Development and deployment of the traffic
emission model in Bolzano
Ing. Gianluca Antonacci, Ph.D.
CISMA Srl, Bolzano / Bozen
Introduction
Aim of the work:
Provide a calculation method to estimate emission from traffic
source in the town of Bolzano
Logical scheme:
Traffic data distribution over the road network pollutant
emission calculation pollutant dispersion in the atmosphere
Field of application:
municipality of Bolzano, on major road network (in principle
extensible, as the method is implemented in generalized way)
Model scheme
Meteorological data
- Wind velocity and direction
- Stability class (turbulence)
Traffic data
Topographical data
- digital elevation map
emission
calculation
Emissions
Dispersion model
Pollutants concentration
Methodologies for emission calculation
Known and established methodologies for pollutant emission from
traffic and their peculiar characteristics caratteristiche
Mobile6 (USA, Canada) not easily adapted to EURO
classification
ARTEMIS (France) most detailed but difficult to apply at a
practical level
HBEFA (Austria, Switzerland, Norway)
COPERT IV (Italy, UK, Greece, Spain, Germany) compliant to
EMEP standard and compatible with the methodology already
adopted from the Province of Bolzano for itsemission inventory
was chosen for the INTEGREEN project
COPERT methodology
Fuel Variables
- consumption
- specifications per fuel type
Activity Data
- number of vehicles per vehicle category
- distribution of vehicles into different
exhaust emission legislation classes
- mileage per vehicle class
- mileage per road class
COPERT procedure, as
Adopted by European
Environmental Agency
2000
1875
1750
1625
1500
1375
EF=f(speed)
1250
1125
1000
875
750
625
500
0
10
20
30
40
50
60
70
80
90
100
v [km/h]
Driving Conditions
- average speed per vehicle type
- average speed per road
Other Variables
- climatic conditions
- mean trip distance
- evaporation distribution
Emission Factors
- per type of emission (hot, cold,
evaporation)
- per vehicle class
- per road class
110
120
130
Emission factors
Emission
factors
Diesel passenger
cars
FE COPERT
IV - AUTOVETTURE
DIESEL
Emission
factors
Gasoline passenger
FE COPERT
IV -
AUTOVETTURE
BENZINA cars
2,5
2,5
2,0
Gasoline
Benzina
<1,4 l
Gasoline
Benzina 1,4 2,0 l
Gasoline
Benzina
>2,0 l
1,5
Diesel <2,0 l
Diesel >2,0 l
1,0
0,5
0,5
1,5
0,0
Euro 0
Euro I
Euro II Euro III Euro IV Euro V Euro VI
Euro 0
Euro I
Euro II
Euro III
Emission
factors
Duty
FE COPERT
- VEICOLI
COMMERCIALI
PESANTI
COPERT
IV IV
EF
Light
Heavy
DutyVehicles
Vehicles
20
18
16
14
Diesel <7,5t
Diesel 7,5 - 16t
Diesel 16-32t
Diesel >32t
12
10
8
6
4
2
0
Euro 0
Euro I
Euro II
Euro III
Euro IV
Euro V
Euro VI
Euro IV
Euro V
Euro VI
COPERT algorithm (1/2)
COPERT formulation was adapted to be general and to deliver
results with acceptable calculation time can also be used in
real-time application
All the emissions are expressed in the generalized version with a
simple formula
EF = (A+B*v+C*v)/(1+D*v+E*v)
EF [g/(km*veh)] emission factor
v [km/h] mean vehicle speed on the road stretch
COPERT algorithm (2/2)
2 types of original EF curves:
- continuous formulation OK
- not continuous or step-wise formulas problems
if we want to derive continuous (smooth) results
The interpolation solves and simplifies the issue
and in most cases the difference with respect to
the original formulation is negligible
Calculation method
The work was split into 2 phases
1. Derivation of continuous curves through min-square
adaptation (146 vehicle classes x 11 pollutants = 1606 formulas)
2. Creation of a calculus software taking advantage of the
derived EF database, in three versions:
A stand-alone software able to read / write files in industry
standard SHP+DBF, which can be used in a common GIS
environment (es. Arcview)
An excel/openoffice compatible spreadsheet, intended for
quick calculation of overall emissions
A full database version intended to be used in real-time
application, based on Postgresql + Python scripting
Pollutants
Considered pollutants:
NOx (nitrogen oxides), CO (carbon monoxide), VOC (volatile
carbone organic compounds), NH3 (ammonia), PM10 (fine
particulate matter), TSP (total suspended particulate matter),
SO2 (sulphur dioxide), CH4 (methan), HC (unburnt
hydrocarbons)
And moreover
CO2 (carbon dioxide) & FC (fuel consumption) which are directly
bound each other
TEM (stand-alone program)
Example: traffic data (input)
Example: emission data (result)
TEM (spreadsheet version)
Compact quick-use
version: the front end
is very simple
Yellow = input
Blue = output
All the calculation is
hidden to the user
TEM (database version)
Input data
How do we ingest traffic data in Traffic Emission software?
When using TEM in real-time mode traffic assignment on the basis of measured
data at traffic station should be ingested. There are some possibilities:
Use a scaling coefficient to derive the number of vehicle transit on each road
stretch with reference to the nearest traffic station needs to be calibrated at
least once (present simplified method)
Adopt simple traffic model based on traffic data to be run on-the-fly before the
emission estimate
Use a complete traffic data (to be verified if this is feasible in a real-time
environment)
Next steps (1/2)
Next step (task foreseen for
year 2014) is to couple a
dispersion model able to
ingest emission data from
TEM, meteorological data
from weather station and
morphological data...
and deliver quasi-real time
maps
of
pollutant
concentrations at the urban
scale with a grid resolution of
50-100m
Next steps (2/2)
From a first overview in specialized literature we propose the adoption of one
of the following air quality models, both open source and therefore adaptable
to our case:
- CALINE (U.S.EPA)
- AUSTAL (D-BUWA)
AUSTAL is more precise and evolved, the state of the art of local scale air
quality dispersion model for urban areas; on the other hand very demanding
in term of computational resources hardly to be used in real time,
especially in the hypothesis of scaling up of the computed road network
CALINE is more simplified and flexible, can also be adapted to peculiar
situations such as emissions in urban canyon, over bridge, in parking lots,
etc...+
A comparison test with both test is foreseen and after the functionality check
the best in terms of computational time and accuracy will be chosen.
Thanks for your attention
