Rudyard Kipling’s description of Great Britain’s painful experience in the Boer War could easily have been written for Europe’s troubled recent experience in vehicle emissions control, air quality and associated public health.
The forty million reasons are the forty million diesels still circulating on the roads of the UK and Europe, emitting more than treble the regulatory limit for nitrogen oxides (NOx) in real-world operation1. Of these, an estimated eight million are Euro 6 diesels. None of these are restricted from a single city in Europe, or taxed in a penal way. In Germany, rather than legislating to clean them up, the government has legislated to prevent them being targeted2. The original source of the problem – the European Union regulatory framework – has been directed not at cleaning up this lasting problem, but only at new cars. NOx emissions from this latest generation of vehicles – certified under the Real Driving Emissions (RDE) regulation – account for just 1% of the total, which leaves the 99% coming from previously sold vehicles. The centrally directed, improved, regulation has hardly touched the problem.
Unless these older vehicles are cleaned up, poor air quality will persist. With the average lifespan of a vehicle in Europe being 14 years, and high-emitting vehicles having been sold through to at least 2018, the problem is not about to disappear automatically through fleet turnover. A particularly egregious example is the Mercedes Citan 1.5-litre diesel van from 2019, which emitted 902mg/km of NOx in real-world driving – 8.6 times the regulatory limit – and which may remain on the road long after the UK’s 2030 ban on new internal combustion engines.
Nevertheless, it is not every single pre-RDE vehicle that is problematic. Contrast the Citan with the 2.0-litre Volkswagen Crafter from the same year, which emitted 53mg/km or 58% below the limit, to see that not all vehicles need restricting to address air quality today. Just over a third – 36% – of all vehicles on the UK roads, for example, create an estimated 87% of all NOx emissions, according to our modelling. A third of all emissions are accounted for by the 12% of dirtiest vehicles. In principle, the most efficient, fairest and least distorting way to bring air quality into compliance is to target those vehicles first.
Rather than focusing on making cleaner vehicles even cleaner, Emissions Analytics has been involved in a project for the last three years to test and rate all vehicles currently on the road, to get to the truly dirty vehicles. The project has been led by the AIR Alliance, and has brought together many parties in the spirit of co-operation. Emphasising the important, urgent and persistent nature of the problem was the landmark legal ruling in the UK in 2020 that the death of Ella Kissi-Debrah can be partially attributed to road pollution4.
Contrary to reports, the arrival of battery electric vehicles has not cured urban air quality problems. Perhaps they will, but market penetration of these vehicles will probably not be high enough until the next decade to make a measurable difference. In the meantime, legal contraventions of air quality standards will continue and can only realistically be addressed by cleaning up the existing fleet of internal combustion engines. Add to this the Covid-19 effect, which has seen a robustness in the used car market5 and early evidence of a switch away from public transport. As a result, older, high-emitting diesel vehicles could become more valuable and be kept on the road longer than expected.
Emissions Analytics is one partner of many in the AIR Alliance. Its first act was to bring together top experts and academics, including Dr Norbert Ligterink from TNO in the Netherlands, Professor Helen ApSimon from Imperial College London, Dan Carder from West Virginia University Institute of Technology in the US – who was responsible for uncovering Dieselgate – and Dr Xavier Querol from IDAEA-CSIC in Spain. The group is chaired by Dr Marc Stettler from Imperial College London.
In contrast to many one-off or time-limited initiatives researching this area, Emissions Analytics and AIR Alliance decided to create an on-going programme of testing and rating vehicles. Furthermore, they wanted to create a legal basis for independently rating urban NOx emissions. This was achieved through the publication of a standardised method – CWA17379 – as the result of almost eighteen months work through the Comité Européen de Normalisation (CEN) together with the automotive industry, cities, academics and lobby groups. From this, the AIR Index (www.airindex.com) was born.
How is all this different from the regulations? It is a voluntary movement bringing together a wide range of parties from different points of view interested in genuinely solving the urban air quality problem. Where the rules of multinational organisations have failed, on-the-ground co-operation has created a real-world ratings system with a legal basis.
With the standardised method coined, data was gathered from multiple sources. All data was subjected to strict quality criteria and tested against dynamic boundary conditions – including the speed and acceleration of driving – to test for validity under the method. Data flowed not just from Emissions Analytics, but from authorities and even manufacturers themselves. Results fully compliant with this method are live and free to access on the AIR Index website. In total, well over a thousand models have been tested and four thousand hours of data collected, to the value of over $16 million.
Despite all this data, the challenge with any system that is not by government fiat is achieving coverage of the whole vehicle parc. It is easy to test a representative sample of vehicles, but not to cover the long tail of models. The AIR Index has not achieved that yet, but is now close. In fact, ratings are already being published for 34,575 model variants.
With such a large repository of data, it has been possible to model and predict the emissions of vehicles similar to those already tested. Using machine learning techniques it is possible to predict with an impressive degree of accuracy the missing results. The model is trained on all the PEMS data drawn from the multiple sources. The validation set is then the fully compliant CWA17379 results, which are of course excluded from the training set. Using this approach, the average error between compliant and modelled values for distance-specific NOx is just 2.6%3. As a result, a large majority of predicted ratings can be validated through blind testing to deliver the correct AIR Index rating on its ‘A’ to ‘E’ scale. For clarity, fully compliant tests and modelled results are always clearly labelled and disclosed as such. If manufacturers, or others, disagree with their predicted ratings, conducting a full test to CWA17379 would lead to the result over-riding the predicted value.
By combining the fully compliant and predicted ratings, it has been possible to create coverage of at least 90% of Euro 5 and Euro 6 models, as shown broken down by powertrain in the table below. These figures include all manufacturers except very low volume niche producers.
Further testing planned may move total coverage to 95% soon.
The result of this work is that there is now a parallel system to the official, centrally directed, certification system that can be adopted by cities across Europe. As a result, cities now have a live, practical tool that can be put immediately to the task of managing the existing fleet to bring NO2 concentrations into compliance.
Analysing the results, we can identify the greatest sources of NOx emissions on UK roads, segmented by Euro stage and AIR Index, as shown in the table below.
Therefore, there is a group of Euro 5 vehicles, with an ‘E’ rating on the AIR Index, that accounts for a third of total emissions, while only representing 12% of vehicles on the road. London’s Ultra Low Emission Zone (ULEZ) has been relatively successful because it has applied an access charge for Euro 5 diesels and earlier, but this is not the most efficient solution as some clean cars are restricted. Were the ULEZ to be based on restricting only vehicles with ‘D’ and ‘E’ ratings on the AIR Index, NOx could be reduced by a further 6% points, while charging 2% fewer cars.
The group of pre-RDE Euro 6 diesels with ‘D’ or ‘E’ AIR Index ratings is smaller than the Euro 5 group, but still important at 9% of total emissions. What is even more significant about these vehicles is that they form the ‘swing group’ that can make the difference between air quality non-compliance and compliance.
In a research report from January 2020 from Imperial College London6, the AIR Index emissions values were combined with standard dispersion models to show the impact of different ULEZ strategies on ambient NO2 concentrations and the number of legal excedances. If the current strategy were applied to the enlarged ULEZ it was shown to bring most but not all locations into compliance with the 40μg/m3 limit, compared to a majority being non-compliant prior to the ULEZ. Switching to ULEZ restrictions based on ‘D’ and ‘E’ on the AIR Index scale, irrespective of Euro stage, was shown to be sufficient to bring all areas into compliance. In other words, the pre-RDE Euro 6 diesels make the difference.
Emissions Analytics is pleased to be involved in this practical, decentralised, collaborative, ground-up attempt to resolve one important element of air quality. On-going testing will move the project closer to complete coverage, and it is set up to welcome an increasing number of data partners. As the supply of data increases, so the accuracy of the database improves.
The only element missing is an honest reckoning by the authorities as to the systemic weaknesses that led to these problems. Without that, the weaknesses may resurface. For the sake of Ella, heed perhaps should be taken of Kipling:
Let us admit it fairly, as a business people should,
We have had no end of a lesson: it will do us no end of good.
Footnotes:
Transport & Environment, Cars with engines: can they ever be clean?, September 2018
Efficient control of NOx from diesel cars and ULEZ reduction of NO2 concentration, ApSimon, H., Oxley, T., Mehlig, D., Woodward, H., Stettler, M., Molden, N.