Towards Zero CO$_{2-}$, NOX- and PM$_{10-}$ Emissions by Passenger Cars: Technology & Behaviour

Climate change is a complex biophysical/biochemical process, which is still not fully understood. Most researchers acknowledge that the natural balance between sources and sinks of greenhouse gases is increasingly out of tune due to steadily increasing emissions of the various greenhouse gases by human activities [1]. The paper will deal with a major source of greenhouse gas emissions: The emissions are from combustion of fossil fuels in internal combustion engines (ICEs) of passenger cars. Hence, it deals with exhaust emissions and not with emissions of particles from wear of car brakes or tyres, which are frequently overlooked.

Road transport has a share of well over $25 \%$ in the final energy consumption in countries belonging to the Organisation for Economic Co-operation and Development (OECD) [2]. Passenger cars (Table 1) dominate road transport in terms of vehicles and kilometre (km).

Currently nearly all cars have engines that run on fossil fuels. This is a rather inefficient way of converting energy into motion because only $12-30 \%$ of the combusted fuel is available as kinetic energy [3]. The remainder is wasted as heat and gaseous pollutants, in particular $\mathrm{CO}_2, \mathrm{NO}_{\mathrm{x}}$ and $\mathrm{PM}_{10}$. $\mathrm{CO}_2$-emissions are directly linked to climate change [4]. $\mathrm{NO}_{\mathrm{x}}$ and $\mathrm{PM}_{10}$ are key air pollutants that have a negative impact on human health [5] and nature [6].

There is a relation between climate change and air pollution via fuel consumption [7]. This explains why regulators require car manufacturers to reduce both [8]. It also explains our interest in ways to reduce these pollutants simultaneously.

The relation between fuel consumption (reduction) and $\mathrm{CO}_2$-emissions is linear, but that is not the case for $\mathrm{NO}_{\mathrm{x}}$ and $\mathrm{PM}_{10}$. A reduction of $\mathrm{x} \%$ in fuel consumption may then lead to $\mathrm{x} \%$ reduction in $\mathrm{CO}_2-, \mathrm{y} \%$ in $\mathrm{NO}_{\mathrm{x}}-$ and $\mathrm{z} \%$ in $\mathrm{PM}_{10}$-emissions. This of course complicates (technical) emission reduction strategies.

Climate change mitigation means reducing $\mathrm{CO}_2$ and other greenhouse gases. Options to achieve this are in particular [see e.g., [9]]:

(1) Reduce the total number of car km in a year by reducing the number and/or length of car trips. Typical options are

(a) Reduce the need to travel: Work from home and use electronic means to stay connected;

(b) Travel together by car (share a car);

(c) Travel differently: Use public transport or active modes (walking, cycling).

(2) Increase the fuel efficiency of cars;

(3) Increase the use of low-carbon fuels;

(4) (Partial) replacement of carbon-based fuels by electricity (from renewable sources);

Changes in behaviour may be voluntary or mandatory. Technical progress, business mobility policies (no leased cars, business travel cards) and integrated government policies (spatial planning, investments in infrastructure and public transport, traffic management and mobility management) may support such behavioural changes.

The purpose of the paper is to discuss a series of options to reduce emissions of passenger cars to zero in The Netherlands by 2030 (reference year) compared to the base year 2010. The paper explores many of the recommendations for further research of our earlier paper [10].

The following questions were addressed:

1. How much $\mathrm{CO}_2, \mathrm{NO}_{\mathrm{x}}$ and $\mathrm{PM}_{10}$ did passenger cars emit in 2010?

2. How much will this be in 2030?

3. What would these figures be if electric cars become mainstream in 2030?

4. What would the impact be of sustained urbanization on these emissions?

5. How would a greener power mix in electric power plants affect the emissions of $\mathrm{CO}_2$, $\mathrm{NO}_{\mathrm{x}}$ and $\mathrm{PM}_{10}$ by electric cars?

The Netherlands is a densely populated small country; in 2010, it had 16.6 million inhabitants and an average population density of around 500 inhabitants per $\mathrm{km}^2$. Many large cities have 5,000 or more inhabitants per $\mathrm{km}^2$ [11]. Car mobility approached 98 billion km driven with 7.6 million cars in 2010 (driver only, add $40 \%$ for passenger km ). In comparison, there were 17.13 million inhabitants and 8.2 million cars in 2017 [12]. The number of car km was the basis for the fuel and emission calculations in this paper. Relevant estimations can be found in Table 1.

To estimate car km in the year 2030, a set of assumptions was applied. There is a direct relation between individual wealth, car ownership and car use. The first assumption is therefore related to the average individual household income. We assume that there is no net economic growth plus a decrease in purchasing power of an average consumer household in the period 2010-2020. The arguments are that the economic downturn of 2008-2014 and recovery since 2014 balance each other out, while an economic cool-down should also not be ruled out around 2019-2020. The number of households also continues to increase and hence less income per household. These factors stabilize car use in 2010-2020. Second, an annual growth in car km of $2 \%$ is expected for 2020-2030. Third, the division of engine types stays the same. Fourth, cars with ICEs still dominate in 2030. This scenario was input to our simulation model (see Section 3), which has shown in Table 2. The lack of electric cars in this table seems strange given the values in Table 3. The explanation for this outcome can be found in Section 2.1.1.

Assumptions regarding urbanization will be added in Section 4.

There were no electric vehicles (EVs) in The Netherlands in 2010. This changed after a new national tax regime and public and private initiatives to provide (a growing number of) charging facilities. By 2016, 86,200 EVs were in use. This small number can be explained by three factors: Price, low-density charging network and limited range.

The normal purchase price of an EV is way above the price of a similar non-electric (conventional) car. For private owners, this is too expensive, while the tax regime in The Netherlands favours business owners; they own $99 \%$ of all EVs. Business users tend to drive (much) more km than private users. The range of an average fully electric vehicle (FEV) (50-100 km per full charge) is very small compared to an average car with a conventional engine ($700+$ per full tank of fuel). This creates range anxiety: will the car battery last until the next charging point? Those who buy an EV favour a dual engine design (diesel-electric or petrol-electric). This is usually a plug-in hybrid electric vehicle (PHEV) (Table 3).

PHEV have lower $\mathrm{CO}_2$ - but higher $\mathrm{NO}_{\mathrm{x}}$ - and $\mathrm{PM}_{10}$-emissions than very fuel-efficient petrol cars [15]. The subsidy for PHEV could be regarded as ineffective.

To estimate the corresponding tail pipe emissions, car km given in Table 1 were multiplied by the average fuel consumption and emission factors per fuel type. Table 4 provides the emissions on a tank-to-wheel (TTW) basis.

Policymakers like to have economic growth and lower emissions. Such a decoupling of economic growth and emissions has taken place as technical progress allowed a reduction in energy intensity of many products and increase of energy efficiency of (industrial) activities in advanced economies. The decoupling had a relative impact because the consumption of fossil fuels still grew, but at a lower pace than the economy. There is also a major caveat, as part of the lower growth in fuel consumption can be explained by outsourcing of production to low-wage countries, which are among the countries with the fastest growing economies and growing users of fossil fuels and emissions. A relative reduction still means an increase in greenhouse gases. For climate change, an absolute reduction is vital [16].

In the above estimations, we assumed no decoupling for car use; km and emissions grow with the same 21.9%. This seems conservative given that the latest model-based estimations indicate that $\mathrm{CO}_2$-emissions of an average new car have been reduced from 181 to $106 \mathrm{~g} / \mathrm{km}$ between 1998 and 2016 [17], while EU regulators aim to reduce these even further. Technical progress seems to work then or does it not? An important caveat is that fuel consumption and emission models cannot replace real measurements. What further complicates the situation is that there are many other interacting factors and developments:

• Higher safety standards lead to larger and heavier cars, which is not compensated by the use of lighter car parts [18];

• A demand for more luxury accessories leads to more on-board consumers of electricity, in particular air-conditioning, entertainment and navigation systems [18];

• Ageing of the car fleet extends replacement intervals: 80% of the cars sold yearly in The Netherlands are used [19];

• EU standards for fuel consumption and emissions and measurement procedures were developed in co-production with major car manufacturers. Tests could be manipulated. This explains a gap of 35% between test cycle data and data from real driving tests; hence, published fuel efficiency and emission data are too low [17].

The earlier estimates for 2030 can be explained by the growth of the car fleet plus these factors. The future is of course uncertain. As various examples have shown, technically fuel use and emissions of an average car can be much lower than the actual values, but will this also happen given these counteracting developments?

With this analysis, questions 1 and 2 have been addressed.

The urgency of $\mathrm{CO}_2$-emission reduction becomes clear from the UN conference in Paris in 2016. A total of 195 countries signed the Paris Agreement. It sets targets for national $\mathrm{CO}_2$-reduction policies in order to prevent the threat of a prolonged rise of the global average yearly temperature of $2 \%$ above pre-industrial levels, an out-of-control scenario. A reduction to $1.5 \%$ is advised to prevent islands from being taken by the oceans [20]. The world is now approaching the (short term) $1 \%$ average growth [21]. It may reach $2.7 \%$ in a business as usual scenario [22].

$\mathrm{CO}_2$-emissions from transport are not the only source of climate change (Table 5). There are also methane emissions from agriculture and production and consumption of natural gas. The Netherlands is also a major trader in oil derivatives.

It becomes apparent that the growth of the population, the economy and international trade lead to higher $\mathrm{CO}_2$-emissions of human activities until 2010. Around that year, the earlier mentioned decoupling of economic growth and emissions must have started.

Since the 1980s of the past century, the political landscape has changed considerably in many countries in Europe, including The Netherlands. Before this period, the political majority and most people were in favour of a substantial role for government in Dutch society. Both the national and the provincial governments were involved with spatial planning and care for the environment. Then, a shift to the right took place, which led to a smaller role for government and a larger role for free market competition. This political wave has been labelled neoliberalism, to distinguish it from the classic liberalism movement, which happened more than a century ago in this and other European countries, when many monarchs lost their position and the national states were constituted in Europe [24]. This 'business first' policymaking has had profound implications for the environment. Major cuts were made in budgets for nature and environment, including research and protection [25]. Subsequent ministers of the Ministry of Transport (& Environment called officially) expanded the highway network and increased maximum speed from 100 to $120-130 \mathrm{~km}$ on many highways. This Ministry tried to remove environmental zones and $80-\mathrm{km}$ speed limits on main roads crossing cities, arguing that these zones were limiting business and led to more congestion and air pollution. This argument was proven to be incorrect by research [26]. The local measures remained after it lost the legal battle with local governments and citizens living alongside these roads.

Air quality in The Netherlands has improved in the past decades due to de-industrialization and cleaner technology in many sectors. Yet, the country still scores average on air quality in Europe. The wind transports pollution from other countries, the North Sea and sea shipping. This background level cannot explain the rather poor air quality along many streets in major cities, however [27]. Cars and other motorized transport are its most likely source.

The Agreement [28] means that The Netherlands has to reduce its $\mathrm{CO}_2$-emissions by $50 \%$ by the year 2030 and even $85-95 \%$ by 2050 in a $2 \%$ scenario. In a $1.5 \%$ scenario, even a $100 \%$ cut is needed; all values compared to the 1990 situation. $\mathrm{CO}_2$-emission reduction targets influence all economic activities, but the implications per economic activity may vary, depending on the actual share in emissions, the way public and private decision-makers deem it necessary to set a lower or higher target for specific activities and the time frame in which such transformation should take place. Politics is a matter of balancing interests and taking responsibility. Who do politicians actually represent in what future analysts looking back probably would call 'Climate Gate'?

The climate problem has a long history. Table 5 shows that it took many years to substantially reduce or stabilize $\mathrm{CO}_2$-emissions by human activities.

The Paris Agreement asks for a (complete) decarbonization of the economy. This means heating and cooling without natural gas, differential pricing for energy and electric road transport [29]. Such a transition is very complex and expensive considering the short time span and until now very modest and fluctuating financial incentives by the national government, which creates uncertainty for long-term investors [30].

The following research tools were used: desk research, a dedicated simulation model, scenarios (Section 4) and (quantitative) evaluation (Section 5). The scope of the article by Vleugel and Bal [10] was broadened to allow a more complex and larger scenario experiment. That experiment was the basis for the simulation model. These changes had profound implications. The research scope and questions were expanded. Some new input data were needed. The structure and functionality of the simulation model were adapted. Finally, the evaluation section and conclusions changed.

An MS$^©$ Excel$^©$ model was built to estimate the energy consumption and emissions of a fleet of passenger cars with different engine-fuel types. The model calculates total fuel consumption and emissions based on the estimated number of km driven per year, emission factors and average fuel consumption per km . It consists of the following items:

- A data entry and calibration module. This allows changes in user data (e.g., composition of car fleet, fuels used) and technical data (see Table 6 for a subset, vehicle brands and types);

- In the solver module policy, scenarios are entered as constraints in a linear programming exercise;

- Matrices with data about fuel consumption, emission factors (EF) and TTW-wheel-to-wheel (WTW) conversion;

- A choice box to allow estimation of the impact of different degrees of urbanization on the number of car km driven;

- A choice box to estimate the impact of a different power mix to charge EV batteries;

- A choice box which adds dynamism in terms of growth of the car fleet and structural change in the power mix ('greener mix').

Numerous academic and professional data sources were combined as in the earlier papers. This prevents reliance on biased data from car manufacturers or governments.

The averages given in Table 6 are key data to estimate the impact of scenarios on emissions (Section 4). For a complete picture, from here on, these emissions will be expressed on a WTW basis. TTW values were multiplied with a 'WTW-factor' [32]. No such multiplier was found for $\mathrm{NO}_{\mathrm{x}}$ - and $\mathrm{PM}_{10}$-emissions in the literature. $\mathrm{CO}_2$-multipliers were used to adapt all TTW values.

The value of scenario experiments is that policy packages of different composition can be tested without the complexity and costs of in vivo experiments. Scenarios with substantially more FEV, urbanization and a greener power mix are fed into the model to estimate changes in car km, fuel consumption and emissions.

Question 3 is about the impact of substantially more EV on $\mathrm{CO}_2$-emissions against the background of a $-50 \% \mathrm{CO}_2$ target for 2030 (Section 2.4.4). To answer this question, scenarios with following assumptions were tested in the model:

- A uniform reduction % in $\mathrm{CO}_2$-emissions for all activities;

- A substantial share of FEV in 2030. They replace petrol, diesel and LPG in each scenario, but to a different extent. In all scenarios, FEV are powered by electricity from a mix of fossil and renewable sources (hence the term 'grey power mix');

- No PHEV because of their emission profile (see Section 2.1.1);

- (Partial) phasing out of ICE in 2030;

- Phasing out of fossil fuels in 2030;

- Continuation of growth in car km (21.9% between 2010 and 2030);

- No significant improvement of fuel efficiency at car fleet level between 2010 and 2030. Fleet replacement is a slow process, and it is unclear what the next technology cycle might bring, when it is road-worthy and what its impact will be by 2030 .

Five scenarios were developed to allow an extensive sensitivity analysis (Tables 7 and 8): how does a change in fleet composition affect emissions of $\mathrm{CO}_2, \mathrm{NO}_{\mathrm{x}}$ and $\mathrm{PM}_{10}$?

Partial replacement of ICE (Scenarios A and B)

In Scenario A, policymakers aim to replace 25% of all cars with ICE by FEV. If realized, then $\mathrm{CO}_2$ - and $\mathrm{NO}_{\mathrm{x}}$-emissions will rise. This contrary effect can largely be explained by the growth of the car fleet, the grey power mix and the assumption that cars will not become more fuel-efficient. If $87 \%$ of the petrol cars were used privately (Table 1), then the businessoriented subsidy for the purchase of an EV would hardly impact the petrol car market. It might affect the diesel car market because $39.6 \%$ of these are owned by a business. Hence, expanding the current tax regime to private car owners would make FEV much more affordable for the general public; hence, the changes in fleet composition would become more feasible, provided that the current limitations of using an FEV (see Section 2.1.1) are resolved. Scenario B contains a faster replacement of petrol ICE by FEV. Its main impact is a reduction in $\mathrm{PM}_{10}$-emissions.

Double A (Scenario D)

Scenario D is more rigorous than Scenario A, but the reduction in $\mathrm{CO}_2$-emissions is limited. It is more beneficial to focus on a vast and rapid reduction of petrol ICE in favour of FEV (Scenarios B and C). Table 8 presents the results.

Full throttle (Scenario E)

Scenario E (Table 8) shows that it is possible to reduce $\mathrm{CO}_2$-emissions by more than $35 \%$ when only FEVs are in use on Dutch roads. Local emissions are also substantially reduced.

The results reflect the earlier mentioned non-linear relation between $\mathrm{CO}_2$ - and $\mathrm{NO}_{\mathrm{x}}$ - and $\mathrm{PM}_{10}$-emissions (see Section 1.1). Especially, the $\mathrm{NO}_x$ values are hardly reduced when many ICEs remain on the road and FEVs remain powered by grey electricity. Scenario E scores best in regard to the target for 2030.

This discussion addressed question 3.

Use of transport modes is a function of spatial density. In The Netherlands, people living in urban areas own less cars and drive less km than people living in rural areas [33]. Distances between activities are smaller in cities than in rural areas. Public transport services, cycling and pedestrian routes are also more prolific, at least in a country like The Netherlands. In urban areas, use of public transport or active modes does not take as much additional time compared to cars, as it would do in rural areas. Car use is also restricted in urban areas by traffic measures, fees and limited space for parking.

Scenario F is used to address question 4. It contains the following assumptions:

- The same car fleet as in Scenario E (Table 8);

- Urbanization reduces the growth in (driver-only) car km. Hence, in scenario 2030 F1 and 2030 F2, car use will be lower than in scenario 2030E but still higher than in 2010..

Demographic studies mention a continuation of urbanization due to migration from rural areas and immigration. The urbanization rate (the % of people living cities) for 2010 is an estimate based on reference NKWK [34]. This leads to the values given in Table 9.

As expected, urbanization helps to reduce emissions even further. This impact could be enhanced if the number of cars is reduced by shared car ownership [35]. This answers question 4.

The power mix (the actual mix of fuel sources [coal, gas, biofuel, wind and solar] used to produce electricity in power plants) is a complex topic. In the previous scenarios, a power mix corresponding to a $\mathrm{CO}_2$ emission of $464 \mathrm{~g} / \mathrm{kWh}$ was used as estimate, but this value may be too optimistic. The energy market is very dynamic, which leads to volatility in the power mix. Growing imports of cheap coal have replaced cleaner natural gas as fuel source in electric power plants. At the same time, there is more electricity from renewable sources, produced locally and imported. Renewable sources are not very well manageable due to the natural fluctuations in these resources. This explains why the power mix may change year by year. The following alternatives were explored (Table 10):

- Scenario G1: Scenario F1 (urbanization 55%) + 25% greener power mix;

- Scenario G2: Scenario F2 (urbanization 60%) + 25% greener power mix;

- Scenario G3: Scenario F1 (urbanization 55%) + 50% greener power mix;

- Scenario G4: Scenario F2 (urbanization 60%) + 50% greener power mix.

To reduce the earlier mentioned non-linear complexity, only the impact on $\mathrm{CO}_2$-emissions was estimated. A greener power mix is already in the pipeline due to public and private investments in wind and solar power in The Netherlands and abroad. A key issue is the stability of the grid. Renewable energy sources, like sun, wind and water, have a fluctuating production. They also cannot be stored in large quantities, yet. Energy management becomes key to its success [36]. This issue is outside the scope of this paper. A dependency on fossil fuels remains. Nonetheless, a greener power mix helps to reduce emissions further along the line of the previous scenarios.

Taking all previous measures to the extreme leads to a major reduction in $\mathrm{CO}_2$-emissions by passenger cars of -69% (2010 base) and 75% (2030 base) respectively. This answers question 5.

The ambitious $\mathrm{CO}_2$ mitigation target requires a complete replacement of the Dutch car fleet by FEV.

By considering the WTW impact of cars, in essence, the entire chain must be addressed. This indicates that a much higher effort is needed to reduce the total level of car emissions.

Finally, it is important to realize that emissions of other road users such as vans, trucks, buses and coaches should also be included in the analysis to understand the full scale of the climate change mitigation challenge as far as road transport is concerned.

A simulation model was used to quantify a rich set of scenarios. To achieve the -50% target for $\mathrm{CO}_2$-emissions for passenger cars, all cars should become fully electric, urbanization should continue as expected and the power mix should become substantially greener. This would require mass production of electric cars instead of the current small series. Still, it may take about two decades with the current average car life cycle of 18 years before full electrification could be achieved. Another issue that has not been mentioned is the impact on the electricity grid and the need to develop a large-scale charging network. The scenarios regarding urbanization are in line with expectations. But, local policymakers have small budgets for alternatives like public transport. Bicycle use is growing so fast that cities start to experience congested cycling lanes and lack of bicycle parking facilities. Hence, it remains to be seen if car km can be reduced and in what manner in and around cities. The fast greying of the population (not considered) is also a factor that will have its impact on passenger car km. A greener power mix is likely with current investments. National transport policy should become more environment oriented. An extension of decades of 'neoliberal' transport policy would mean more roads and more car km, voluntary transition to EV and lower budgets for alternative modes of transport. The feasibility of the extreme scenario is, therefore, rather uncertain, but less extreme scenarios also lead to a significant reduction in $\mathrm{CO}_2$-emissions.