NORTRIP- 2010 11 16

1

SMHI road dust emission model Gunnar Omstedt, SMHI

•model description and results

•improvements of existing model

•new parameters

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

NORTRIP- 2010 11 16

2

Model description

)(** int, dustelfe erwrefPMfq

suspensionf

suspensionf

directf

totf eee

Winter

Summer )(* , dustefe summerrefPMfq

suspensionf

fq is the source strength reduction function due to the moisture content of the road (0-1) l is the amount of dust on the road (0-1)

)(,

dustrefPMf

e is a reference emission factor

NORTRIP- 2010 11 16

3

l is the dust layer (relative units)

ll SiSot

l

)(

Sources Sinks Sanding Vehicle-induced

suspension Traffic related road wear Run-off with precipitation

g (mm) is the moister amount on the surface

gg SiSot

g

)(

Sources SinksPrecipitation Run-off

Evaporation

Equations 1 and 2 are coupled due to the parameter describing vehicle-induced suspension.

If the surface is wet, only a small fraction of l is emitted, then l is increasing.

If the surface is dry and l is large, a large fraction of l is emitted, then l is decreasing.

(1)

(2)

NORTRIP- 2010 11 16

4

Referens emissionfactor:

the baseline for the model is givenby scaling, using NOx as a tracer

backgroundNOx

roadsideNOx

backgroundPM

roadsidePMNOx

fPMf

CC

CCee

NORTRIP- 2010 11 16

5

Model description: the fq parameter

htt errgg *),0.1min( 1 t denotes present hour and t-1 previous hourg street moister content (mm)rr is the hourly precipitationeh is the hourly reduction factor for g due to evaporation

)*exp( ph Eke Ep is potential evaporation; k empirical factor found to be k=0.075

NORTRIP- 2010 11 16

6

Model description: the fq parameter

))1(

(*3600

a

p

an

p

rL

cerR

E

Rn is net radiation, ra aeodynamic resistance Δe the saturated vapour pressure deficit

gfq *93.01

Input data: (global radiation), cloud cover,windspeed, humidity and temperature

fq is the source strength reduction function due to the moisture content of the road

NORTRIP- 2010 11 16

7

Model description: the road dust layer -sources

Two kinds of particles are put into the model:•particles caused by the wear of pavement; •particles from traction sand

sw lll

l total dust layer (0,1)lw road wear dust layer (0,1)ls sand dust layer (0,1)α share of road wear particles in the total dust layerβ share of sand particles in the total dust layer

susprfststtwtw ffakll ),0.1min( 1,,

dayssandingnotifffll

dayssandingifffN

ll

susprftsts

susprftsts

),0.1min(

)1

,0.1min(

1,,

1,,

The road wear dust layer

The sand dust layer

ast share of studded tyres (%)kst normalising factor, 0.03frf hourly decay of road dust layer due to runoff by rain (0,1)fsusp hourly decay of road dust layer due to suspension (0,1)

N number of sanding days during a winter season

NORTRIP- 2010 11 16

8

Model description: the road dust layer -sources

The road wear dust layer

The sanding dust layer

Sanding occasions are calculated based onmeteorological data. (Sanding is assumed to occur during slippery conditions).

NORTRIP- 2010 11 16

9

Model description: the road dust layer - sinks

qdecaysusp fkf *1fsusp is the relative hourly decay of the dust depot due to suspensionkdecay is an empirical parameter (0.001,decresing the dust layer to half of its value within about one dry month)

a) suspension

b) run-off

)0.,1*)(,0.0max( htt errgRF RF is the hourly water run-off and is used tocalculate the hourly decay of road dust layer due to runoff by rain (0,1) , frf

95.0

)2/()2(*05.01,2

0.12

max

maxmax

rf

rf

rf

fRFrrif

RFRFfRFrrif

frrif

RFmax is maximum hourly water run-off (10 mm)

NORTRIP- 2010 11 16

10

Model description: road dust layer and final results

2-Jan 2-M ar 1-M ay 30-Jun 29-Aug 28-O ct 27-D ec

0

0.2

0.4

0.6

0.8

1

dust

laye

r dust layerl tota llw road w ear

ls sand

2-Jan 2-M ar 1-M ay 30-Jun 29-Aug 28-O ct 27-D ec

0

200

400

600

800

1000

ef(d

ust

) m

g/v

km -

mod

el

2 -Jan 2-M ar 1-M ay 30-Jun 29-Aug 28-O ct 27-D ec

0

200

400

600

800

1000

1200

1400

ef(

dus

t) m

g/vk

m -

trac

er m

eth

od Results using also a dispersion model

kmmg/veh216method)tracer

km mg/veh 209model)

(

(

suspensionf

suspensionf

e

e

NORTRIP- 2010 11 16

11

Examples of results

Street Height of buildings (m) side 1/side 2

Width of the street

(m)

Traffic intensity

(vehicles/day)

Heavy duty

vehicles (%)

Antiskid treatment

Share of studded

tyres (%)

Sundsvall/Skolhusallen 10/1 20 20000 4 sand 90 Uppsala/Kungsgatan 20/0 18 18000 5 sand 76

Stockholm/Sveavägen 25/25 33 29100 4 salt 75 Stockholm/Norrlandsg 25/25 15 14800 4 salt 75 Stockholm/Hornsgatan 24/24 24 35000 5 salt 75 Malmö/Amiralsgatan 25/25 21 23000 10 salt 30

NORTRIP- 2010 11 16

12

Comparison of measured (+) and modelled (solid line) daily mean concentrations of PM10 (µg/m3) at Hornsgatan for the year 2004. (b) The same results as (a) but sorted according to size

Hornsgatan 2004

NORTRIP- 2010 11 16

13

1-Jan 1-M ar 30-Apr 29-Jun 28-Aug 27-O ct 26-D ecYe a r 2 0 0 0

0

10

20

30

PM

2.5 s

tree

t-P

M2.

5 ro

of

(µg

/m3 )

N um ber of datapoints=356Average m easured=5.4Average m odelled=5.5r=0.72

1-M ar 15-M ar 29-M ar 12-Apr 26-Apr 10-M ay 24-M ay 7-Jun 21-JunÅr 2 0 0 5

0

5

10

15

PM

2.5 s

tree

t-P

M2

.5ro

of

(µg

/m3 )

N um ber of datapoints=100Average m easured=2.4Average m odelled=2.4r=0.61

a)

b)

The model can also be used for PM2.5 by changing the reference emissionfactor

Comparison of measured (+) and modelled (grey solid line) local contribution to daily mean PM2.5 concentrations (µg m–3) at two different streets in Sweden with different shares of studded tyres: (a) Hornsgatan in Stockholm in 2000 with approximately 75% studded tyres in the winter, (b) Amiralsgatan in Malmö for about four months in 2005 with approximately 30% studded tyres in the winter.

Other results- PM2.5

NORTRIP- 2010 11 16

14

1-Jan 1-M ar 30-A pr 29-Jun 28-A ug 27-O ct 26-D ecY e a r 2 0 0 3

0

10

20

30

40

50

PM

10st

reet

-PM

10ro

of (

µg/

m3 ) N um ber o f da ta po in ts= 344

A verage m easured=7 .4A verage m odelled=7 .1r=0 .84

0 10 20 30 40 50

PM 1 0 (m e a su re d -b a ckg ro u n d ) µg /m 3

0

10

20

30

40

50

PM

10

- m

od

elle

d (

µg

/m3 )

m easured- background m odelledyearly m ean 7 .4 7.1 90-percentile 13.2 12.398-percentile 17.2 16.9

PM10 Jagtvej year 2003PM10 Jagtvej year 2003

Comparison of measured (+) and modelled (solid line) daily mean concentrations of PM10 (µg/m3) at Jagtvej / Copenhagen for the year 2003.(b) The same results as (a) but sorted according to size. Model results are based on measured NOx concentrations.

Other results- no studded tyres

kmvehmgeee suspensionf

directf

totalf /129

Ketzel, M., Omstedt, G., Johansson, C., During, I., Pohjola, M., Dieterman,O., Gidhagen, L., Wåhlin,P., Lohmeyer, A., Haakana, M. and Berkowicz, R., 2007: Estimation and validation of PM2.5/PM10 exhaused and non-exhaused emissionfactors for practical street pollution modelling. Atmospheric Environment 41, 9370-9385.

NORTRIP- 2010 11 16

15

Type of stations 19 streets and close to roads stations 21 urban background stations

Andersson, S. och Omstedt, G., 2009: Validering av SIMAIR mot mätningar av PM10, NO2 och bensen. Utvärdering för svenska tätorter och trafikmiljöer avseende år 2004 och 2005. SMHI Meteorologi, Nr. 137.

Model validation of SIMAIRroad

NORTRIP- 2010 11 16

16

Results- streets and roads

The uncertainty of modelling estimation is defined as the maximum deviation between the measured and calculatedconcentration levels for 90 % ofindividual monitoring points, without taking into account the timing of the events. The average annual modelling uncertaintyfor PM10 is defined as ±50%

Comparison with EU Air Quality Directive targets

MRPE annual mean=0.38 ; MRDE annual mean=0.24MRPE 90-percentile =0.42

p

pp

O

MORPEMRPE

max)max(

Fairmode http://fairmode.ew.eea.europa.eu/

LV

MORDEMRDE LVLV

max)max(

NORTRIP- 2010 11 16

17

Improvements

Improve input data; study tyres, wintersanding / salting, street sweeping

• The computation of the sanding days is currently based on the meteorological parameters. This treatment could be improved by using the predictions of so-called road weather models.

• The cleaning of road surfaces is not considered in the present version of the model, although it can potentially have a major impact on the suspension emissions.

• The modelling of the influence of precipitation involves also challenges, especially in case of short-term, spatially limited and weak rain-and snowfall.

Mari Kauthaniemi et al., 2010

Include snow in the model; experimental data from northern Sweden (Umeå) and Finland can help us here

NORTRIP- 2010 11 16

18

New parameters

a) Studded tyres

b) Vehicle speed

c) CMA

d) Street sweeping

NORTRIP- 2010 11 16

19

new parametersa) studded tyres

referencewearroad

tyresstuddedwithoutfwearroad

tyresstuddedwithoutfref

etyreaetyreaRW

RWg )(%)*/()(%)*()( ,,

Suggestion: include a new function g, where g is a normalised function for the road wear,

use a linear relationship based on Norman and Johansson (2006) for dry conditions

wearroadtyresstuddedwithoutPMe , the road wear without studded tyres, tyre(%) is the relative proportion of

cars using studded tyres and a is a constant.

Typical values:

wearroadtyresstuddedwithoutPMe , 49 mg/ veh km

g with 75% studded tyres 1

g with 30% studded tyres 0.54

g with 0 % studded tyres 0.23

NORTRIP- 2010 11 16

20

Process Emission factor (mg vkm–1)

Notes

Exhaust 25–79 Share of heavy duty vehicles varies between 2 and 20%

Road wear, studded tyres 99–240 Share of studded tyre use varies between 45 and 92%

Road wear, un-studded tyres 36–57 Vehicle component wear estimated at 10 mg vkm–1

Winter sanding/salting 50–61 Uncertain estimation

Vehicle components wear 8–12 Share of heavy duty vehicles varies between 2 and 20%

http://simair.smhi.se/luftkvalitet/documents/meteorologi_134.pdf

NORTRIP- 2010 11 16

21

new parametersb) vehicle speed

)(refV

Vf

include a new function f, where V is vehicle speed and Vref

is the reference vehicle speed

What do we know about f ?

Measurements in lab. (Gustafsson et al,2005): an increase by a factor of 4 for an increase in vehicle speed from 40 to 80 km/h for studded tyres. Omstedt et al. (2005)similar emission factors for a Swedish highway (110 km/h) as those for Hornsgatan( 40 km/h)

Johansson et al.(2008):an increase in emission factors of about 20% from 40 km/h (streets inside Stockholm) to 80 km/h( Essingeleden)

Similar results from other studies, emission factors for motorways are not extremely highcompare to emission factors for city streets

NORTRIP- 2010 11 16

22

new parametersc) CMA

What do we know about CMA?timescale,..

•keep the surface wet for same days or include a new parameter similar to fq

•do some sensitivity analyses to find reasonable time scales

NORTRIP- 2010 11 16

23

new parametersd) street sweeping

•decrease the dust layer at selected time (probably rather different results due to selected time)

•check the timescale for the increase of the dust layer due to studded tyres

•do some sensitivity analyses

NORTRIP- 2010 11 16

24

Databases and models used by the SIMAIR system

Gidhagen, L., Johansson, H. and Omstedt, G., 2009: SIMAIR - Evaluation tool for meeting the EU directive on air pollution limits. Atmospheric Environment, 43, 1029-1036.Omstedt,G., Andersson,S., Gidhagen, L. and Robertson, L., 2010. New model tools for meeting the targets of the EU Air Quality Directive: description, validation and evaluation of local air quality improvement due to reduction of studded tyre use on Swedish roads. Accepted ( Int. J. Environment and Pollution)