Mais conteúdo relacionado
Semelhante a Air Quality Modelling of new and Emerging Vehicle Technologies- What can they Deliver in Scotland? (20)
Air Quality Modelling of new and Emerging Vehicle Technologies- What can they Deliver in Scotland?
- 1. 1© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
Clean Air Zones in Scotland, what
can current technology deliver?
Scott Hamilton, PhD
Nicola Masey, PhD
Ricardo, UK
Dispersion Modellers Users
Group conference, London, 2018
10m resolution air quality model of most of Scotland’s urbanised areas >30,000 km2
- 5. 5© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
Topics for this talk
• No legislation (its almost home time after all)
• Main focus is large scale/high resolution air quality
modelling.
• Technical description of the RapidAIR model
• How it can be used to model big policy
interventions (like CAZs!)
• A case study I modelled in our RapidAIR system
for this event- which hopefully shows the benefits
of a national scale high resolution model set up
consistently for all areas
• “What is the best we can expect from available or
soon to be available engine technologies?”
• Some thoughts/explanations on the findings
- 6. 6© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
Caveats / limitations
• Average speed based emissions model
(COPERT5)
• Meteorology is steady state obviously not the
case for such a large domain
• Assumption that emerging engine tech will
reduce emissions as forecast
• A very simple view of the future – no change in
other source types
• Main focus is PM10 because of our really low
annual mean standard- also we are only
modelling primary emissions of PM
• Road abrasion, brake wear, tyre wear- very
simple treatment
- 7. 7© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
What is RapidAir® ?
Dispersion model which automates much of the workflow for
dispersion modelling for road sources in large areas.
Some neat things it does:
• Traffic emissions model (COPERT5)
• Road dispersion model (based on AERMOD)
• Methods developed from USEPA guidance
• Street canyon options (with auto allocator)
• Area source model (AERMOD)
• Unlimited domain size and number of sources
• Met data- automation of sourcing / handling
• Automatic integration of background values
(in the UK)
• Model calibration and scaling
• Lots of utilities
• NOx NO2 chemistry options (Defra, Jenkin)
• Interactive plotting
• Automatic reporting (… of the boring stuff)
• Runs on any modern OS
• Its REALLY fast
https://ee.ricardo.com/air-quality/city-scale-air-quality
- 8. 8© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
RapidAIR UI- runs on any office computer
• RapidAIR uses open source
python libraries- typical city
run takes a few min (Win, i5,
16GB RAM)
• Most processes are
automated by controllers
written in pure python
• Pandas, GDAL, Numpy,
python 2.7 standard library
• The UI is developed in the
popular QT framework (QGIS
is built in this too)
We use the Continuum Analytics Anaconda python distribution.
This has transformed the way I work! https://www.anaconda.com/download/
This distribution has become something of a standard in scientific computing, in areas like machine
learning, astronomy, meteorology and other data intensive scientific disciplines.
- 9. 9© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
Data
RapidAir NO2 model
RapidAir NO2 in a GIS
RapidAir NO2 in a GIS
RapidAir NO2 in Google Earth
RapidAir NO2 in Google Earth
Mostly it’s a dispersion model we’re using in large
urban areas- this is London
- 10. 10© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
UK Projects- example
Ricardo recently developed a
RapidAir model for Fife Council. The
project was funded by the Scottish
Government air quality grants
scheme.
The model has a resolution of 3m
(>300million prediction points) and
covers the whole Kingdom. Data
products include common GIS
formats, Google Earth layers,
interactive report including OpenAir.
- 11. 11© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
NO2 annual mean concentrations in building footprints, 2008
UK example, concentrations in building footprints
- 12. 12© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
BAU POLICY
Beijing NOx from road traffic, 2014
International example, testing the effect of city policy
- 13. 13© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
Scotland is embarking on the
Clean Air For Scotland
initiative.
So we modelled some
technology enhancements to
get a feel for what they might
deliver in reducing NO2 and
PM10
- 14. 14© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
What if Scotland was a Low Emission Zone?
• New air quality modelling techniques allow us to very quickly look at very large
areas of the country in a single model run. For the purposes of today I modelled
a high level “what if” scenario for most of Scotland which is designed to
stimulate discussion around what engine technology improvement can deliver in
the next decade or so
• The case I’m going to show you is not concerned with detailed LEZ schemes or
boundaries, but simply asks the question, what’s the best we can expect from
currently available, or soon to be available engine technologies?
• Baseline is the 2016 fleet mix from the DfT, including fuel use split, technology
mix, vehicle weight categories, Scotland specific information where applicable.
• Test case 1 involves setting all light and heavy vehicles to engine technologies
which are either best currently available, or best that will soon be available (e.g.
Euro 6 D)
• Test case 2 asks what we could expect from a radical reduction in diesel
vehicles in the light fleet (to zero!)
- 15. 15© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
Some important clarifications…
• The modelling in this talk took less than two days and
was done especially for this and another talk.
• It is based on openly available data so a lot of effort was
avoided
• This is mainly to show what can be done with modern
dispersion modelling methods to get fast answers to wide
ranging policy measures
• A key benefit is complete consistency between towns and
cities as they are in the same model run (its also really
fast)
- 16. 16© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
What if Scotland had a single very large LEZ?
Model domain
• 400 million discrete points
• 10m resolution
• Stretches from South Ayrshire in
the south west to Aberdeen in the
north East
• Modelled in our RapidAIR suite-
run time ~150 sec
• Emissions modelled in our
RapidEMS module (140,000 links
in the UK wide model, run time a
few seconds)
• In Scotland we have a ‘special’
interest in PM10, out standard is
18 µgm3
- 17. 17© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
Emission modelling- NOx
All emissions were modelled in our RapidEMS system, which uses the COPERT V coefficients in
a computationally efficient manner. There are 140,000 road links in the model (though most of
these are in the rest of the UK)
- 18. 18© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
Emission modelling- PM10
All emissions were modelled in our RapidEMS system, which uses the COPERT V coefficients in a computationally
efficient manner. We can calculate emissions on 1 million road links in 1 minute.
There are 140,000 road links in the model (though most of these are in the rest of the UK)
- 22. 22© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
Results at roadside- concentrations in a transect
This tool in QGIS is sampling
the road concentration grid
from RapidAIR that match the
transect
- 23. 23© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
NO2 cross road profile of concentrations- M74 in Glasgow
baseline
best available tech
BAT plus no diesel LDV
Exposure zone
~150m
0
80
- 24. 24© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
NO2 difference plot- central belt (base minus BAT)
NO2
reduction
ugm3
Very highest concentrations are attenuated by up to 28 ugm3, more typically
roadside concentrations reduced by about 5 to 10 ugm3.
- 25. 25© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
NO2 results- explained
• New iterations of Euro 6 are expected to deliver very substantial reductions in
NOx and hence NO2
• Traffic is usually the dominant source so reductions in emissions in this sector
are very apparent in overall concentrations
• Hence if the new Euro standards deliver, significant reductions in NO2 should
follow
• Going beyond BAT towards removing diesel completely from the LDV fleet
delivers more benefits, but these are quite marginal compared with “factored
in” improvements that should follow Euro 6 improvements
• What is remote sensing telling us about NOx emissions from the UK road
fleet?
- 30. 30© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
PM10 cross road profile of concentrations- M74 in Glasgow
baseline
best available tech
BAT plus no diesel LDV
Line is obscured
Exposure zone
~150m
- 31. 31© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
PM10 difference plot- central belt (base minus BAT)
PM10
reduction
ugm3
Very highest concentrations are attenuated by up to 5 ugm3, more typically
roadside concentrations reduced by about 0.5 to 1 ugm3.
- 32. 32© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
PM10 results- explained
• New emission standards will not deliver significant reductions in exhaust
emissions compared with existing vehicles
• The biggest road traffic component in future years will be tyre, brake and road
wear particles (in Scotland its already 70% or more of the total). I don’t foresee
a step change in the prevalence of these technologies. Electric cars still need
roads, tyres and brakes after all.
• Traffic is usually not as dominant a source so reductions in emissions in this
sector are not as apparent in overall concentrations.
• Hence even for ambitious LEZ schemes we might not expect to see much
change in PM10 concentrations- though we would expect some benefit
• Going beyond BAT towards removing diesel completely from the LDV fleet
does not change PM10 concentrations much at all
- 33. 33© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
PM10 results- problems ahead for policy makers….
• The most uncertain components of PM emissions are
becoming the most important
• Effort is needed to improve the emission factors for brake,
tyre, road wear, resuspension
• The emission factors need to be localised- the same values
(usually EMEP CORINAIR) are used in Athens as in
Inverness
• The UK as a whole needs to understand road dust
resuspension better. Again, is it plausible that every city has
the same dust loading per m2 of road? Rainfall varies a lot
in the UK….
• In Scotland there is a particular issue with particulates. Our
standard is the lowest in the world, but we understand very
little about some of the main emission sources driving
concentrations. Engine technologies will help NOx/NO2,
particles not so much.
- 34. 34© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
https://www.informationisbeautifulawards.com/showcase/1614-3d-visualisation-of-air-pollution-in-glasgow
http://www.trudihannah.com/#/pollutionvisualisation/
Results from a Ricardo collaboration with
Trudi Hannah at Glasgow School of Art.
- 35. 35© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
To conclude
• New modelling methods allow policy that affects large areas to be modelled in a
holistic and consistent manner. Its also very time efficient, both in computation and
man hours.
• The case study prepared for this event suggests the following:
• Plausible improvements in NOx from better vehicles has the potential to deliver
quite large reductions in NO2 in our Scottish towns and cities. This is mainly
because road traffic NOx is still the dominant source driving high concentrations
• Going beyond BAT and removing diesel vehicles delivers additional gains but
these are more marginal for NOx
• The same vehicle improvements reduce PM10 as well but by a smaller amount
• The reduction in PM10 is still significant given the health benefits
• As exhaust emission reduce the non-exhaust component becomes dominant. It
is possible therefore that Scotland could still see PM10 concentrations over the
18ugm3 threshold, with the emissions arising from brake, tyre and road particles.
• Even with electrification, if material science and automotive design can’t reduce
these uncontrolled releases we may still have a domestic compliance issue for
PM10 in Scotland.
• Localisation of emissions of NOx and PM appears to be necessary given the
insights from remote sensing and the importance of non-exhaust emissions
- 37. 37© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
Scott Hamilton, PhD
Technical Lead, Air Quality Modelling
Ricardo Energy and Environment
scott.hamilton@ricardo.com
- 40. 40© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
CMAQ NO2
City RapidAir® model
Neighbourhood RapidAir® model
Street Level RapidAir® model
National scale CMAQ model
Nesting RapidAir® in CMAQ model- UK example
Boundary conditions to the city scale model come from
CMAQ- meteorology from WRF can be used in both for
consistency
- 41. 41© Ricardo-AEA LtdRicardo Energy & Environment in Confidence
RapidAIR also includes ‘Notebooks’
• The Jupyter notebook format is
fuelling something of a revolution in
sharing scientific work. Jupyterlab is
the latest iteration, its amazing.
• We can use the technology almost as
a retraceable blockchain of steps
taken in a modelling project
• We use the format to provide worked
examples and training in using the
model
• The notebook has cells, each of which
contains either text (as Markdown),
python code, or outputs from running
the code. Videos can be embedded,
as can hyperlinks.
• An entire workflow can be saved and
shared in seconds in a completely
reproducible manner
Code execution
Output beside the code that created it, so
reproducibility is enhanced