AirWare On-line Reference Manual
| ||Release Level || 6.1 |
| ||Release Date ||2012 05 |
| ||Revision Level ||1.0|
Last modified on:
Saturday, 1-Dec-12 11:36 CET
Emission Control Optimization: a summary
Optimization methodology, data and tools
The impact of emissions on regulatory compliance and several additional criteria such as population exposure
and costs of control are extremely complex, non-linear, dynamic, and spatially distributed,
driven to a considerable part by (uncontrollable) meteorological phenomena.
This rules out the use of classical, gradient-based optimization.
The alternative is a multi-stage, basically heuristic approach to multi criteria optimization
of non-differentiable systems. The basic step, in a simplified presentation are as follows:
- We define a set of measurable criteria that describe the decision problem:
in short, we want to improve compliance, minimize exposure, and keep the costs
of control measures low. This defines the preference structure,
the goals and objectives and the constraints in terms of measurable criteria.
- The next steps identifies possible strategies to meet these objectives;
this consists of two related components:
The first component consists of the structuring the emission inventory to define
individual emission sources and groups of similar (in terms of applicable control measures)
- the identification of controllable sources of emissions;
- the identification of measures, policies and technologies that can be applied
to these sources to reduce emissions.
The second component is the identification of possible emission control measures
in terms of their efficiency, costs, and applicability to these sources.
With these two components, we can now define optimization scenarios
that consists of
- combinations of emission sources and control measures
- within a framework of:
- preference structures,
- scenario specific choices (period, global constraints, maximum number of runs,
minimum number of feasible solutions, specific preferences (weights)
for the use of individual technologies).
- The scenarios now apply the selected measures to their respective emission sources
or source classes, and evaluate the system performance (emissions and costs) for these combinations.
These primary optimization scenarios now generate sets of feasible (which include possibly
unconstrained) solutions. These solutions are further analyzed and classified in the
DMC (discrete multi-criteria DSS).
By normalizing all solutions between NADIR and UTOPIA (the theoretically possible worst
and best possible outcomes) each candidate solution can be ranked
according to its (multi-dimensional Euclidean) position between
NADIR and UTOPIA: the closer to UTOPIA, the better.
With the DMC tool, the set of candidate solutions (to the primary emission reduction
problem) can now be structured by:
- Toggling on or off (including or excluding, considering or ignoring) any one of
the superset of criteria from the original preference structure;
- Introducing ex post constraints, i.e., moving upper and lower bounds
on any and all performance variables (criteria) to define sub-sets of candidate
solutions that better correspond to the objectives of decision makers, stakeholders, their proxies;
- Defining an alternative reference point: the reference point is the target goal of the optimization.
The default reference point is UTOPIA. This target can be modified by switching on or off any
of the criteria (dimensions), or by moving the reference point along any or
all of the dimensions considered: this, indirectly, defined criteria specific preferences or
weights without having to formulate explicit trade-offs or equivalences.
The trade-off between criteria is expressed by defining a most attractive (
if unattainable) solution.
The definition of constraints and a reference point allows to filter the set of candidate
solution into feasible (meets all criteria) and infeasible (violates one or more of the constraints)
and non-dominated (pareto optimal) and dominated solutions (for definitions, see below).
The most promising (feasible and non-dominated) candidates can now be subjected
to a final step in the analysis: their resulting emission scenarios are used with
the fate and transport model (for the corresponding meteorological scenarios implied
by the period selected) to test the impacts (performance variables or criteria)
after adding the additional effects of transport and dispersion.
The resulting solutions can now again be structured/filtered by the DMC DSS tool
to arrive at one or more final compromise solutions as the basis of any air quality management policy.
The "measure of success" is the comparison of the improved regulatory compliance
(and all other impact related criteria) with the baseline scenarios and
monitoring data analysis for a baseline period.
This provide a clear and quantitative measure of achievable improvement