timo elolähde 1 traffic model system and emission calculations of the helsinki metropolitan area...
TRANSCRIPT
Timo Elolähde
1
Traffic Model System and Emission Calculations of the Helsinki Metropolitan Area Council
Timo Elolähde 4
Definitions of areal divisions
YTV area includes the cities of Helsinki, Espoo, Vantaa and Kauniainen.
Surrounding areas include eight municipalities around the YTV area.
Helsinki region = YTV area + surrounding area = 12 municipalities
Metropolitan area is used to describe an area contained within approximately a 100 kilometre radius from Helsinki. It consists of 72 municipalities.
Timo Elolähde 5
Tampere
Mikkeli
LahtiHämeenlinna
Tammisaari
Kotka
Hämeenlinna
Lahti
Kotka
YTVTammisaari
PKS
47,000 commutersin 1980
88,000 commutersin 1990
108,000 commutersin 2002
YTV
Turku
YTV
Proportion of commuters in the municipality’s work force
Over 35 %10 - 35 % 2 - 10 %
Commuting in the Helsinki Metropolitan Area 1980–2002
6Timo Elolähde
Population Jobs 31.12.2004 31.12.2003 in YTV area
HelsinkiEspoo
Kauniainen
Vantaa
559 000
369 000
227 400
104 000
8 500 2 700
185 40095 000
980 300
0
200000
400000
600000
800000
570 700
Population and the number of jobs in the YTV Area
Timo Elolähde 8
Journeys made daily by public transport and by car within the YTV area
Journeys (1000/day)
66
53
42 39 39
0
25
50
75
100
1966 1976 1988 1995 2000 2005
Share taken by public transport (%)
(38)
0
500
1 000
1 500
1966 1976 1988 2000 2005
Private car
Public transport
1995
Timo Elolähde 10
Traffic model system
Traffic is divided into three parts• internal trips made by the
inhabitants of the region• trips generated by Helsinki-
Vantaa airport (air passengers and employees)
• external trips (cars only)• freight transport (vans and
lorries)
•Modes• walk, bicycle• public transit• car (as driver or passenger)
Trip categories• home-based work trips• home-based school trips• other home-based trips• non-home-based trips
•Time periods• morning peak hour• average hour of the day• evening peak hour
Timo Elolähde 11
Traffic model system
Tools• Emme/2 macros (contain Unix file handling commands)• SAS programs (preparation of input, writing some macros)• FORTRAN programs (summary of results)• Unix scripts (renaming output files)
Timo Elolähde 12
Feedback in the four-step model system
internal trips airport trips
trip generation trip generation
destination choice destination choice external trips
mode choice mode choicecommercial trips
total demand and route choice
Timo Elolähde 13
Model types
trip category trip generation mode choice destination choice
home-based work trips
trips / person working
logit model logit model
home-based school trips
trips / person of school age
distance table (distribution)
logit model
other home-based trips
trips / inhabitant logit model logit model
non-home-based trips
trips / inhabitant logit model logit model
Timo Elolähde 14
Logit model and logsum
probability of alternative i
.
1
J
j
V
V
ij
i
e
eP
logsum = ln (
J
j
V je1
)
where
where,2211 niniii xxxV
k. ealternativin variableof value=x
n),1,2,=(j variableoft coefficien
ealternativ offunction benefit
jk j
jj
i
x
x
iV
Timo Elolähde 15
Variables used in models
•Mode choice models• nr of transfers, transit• travel time, transit or car• travel cost, transit or car• parking place availability
(arriving trips / parking place)• parking cost• cars/household• ln(distance), walk or bicycle• distance 0-5 km, walk or bicycle• distance 5-10 km, walk or
bicycle• dummy variables
•Destination choice models• logsum of mode choice• scale factor (inhabitants, jobs)• ln(jobs)• dummy variables
Timo Elolähde 16
Mode combinations possible
Influence of the number of modes (ms149, ms199, ms249, ms299)
on text registers and description fields of matrices
(e.g. ”morning peak %t2% work trips”)
text register 3 4 -4 5
t1 Walk+bicycle Walk+bicycle Walk Walk
t2 transit Bus+tram transit Bus+tram
t3 Car Car Car Car
t4 NO BIKE NO BIKE bicycle bicycle
t5 NO RAIL Heavy rail NO RAIL Heavy rail
Timo Elolähde 17
Principles applied in coding macros
• The same selection of possible variables in all models (except school trips)
• No constants in the model formulas but the coefficients of the models are in scalars
• Systematics in matrix numbers• If a variable is not in the model, its coefficient is zero• Only the number of the first input matrix is given as a macro
parameter, other consecutive numbers are calculated (e.g. nr of transfers in matrix %2%, transit time in matrix r2=%2%+1)
• Logical scalars (school trip models in macro school_%ms250%.mac, where ms250=96 or ms250=2001)
Timo Elolähde 18
Scalars containing the coefficients
model variables coefficientshome-based other home- non-home- work trips based trips based trips
destination logsum ms106 ms156 ms206destination scale factor ms107 ms157 ms207destination ln(jobs) ms108 ms158 ms208mode dummy, walk ms110 ms160 ms210mode dummy, bus+tram ms111 ms161 ms211mode dummy, car ms112 ms162 ms212mode travel cost, heavy rail ms115 ms165 ms215mode travel cost, bus+tram ms116 ms166 ms216mode travel cost, car ms117 ms167 ms217mode parking ratio ms118 ms168 ms218mode parking cost ms119 ms169 ms219
Timo Elolähde 19
Writing an Emme/2 macro with a SAS program
Why?
Do you want to copy and paste this section 24 times and edit the parts which are underlined?
Solution: Give the changing part as data cards and write the rest of the macro with a SAS program (or with some programming language).
1 y ms311 y wt24h home-based work trips~?q=1 y
mf301
y gn01,gn04
o
+ +~?b=1 2
Timo Elolähde 20
Essential parts of the SAS programfilename outfi2 'K:\Emme2\summary_matr_demo2.mac';
data matr;length mxnro msnro $ 5 name $ 6 descr $ 40;input mxnro $ 4-8 msnro $ 10-14 name $ 16-21 descr $ 23-62 ;cards; mo09 ms301 nrinha total nr of inhabitants mf301 ms311 wt24h home-based work trips 24h ms999 last line ;
data _null_;
set matr;
file outfi2;
if _N_ = 1 then do; put "~#" / "~#** calculate sums of vectors" / " 3.21" ;end;
Timo Elolähde 21
Essential parts of the SAS program
nro = substr(msnro,3,3);if (nro ne '999') then do; put "~# *** matrix " _N_ " *** " ; put " 1" / " y" / msnro $ 2-6 / " y" / name $ 2-7 / descr $ 2-41 / "~?q=1" / " y" // mxnro $ 2-6 /// " y" ; if (substr(mxnro,1,2) = 'mo') then put " gn01,gn04" // " +" ; else if nro in ('311') then put " gn01,gn04" // " o" // " +" / " +" ; put "~?b=1" / " 2" ;end;
if nro = '999' then do;
put " q" / "~#** output the list of scalars" /
" reports=summary_matr_demo.txt" /
" 3.14" / " 2" / " ms" / "~?b=1" / " 2" / " q" /
" reports=%1%" / "~/ *** summary_matr_demo.mac ***" ;
end;run;
Timo Elolähde 23
Traffic surveys
Internal trips• trips made by the inhabitants of the YTV area (four cities)
during one day (24 h) in autumn 2000• personal trip diary interview, 8,666 persons and 28,553 trips
Trips generated by Helsinki-Vantaa airport• 875 air passengers and 801 employees (flying and non-flying)• survey made in autumn 2001
External trips and freight transport• origin-destination study made in autumn 1988
Timo Elolähde 24
Model estimation
Internal trips• estimation made by Ms Nina Karasmaa (Helsinki University of
Technology, Transportation Engineering)• Alogit program• More than 50 model sets were estimated and tested• Differences e.g. in number of modes and model hierarchy
(mode choice after destination choice or vice versa)• Three modes in the model set selected.
Trips generated by Helsinki-Vantaa airport• estimation made by Mr Jyrki Rinta-Piirto (Strafica Ltd)
External trips and freight transport• models estimated in 1990 are based in changes in land use.
Timo Elolähde 26
”Minor” problem in emission calculations
• Traffic models produce demand matrices for three weekday hours.
• Finnish Meteorological Institute needs emissions for every hour of the year for dispersion calculations.
Timo Elolähde 27
Principle of emission calculations
auto demand matrices (car+van, lorry) transit demand matrices* morning peak * morning peak* average hour of the day * average hour of the day* evening peak
transit assignment (speed, @voltr)regression models * two hours
auto demand matrices emissions of bus links* 10 weekday hours (fuel, CO2, SO2, CO, NOx, PM, HC)* 7 Saturday hours and rail links (@energy, CO2, NOx, PM)* 7 Sunday hours * morning peak
* average hour of the dayauto assignment (speed, volau, volad)* 10+7+7 hours regression models
emissions of auto links and centroids emissions of bus and rail links(fuel, CO2, SO2, CO, NOx, PM, HC) (CO2, SO2, CO, NOx, PM, HC)* 10 weekday hours * 24 weekday hours* 7 Saturday hours * 24 Saturday hours* 7 Sunday hours * 24 Sunday hours
copying or interpolation
emissions of other hours* 14 weekday hours* 17 Saturday hours* 17 Sunday hours
total emissions
Timo Elolähde 28
Emission calculations
Tools• Emme/2 macros (contain Unix file handling commands)• FORTRAN programs (copying or interpolation from link data
of 10+7+7 hours to 14+17+17 hours and summary of results)• Unix scripts (dialog of FORTRAN run, renaming output files)
Emission factors• fuel consumption, CO2, SO2, NOx, particles (PM), CO, HC• polynomial functions of average speed (from assignment)
Timo Elolähde 29
Examples of emission factors:NOx emissions of cars and vans
0
0,2
0,4
0,6
0,8
1
1,2
1,4
10 20 30 40 50 60 70 80 90 100 110 120
average speed (km/h)
em
iss
ion
(g
/km
/ve
h)
car kat 2005
car diesel 2005
van diesel 2000
car kat 2030
car&van diesel 2025
Timo Elolähde 30
Examples of emission factors:NOx emissions of trucks and buses
0
2
4
6
8
10
12
14
16
10 20 30 40 50 60 70 80 90 100
average speed (km/h)
em
iss
ion
(g
/km
/ve
h)
trailer truck 2000 (EU 2)
single-unit truck 2000 (EU 2)
bus 2000 (EU 2)
trailer truck 2025 (EU 5)
single-unit truck 2025 (EU 5)
bus 2025 (EU 5)
Timo Elolähde 31
Examples of emission factors:CO2 emissions of cars and vans
0
50
100
150
200
250
300
350
10 20 30 40 50 60 70 80 90 100 110 120
average speed (km/h)
em
iss
ion
(g
/km
/ve
h)
car kat 2000
car diesel 2005
van diesel 2000
car kat 2025
car diesel 2030
van diesel 2025
Timo Elolähde 32
Examples of emission factors:CO2 emissions of trucks and buses
0
500
1000
1500
2000
2500
10 20 30 40 50 60 70 80 90 100
average speed (km/h)
em
iss
ion
(g
/km
/ve
h)
trailer truck 2000&2025
single-unit truck 2000&2025
bus 2000&2025
Timo Elolähde 33
Proportions of vehicle types in emission calculations (volau)
4) emission factors for average petrol car in 2000 (43 % non-kat, 52 % EU0-2, 5 % EU3, 0 % EU4-5)5) emission factors for average petrol car in 2025 ( 0 % non-kat, 0 % EU0-2, 25 % EU3, 75 % EU4-5)
percentage2005 2030 in scalar
cars and vans cars, non-kat 0 0 ms80 cars, kat-1995 80 4) 0 ms81 cars, kat-2020 - 85 5) ms82 cars, diesel 1995 10 0 ms83 cars, diesel 2020 - 5 ms84 vans, diesel 1995 10 0 ms69 vans, diesel 2020 - 10 ms85total 100 100
Timo Elolähde 34
Proportions of vehicle types in emission calculations (volad and bus)
percentage2005 2030 in scalar
trucks single-unit trucks EU 0-2 70 5 ms86 trailer combination trucks EU 0-2 30 2 ms87 single-unit trucks EU 4-5 - 65 ms88 trailer combination trucks EU 4-5 - 28 ms89total 100 100
buses LPG or CNG buses buses in Helsinki EU 0-2 100 0 ms81 regional buses EU 0-2 100 0 ms83 buses in Helsinki EU 4-5 - 100 ms82 regional buses EU 4-5 - 100 ms84
Timo Elolähde 35
Regression models in emission calculations
• The regression models have been estimated using volume counts on four cordon lines.
• For auto assignment, the volumes (car+van and truck) for each hour of the day (10+7+7) are used as regressands and three forecasted hours (morning peak, evening peak and an average hour of the day) as regressors of the model. The models are used for calculating the demand matrices for each hour.
• For transit assignment, the bus volumes for each hour of the day (3*24) are used as regressands and two forecasted hours (morning peak and an average hour of the day) as regressors of the model. The models are used for calculating the link volumes and emissions for each hour.
Timo Elolähde 36
Emission calculations
• emission on regular link [kg/h] = volume [veh/h] * length [km] * emission [g/km/veh] / 1000
• cold starts (three classes of motor temperature) and emissions of connector links handled as emissions of the area (in the centroid)
• example of copying and interpolation of the emission (from 10+7+7 hours to 14+17+17 hours)
hour hour weekday emission 4am- 5am 4- 5 EMIS_WD_23_5 5am- 6am 5- 6 EMIS_WD_23_5 6am- 7am 6- 7 (EMIS_WD_23_5 + EMIS_WD_7)/2. 7am- 8am 7- 8 EMIS_WD_7 8am- 9am 8- 9 EMIS_WD_8 9am-10am 9-10 EMIS_WD_9_1310am-11am 10-11 EMIS_WD_9_13
Timo Elolähde 37
Principle of emission calculations (repeated)
auto demand matrices (car+van, lorry) transit demand matrices* morning peak * morning peak* average hour of the day * average hour of the day* evening peak
transit assignment (speed, @voltr)regression models * two hours
auto demand matrices emissions of bus links* 10 weekday hours (fuel, CO2, SO2, CO, NOx, PM, HC)* 7 Saturday hours and rail links (@energy, CO2, NOx, PM)* 7 Sunday hours * morning peak
* average hour of the dayauto assignment (speed, volau, volad)* 10+7+7 hours regression models
emissions of auto links and centroids emissions of bus and rail links(fuel, CO2, SO2, CO, NOx, PM, HC) (CO2, SO2, CO, NOx, PM, HC)* 10 weekday hours * 24 weekday hours* 7 Saturday hours * 24 Saturday hours* 7 Sunday hours * 24 Sunday hours
copying or interpolation
emissions of other hours* 14 weekday hours* 17 Saturday hours* 17 Sunday hours
total emissions