PowerPoint-Pr

1
Multi-Pollutant Multi-Effect Modelling of
European Air Pollution Control Strategies - an
Integrated Approach Case Studies Joint
Optimisation Results
2
The MERLIN team

• IER University of Stuttgart (Co-ordinator)
• Norwegian Meteorological Institute (met.no)
• Aristotle University of Thessaloniki,
  Laboratory for Heat Transfer and Environmental
  Engineering (AUT/LHTEE)
• University College London (UCL)
• ECOFYS Energy and Environment
• Institute for Ecology of Industrial Areas (IETU)
• Energy Research Center (ERC) of Ostrava
  Technical University
• National Institute of Meteorology and Hydrology
  (NIMH)
• University of Ploiesti (Ploiesti, Romania)

3
Objectives
Development and application of methodologies and
toolsfor an integrated assessment of European
air pollution control strategies
Features

• multi-pollutant, multi-effect assessment
• cost-effectiveness and cost-benefit analysis
• application of advanced optimisation methods
• inclusion of non-technical measures
• macroeconomic effects and distributional
  burdens of air pollution control
• inclusion of new member states

4
Acidification
Global Warming
Eutrophication
N2O
SO2
CH4
NOx
CO2
NH3
CO
NMVOC
Urban Air Quality
Particulate Matter (PM2.5 / PM10)
TroposphericOzone
primary secondary Aerosols
Multi-Pollutant Multi-Effect Analysis
5
MERLIN Model Framework
Databases compiled for 2000, 2010, 2020

• Optimal Strategies, including
• Emissions by country
• Concentrations by gridcell
• Abatement costs by country and sector
• Avoided damage costs
• ...

SADB
MDB
Stock Activities
MeasureData
Scenario-ToolScenarion development and data
management
OMEGA
Changing Stock and Activity by implementing
measures
Stock Activities
MeasureData
Optimisation
SADB
MDB
Scenarios
Modified databases

• Emission Scenarios for future years, giving
• Emissions by country and sector
• Implementation degrees of measures
• Changes in emissions relative to the base case
  by country and sector

6
Measure-Matrix-Approach
Costs of implementation(typically with reference
to Stock or Activity)
Measure-Database (MDB)
Reference Description DEF1 DEF2
.. DEFn Costs
Meta-Information
techn. measures(affecting EFs)
non-tech. measures(affecting S or A)
unique ID
Information on imple-mentation,
interdependencies, i.e. AND, OR, XOR, ...
Stock-Activity-Database (SADB)
Reference Description Stock (S) Activity (A)
EF1 EF2 .. EFn
Meta-Information
e A EFi
EF Emission Factor S Stock (e.g. of
vehicles) A Activity (e.g. km/yr) e source
emissions E source-group emissions
E S e
7
State of Work

• Stock, activity data and emission factors for
  2000 and the baseline scenario 2010 established
  and checked by comparing it with EMEP 2000 and
  the CAFE baseline 2010 and UNFCCC 2010 data
• About 1000 measures per country ( 25 000
  measures) implemented, measure cost data
  currently validated against literature and other
  projects, e.g. EGTEI, IIASA RAINS)
• A large number of model scenario runs have been
  carried out for testing assessing the full
  model system, including different meteorological
  conditions based on source-receptor relationships
  for 1997, 2000, and other years to come

8
175 bill.
123 bill.
Impact Assessment Assessment of health impacts
due to emissions in EU25 for 2000 and 2010
baseline (estimates including local damages)
9
Model results (1) Ozone maps for a
health-related strategy aiming at compliance with
AOT60 (i.e. 0 ppb.h exceedance)
DAOT40f
DAOT60 for a health-oriented Ozone scenario
Exceedance of AOT 60 base case 2010 64 ,
resulting health impact scenario 25 of grid
cells
DAOT40c
Total costs 92 billion , emission reduction
2.97 Mt of NOx, 2.51 Mt NMVOC
10
Model results (2) PM2.5 and PM10
Analysis of a joint scenario to achieve
compliance with AOT60, PM10/2.5 concentrations
(threshold set to 10 mg/m3) and Kyoto targets for
GHG emissions (as CO2-equivalent) with regard to
PM2.5 (above) and PM10 (below) impacts. (1997
meteorology, EU25 countries)
11
Model results (3) Case study road transport PM
in Germany

• Investigating additional options to reduce PM10
  from road transport sources until 2010, how far
  can M(T)FR be taken?
• 3 approaches (technical measures only tech
  switch measures tech, switch and
  non-technical measures)
• non-technical measures modeled based on
  activity changes (not driving, switching to
  public transport, long term responses excluded)
• costs for non-technical measures derived from
  dead weight loss
• assuming retrofitting of existing vehicle stock
  and new vehicles sold equipped with PM control
  equipment from 2006 on
• Some results
• total costs for Passenger Cars (diesel) 1.3
  bill. /a
• ? 7.1 Mill. PCdiesel in Germany projected by
  2010
• roughly 184 /vehicle year
• total costs for TECH scenario about 3.8 bill.
  /a
• up to 17.9 bill. /a for tech switch nontech

12
Model results (4) Case study road transport PM
in Germany
mill.

• Costs in Germany vs. benefits from reduced
  emissions of PM10, NOx and VOC in EU25
• PM10 reduction on top of 2010 BAU (? 2010 CAFÉ
  Baseline) in the sector Road Transport
• reduction in total PM10 emissions from 8.3 to
  13.7

13
Model results (5) Case study road transport PM
in Germany
TECH
TECH SWITCH
TECH SWITCH NONTECH
Dmax ? 0.44mg/m3
Dmax ? 0.69mg/m3
Dmax ? 0.70 mg/m3

• Resulting changes in primary PM concentrations
  for the three cases investigated

14
Model results (6) Case study road transport PM
in Germany

• Most efficient measures to reduce 10 kt from
  road transport sources in Germany?

Technology Name of Measure Heavy duty vehicles
pre EURO Diesel Oxidation Catalyst, base
metal Heavy duty vehicles EURO I DPX Catalyzed
Particulate Filter Heavy duty vehicles EURO
II DPF PM sinter metal filter catalyst (open
system) Heavy duty vehicles EURO III DPF PM
sinter metal filter catalyst (open system) Urban
Buses pre EURO Modern Diesel Oxidation Catalyst,
precious metal Urban Buses EURO I Modern Diesel
Oxidation Catalyst, precious metal
15
Model results (6) Case study road transport PM
in Germany

• Optimised for PM10 reduction !

Technology Name of Measure Heavy duty vehicles
pre EURO Diesel Oxidation Catalyst, base
metal Heavy duty vehicles EURO I DPX Catalyzed
Particulate Filter Heavy duty vehicles EURO
II DPF PM sinter metal filter catalyst (open
system) Heavy duty vehicles EURO III DPF PM
sinter metal filter catalyst (open system) Urban
Buses pre EURO Modern Diesel Oxidation Catalyst,
precious metal Urban Buses EURO I Modern Diesel
Oxidation Catalyst, precious metal
16
Conclusions Outlook

• main issues usually availability and access to
  data for stock, activities and measures adding
  and improving input data will further improve the
  quality of results
• main advantages of the measure-matrix approach
  and the GA optimisation are the scalability from
  individual measure assessment to sectoral to
  country/regional analysis, both for
  cost-effectiveness and cost-benefit assessment
• detailed analyses of model results indicate
  complex interactions between measures in
  particular for stationary sources integrating
  more sophisticated ways to account for costs of
  early de-commissioning of equipment will improve
  the assessment of relative cost-effectiveness of
  measures further
• extension to cover heavy metals, benzene and
  direct pollutant releases into water and soil is
  currently in progress, as well for non-CO2 GHGs
• long-term plans include investigating
  parallelisation, web-accessability and dynamic
  modelling over longer time periods, allowing for
  improved assessment of indogeneous and
  exogeneous changes in stock, activities and
  measure implementation (e.g. duration of
  implementing)

17

• Contact
• Stefan Reis sr_at_ier.uni-stuttgart.de
• http//www.merlin-project.info
• http//www.ier.uni-stuttgart.de

TFEIP ESPREME Workshop on Heavy Metals and
POPsRovaniemi/Finland, Oct 18/19
2005 http//espreme.ier.uni-stuttgart.de/workshop

Project on Natural and Biogenic Emissions
(NATAIR) http//natair.ier.uni-stuttgart.de
Thank you for your attention!
18
Is the global optimum reached?

• This can not be proven (for so-called np-hard
  problems), however it is likely, that the found
  solution is close to the optimum. This is
  achieved by
• Inclusion of special strategies as subsets of the
  starting population, to direct the search to
  regions of the solution space, which indicate
  potential for local/global optima.
• enhancement of the fitness of young solutions,
  i.e. leaving the mutation operator enough time to
  improve them locally, so they are not prematurely
  suppressed by older ones.
• Use of diversity increasing operators, preventing
  the search to ignore promising regions too early
  in favour of few strategies with high fitness
• Simulation of SINEs (short interspersed elements)
  to provide points to the crossover operator where
  cutting is done with increased probability.

see e.g. Reis S , Nitter S, Friedrich R (2004)
Application of Genetic Algorithms for the
Optimisation of Multi-Pollutant Multi-Effect
Problems. In Pahl-Wostl, C., Schmidt, S. and
Jakeman, T. (eds) iEMSs 2004 International
Congress "Complexity and Integrated Resources
Management". International Environmental
Modelling and Software Societey, Osnabrueck,
Germany, June 2004.