department of Environmental Science, Aarhus University, Roskilde, Denmark e Environment and Health Protection Administration of the City of Stockholm, Sweden f Nordic Envicon Oy, Helsinki, Finland

g Swedish National Road and Transport Research Institute (VTI), Sweden h Swedish Meteorological and Hydrological Institute (SMHI), Norrkoping, Sweden

HIGHLIGHTS

• A coupled road dust and surface moisture model is presented and applied.

• The surface moisture strongly determines the temporal variation of the road dust emissions.

• The model predicts very well the temporal variation of suspended road dust emissions.

• Time scales for suspension are significantly longer than previous estimates.

• The model calculates the contribution of road salting to the emitted PM10 concentrations.

ARTICLE INFO

ABSTRACT

Article history: Received 22 January 2013 Received in revised form 16 April 2013 Accepted 26 April 2013

Keywords: Air quality Particulate matter Non-exhaust emissions Road dust Suspension

Non-exhaust traffic induced emissions are a major source of particle mass in most European countries. This is particularly important in Nordic and Alpine countries where winter time road traction maintenance occurs, e.g. salting and sanding, and where studded tyres are used. In this paper, Part 1, the road dust sub-model of a coupled road dust and surface moisture model (NORTRIP) is described. The model provides a generalised process based formulation of the non-exhaust emissions, with emphasis on the contribution of road wear, suspension, surface dust loading and the effect of road surface moisture (retention of wear particles and suspended emissions). The model is intended for use as a tool for air quality managers to help study the impact of mitigation measures and policies. We present a description of the road dust sub-model and apply the model to two sites in Stockholm and Copenhagen where seven years of data with surface moisture measurements are available. For the site in Stockholm, where studded tyres are in use, the model predicts the PM10 concentrations very well with correlations (R2) in the range of R2 = 0.76-0.91 for daily mean PM10. The model also reproduces well the impact of a reduction in studded tyres at this site. For the site in Copenhagen the correlation is lower, in the range 0.44-0.51. The addition of salt is described in the model and at both sites this leads to improved correlations due to additional salt emissions. For future use of the model a number of model parameters, e.g. wear factors and suspension rates, still need to be refined. The effect of sanding on PM10 emissions is also presented but more information will be required before this can be confidently applied for management applications.

© 2013 Elsevier Ltd. All rights reserved.

* Corresponding author. Tel.: +47 63898164; fax: +47 63898050. E-mail address: bruce.denby@nilu.no (B.R. Denby).

1352-2310/$ - see front matter © 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.atmosenv.2013.04.069

1. Introduction

PM10 (airborne particle matter with diameter < 10 mm) concentrations exceed the European Union (EU) limit values (EC, 2008) in almost all countries in Europe. Up to 21% of the European urban population is currently exposed to PM10 concentrations in excess of the EU daily mean air quality limit value (less than 36 days > 50 mg m~3), and there is little or no downward trend in most cities (EEA, 2012). Sources and processes leading to PM10 are varied but one of the major sources, particularly in some urban environments, is non-exhaust particle emissions from traffic. Unlike exhaust emissions, that are regulated by legislation (EC, 2005), non-exhaust emissions are non-regulated.

In many Nordic cities non-exhaust particle emissions are the main reason for high PM10 levels that can lead to exceedances of these limit values (Omstedt et al., 2011). This is connected to the use of studded tyres and winter time road traction maintenance, e.g. salting and sanding. Importantly for cities where this occurs, the non-exhaust contributions to PM10 associated with traction sanding and salting may be negated when reporting exceedances of limit values in the EU (EC, 2008).

City authorities are required to improve air quality in order to meet the EU standards. Air quality models are regularly applied to assess future scenarios and help with planning these mitigation strategies. For the case of non-exhaust emissions there is a need for a modelling tool that can adequately represent the processes associated with non-exhaust emissions and that can provide the possibility to assess abatement strategies such as speed reduction, studded tyre reduction, street cleaning, dust binding and changes in road maintenance procedures.

Non-exhaust traffic emissions can come from a variety of sources. The largest known emission source on paved roads is road wear through the use of studded tyres, but other sources include brake and tyre wear. The suspension of particles on the road surface (dust loading) that originates from wear, salting, sanding and other fugitive sources can also contribute significantly (e.g. Mathissen et al., 2012; Bukowiecki et al., 2010), particularly in countries where studded tyres are used (e.g. Gustafsson et al., 2008; Norman and Johansson, 2006). An overview and comparison of non-exhaust emissions can be found in Boulter (2005) and Boulter et al. (2006).

Most emission models, used for air quality applications, treat the non-exhaust emissions as emission factors (e.g. EMEP/EEA, 2009; Ketzel et al., 2007). Some may include the impact of surface wetness, usually by some parameterisation related to atmospheric humidity and/or precipitation, e.g. the US EPA model AP-42 (US EPA, 2006) or (Pay et al., 2011), in order to account for the retention of particles on the surface under wet conditions. However, these models do not account for the build up of dust on the road surface and cannot be used to assess mitigation strategies to any effective degree. One model that includes both dust loading and surface moisture as predictive variables is the model from Omstedt et al. (2005). This model has been successfully applied to a range of data (Omstedt et al., 2011) but relies on local measurements to calibrate the model at individual sites. In addition it does not explicitly include process descriptions related to studded tyre wear, vehicle speed, salting, etc. Berger and Denby (2011) presented a more generalised approach to road dust modelling by basing emissions on total wear rates with the intention of reducing the dependency of the model on local measurements. Even so, that model had a simplified approach to surface moisture and did not include the impact of salting or sanding.

This paper, along with a subsequent paper (Denby et al., 2013) referred to as Part 2, presents the newly developed NORTRIP emission model (NOn-exhaust Road Traffic Induced Particle emissions) that has been developed during the NORTRIP project, a cooperative project between the Nordic countries of Norway,

Sweden, Finland and Denmark (Johansson et al., 2012). The model builds upon the work previously carried out by Berger and Denby (2011) and Omstedt et al. (2005), where the concept of surface mass balance for dust and moisture was first developed, and is comprehensively documented in Denby and Sundvor (2012).

The model developed consists of two main parts. These are:

1. Road dust sub-model: This predicts the road dust, sand and salt loading through a mass balance approach and determines the emissions through suspension of these loadings as well as through direct wear of road, tyre and brake sources.

2. Road surface moisture sub-model: This determines road surface moisture essential for the prediction of suspension and the retention of dust and salt from the road surface. A surface mass balance approach is also applied, coupled to an energy balance model to predict evaporation/condensation.

In this paper (Part 1) we provide a description of the road dust sub-model and apply the model to a number of datasets where surface moisture measurements are available. As such we concentrate in this study on the road dust part of the model in order to remove the uncertainty associated with surface moisture modelling. In Part 2 the road surface moisture sub-model is described and the entire model is applied to a larger number of datasets, not limited to roads where surface moisture measurements are available.

This paper is presented in the following form: In Section 2 the road dust processes are conceptually outlined and in Section 3 the road dust sub-model is described mathematically. In Section 4 the application sites are briefly described and results of the modelling are presented. Sensitivity analysis is carried out and some of the most important aspects and results of the modelling are outlined. Section 5 provides a discussion and conclusion.

2. Model concept

One of the most fundamental problems with road dust emission modelling is the complexity and variety of processes. This problem is enhanced by a general lack of monitoring or experimental data to support process descriptions (e.g. dust loading measurements) as well as a lack of input data suitable for representing the processes (e.g. road wear rates, salting and sanding application rates). When measurements of dust loading are made, e.g. Amato et al. (2009) who reported measurements of PM10 mass loading, they are often not related to any traffic or concentration data and cannot be used to help understand the processes involved.

There are some basic elements to the model that need to be described if it is to fulfil its aim as a process based modelling tool to aid air quality planning. These include:

1. Determination of vehicle related wear rates (road, tyre and brake) and their direct emission or accumulation on the road surface. Emissions are dependent firstly on wear and then secondly on size distribution of that wear. These may be vehicle type, tyre type, speed and wear source dependent.

2. A description of emissions through traffic induced suspension of road dust, sand and salt loading. The suspension will depend on tyre types, vehicle speeds, road surface characteristics and most importantly road surface moisture.

3. Calculation of the road dust and salt loading, dependent on the road dust and salt mass balance, is required to estimate suspension rates.

4. Retention of the wear particles and suspended emissions based on road surface conditions, requiring a description of surface wetness. If no observations are available then surface moisture modelling is required using a mass and energy balance

Fig. 1. Schematic outline of the NORTRIP emission model.

approach that includes processes such as evaporation, drainage, vehicle spray, etc.

The key elements of the modelling system, as described in Denby and Sundvor (2012), are presented schematically in Fig. 1.

Within the road dust sub-model a range of processes, as discussed above, are described. Unfortunately not all processes are well enough known, or have sufficient input data available, for them to be implemented in all modelling applications. But since the model is also intended as a conceptual tool for further understanding we also include processes that need further attention. In Table 1 we list the entire range of processes that are included in the complete model

Table 1

Processes included in the NORTRIP emission model as described in Denby and Sundvor (2012). Not all processes are included in the applications in this paper (Part 1) and the subsequent paper (Part 2). Their inclusion in the applications is indicated by a cross.

Process

Applied in part 1

Applied in part 2

Mass balance equation for suspendable x

dust, sand and salt loadings Mass balance equation -

for non-suspendable dust (sand) loadings Addition of salt through x

road maintenance activities Road, tyre and brake total wear and related x

PM size fractions Retention of wear particles on the road surface x

due to surface moisture Removal of the dust loading through traffic x

induced suspension Drainage of the salt loading x

Vehicle spraying of the dust and salt load -

Removal of the dust and salt loading through -

cleaning and snow ploughing Abrasion of the road surface through -

sand (sand paper effect) Crushing of sand into suspendable particles -

Windblown suspension -

Accumulation of dust through atmospheric deposition -Salting and sanding road maintenance -

activity modelling Direct emissions from wear sources x

Suspended emissions from dust and salt loading x

(Denby and Sundvor, 2012) and indicate which of these processes are described in the current paper, Part 1, and the subsequent paper regarding surface moisture, Part 2. Some processes, e.g. windblown suspension and atmospheric deposition, are not included in the current paper as they are of relatively little importance for the current study. Other processes, such as crushing and abrasion due to sand, are too uncertain to be usefully quantified.

3. Road dust sub-model description

In this section the model formulation of the road dust submodel is described. Only brief discussions concerning parameter estimation are included and the reader is generally referred to Denby and Sundvor (2012) for a more detailed description. Only processes applied in this paper are included here.

3.1. Mass balance for dust and salt

The road surface dust mass loading may be separated into different size fractions but only a suspendable size fraction, (<~200 mm), often referred to as silt but termed dust(sus) in this paper, is represented in this modelling study. The delineation at around 200 mm is intended to reflect the size distribution of road wear particles, which have been shown to have sizes mostly <200 mm, peaking at around 20 mm (Snilsberg et al., 2008). Salt is also included in the model as a mass type and it can be divided into different salt types, e.g. sodium or magnesium based salts or acetates. Currently only MgCl2 and NaCl are described in the model.

The different mass types are indexed with with m, where m refers to dust(sus), salt(na) or salt(mg). Contributions from applied sand, from different wear sources and from different salting types are tracked individually within the model. The road mass balance equation for the surface mass Mroad, for different mass types m is written as

VMroad dt

pm _ cm ' road croad

where Proad and Sroad represent the production and sink terms respectively. The model calculates Mroad in g km-1 but this is

usually presented as g m where the road width is used for the conversion.

Road dust production (Pr0ad) is the sum of a number of sources. In the current application this includes only the retention of wear particles on the road surface (Pretention) and the contribution from sanding in the suspendable size fraction, where the term /ssaundjng indicates the suspendable fraction of the total traction sanding mass production (Psanding).

dust(sus) road

_ Pretention + Psanding '/sanding

There is only one source of salt production in the model, that being the addition of salt related to defreezing or dust binding activities.

The removal processes (sinks) are similar for both dust and salt, and are assumed to be proportional to the available mass. We can calculate the sinks (Sprocess) based on appropriate rates (Rprocess) for each process and apply these to all dust or salt masses individually as follows

cm _ iim n

cprocess — Mroad'Rprocess

The sink terms included here are suspension (^suspension) and drainage (Sdrainage). We write the road dust and salt sink terms, indexed with m, for the various processes as follows:

cm _ cm I cm

road — suspension ^ drainage

3.2. Road dust production through wear

A proportion of the dust produced through wear is not removed from the system but is retained on the surface (Pretention), Eqs. (5) and (6). The wear rate per vehicle W|ource (g km veh-1) is specified for each wear source (source = road wear, brake wear or tyre wear), for the different tyre types (t = studded, non-studded winter and summer) and vehicle types (v = light and heavy duty). Under dry conditions the fraction of wear that is immediately suspended, on time scales of an hour or less, is specified by the factor/o,dir-source. Under moist conditions this is modified by the surface wetness retainment factor (fq,source). This last term is dependent on the surface moisture (fq,source = 1 indicates a dry surface) and may be different for road/tyre wear sources than for brake wear, since the latter is not in direct contact with the road surface. The production of dust loading from the different wear sources is then specified as

retention-source

tyre vehicle

Pretention-source ^^ ^ ] N ' ' Wsource

t = st'Wi'su v = he;li

' — /ö'dir-source './¿¡'sourceJ

where Nt,v is the number of vehicles for a specified tyre (t) and vehicle type (v).

Wear rates for specified vehicle and tyre types will depend on a number of factors, e.g. pavement type and vehicle speed. The functional dependency of the road wear rate per vehicle (Wrto"adwear) is given as:

roadwear

0,roadwear "Pave

Vref ,roadwear

The basic road wear parameter (W0'Toadwear), which is user specified for the different vehicle categories (v) and tyre types (t), is valid at the reference speed Vref, roadwear. The power law dependency on vehicle speed (Weh) is taken to be linear (awear = 1) since the

linear speed dependence has been confirmed in a range of laboratory experiments (e.g. Gustafsson et al., 2008; Snilsberg et al., 2008).

For the studded tyre road wear rate (W0t;V)adwear), use is made of the Swedish road wear model (Jacobson and Wagberg, 2007) which takes input data concerning maximum stone size (MS), Nordic ball mill (NBM) value for hardness and percentage of stone size > 4 mm (S>4mm) and determines the basic road wear parameter (W0road-wear) and the pavement type factor (hpave) used in Eq. (7). The Swedish road wear model uses a reference pavement type (ABS16 with porphyry from Alvdalen) with a wear rate of 2.88 g km-1 veh-1 at the reference speed of 70 km h-1. We thus set the reference wear parameter for studded light duty vehicles to W0roadwear = 288 g km-1 veh-1. The road surface parameters used to calculate the pavement type factor for a particular pavement indexed with p is given as:

hpave = 2.49 + 0.144NBMp - 0.069 MSp - 0.017-S>4mm (8)

Swedish roads often use MS = 16 mm with a range of NBM from 5 to 15, for Norwegian roads smaller stone sizes are favoured with similar ball mill values (e.g. MS = 11 mm and NBM = 6) (Snilsberg et al., 2008). In countries where studded tyres are not employed, e.g. Denmark, then even smaller stone sizes may be used. The percentage of stones >4 mm is taken to be 75% in most cases. Typical wear rates are thus 2-5 g km-1 veh-1 but these can vary significantly dependent on the material used, with increasing wear rates for smaller stone sizes. Unfortunately knowledge of these road surface parameters is often lacking and so there can be significant uncertainty in the final wear rate.

Nominal values for the non-studded winter and summer tyre road wear are based on road simulation measurements (Snilsberg et al., 2008; Gustafsson et al., 2008) that indicate that non-studded tyre road wear is around 20-30 times less than studded tyre wear. However, under real world conditions urban roads are rarely clean and smaller stones and gravel may enhance wear from non-studded tyres.

Tyre wear follows a very similar description to the road wear but is not considered to be dependent on the pavement type.

tyrewear

0;tyrewear

( VVeh Y I Vref ;tyrewear

Brake wear is not considered to be dependent on tyre type or on the vehicle speed. It is better determined by braking activity than by vehicle speed, though there may be some relationship between these two. We use a general 'driving cycle' factor that can alter the

basis brake wear parameters. hd

drivingcycle

= 1 is valid for the basis

brake wear parameter provided. Driving cycle type may include highway, urban, congested, etc. and these are represented by the given 'driving cycle' factor using the index d.

Wv Wv ' hd

brakewear — 0,brakewear' drivingcycle

Estimates for the tyre and brake wear parameters in Eqs. (9) and (10) are taken from Boulter (2005).

Heavy duty vehicles (HDV) are expected to contribute significantly more to road wear than light duty vehicles (LDV). In the literature the enhancement of road wear by HDV can be from 5 to 100 (Gehrig et al., 2004; Boulter, 2005), though in many studies it is often difficult to distinguish between suspended and direct road wear. We enhance all LDV wear rates (v = li) by a factor of 5 to specify HDV wear rates (v = he). It is also worth noting that the percentage of studded tyres on HDV is often quite low, or none at all, and so the studded tyre contribution from HDV's may not be significant. The final model parameters are listed in Table A.2.

ï 07 ■ s

Hornsgatan 2006-2007 -Homsgatan 2007-2008 -Hornsgatan 2008-2009 Hornsgatan 2009-2010 -Hornsgatan 2010-2011

0.1 1 10 100 0.1 Suspension rate x 10"6 (veh'1)

Fig. 2. Results of the model sensitivity analysis on Hornsgatan data to determine the basis suspension rate (f0,s right: mean concentrations. The period covers October to June in all cases.

B.R. Denby et al. / Atmospheric Environment 77 (2013) 283-300 35 -I 30 -

E 25 É

15 ■ 10 ■ 5 ■ 0

Hornsgatan 2006-2007 Hornsgatan 2007-2008 Hornsgatan 2008-2009 Hornsgatan 2009-2010 Hornsgatan 2010-2011

1 10 100 Suspension rate x 10"6 (veh'1)

ion). Left: correlation for the daily mean PM10 concentrations and

Fig. 3. Results of the model application at Hornsgatan for the period October 2008 to June 2009. Top: Daily mean modelled and observed concentrations of PM10. Middle: Surface mass loading of suspendable dust. Bottom: Effective emission factor calculated by the model and derived from observations.

3.3. Road dust production through the application of sand

The contribution through sanding (Psanding) is given by the total mass of traction sand (Msanding) distributed on the road at a particular hour (tsanding). The contribution of mass through sanding is distributed over the time step of the model (At = 1 h). The amount of sus-pendable sand (<200 mm) is specified by the factorfading. Units for sanding are generally provided as g m-2 and the conversion factor to g km-1 is included in Eq. (11), assuming all of the sand arrives on the road surface, lane width (blane) and number of lanes (nlanes).

sanding

Msanding tsanding

■1000-nlanesbl a

Some knowledge of the size distribution of the traction sand is thus required. A study carried out on Swedish traction sand, see Denby and Sundvor (2012), indicates that 6% of the sand mass is <200 mm in diameter and that 0.8% of the sand mass is <10 mm (i.e. 16% of the suspendable fraction).

3.4. Road salt production

Winter time salting activities are described by an addition of mass (MiaJtiing), either as dry salt or in solution, for the salt type i. As

with sanding the instantaneous mass increase is spread out over the hour based on the timing, tsalting. Units for salting are provided as g m-2 and the conversion factor to g km-1, assuming all of the salt arrives on the road surface, is included in Eq. (12).

psalt(i) _ Msalting Vbluing

1000 nlanes blane

3.5. Traffic induced suspension

The reduction of road dust and salt loading through suspension is given by:

Ssuspension

suspension

Mm $ Rm

road$ suspension

vehicle

y V Rm't'v .

suspension

Nt'v /

Rm't'V _ um ct'V ict'V yi

suspension nlanes 0'suspension Jsuspensio^j0'suspension^ veh

q;suspension

(groad)

Fig. 4. As in Fig. 3 but for the period October 2009 to June 2010.

Fig. 5. As in Fig. 3 but for HCAB in the period October 2007 to March 2008.

35 ■

30 ■

20 ■

10 ■

■ Observed □ Modelled

29.1 28.7

12.3 12.1

^ is ° st

^ is ° St o S) id S

^ is ° st

Data set

w is o St

Fig. 6. Net mean concentration of observed and modelled PMio for the 7 datasets. For Hornsgatan the period modelled is October—June and for HCAB the period is October—May.

g 80 à

— 70

S 30 E

= 20 ra

■ Observed □ Modelled

S I ¿"2 P s>

Data set

9 I ti 'S

21.8 21.9

Fig. 7. Net daily mean 90'th percentile concentration of observed and modelled PM10 for the 7 datasets. For Hornsgatan the period modelled is October—June and for HCAB the period is October-May.

An important term in Eq. (15) is the suspension factor (fstuspension) which defines the fraction of mass that is removed for the passage of each vehicle. This is dependent on the vehicle speed

) that can

■t,v

suspension

and on a basic user defined suspension factor (f0t;; be specified by vehicle (v) and tyre (t) type, Eq. (16).The suspension rate for salt and sand suspension is taken to be the same as for the suspendable dust though it is possible to specify this separately in the model using the scaling factor (hmsuspension). The suspension rate is given with a power law (asus) dependence on vehicle speed.

J suspension

ft,v /0;

suspension

( Vveh Y" Vref;sus

The processes affecting the suspension are not just the turbulent and mechanical suspension from the road surface, though this is the mechanism by which suspension will finally occur. It is also governed by the availability of the dust, e.g. from road or the shoulder/pavement sources, on the cohesive forces on the dust, e.g.

salt and hygroscopic properties of dust, and also on the surface structure. Experiments with a suspension simulator (Blomqvist et al., 2011) have shown a strong dependence of turbulence induced suspension on the surface texture (macro-structure).

The vehicle speed dependence of suspended particles has been assessed previously using both tower measurements (e.g. Etyemezian et al., 2003) and mobile measurements (e.g. Pirjola et al., 2009). In most cases there appears to be a near linear dependence of the suspension emission factor on vehicle speed. Some experiments have been carried out to determine the suspension rates of dust deliberately distributed on the road surface (e.g. Langston et al., 2008). These indicate that applied dust is quickly removed from the surface, with suspension rates of the order of 10-2-10-3 veh-1. i.e. an e-folding rate of 100-1000 vehicles. Patra et al. (2008) estimated this rate to be 3 x 10-4 veh-1 based on distribution of rock salt on a road in London. Kupiainen and Pirjola (2011) found that traction sanding, added to the surface under dry conditions, increased the suspended emissions by a

SP £, 0.8 ai

0 0.7 «

g 0.6 -

8 0.5 -c n

1 0.4 H

f o.3 H ra u

"■J =>

'S o S.

■ With retention and suspension

□ With retention and no suspension

□ No retention and no suspension

Ï3 2 " o 3

Data set

S®, i-i

Nï > O

Fig. 8. Correlation (R2) of modelled versus observed net daily mean PM10 for the 7 datasets. For Hornsgatan the period modelled is October—June and for HCAB the period is October-May. Shown are the results including or excluding retention and suspension in the model calculations.

35 E 30

■e C

€ 20

■ With retention and suspension

□ With retention and no suspension

□ No retention and no suspension

— 15

° I ¿"2 P ET

is => S K S)

Data set

KJ 00 O »

Fig. 9. Net mean concentration of observed and modelled PM10 for the 7 datasets. For Hornsgatan the period modelled is October—June and for HCAB the period is October—May. Shown are the results including or excluding retention and suspension in the model calculations.

factor of 15 but that the PM10 emissions reduced quickly, over a matter of hours. These high suspension rates result in the suspension of dust on trafficked roads on time scales of an hour or less and these suspended emissions are represented by the 'direct'

be continuously replenished by rain water, drained at the same rate. If this is the case, and the dust and salt is continuously mixed with a given efficiency (hdmrain eff ) in the surface water, then the total sink of mass in the drainage water is given by

Mrmoad f f ,m (groad groad,drainable-mm)N

S3r"n* ~T- 1 - exp - hmr«f -V gr„.d.dr.in.b,.-„.n ' (17)

emissions presented in Section 3.8. Such rates are not commensurate with the suspension seen during and after the studded tyre season where the dust loading and suspension extend well beyond the studded tyre season. It is likely that suspension of freshly laid dust/sand and the retained dust from wear sources during wet periods have a different adhesive quality and distribution, and hence suspension rate. We use model sensitivity analysis to determine suspension rates, Section 4.1. An optimal value of f0 suspension = 5 x 10-6 veh-1 is determined. HDV suspension rates are prescribed as being a factor of 10 higher than LDV rates.

3.6. Drainage

The removal of dust and salt by drainage is related directly to the amount of surface water that is drained from the road groad, drainable. This water will carry with it both dust and salt. The removal of dust and salt requires knowledge of the level of mixing in the drainage water. Salt should be fairly well mixed since it will be in solution. Suspendable dust may not be well mixed and the efficiency of removal for non-suspendable dust by drainage may be very poor (Vaze and Chiew, 2002). To reflect this, a drainage efficiency parameter is used (hmrain_eff) which can range from 1, for the well mixed situation, to 0, when no dust or salt will be removed through drainage.

In Part 2 of this study, that describes the surface moisture drainage process, the surface water layer (groad) that results from precipitation is removed instantaneously to a minimum drainable level (groad ,drainable-min) since the time scale for drainage is much less than the model time step of 1 h. This means that the surface reservoir of non-drainable water (groad,drainable-min) is considered to

where groad is the depth of the water layer over the time step Dt caused by the precipitation (mm h-1).

3.7. Moisture retention factor

The retention factor, which limits emissions and leads to retention of wear particles, is based on road wetness conditions and provides an essential coupling between the road dust and road moisture sub-models. When surface moisture modelling or measurements are available that provide surface moisture levels in mm then the retention factor fq,source is described using a discontinuous linear dependence

fq .source = max(0 ,min[1,1 - gratio .source (18)

(groad - gretention-min .source) gratio.source = "T r

1 gretention-thresh .source - gretention-min . source J

and source is any one of the various emission sources (i.e. direct road, direct brake, suspended road). A threshold value (gretention-thresh.source) defines the upper limit for retention for which fq source = 0 (wet surface, full retention). A minimum retention value (gretention-min,source) defines the minimum surface moisture that inhibits suspension. Below this value fq,source = 1 (dry surface, no retention).

When surface moisture measurements based on conductivity are used then the surface retention parameter is defined in a binary way. i.e. when conductivity is above a certain threshold then the surface is considered moist. This will vary from instrument to instrument.

3.8. Emissions and size distributions

The total emission (Ex) from wear and surface suspension sources for a particular PM size fraction (x), e.g. x = 10 mm, is given by

ex _ ex , e E = Edirect + E

suspension

The direct emissions, from the wear sources, can be written in the following way:

direct

dir-source

dir-source

tyre vehicle

E E Nt 'v'Wo

t = st,wi , su v = he ,li ' fq,source '/pm,dir-source

source

dir-source

The factor fPxM dir source indicates the proportion of the PM in any particular size fraction x and is dependent on the wear source.

The emissions through suspension follow the same form as the sink term for traffic induced suspension, Eqs. (13)—(15), and are the sum of all the suspendable mass types (sus).

suspension

vehicle

vehicleR

t = st,wi , su v = he ,li

m , t, v /x

suspension 'f PM , sus-road

The fraction of both the direct and suspended road wear particles in any Particular size Category fpM dir-roadwear and fpM sus-road) wi" likely be dependent on factors such as pavement type, tyre type and on vehicle speed. Laboratory experiments indicate an increase in the PM10 size fraction with increasing speeds (Snilsberg et al., 2008). We consider the direct road wear and the suspended emission size fraction to be linearly dependent on vehicle speed such that

PM , road = fPM ,ref,road '

0 + CP

PM-fraction

t1 + c

PM fraction

PM fraction

where the subscript road refers to both the direct road wear (dir-roadwear) and the suspended (sus-road) emission size fractions. The coefficient c

PM-fraction

which is the slope of the speed dependence, is derived from data collected in the laboratory by Snilsberg et al. (2008). The reference PMx fraction (fXM ref road) at the reference speed (Vref PM-fraction) has been estimated for a range of tyre types by Snilsberg et al. (2008) to be from 0.10 to 0.29 for PM10, based on sampled road wear particles from behind tyres in laboratory experiments. Estimates of this parameter have also been reported by Gustafsson and Johansson (2012) in the laboratory, based on PM10 emission factors estimated during ventilation in combination with laser scanning of the road surface to determine total wear. This method yields an estimate of around 0.08 for the PM10 fraction of total wear. This parameter is uncertain but may be further constrained by comparison of model with observations, which has been done for a large number of datasets in Denby and Sundvor

(N ™ AJ E

^ -a o '—) ^ aj

^.¡yês S IS

M Id '

LO "t © ©

LO PI © ©

TT TT M oS

00 r^ m Ln

I-v 00 © ©

r^ 00 C^ © '—CNCN o © © ^ II

c© CO CO CO cO(n|s f^CN fl (N nCN nCN (^CNqq

"" ' "" I ro .L « -L "" '

C© C©C© C© C©£S£S

ir<N ¡-(Ni- M ¡-CN irCN<< O O O O O UU

I ©©

J 140 %

£ 120

M f 100

C 80 £

□ Exhaust

□ Suspended ■ Direct

M X o o M X o o M X o o M X o o M X o o

? 3 g® o s S 3 1 w "ÏH ° S o S 00 S M 3 1 V w "ÏH ° S o s 10 S ? 3 g« K » O 3 Data set P 3 g« P s

Fig. 10. Mean emission factors for the direct, suspended and exhaust emissions as calculated with the model for the 7 datasets. No salt included. For Hornsgatan the period modelled is October—June and for HCAB the period is October—May.

(2012). We apply a value off™ foad estimated uncertainty of ±30%.

= 0.18 in the model, with an

3.9. Conversion of emissions to concentrations

When comparing model results to observed concentrations it is necessary to convert emissions to concentrations. As in many such studies, e.g. Omstedt et al. (2005), this is carried out with the help of observed NOx concentrations and calculated NOx emissions. This provides a conversion factor that converts emissions to concentrations. In a similar, but inverse, way this conversion factor can also be used to determine PM10 emission factors from observed concentrations. The conversion factor (/con) for converting emissions to concentrations is given by

[NOXraffic] - [No£ackground] NO£mission

Using this method to determine the dispersion is considered to be more accurate than using a dispersion model but there can still be significant uncertainty in the NOx emissions. Comparisons with 'real world' NOx emission studies, e.g. Carslaw et al. (2011), indicate that for some vehicle classes uncertainty can be up to 30%. Uncertainties due to other factors such as the choice of driving cycle, traffic counts, vehicle speeds, vehicle types, road slope and cold starts will also contribute. For the datasets used in this study it has been estimated that the NOx emission factors have an uncertainty in the range of -20% to +30%.

3.10. Numerical implementation

Most of the model is a direct application of the equations listed in Section 3. For the time integration, which is carried out in steps of 1 h, a short term solution to the mass equations is introduced to ensure stability. For a generalised sink (S = RM) term and production (P) term the mass (M) balance equation can be written as

where the superscript indicates the two measurement sites, traffic and background, needed to determine this conversion.

VM = P - R.M

■ Salt

□ Exhaust

□ Brake wear

□ Tyre wear

□ Road wear

s'a o sr

M X o o s? X n r>

g 3 1 </> 93 N. > o OJ M > o &

SJ o? NJ TO O Ni O |SJ

O QJ O % >sl O oo o

K OJ O 3 P § en

Data set

Fig. 11. Mean contribution to the concentrations, as calculated by the model, for the emission sources of road wear, tyre wear, brake wear and exhaust for the 7 datasets. Salt is included in only 3 of these (Section 4.4). For Hornsgatan the period modelled is October—June and for HCAB the period is October—May.

This has an analytical solution when P and R are constant during the time integration from t0 to t0 + At given by

M(t0 + At) = R + (W0) - R) e-R At for Rs 0 = M(t0)+P$At for R = 0

4. Model application

In Denby and Sundvor (2012) the NORTRIP model was applied to 14 different datasets and a number of the model parameters were derived based on these. In this paper (Part 1) we apply the model only to datasets where surface moisture measurements are available in order to concentrate on the suspension part of the model. Two sites are suitable, one at Hornsgatan, a street canyon site in Stockholm, where five years of continuous measurements (2006— 2011) are available and one at H.C. Andersen Boulevard (HCAB), an open street canyon in Copenhagen, where two years of measurement data are available (2006—2008). In common to both sites are kerbside and roof top measurements of PM10 and NOx, the

availability of meteorological data, surface moisture measurements and, for some years, information concerning salting and sanding activities. Both streets have similar vehicle speeds (44 km h-1). The major difference between the two sites is that studded tyres are used in Stockholm but not in Copenhagen. For Hornsgatan road surface parameters to calculate wear are known (Eq. (8)) and give a value of hpave = 0.83. For HCAB no information is available concerning the road surface parameters and a value of hpave = 4.0 is applied. This value is intended to provide average concentrations similar to those observed. However, the high pavement wear rate has been analysed previously, Wahlin et al. (2006), and was attributed to the poor quality of the HCAB road surface, that is made of steel slag.

4.1. Sensitivity to suspension rates

Before providing model results we present one of several sensitivity analyses that have been carried out with the model to help confine model parameters. The suspension rate applied in the simulations has been derived using a sensitivity analysis on the Hornsgatan dataset. This dataset, with very high temporal

Fig. 12. Results of the model application at Hornsgatan for the period October 2010 to June 2011. Top: Daily mean modelled and observed concentrations of PM10 including the salt contribution. Middle: Surface mass loading of suspendable dust and salt. Bottom: Total mass of salt applied to the road per day.

correlation, see Section 4.2, allows an optimal suspension rate to be derived by maximising the temporal correlation. Results of the sensitivity assessment are shown in Fig. 2. The optimal rates lie between —8 10-6 veh-1. A value of f0, suspension = 5 x 10-6veh-1 has been applied to all model runs in this paper for all tyre types. The mean concentrations are not sensitive to variation in the suspension rates when these are larger than 2 x 10-6 veh-1, with a variation <8% over the optimal range indicated.

4.2. Model results

In Figs. —5 we present three examples, two from Hornsgatan and one from HCAB. In these figures the daily mean modelled PM10 and observed net concentrations (traffic — background) are shown, along with the mass loading on the road surface and the effective emission factor calculated by the model and derived from observations. For Hornsgatan the period modelled starts in October and ends in June, which is intended to cover the studded tyre season from November to April. For HCAB the period presented is determined by the availability of the observational data, from October to March.

For Hornsgatan we see very highly correlated modelled concentrations that are the result of both immediate and retained suspension. The retained suspension contribution is the result of the available mass loading on the surface which reaches a maximum of around 120 g m-2 at Hornsgatan. In 2008—2009 (Fig. 4) a long wet period occurs from mid December to mid March, where the mass loading builds up continuously. Maximum emissions occur at the end of this wet period.

For HCAB the modelled concentrations do not reflect the observations as well. Suspension is much less at this site with a maximum surface mass loading of 18 g m-2. There are no extensive wet periods in which the mass loading builds up and wear rates are less than in Hornsgatan during the studded tyre season.

In Figs. 6 and 7 we present observed and modelled net annual mean PM10 concentrations as well as the net daily mean 90'th percentiles for all the datasets. In Figs. 8 and 9 the impact of surface retention and suspension on the correlation and net mean concentrations is shown. The results presented show model runs with 1) both surface retention and suspension, as in Figs. 3 —7 (normal application of the model), 2) with no suspension but including retention and 3) without suspension or retention (continually dry

Fig. 13. As in Fig. But for HCAB for the period October 2007 to March 2008.

surface). Numerical results for each dataset, along with traffic and surface wetness information are provided in Table 2.

When both retention and suspension are included, the comparison indicates an extremely high correlation (Fig. 9) of modelled and observed net PM10 concentrations for Hornsgatan (R2 = 0.77— 0.91) as well as consistent predictions of means (Fig. 7) and per-centiles (Fig. 8) over the 5 year period. This is significant because in 2010 a studded tyre ban was introduced on Hornsgatan and this reduced the number of cars with studded tyres by a factor of two (Table 2). The impact of retention, caused by surface moisture, is most clearly seen in the temporal correlation. For Hornsgatan the correlation is dramatically reduced by the exclusion of the surface retention (Fig. 8). The mean concentrations are less affected by this as these reflect the total wear independent of the timing. On the other hand, when retention is included but suspension is repressed then correlations during years without long wet periods (first 3 years) are still quite high. However, in this case retention without suspension leads to a loss of suspended wear mass and the mean concentrations are significantly under predicted as a result (Fig. 9).

For HCAB the correlations are lower, less than 51% of the variability is explained by the model, and the suspension part of the

model plays a less significant role, reflected in only a slight reduction in correlation with the exclusion of surface retention. Mean concentrations are hardly affected by the exclusion of retention, however there is an underestimate when suspension is not included. This low dependence on retention is closely associated with the surface wetness, since HCAB is seen to be wet only 20%—29% of time, compared to Hornsgatan that has road wetness frequencies from 43% to 56% (Table 2).

4.3. Source contributions

The model can be used to quantify the various source contributions to the emissions as well as indicating the contribution from direct emissions, immediate suspension under dry conditions, and suspended emissions, resulting from accumulation of mass on the surface. Using the model runs presented in Section 4.2 we show in Fig. 10 the mean PM10 emission factor for the direct, suspended and exhaust emissions. In Fig. 11 we show the source contributions to the total PM10 concentrations.

Wahlin et al. (2006) carried out receptor modelling of chemically analysed filters gathered at the same site at HCAB for four two

Fig. 14. Results of the model application at Hornsgatan for the period October 2010 to June 2011 including sand. Top: Daily mean modelled and observed concentrations of PM10 including the sand contribution. Middle: Surface mass loading of suspendable dust and suspendable sand. Bottom: Total mass of suspendable sand applied to the road per day.

week periods in 2003 and 2004. Their analysis indicated average emission factors for road wear of 83 mg km-1 veh-1, which is similar to the emission factors indicated by the model here. This supports the high road wear rate chosen for this site. It also indicates that the choice of pavement material may be very important for the PM10 concentrations.

Filter sampling, chemical analysis and receptor modelling of PM10 using COPREM (Wahlin, 2003) has been carried out in Hornsgatan over four sampling periods scattered in the period 2006—2007 (Sjodin et al., 2010). In total 28 daily samples were made and the sources of exhaust, road wear and brake wear were identified to contribute with 12.7%, 73% and 2.4% of the total PM10 mass respectively. For the same year, but not coincident period, the model predicts (see Fig. 11) contributions for the same sources of 10.6%, 76% and 5.5%. This supports the model calculations to some degree, particularly for the road wear component.

4.4. Salt and sand contributions

The model allows the application of both salt and sand (Sections 3.3 and 3.4) to the road surface as part of winter road maintenance activities. For Hornsgatan 2010—2011, information concerning days of salting and sanding is available, as is salting for both years at HCAB. In the model formulation presented here salt in solution may be removed by drainage processes, even without the implementation of a complete surface moisture model, as long as the precipitation is known. Results of the application of salt to both Hornsgatan (2010—2011) and HCAB, using a drainage efficiency of hsalt-eff = 0.5, are shown in Figs. 12 and 13. The average contribution of salt to PM10 is shown in Fig. 11. In both cases the addition of salt slightly increases the correlation, from 0.76 to 0.81 for Hornsgatan and from 0.39 to 0.44 for HCAB (Table 2). Salt also adds to the emitted concentrations with an addition of 1.0 mg m-3 for Hornsgatan and 1.2—3.3 mg m-3 for HCAB. These additions are from 6 to 16% of the total concentrations. Though there are no measurements of salt concentrations available for the periods modelled, Wahlin et al. (2006) estimated salt contributions to PM10 at HCAB to be 23% for a different period in 2003 and 2004.

Sand may also be applied to the road surface in the model, even though there is significant uncertainty in its rate of application, in its size distribution and in the mechanisms for its removal. For Hornsgatan 2010—2011 sand has been modelled assuming that 1% of the total sand mass is suspendable (i.e. <200 mm), Fig. 14. This is a lower value than indicated from direct measurements of the traction sand size distribution (Section 3.3), which was measured to be 6%. The 1% value used here leads to a contribution from sanding to the PM10 concentrations that is realistic in comparison to measured concentrations. In this case sand contributes with an average of 4.3 mg m-3, a 30% increase in PM10 concentrations, and leads to a slight increase in the correlation from 0.76 to 0.78. Clearly this result will depend on the suspendable portion of the sand and the application method but it does provide a first indication of how sanding can be included quantitatively in the model processes.

5. Discussion and conclusion

In this paper the road dust sub-model of a coupled traffic induced non-exhaust emission model is presented and is applied to two sites with 7 years of data where measurements of road moisture are available. The model performs very well on data from Hornsgatan in Stockholm where studded tyres are the major source of wear. The model successfully predicts the impact of a reduction in studded tyres and explains from 76 to 91% of the variability in the

daily mean concentrations for the five years assessed. The application of the model to H.C Anderson Boulevard in Copenhagen, where studded tyres are not in use, is less successful but can still explain from 44 to 51% of the variability seen.

It is doubtful, given the available data and the accuracy of the NOx emissions and the surface moisture measurements, that a higher correlation could be achieved for Hornsgatan for most of the years. This high correlation occurs without significant removal processes for road dust related to drainage or vehicle spray and without any other production process such as sanding or salting. Correlation tends to be poorer for the years that underestimate the mean concentrations, and this may indicate that additional sources such as sand, salt or other fugitive sources may be responsible for the missing mass.

The surface moisture, and subsequent retention and suppression of suspension, is the most important factor governing the timing of the emissions. This was clearest at Hornsgatan where three conductivity measurements on trafficked road lanes were used to determine surface wetness. At HCAB the modelled correlation was poorer than at Hornsgatan. It is possible that the road wetness data, taken in the less trafficked bus lane of the road using a sonic based sensor, may be a poor indicator for the traffic lane road wetness. Moreover, the possibility exists that other emission sources near the site are impacting on the measured concentrations and these are not correlated to traffic.

The model has also been applied to assess the impact of salting and sanding on the mass loading and on the emissions. Including salt in the model slightly improves the correlation and led to a 6— 16% increase in the mean concentrations. The application of sand to one of the datasets shows that sand could have a significant impact on the emissions but this aspect will need further attention, particularly in the removal processes for sand, if the contribution of sand to PM concentrations is to be properly quantified.

Uncertainties are large for a number of parameters and these have been necessarily confined using observational data. In Denby and Sundvor (2012) model parameters were estimated using data from seven different sites, including the two presented here, to constrain suspension rates, the fraction of PM10 in the total road wear and a number of surface moisture parameters not included in this study. The current set of model parameters must be considered to be best estimates. Based on current knowledge of these parameters we estimate a total model uncertainty of approximately ±40% for long term means (including uncertainties in wear rates, NOx emissions, suspension rates, road surface types and traffic input parameters). Further measurement campaigns that can concurrently determine the source contributions, the surface mass loading, the surface wetness, etc. are required in a range of environments to further refine the model parameters and to test the model concept.

The results presented here are based on the use of measured surface wetness, which has been shown to dominate the temporal variation of the PM emissions. For most applications surface wetness measurements are not available and so a surface moisture model is also required. This aspect will be addressed in Part 2 of this study (Denby et al., 2013) and provides the possibility for including more removal terms and for the impact of salt on the surface moisture vapour pressure to be included.

Acknowledgements

This work has been carried out within the Nordic Council of Ministers Project NORTRIP (BLS-306-00064) with substantial financial support provided by the Norwegian Climate and Pollution Agency (KLIF 4011009). The measurements at HCAB are funded by the Danish Environmental Protection Agency.

Appendix A. Model variables and parameters

Table A.1

Variables used in the road dust emission sub-model.

Variable

Variable

Description

M[oad pm

road cm

road /o,dir-

/q,sour

drivingcycle

Vref, sus

Vref , roadwear

Vref,tyrewear

Vref,PM —fractio awear

Msanding

nlanes f sus sanding

Msalt(i)

salting tsalting

Jo , suspension

0 , suspension hsus

tsanding

suspension /PM ,dir source /PM ,sus road /PM ,ref ,dir source

groad,drainable-min gretention-thresh.source

gretention-min,source

g km—1 or g m—2 g km—1 h—1 g km—1 h—1 0-1

veh h—1 g km—1 veh—1

km h—1 km h—1

km h—1

km h—1

gm 2 h

veh—1

g km—1 h—1 g km—1 h—1

g km—1 h—1

(km h—1)—1

mm mm mm

IP P D D IP

IM, IP

IM, IP

IP IP P

IM IM IP

D or IT D or IT IP IP IM

D or IT

D IP IP IP

P IP IP

Model time step

Road surface mass (dust, sand or salt) loading for the mass type m

Production of road surface dust or salt for the mass type m

Sink (removal) of road surface dust or salt for the mass type m

Fraction of wear that is directly emitted

to the air and not retained on the surface

Surface retainment factor based on the surface moisture. Wear

and dust loading is retained on the surface when this has a value

of zero. (source = road-tyrewear, brakewear and suspension).

Number of vehicles per hour with the specified

tyre types (t) and vehicle type (v)

Basis wear factor for different tyre types (t), vehicle types (v)

and wear (source = roadwear, tyrewear, brakewear)

Pavement type factor, used to adjust the basic

wear factor for road wear only (W0'loadwear)

Driving cycle factor, used to adjust the basic

wear factor for brake wear only (W0jrakewear)

Vehicle speed for the different vehicle types (v)

Reference vehicle speed at which

the suspension factor (/0%spension) is valid

Reference vehicle speed for which

the road wear parameter (W0'loadwear) is valid

Reference vehicle speed for which

the reference PM fraction (W^vt

r) is valid

Reference vehicle speed for which the reference

PM fracti°ns ^M' ref' dir source and & ' ref ;sus road-"

Power law index for the vehicle speed dependence of road and tyre wear.

Power law index for the vehicle speed dependence of road suspension.

Mass of sand applied to the road during a traction sanding event at time tsanding

Width of a single traffic lane on the road

Number of lanes on the road

Fraction of traction sanding mass that is classified

in the model as suspendable (<200 mm)

Mass of salt applied to the road during a road salting event for salt type i

Timing of the salting event. Input or derived by salting rules

Basis rate of suspension per vehicle for the given tyre (t) and vehicle (v) type

Scaling factor to adjust the basic suspension rates for the different mass types m

Scaling factor to adjust the basic suspension rate

that can be specified per site. Default is unity.

Drainage efficiency factor for dust and salt

Timing of sanding events. Input or derived by sanding rules

Total non-exhaust emissions in the size fraction

Total non-exhaust direct wear emissions

in the size fraction (source = roadwear, tyrewear, brakewear)

Total non-exhaust suspended emissions in the size fraction

Proportion of the direct wear mass in the size fraction .

Proportion of the suspended mass in the size fraction .

Reference value for the proportion

of the direct wear mass in the size fraction .

Reference value for the proportion

of the suspended mass in the size fraction .

Slope of the vehicle speed dependence for the proportion

of the suspended mass in the size fraction .

Index for dust types used in the model. Suspendable

dust from wear dust(sus), suspendable sand dust(sus-sand),

non-suspendable sand dust(non-sus)

and the two salt types salt(na) and salt (mg).

Index for tyre types: Studded (st), winter

non-studded (wi) and summer (su)

Index for vehicle types: Light (li) and heavy (li)

Water mass on the road surface

Non-drainable road water mass

Threshold value defining the upper limit for retention,

above which full surface retention is achieved

Threshold value defining the lower limit for retention,

below which no surface retention is achieved

source

hdmrain eff

dir source

PM ref sus road

PM fraction

a Variable types are defined as prognostic (P), diagnostic (D), model input parameter (IP), site specific input metadata (IM) or site specific input time series (IT).

Table A.2

Model parameters for the road dust sub-model used for the model runs described in the text.

Wo.roadwear (g km 1 veh 1) Studded tyres (st) Winter tyres (wi) Summer tyres (su)

Road wear Heavy(he) Light (li) Reference speed Vref.roadwear (km h"1) Power law factor for road wear awear 14.4 2.88 70 1 0.75 0.15 0.75 0.15

Wo,tyrewear (g km"1 veh"1) Studded tyres (st) Winter tyres (wi) Summer tyres (su)

Tyre wear Heavy (he) Light (li) Reference speed Vref.tyrewear (km h"1) 0.5 0.1 70 0.5 0.1 0.5 0.1

Wo,brakewear (g km"1 Veh"1) Studded tyres (st) Winter tyres (wi) Summer tyres (su)

Brake wear Heavy (he) Light (li) Reference speed Vref,brakewear (km h"1) 0.05 0.01 70 0.05 0.01 0.05 0.01

Name h pave (p) hdrivingcycle (d)

Pavement and driving cycle type scaling factor Hornsgatan 0.83 HCAB 4 1 1

f0,suspension (veh ) Studded tyres (st) Winter tyres (wi) Summer tyres (su)

Suspension parameters Heavy (he) Light (li) 5.0 x 10~5 5.0 x 10~6 5.0 x 10~5 5.0 x 10~6 5.0 x 10~5 5.0 x 10~6

Reference speed Vref,sus (km h"1) Power law factor for suspension asus 50 1

ho,sus h0,sus-sand h0,salt fsus-sanding f0,dir-source 1 1 1 0.01 1

Wear parameter PMtsp PM10 PM2.5

Fractional size distribution of emissions fPM,ref,roadwear °.5 fPM,dir-tirewear 0.5 fPM,dir-brakewear 1 fpM,ref,sus-road 0.5 Reference speed 50 Vref,PM-fraction (km h_1) CPM-&action(kmh"1)"1 0.012 0.18 0.1 0.8 0.18 0.008 0.01 0.5 0.008

Efficiency parameter Suspendable dust Salt

Efficiency factors hdrainage-eff 0.0 0.5

Parameter Road

Drainage and retention parameters gdrainable (mm) gretention-thresh (mm) gretention-min (mm) 0.6 0.1 0.04

Appendix B. Supplementary material

Supplementary material related to this article can be found at http://dx.doi.org/10.! 016/j.atmosenv.2013.04.069.

References

Amato, F., Pandolfi, M., Viana, M., Querol, X., Alastuey, A., Moreno, T., 2009. Spatial and chemical patterns of PM 10 in road dust deposited in urban environment. Atmospheric Environment 43 (9), 1650-1659.

Berger, J., Denby, B., 2011. A generalised model for traffic induced road dust emissions. Model description and evaluation. Atmospheric Environment 45, 36923703.

Blomqvist, G., Gustafsson, M., Bennet, C., Halldin, T., 2011. PM10 suspension of road dust is depending on the road surface macro texture. In: Proceedings of the European Aerosol Conference, September 2011, Manchester. URL: www. eac2011.com.

Boulter, P.G., 2005. A Review of Emission Factors and Models for Road Vehicle Non-exhaust Particulate Matter. TRL Limited, Wokingham, UK. (TRL Report PPR065). URL: http://uk-air.defra.gov.uk/reports/cat15/0706061624_Report1__Review_ of_Emission_Factors.PDF.

Boulter, P.G., Thorpe, A.J., Harrison, R.M., Allen, A.G., 2006. Road Vehicle Non-exhaust Particulate Matter: Final Report on Emission Modelling. TRL Limited, Wokingham, UK. (TRL Report PPR110). URL: http://uk-air.defra.gov.uk/reports/ cat15/0706061624_Report2_Emission_modelling.PDF.

Bukowiecki, N., Lienemann, P., Hill, M., Furger, M., Richard, A., Amato, F., Prévôt, A.S.H., Baltensperger, U., Buchmann, B., Gehrig, R., 2010. PM10 emission factors for non-exhaust particles generated by road traffic in an urban street canyon and along a freeway in Switzerland. Atmospheric Environment 44, 2330-2340.

Carslaw, D.C., Beevers, S.D., Tate, J.E., Westmoreland, E.J., Williams, M.L., 2011. Recent evidence concerning higher NOx emissions from passenger cars and light duty vehicles. Atmospheric Environment 45, 7053-7063.

Denby, B.R., Sundvor, I., 2012. NORTRIP Model Development and Documentation. Norwegian Institute for Air Research. (NILU OR 23/2012). URL: www.nilu.no.

Denby, B.R., Sundvor, I., Johansson, C., Pirjola, L., Ketzel, M., Norman, M., Kupiainen, K., Gustafsson, M., Blomqvist, G., Kauhaniemi, M., Omstedt, G., 2013. A coupled road dust and surface moisture model to predict non-exhaust road traffic induced particle emissions (NORTRIP). Part 2: surface moisture and salt impact modelling. Atmospheric Environment. in preparation.

EC, 2008. Directive 2008/50/EC of the European Parliament and of the Council of 21 May 2008 on ambient air quality and cleaner air for Europe. Official Journal L 152, 1-44,11.6.2008.

EC, 2005. Directive 2005/55/EC of the European Parliament and of the Council of 28 September 2005 on the approximation of the laws of the Member States relating to the measures to be taken against the emission of gaseous and par-ticulate pollutants from compression-ignition engines for use in vehicles, and the emission of gaseous pollutants from positive-ignition engines fuelled with natural gas or liquefied petroleum gas for use in vehicles. OJ L 275, 1-163, 20.10.2005.

EEA, 2012. Air Quality in Europe - 2012 Report. Report No 4/2012. European Environment Agency. URL: http://dx.doi.org/10.2800/55823 http://www.eea. europa.eu/publications/air-quality-in-europe-2012.

EMEP/EEA, 2009. EMEP/EEA Air Pollutant Emission Inventory Guidebook. EEA Technical report No 9/2009. URL: http://dx.doi.org/10.2800/23924 http:// www.eea.europa.eu/publications/emep-eea-emission-inventory-guidebook-2009.

Etyemezian, V., Kuhns, H., Gillies, J., Chow, J., Hendrickson, K., McGown, M., Pitchford, M., 2003. Vehicle-based road dust emission measurement (III). Atmospheric Environment 37, 4583-4593.

Gehrig, R., Hill, M., Buchmann, B., Imhof, D., Weingartner, E., Baltensperger, U., 2004. Separate determination of PM10 emission factors of road traffic for tailpipe emissions and emissions from abrasion and resuspension processes. International Journal of Environment and Pollution 22, 312-325.

Gustafsson, M., Johansson, C., 2012. Road Pavements and PM10. Trafikverket report 2012:241. URL: http://publikationswebbutik.vv.se/upload/6939/2012_241_ Road_pavements_and_PM10.pdf.

Gustafsson, M., Blomquist, G., Gudmundsson, A., Dahl, A., Jonsson, P., Swietlicki, E., 2008. Factors affecting PM10 emissions from road pavement wear. Atmospheric Environment 43, 4699-4702.

Jacobson, T., Wägberg, LG., 2007. Developing and Upgrading of a Prediction Model of Wear Caused by Studded Tyres and an Overview of the Knowledge of the Factors Influencing the Wear - Version 3.2.03. The Swedish National Road and Transport Research Institute, Linköping. (VTI notat 7-2007) (in Swedish). URL: http://www.vti.se/en/publications/developing-and-upgrading-of-a-prediction-model-of-wear-caused-by-studded-tyres-and-an-overview-of-the-knowledge-of-the-factors-influencing-the-wear-version-3203/.

Johansson, C., Denby, B.R., Sundvor, I., Kauhaniemi, M., Härkönen, J., Kukkonen, J., Karppinen, A., Kangas, L., Omstedt, G., Ketzel, M., Massling, A., Pirjola, L., Norman, M., Gustafsson, M., Blomqvist, G., Bennet, C., Kupiainen, K., Karvosenoja, N., 2012. NORTRIP: NOn-exhaust Road TRaffic Induced Particle Emissions. Development of a Model for Assessing the Effect on Air Quality and Exposure. Department of Applied Environmental Science, Stockholm University, Stockholm. (ITM-report 212). URL: http://www.itm.su.se/documents/ publications/NORTRIP_final_report_20120625.pdf.

Ketzel, M., Omstedt, G., Johansson, C., Düring, I., Pohjola, M., Oettl, D., Gidhagen, L., Wählin, P., Lohmeyer, A., Berkowicz, R., 2007. Estimation and validation of PM2.5/PM10 exhaust and non-exhaust emission factors for practical street pollution modelling. Atmospheric Environment 41, 9370-9385.

Kupiainen, K., Pirjola, L., 2011. Vehicle non-exhaust emissions from the tyre-road interface — effect of stud properties, traction sanding and resuspension. Atmospheric Environment 45, 4141—4146.

Langston, R., Merle Jr., R.S., Etyemezian, V., Kuhns, H., Gillies, J., Zhu, D., Fitz, D., Bumiller, K., James, D.E., Teng, H., 2008. Clark County (Nevada) Paved Road Dust Emission Studies in Support of Mobile Monitoring Technologies. Clark County Department of Air Quality and Environmental Management, Las Vegas, nV. URL: http://www.epa.gov/ttnchie1/ap42/ch13/related/Final_Test_Report.pdf.

Mathissen, M., Scheer, V., Kirchner, U., Vogt, R., Benter, T., 2012. Non-exhaust PM emission measurements of a light duty vehicle with a mobile trailer. Atmospheric Environment 59, 232—242.

Norman, M., Johansson, C., 2006. Studies of some measures to reduce road dust emissions from paved roads in Scandinavia. Atmospheric Environment 40,6154—6164.

Omstedt, G., Bringfelt, B., Johansson, C., 2005. A model for vehicle-induced non-tailpipe emissions of particles along Swedish roads. Atmospheric Environment 39, 6088—6097.

Omstedt, G., Andersson, S., Gidhagen, L., Robertson, L., 2011. Evaluation of new model tools for meeting the targets of the EU Air Quality Directive: a case study on the studded tyre use in Sweden. International Journal of Environment and Pollution 47, 79—96.

Patra, A., Colvile, R., Arnold, S., Bowen, E., Shallcross, D., Martin, D., Price, C., Tate, J., ApSimon, H., Robins, A., 2008. On street observations of particulate matter movement and dispersion due to traffic on an urban road. Atmospheric Environment 42, 3911—3926.

Pay, M.T., Jimenez-Guerrero, P., Baldasano, J.M., 2011. Implementation of resuspension from paved roads for the improvement of CALIOPE air quality system in Spain. Atmospheric Environment 45, 802—80 .

Pirjola, L., Kupiainen, K.J., Perhoniemi, P., Tervahattu, H., Vesala, H., 2009. Non-exhaust emission measurement system of the mobile laboratory SNIFFER. Atmospheric Environment 43, 4703—4713.

Sjodin, A., Ferm, M., Bjork, A., Rahmberg, M., Gudmundsson, A., Swietlicki, E., Johansson, C., Gustafsson, M., Blomqvist, G., 2010. WEAREM Wear Particles from Road Traffic — A Field, Laboratory and Modeling Study. IVL Report B1830. URL: http://www.ivl.se/webdav/files/B-rapporter/B1830.pdf.

Snilsberg, B., Myran, T., Uthus, N., 2008. The influence of driving speed and tires on road dust properties. Doctoral thesis. In: Snilsberg, B. (Ed.), Pavement Wear and Airborne Dust Pollution in Norway. Characterization of the Physical and Chemical Properties of Dust Particles. Norwegian University of Science and Technology, Trondheim. (Doctoral theses at NTNU, 2008:133).

US EPA AP 42,2006. Miscellaneous Sources, fifth ed., vol. I. http://www.epa.gov/ttn/ chief/ap42/ch13/ (chapter 13).

Vaze, J., Chiew, F.H.S., 2002. Experimental study of pollutant accumulation on an urban road surface. Urban Water 4, 379—389.

Wahlin, P., Berkowicz, R., Palmgren, F., 2006. Characterisation of traffic-generated particulate matter in Copenhagen. Atmospheric Environment 40, 2151—2159.

Wahlin, P., 2003. COPREM — a multivariate receptor model with a physical approach. Atmospheric Environment 37, 4861—4867.