Optimising Car Life for Minimum CO₂ Emission

This paper is founded on research done by the author for SAE-A's submission to a government committee looking to set Australia's long-term fuel consumption standards. The author was, in 2004, responsible for advising the government on the $2010 \mathrm{CO}_2$ standards for all light duty vehicles $[1,2]$. In that work, the lowest attainable $\mathrm{CO}_2$ emission at reasonable cost was determined. However, no consideration was given to any impacts of the non-fuel $\mathrm{CO2}$ contributions to whole of life emissions. The non-fuel-related $\mathrm{CO}_2$ emissions have grown from $16 \%$ in 1980 [3], but despite initiating policies to reduce manufacturing energy consumption [4], it had grown to $20 \%$ by 1997 [5] largely as the result of reduced vehicle fuel consumption in meeting government targets caused by perceived energy shortage and the 'energy crisis' initiated by the oil shocks of 1973 and 1979. Although attempts have been made to include manufacturing energy as a vehicle design objective, only small reductions in embedded energy appear to be possible[6].

Thus, today it now represents about $30 \%$ of total $\mathrm{CO}_2$ assignable to light duty vehicles and therefore cannot be ignored [7]. This proportion will increase as the fuel efficiency of vehicles continues to improve more rapidly than the remaining energy inputs (sometimes with an increase in manufacturing energy as with hybrids).

The object of this paper is to compare the fixed and variable energy content in passenger car life cycles and to determine what is the optimum service life to minimise greenhouse gas emissions. The problem is treated for the EU market, although the method will be applicable to other markets.

For this investigation, it is assumed that the only $\mathrm{CO}_2$-producing components are the manufacturing energy and embedded materials energy offset by recycled materials and parts, together with the in-service fuel and consumables sourced CO2-e. It is worth noting that the in-service consumables other than fuel are of second-order importance and their variation in consumption over the vehicle life or vehicle technology is assumed to be constant irrespective of age [5]. The assumption is made that the fuel consumption (use per distance travelled) does not vary as the vehicle ages [7].

Thus, the problem reduces to finding.

where $C O 2_{ {fuel }}(y)$ is part of a series in $\mathrm{CO}_2$ or fuel consumption that for new cars progressively that reduces (e.g. 3%/y) as manufacturers meet the requirements of reduced fuel consumption for compliance with EU regulations.

n is the number of years of service life before it is replaced by a new vehicle and $a$ is the number of life cycles c studied.

Accounted for in modelling is that in the early years of use vehicles typically travel more than in later life as found in the ABS SMVU [10]. This variation shown in Figure 1 is quite consistent with the mean travel and age of the EU population [8, 9]. Subjectively, this can be explained as many new cars are used for business purposes early in their lives, and as second or third hand cars used less for only local trips late in their lives when reliability and other defects make them less attractive for long journeys. Moreover, these data represent an average car, and thus include scrapping (retiring) cars from the fleet through crashes and then later owner decisions to send the car to recycling, which significantly increases after the median age.

The relative importance of the 'front end' manufacturing and materials energy or embedded energy (and corresponding $\mathrm{CO}_2$-e emissions) can be expressed as a ratio to the average annual energy/emissions from fuel use. This variation is given in Figure 2. It can be seen that between 1978 and 2012 embedded $\mathrm{CO}_2$-e rose from $15 \%$ to $31 \%$ in the whole of life analysis. During this time, the overall life cycle $\mathrm{CO}_2$-e fell by just over $40 \%$ for the class-E-sized car. For the Battery Electric Vehicle (BEV), there is a further 10% reduction, but the hybrid of the day had another 20% reduction or at 40% of the baseline 1978 car. Results for 2012 class B size conventional and BEV cars are given. Note worthy are the relatively high ratios of embedded versus annual use sourced $\mathrm{CO}_2$-e for the BEV cars because of the significant emissions from battery production.

A graphical example of the application of Eq. (1) is given in Figure 3. To illustrate the concepts in the example, it is assumed that the initial $\mathrm{CO}_2$ is four times the first year in-service $\mathrm{CO}_2$ (or equal to 4 years of fuel use). A constant average travel per year is selected (e.g. assumed to be $15,000 \mathrm{~km} / \mathrm{y}$ ). The vehicle chosen emits $200 \mathrm{~g} / \mathrm{km} \mathrm{CO}_2$ at the beginning and throughout its life. This vehicle is sold into a market where replacement vehicles have a $2.5 \% / \mathrm{y}$ improvement in fuel consumption year by year, so any replacement vehicle will have lower fuel consumption; its reduction depends on its time of delivery into the market. From the three lifetimes illustrated, the best $\mathrm{CO}_2$-e reduction for somewhere between 15 and 20 y lifetimes. The $\mathrm{CO}_2$ mitigated is only a few per cent.

This vehicle is now sold into a market where replacement vehicles have a $7.5 \% /$ y improvement in fuel consumption and embedded energy of three times in the first year use. Best $\mathrm{CO}_2$-e emission reduction is between from a 10 and 15 y life to scrap with around $25 \%$ reduction in $\mathrm{CO}_2$-e from continuous use of the original vehicle, after 30 y as seen in Figure 4.

Figure 5 shows the EU trend in fleet average fuel consumption [12] and the projection to meet the 2020 target. It is apparent that the existing trend, changing at about $4.5 \%$ reduction in fuel consumption and therefore $\mathrm{CO}_2$-e emissions per year, if maintained will meet the EU target of $95 \mathrm{~g} / \mathrm{km}$. Likely, industry will achieve lower because manufacturers have to individually meet the target and will need to do so with a safety margin. This is illustrated by the fleet average of about 120 versus the $130 \mathrm{~g} / \mathrm{km} \mathrm{CO}_2$ required for 2015. Therefore, it is highly likely that an improvement rate close to $5 \%$ will continue.

Modelling of a class $E$ car has been done using Eq. (1) with the variable km of travel per annum given if Figure 1 and the ratio of embedded to year 1 emission of $\mathrm{CO}_2$-e for the range seen in Figure 2. The results, in Figure 6, show the surface for minimum $\mathrm{CO}_2$-e as it varies with this ratio and the annual rate of improvement in fuel consumption (and $\mathrm{CO}_2$-e) to give the corresponding optimum vehicle service life. It can be seen that for the low rate of fuel consumption reduction in the period prior to 2007 the optimum vehicle life was around 17 years, whereas for current cars with a nearly $5 \%$ improvement in fuel consumption per year the optimum life reduces to 10 years.

Based on the assumption that hybrids and battery electric vehicles (BEV) improve at the same rate as the market a whole, their optimum life is longer 12 and 18 years, respectively, because of higher embedded energy (and $\mathrm{CO}_2-\mathrm{e}$) lower energy consumption in service. If their improvement rate is different, the surface allows alternative conclusions to be drawn.

The results for the smaller class B cars show a similar shaped surface with optimum lives of 14 years for older vehicles and for 12 and 17 years as the optimum for current and BEVs, respectively.

Inclusion of the embedded materials and production energy, with allowance for recycling, significantly delays the time of benefit from introducing a new fuel efficient car fleet. Unsurprisingly, the lower the in-use energy consumption (lowest for BEVs), the longer the cars need to operate in service to amortise their high embedded energy through their service life, which is a consequence of the low energy density of the battery and their relatively higher mass, even though more weight saving technologies are found in BEVs. The longer optimum life suggested for BEVs and hybrids leads to the question of whether these batteries have the durability to last perhaps twice the normal warranted battery life of ten years. Also for BEVs, the well-proven assumption, that in-service energy consumption does not vary with age, may not be valid as battery efficiency deteriorates with age as does battery available kWh. This is in contrast to the parallel hybrid, which in Figure 2, already has the lowest lifetime energy consumption and is optimum when retained for a service life of 13 years as seen in Figure 6 for class E cars.

It can be hypothesised that as the law of diminishing returns takes effect, that the per cent rate of improvement in fuel consumption will reduce, and that the optimum life will move back to higher years of in-service. In the market place, the resale price of the Toyota Prius has held up at least as well as the conventional C class cars, indicating that relatively less demanding use compared with a BEV may enable battery replacement free lives of the high teen years. This along with the prospects of significantly increased improvement possible in the internal combustion engine as found in Formula 1 racing cars for example make a strong case for the encouragement of mild hybrid uptake in the market.

It has been the province of automobile clubs to give advice to members on how long cars and vans might be used for life-time minimum per km travel cost; often slated as about 17 years. A similar approach has been applied to determine the optimum life of vehicles to minimise their life-time $\mathrm{CO}_2$-e when their manufacturing and materials energy (including recycling) is added to their life-time fuel use emissions. If the life-time is short, there will be a need to enact legislation and incentives to terminate the life of a vehicle. 'Cash for clunkers' was a marketing ploy in the past. It is concluded:

1. The results here suggest for the European market, with the needed rapid improvement in vehicle fuel consumption to meet the EU 2020 target $\mathrm{CO}_2$ emission regulation, that normal E and B class vehicles have an optimum life now of about 10 and 12 years, respectively. Mechanically many of these vehicles could continue to operate through to 15 to 20 years. However, as the rate of fuel consumption reduction reduces in the future, as it must, the optimum life will increase perhaps to 17 to 20 years.

2. It is recommended that some form of government policy be implemented to enable this process, of changing vehicle life-time, in the decades ahead in support of the most rapid achievement of greenhouse gas emissions reduction.

3. If battery electric vehicles (BEV) follow similar usage patterns to regular cars, their optimum life needs to be longer at the present time, at around 18 years, in a market with $5 \%$ improvement energy consumption per year. If their annual travel is less than that of the fleet average, as it is likely to be, the time in service should be even longer. This seems to be in conflict with early adopters and fleet buyers who are major purchasers of BEV cars, since the whole of car BEV image is important and the support for interior/infotainment systems etc in the very long term is unlikely. Therefore, it may be anticipated that owners are likely to be early rather than late scrapping their cars.

4. On the other hand, hybrids, even the simpler parallel hybrid used as the example here, looks to be the outstanding technology with an optimum life-time in the present market of 13 years and likely increasing to high teen years as technological improvements diminish.

5. These findings indicate that governments should subsidise hybrids, instead of the present support for all electrics i.e. BEVs.

6. Finally, it is obvious that these findings depend on the energy mixes in vehicle production and materials, and the proportion of $\mathrm{CO}_2$-free electricity produced. Here, these values are based on the best available information. If vehicle production involved no $\mathrm{CO}_2$-e emission, these findings would be overturned. However, it is very unlikely that major components of the car such as steel, tyres etc will be made without carbon emissions.