“Fleet expansion of natural gas private cars in Italy through 2020: is it worth it?”

Mr. Giorgio Ballardin

Scuola Mattei – Eni Corporate University

ABSTRACT: Several considerations lead to consider natural gas as a viable alternative to liquid fuels. First of all, it represents an option in curbing gaseous pollution in the transport industry. Secondly, a more extensive use of it would also reduce dependence on foreign oil, improving security of energy supply. These is just one side of the coin. The other one entails building and maintaining a refueling stations network carrying, consequently, some financial requirements. This last implication has a pivotal role in deciding if methane for automotive purposes is a convenient solution. Therefore the final aim of this work is to compare its advantages, mainly deriving from avoided pollution, with economic rationale. A system dynamics model is built and taken as reference for all quantitative assertions. The model, named CH4aCleanerAir, contains data referring to two scenarios: scenario Business as Usual and Scenario A. In Scenario A, several simulations are run in order to understand how some variables behave according to variations in methane fuelled fleet magnitude. Are new infrastructure needs compatible with existing refueling infrastructure? To which extent pollution cutbacks matter? By analyzing the scenarios’ gap numerous remarks follow. Within the model, global emissions, local emissions and infrastructure needs are treated separately. This allows to have separated outcomes for all the major aspects of these issues. The quantitative results are the basis for the final assessment, that is twofold. The first approach is grounded on the externalities theory. The second one tries to evaluate monetarily CO2 emissions. Both gather in the last momentum of the work, i.e. the comparison with financial requirements for the infrastructure. Final remarks end the analysis, underlining what are the points that the research finds out. Regardless that several simulations were run, particular attention was paid on a reference scenario drawn by the European Commission. As long as global emissions are concerned, beneficial effects of using more methane in transport seem to be tiny. Local emissions suggested two sets of pollutants would behave differently. CO, HC and O3 would dramatically decrease where NOX and soot would barely change. This would have a different impact depending on where fleets would be introduce. Urban area are likely to benefit more than countryside areas. Road transport would significantly relies on liquid fuels. Hence, it would not change dramatically the present situation. The comparison with infrastructure needs tells us that costs to build the new network are still constantly higher than the assessed advantages from the avoided pollution. This consideration gives an additional hint of how much we should invest in this option rather than on other alternative fuels.

Mr. Giorgio Ballardin Scuola Mattei – Eni Corporate University

Via S. Salvo, 1 San Donato Milanese 20097 Milan ITALY

giorgio.ballardin@enicorporateuniversity.eni.it Phone number: 0039 520 57915

1. Foreword 1.1 Context On December 2003, the European Commission drew a plan involving the deployment of alternative fuels in road transport. Natural gas (NG or CH4) played a major role in this non-binding report. Every single EU Country should have by 2020 a 10% share of natural gas-fuelled cars, passing through 2% in 2010 and 5% in 2015. Honestly, it seems like a challenging gamble since most EU Countries have an almost non-existent CH4-fuelled fleet. Alternatively, Italy already has both a sizeable fleet and a good knowledge of how to use methane for automotive purposes. It, consequently, would be easier in the Italian context to attain this proposal. At present, the Italian natural gas refueling network is growing at a fairly fast rate. There are already plans to build new stations, especially in the less-served geographical areas. This phase began in the 90’s when environmental issues were extensively felt and influenced by public opinion. 1.2 Issue’s importance The advantages deriving from expanding the NG-fuelled car share are manifold. First of all, it will help to lower both the global and the local emissions. Since car fleet mileage has continued to grow, the present, already alarming, situation could become even harsher. There are many ways to fight this phenomenon: implementing car sharing and car pooling policies, supporting an extensive use of public transportation, to mention only a few. But widening the use of low harmful fuels would undoubtedly be one of the most effective way to tackle this problem. Then it would allow reduced dependency from oil derivatives. Diversifying fuel types would mean having the possibility to increase the security of supply. Recently this issue has gained importance because of the international political situation. Italy could decide to re-allocate its purchases of natural gas, favoring Countries that are considered more politically stable. By supporting the development of this industry, new employment opportunities would be created. There would be the need to hire additional people in the filling stations, for example. On top of that, new research would be required to develop NG engine-related technologies. This would allow Italy to gain a competitive advantage in the future marketplace by managing this sort of knowledge. In the early stages in the industry, Italy was a pioneer in this market and now it has the opportunity to revitalize that role. 1.3 Final aim The aim of the present work is to compare environmental beneficial effects and financial requirements of expanding the present Italian natural gas-fuelled fleet in Italy up to 2020. The basic perception driving the researcher to implement this model is manifold. What is the magnitude of investments required to build an appropriate refueling network? How much pollutant emissions can we save by backing this fuel? Is the network capable of supplying this quantity of natural gas? Other, more qualitative, considerations should be taken into account. For instance, improved energy supply security has a substantial role in this issue. But this model was meant to be primarily quantitative, so more attention will be paid to analyzing these aspects. Ultimately, this work will focus on what are the implications of having a natural gas fleet. It is not meant to be stand alone research. Indeed, it is supposed to support Government decisions when called to choose the optimal “fuel portfolio” regarding private cars in the years to come. It aims, therefore, at shedding some light over one of the many technical solutions. After making the industry’s situation clearer, the set of information will hopefully be more comprehensive, driving a more solid fuel policy.

1.4 Expected results As already mentioned, the two main factors that will be taken into account will be the infrastructure costs and the avoided gaseous emissions. Infrastructure costs mean, in this context, building new refueling plants than the Business As Usual case and connecting them to the existing network. Regarding gaseous pollution, the issue will be divided into two main parts: global and local emissions. These quantitative results will have to be compared with the outcomes of similar research. The researcher would expect to find that global emissions from natural gas-fuelled vehicles are lower but not enough to justify the additional investments1. What really would drive the decision to support natural gas is the significantly lower local emissions. Quantifying this aspect would mean having a solid basis to assess if the additional costs are worth the initiative. 1.5 Possible way to exploit the data The possible scenarios coming out of this paper are meant to be used by whomever is in charge of defining the fuel mix in the road transport sector. While assessing how to allocate the shares, one has to know exactly which are the quantitative effects deriving from all of the considered options. This work will sort out the main information regarding the natural gas option. In other words, it is supposed to provide a better knowledge of which are the actual advantages and disadvantages of NG for automotive purposes.

2 The Italian context The very early stages of CH4 use in the transport industry in Italy were led by political reasons. The Fascist Government, forced by a Society of Nations embargo, issued a bill in the mid-30s compelling all private and pubic buses to be fuelled with CH4 by January 1st, 1938. That, of course, boosted investments in both infrastructure and vehicle-related technologies making Italy a pioneer in this field2.Ultimately, seventy years since then we’re still tackling energy security as a major issue. International players have changed, but the concern is the same and, admittedly, with stronger intensity. Two other turn points made Italy rely more and more on this technology: the 1970’s oil crises and a renewed and deeper environmental concern in the 1990’s. That process seemed to slow down after WW2, regaining trust only as a feasible alternative to conventional fuels. This long history of retrofitting gasoline-fuelled vehicles is confirmed by Italy’s rank on an international basis. Having the forth biggest fleet, as of 20023, the infrastructure ranks fifth worldwide in terms of installed fuelling plants. As of May 2004, Italian automotive CH4 network consisted of 458 refueling plants4 for private purposes. Though more then two thirds of them (307) are concentrated in only 5 out of 20 Italian Regions5. Even if these areas count as 39% of the total Italian population, this figure still remains remarkable. The most recent projects for new plants tend to even this concentration, envisaging new constructions where the dearth is more evident (e.g. in the southern Regions). This would allow the network to have a more homogeneous reach for all end users. When thinking of expanding the refueling network, other concerns remain. The first one is how to forecast the proportion of mono- and multi-fuel stations selling NG. Mono-fuel plants have more refueling modules but are more expensive since the main facility has to be built on purpose. Moreover, mono-fuel stations employees are better equipped to deal with CNG technology, being servicing a key factor in fostering the growth of CH4 fuelled cars sales. The 1 Since CO2 emissions, for example, are only about 25% lower than gasoline exhaust on a well-to-wheel basis. 2 Rinolfi and Cornetti (2004). 3 Source: "The Gas Vehicles Report", February 2002. 4 Source: www.federmetano.it. 5 The five regions are: Lombardy, Venetia, Emilia-Romagna, Tuscany and the Marche.

majority of the fleet is retrofitted by refueling stations-related garages and most of the repairs are done by them as well. On the other hand, some gasoline stations can be modified, becoming multi-fuel. It is cheaper and it can rely on the existing infrastructure. Indeed, the optimal strategy would be to harness what’s already in place, rather then building a parallel infrastructure. This holds particularly true for Italy, who outnumbers by far the average Europeans car/refueling plants ratio. Secondly, a significant running cost is the power needed to pump methane up to the right pressure, i.e. about 220 bar. If the plant is located close to a town it’s likely to have very low network pressure, requiring a need for more power. But filling stations closer to urban areas represent a better appeal on drivers. Conversely, outside the cities the so-called primary network would flow CH4 at a close-to-desired pressure. This implies a lower use of energy but plants relatively out of reach for the majority of end users. Last but not least, another factor that will strongly influence the infrastructure is the on-going liberalization. Indeed, in the past all CH4 refueling plant owners were compelled to sign a contract with the then monopolist, ENI. All contractors were granted the same conditions. This, for example, meant that no matter how big the annual CH4 supply, there was a unique price for CH4/cubic meter purchased. Now, instead, everybody can negotiate with whichever provider she/he might prefer, creating the potential to have differentiated contract conditions (influenced mainly by bargaining power). This new element has added a source of uncertainty and, therefore, risk for those who are interested in starting up a new refueling plant. When asked to make a prediction of future costs in this sense, right now it’s fairly challenging to answer. Hopefully, as time goes by liberalization will bring lower price levels for everybody, offsetting the abovementioned shortcoming.

3. The model “CH4aCleanerAir” The model “CH4aCleanerAir” has a core part, named “Fleet and cars forecast”, that envisages the future development of both natural gas-fuelled cars fleet and the related refueling plants network. It serves as an engine for all the other sections that, therefore, can be seen as its branches. The original idea, as already mentioned, was to analyze effects on the various aspects involved in this transport option. For example, trying to compute the required refueling plant investments triggered by an hypothetical, strongly expanding methane-fuelled fleet. Likewise, it might be interesting to know how the present traditional fuel plants’ infrastructure could back the expansion of automotive NG. Combining the previously mentioned data, a prediction regarding local and global avoided emission will be drawn. The model foresees two scenarios: Business As Usual (BAU) and Scenario A. Intuitively, in the BAU scenario nothing changes from the present situation. This means, for instance, that the CH4 fuelled car fleet will remain the same, as share of total, up to 2020. In scenario A, instead, several conditions will be changed. By modifying the underlying drivers of the model, the researcher would like to understand which are the leverages that could make natural gas an efficient solution for the automotive industry’s development. Taken as given, the NG fleet forecast needs a plan that will maximize its positive effects, e.g. minimizing pollution. From time to time, single variables will be changed drawing hypothetical, but feasible, future states of the world. These aspects are fully explained in the following paragraphs. Most of them are, from a quantitative point of view, treated separately, but each will jointly contribute to the final conclusion of this work. 3.1 The core section: “Fleet and plants forecast” This part will provide an up to 2020 forecast of the CH4-fuelled cars fleet and the related number of required plants (both mono- and multi-fuel). The development of the CH4 fuelled cars stock and the “cars per refueling plant” variable will be defined and provide the number of required plants. The underlying assumption, therefore,

is to know this in advance and therefore resize the present infrastructure according to future needs and the construction time6. The researcher takes into account only three fuels: gasoline, diesel and natural gas. This is justified by the fact that GPL fuelled vehicles play a minor role. In the future, likely, they will be taken over by the remaining fuels because the use of these cars are becoming relatively more expensive. Both electric and hydrogen fuelled vehicles pose serious problems when asked to draw a solid, quantitative forecast. They are not added because they would represent an additional source of uncertainty. In order to have a prediction about the overall fleet up to 2020, an econometric regression is implemented. An S-shaped logistic curve is fitted into the time series with different ceilings where to tend to. At the end only five regressions’ parameters are considered (with 35, 38, 41, 44, 47 million cars as ceilings, respectively)7. This research wants to cover very different fleet evolution and alternate them in Scenario A. Present Italian private car fleet consists of 34.310.446 cars, 430.000 of which are natural gas-fuelled.

Graph no. (1): Upper part: “Auto” refers to the actual fleet data where N41 is the logistic regression (N because parameters were non-linearly estimated and 41 is the chosen ceiling). Lower part: number of NG-fuelled cars up to 2020 (1) and overall fleet forecast (2).

6 Experts suggest to use an average year-long lag construction time. By knowing this in advance, it allows to satisfy the future demands for refuelling plants. 7 The R2 of these regressions ranges from 99.3% (35M ceiling) to 97.9% (47M ceiling). A non-linear estimation of the logistic function parameters, β0 and β1, was carried out.

A basic assumption is that natural gas-fuelled cars cannibalize only gasoline-fuelled cars and not, for example, diesel-fuelled cars. It’s important to bear this in mind when interpreting the final results, particularly those regarding local pollution. This hypothesis is based on the fact that diesel car retail prices are becoming more competitive with gasoline retail prices. In order to compete with diesel on fuel prices, natural gas has to take over gasoline, which, in relative terms, is more expensive. In addition to this there is a technical reason. The easiest and cheapest way to retrofit a car, so far, is starting with a car that is gasoline-fuelled rather than one that is diesel-fuelled.

Graph no.(2): Main components of the fleet and plants forecast section. Refueling plants in 2004 and 2005 are taken as given, locking the model in its first two steps. Whenever time (t+1) plants required exceeds time (t) existing plants, new plants by the same amount are automatically ordered for the year to come. Hence, “new plants” is to smooth the forecasted additional time (t+1) need of plants. Intuitively, all “new plants” are made to flow into the existing plants stock at the following period, when they actually become refueling plants. That is why if the required plants series monotonically goes up, the “existing plants” series mimics the same pattern, lagged by one period of time (i.e., one year)8. Indeed, if the number of required plants stalled, there would be a perfect match between demand for infrastructure and existing filling plants. Here are the formulae and a close-up of the model that show how these concepts are expressed in algebraic terms: New plantst = max [0,(Required plantst+1-Existing plantst)] (1) Existing plantst+1 = Existing plantst + New plantst (2)

8 The required refuelling stations variable was “dragged” one period in advance. By anticipating this variable, one could know in advance how many future stations are needed therefore starting construction in time to satisfy future demand.

Graph no. (3): Model close-up about the generation of new filling plants. The model is constrained not to allow a decrease in the stock of natural gas refueling stations9. This assumption is supported by the fact that a Government, after having spent so many efforts in expanding it, wouldn’t allow the network to shrink,. Moreover, common wisdom would suggest that it is not reasonable to decommission refueling plants when forcing short term downward fluctuations. If political power was to surrender to market forces, it would give an impression of ambiguity about the way it leads policy. In other words, Government has to be supportive of this initiative, even if there’s a shrinking fleet, allowing the potential of a network to exist in any case. A non-negativity condition is, hence, set: if the number of required plants is below the level of existing ones, nothing happens. During the simulations, the modeler can change the final CH4 fuelled cars share and the final “cars per refueling plant” variable estimate. The latter was fine tuned according to suggestions collected by industry’s experts10. By 2020, the variable was made to reach 700 cars per refueling plant. An important task, then, is given to sorting out how many refueling stations would have to be either mono- or multi-fuel. This issue holds particular importance because of the different construction and running costs of the two11. Everybody in the industry agrees that we will have an increasing amount of multi-fuel stations. Consequently, the modeler draws a quadratic function between the present data and the assumed one, up to 2020. These shares determine how many mono-fuel and multi-fuel plants will be built each year. A slider was linked to the final value, in order to see how the overall infrastructure cost would vary. The outcomes are two-fold regarding the number of refueling plants associated each year. They are used as a check for the stock of refueling plants and as a basis to compute the amount of investments required each year.

9 For example, tearing them down or even closing them temporarily. 10 Mr. Mariani, from Metauto, and Mr. Tozzoli, from Federmetano. 11 Multi-fuel refuelling stations range between 100.000 and 150.000 €. Multi-fuel stations range between 300.000 and 600.000 €.

3.2 Infrastructure investments and running costs Infrastructure investments mean, here, expenditures related to the construction of new plants. Running costs, instead, are services or goods meant to be used within the year. They both refer to additional costs over the BAU scenario. Therefore, there is no surprise with regard to the existing infrastructure. This is justified by the fact that all the work is grounded on a single question: “what if we expanded the fleet and the refueling infrastructure?”. Investments required to build a multi-fuel rather than a mono-fuel refueling plant are different. Hence, average values, 125.000€ and 400.000€ respectively12, are assumed. They could vary, for example, with the distance from the network13, but it is extremely difficult to determine the distribution of this random variable. Most likely, the majority of new refueling plants will be built close to urban areas. This would mean they would be closer, on average, to CH4 pipes and, likely, procure lower connection costs. But it still remains, too, an ambiguous consideration to fit in a quantitative model like this. It is nevertheless worth mentioning because it will drive costs in any case, even if it’s not easy to compute. The latter fact has an influence on running costs. The closer to towns, the likelier there will be a secondary, lower pressure CH4 network close by. This significantly affects the power required to compress the gas to the desired pressure. There is, thus, a trade off regarding the distance from urban areas between connection costs and the power bill. Yearly power bills are the result of three factors: cost of power per kwh employed, energy required to deliver a cubic meter of CH4 and the average cubic meter sales per year. The first figure is estimated to be 2 Eurocents per kwh. Then, on the one hand, energy required to pump the methane (as estimated by Mr. Mariani) is approximately 0.1-0.15 kwh/cubic meter of CH4. Other estimates claim 0.1-0.35 kwh/cubic meter of CH4 range as realistic. 0.15 kwh/cubic meter of CH4 is taken as representative. Finally, average cubic meter sales per year is the result of total CH4 used for automotive purposes divided by the number of plants for each year. Other costs are also taken into account. Regarding labor, the researcher assumes that both types of plants need two employees to work. On average, the per employee annual cost for this kind of labor is estimated to be 25.000€/year, all inclusive. The third cost included in the estimates is maintenance that, in 2004, is approximately 10.000€/year. A basic assumption on inflation is made in order to express values in their real terms. An average 2% per year is, therefore, included. This hypothesis makes sense when thinking about the constraints posed by the European Central Bank concerning monetary policy. In order to run a Net present Value (NPV) calculation, estimates regarding discount rates are needed. According to this, a flat curve, at a 3% level is assumed. 3.3 Global emissions Since natural gas is supposed to cannibalize only gasoline, in both scenarios diesel-fuelled cars have the same share over the fleet. This allows the researcher to single out the effect deriving from this kind of substitution. The diesel car path is taken as given and, consequently, has no effect on how scenario A evolves. Global emissions deals only with CO2 and therefore calculations are fairly straightforward. In order to work out the total amount of CO2 emitted per year, two sets of data are needed: the overall number of the fleets, sorted by fuel, and the CO2 emission per car per kilometer. One can not weigh emissions by each individual fuel because these data are expressed in terms of kilometer traveled and the yearly mileages vary among the fleet. That’s why weights are used in term of kilometer traveled by the single fleet. Then the BAU scenario estimates are carried out, checking their consistency with real data. Subtracting tons of avoided CO2 in scenario A from total BAU emissions gives scenario A

12 Experts suggested to take these two figures as reasonable estimates. 13 Connections costs are about 300-600€ per meter. They were included in the overall construction costs.

emissions. In order to make it in relative terms as well, global emissions reduction are related to total BAU global emissions. This gives us the extent to which policy can help to reduce pollution. The percentage reduction of CO2 from CH4 compared to gasoline was meant to be in well-to-wheel terms (Life Cycle Assessment). A 25% LCA CO2 reduction estimate was taken from the “Market development for alternative fuels” report by Alternative Fuels Contact Group presented to the European Commission on December 2003. Then the annual amount of CO2 tons saved serves as a basis for computing where the theoretical monetary savings come from. In order to implement this, 20€/ton was estimated to be a reasonable monetary value for CO2 avoided emissions.. This data has been adjusted according to inflation dynamics, as well. 3.4 Local emissions The basic weights used in this part are total kilometers traveled by each fleet14 over kilometers traveled by the overall fleet. Combining these weights with the percentage reduction, sorted by pollutants, gives the percentage reduction of local emissions comparing the two scenarios. As with global emissions the reductions are computed taking into account only the effects of introducing additional CH4 fuelled cars. Basically, the question the researcher wants to ask is: how much local pollution can we save by implementing this transport policy? Percentage reductions figures are used. Then, an index is built where the reductions in relative terms were weighted by how much every single fleet contributed to the annual mileage. Therefore, this method gives a number that doesn’t have a meaning in absolute terms, but it does in relative ones. Indeed it allows the modeler to compare how these two indexes performed in the two scenarios. Finally, working out the deviations is straightforward. 3.5 Externalities estimate and Net Present Value An interesting comparison would be the one between the actual costs incurred from expanding fleet and the monetary evaluation of avoided pollution. These considerations give a glimpse of how they could balance off each other. The additional costs taken into account are those related to the infrastructure implying the construction of new filling plants. Running costs are also computed but not considered because the focus of the issue and the subsequent policy is long term. As a result, knowing the cash flow, related to this aspect, is meant only to provide additional information. Beneficial effects are quantified in two manners. The first one is to evaluate the avoided emissions of CO2, provided that a value for that is easily definable. Conversely, urban gaseous pollution can not carry it out since there are fewer reliable estimates. On top of this, a more comprehensive measurement is needed. Consequently, it sounds reasonable to use tools belonging to the externalities literature15. These, indeed, would embed both local and global emissions impacts on both the environment and human health. Of course, outcomes from the two latter approaches will be assessed conditionally according to the employed methodologies. Ultimately, NPV will turn out to be useful when comparing the cash flows of interest. The absolute values don’t have an own meaning. They are only aimed at understanding if there is an off chance that natural gas might be successfully harnessed as an alternative fuel in road transport.

14 I.e., the sub-fleets employing different fuels. 15 Data from Carslaw et al. (1995) quoted in a IANGV report were employed (and adjusted for inflation). They refer to the UK transport system, but since there were not similar data for Italy, the modeller would rather take these as an approximation.

4. Results Many simulations were run, changing the evolutions of the variables considered over the time span. But, ultimately, the main variable to take into account is the percentage of CH4 fuelled cars at 2020. Since the European Commission adopted 10% as reasonable by then, the modeler will focus on the implications of this assumption. The final share of CH4 equipped filling stations was approximately 17.6%. The already quoted study, “Market development of alternative fuels” on page 24, deemed 25% to be realistic. Of course, in that work there is a similar assumption that 10% of all vehicles will be NG by 2020. Otherwise, they could not be compared. Therefore, assuming 700 NG-fuelled cars per NG filling plant, as done in the model, gives an outcome that is compatible with that report. Since diesel-fuelled cars will get close to parity from the present 21.6%, in terms of stock, their sales has to grow really fast in the years to come (since the latter is a flow variable). The same holds true for methane, whose lead on the gasoline-fuelled cars sales is getting slimmer and slimmer. According to this same study, natural gas demand for automotive purposes will represent 5% of the total gas demand by 2020, on a EU basis. The models suggests a 6,3% by the same date for Italy. The underlying assumption here is that, in Italy, overall gas demand will grow from present ca. 80 Gm3 of CH4/year up to 110 Gm3 of CH4/year by the end of the period. The broad deployment of natural gas, in the coming years, will mainly come from the power industry. Since Italy seems far on its way to complying with Kyoto’s regulations, there’s the actual need to use technologies, such as combined cycle generation plants, to lower GHGs emissions. Most of these technological options foresee the deployment of natural gas. That is why overall demand has the potential to be significantly boosted. 4.1 Global emissions The CO2 emissions avoidances are very close to the expectations. The ending figure is a 2,13% decrease in scenario A than in the BAU scenario. Intuitively, if the share of NG-fuelled cars is 10% by 2020 and the LCA global emissions has been 25% lower than gasoline, a 2,5% decrease would sound reasonable, other things being equal. The reason why it’s 0,57% lower is because NGVs usually have lower mileage than diesel-fuelled cars (as assumed). This means that diesel-fuelled fleet will weigh more in 2020 that its actual share, partly absorbing the beneficial effects coming from NG vehicles. When thinking in absolute terms, instead, an almost 3 millions tons of CO2 could be avoided by 2020. The problem dealing with CO2 from transports, and in particular road transport, is that it’s the industry with the fastest growth rates lately. Technological innovations or developments, such as this one, seem to have a hard time offsetting this booming sector. Clearly, NGVs can only play a part in the wider policy of global emissions reduction in transports. Since the policy will be about a portfolio of solutions to be carried out, one should consider whether NGVs are an efficient solution, from a monetary point of view. There’s no doubt about the positive effects that might come from introducing this sort of policy. The problem lies on deciding how to allocate investments. In other words, it’s more correct to compare the avoided emissions with the investments required to build the related infrastructure, in this case. There might be other ways to reduce CO2 that are equally good or even better. This would lead to a decision, for instance, to not have a 10% expectation by 2020 but, say, 5% or 7%. Natural gas is an hydrocarbon and therefore nobody expects it to dramatically cut off CO2 emissions, in well-to-tank terms. But the additional value this work is aiming at is to depict by which extent it could be a preferable solution over others.

15.10 gio 5 ago 2004

CO2 av oidances in absolute and relativ e terms

Page 12004.00 2008.00 2012.00 2016.00 2020.00

Years

1:

1:

1:

2:

2:

2:

0

1500000

3000000

0

2

3

1: CO2TonsAv oided 2: CO2 PercRed

1

1

1

1

2

2

2

2

Graph no. (4): CO2 avoidances in absolute and relative terms. 4.2 Local emissions As previously mentioned in the descriptive part of the model, the reduction regarding local emissions has been implemented considering the national flow of emissions. Therefore it hasn’t taken into account any possible skewed impact from these reductions. For example, it’s reasonable that major cities will be the first to adopt natural gas-fuelled fleets. Indeed the final aim consists of having an idea of what would be the contribution of natural gas in having a greener private cars fleet. Assessing the different influences that it could have on specific areas goes beyond the scope of this work. That kind of study would be even more technical when asked to define what the actual plans would be in order to build an infrastructure in predetermined cities. Here, instead, the modeler is dealing with what comes first: evaluating if NGVs are seemingly a favorable solution on a national level. Then there will be room to decide where NGVs are best able to demonstrate their strengths. Here follows the data with which the researcher started. They were exploited by creating two parallel indexes, for the BAU and A scenarios respectively, to ultimately assess deviations.

% CO % HC % NOX % Soot % O 3

NG vs Gasoline -75 -82 -72 0 -88NG vs Diesel 138 -40 -95 -100 -50

Natural Gas Vehicles emissions

Table no. (1): Source: FederMetano, www.federmetano.it. Aside from CO2 emissions, there are significant reductions in the other gaseous pollutants. It’s always necessary to analyze these outcomes bearing in mind that by 2020 only 10% of the fleet will be fuelled with natural gas. On the one hand, there are three pollutants where natural gas for automotive purposes performs particularly well: CO, HC and O3. This is because NG contributes extremely low emissions when compared with gasoline-fuelled cars. On the other hand, in term of soot and NOX, the effect is weaker. The reason for this is that the major contributor is the diesel-fuelled fleet. Since the modeler assume that no diesel cars are replaced by natural gas ones, the pattern of these two pollutants seems to be strictly bound to the increase in diesel cars. These assertions lead to an obvious conclusion: if natural gas has to substitute either gasoline or diesel, there would be a trade off in terms of which local pollutants to reduce. By taking over gasoline cars, the advantage would be in lowering CO, HC and O3 emissions. Instead, by

substituting diesel cars major reductions would be in soot and NOX flows. By having a more balanced mix of cars to cannibalize there would be a smoother, lower than average reduction that would involve all local pollutants. As of 2004, diesel-fuelled cars offer a wide range of cars to choose from and retail prices are comparable to gasoline-fuelled cars. In Italy, moreover, diesel is cheaper than gasoline, making it an attractive purchase. These are the reasons hindering the substitution of diesel cars with NG cars. The supplanting of gasoline is a constraint because it is cheap. Boosting drivers to buy a NG car over a diesel-fuelled one would require more incentives. Therefore, having a more mixed solution in terms of substituting cars might be too expensive. It would be cheaper just to substitute gasoline-fuelled cars.

Year CO % reduction

HC % reduction

NOX % reduction

Soot % reduction

O3 % reduction

CH4 fuelled cars (%)

2004 0,00 0,00 0,00 0,00 0,00 1,262005 0,00 0,00 0,00 0,00 0,01 1,562006 0,36 0,38 0,18 0,13 0,42 1,902007 0,76 0,80 0,37 0,27 0,89 2,262008 1,20 1,27 0,58 0,43 1,42 2,662009 1,69 1,78 0,80 0,60 1,99 3,102010 2,22 2,33 1,04 0,78 2,61 3,562011 2,80 2,94 1,29 0,98 3,30 4,062012 3,41 3,57 1,52 1,15 4,01 4,592013 4,08 4,24 1,75 1,31 4,78 5,152014 4,80 4,95 1,94 1,45 5,61 5,742015 5,57 5,72 2,13 1,56 6,49 6,372016 6,40 6,53 2,33 1,69 7,43 7,032017 7,27 7,39 2,60 1,90 8,43 7,722018 8,13 8,22 2,75 1,96 9,39 8,442019 9,04 9,09 2,94 2,06 10,40 9,202020 10,13 10,11 3,15 2,21 11,62 10,00

Table no. (2): Local emissions reductions sorted by year and gaseous pollutant. 4.3 Infrastructure costs The model is connected to projects running in 2004 that will be completed by 2005. At that time, the ratio of cars per filling plant was set to be downward sloping, starting from about 1000, which is the present value. This implicitly creates an over supply of plants when no additional filling plants are required for the second year, i.e. 2005. The increasing fleet, by 2006, recreates a positive gap between the number of required and the number of existing plants, therefore triggering the need for new construction. This gives the modeler a hint. If in the BAU case the fleet was not to increase, the network would outnumber the actual requirements in the years to come. Bluntly, with an unchanged fleet by 2005 and with the new plants, the cars-over-filling-plants figure would drop to approximately 858, from 939. Or otherwise, the fleet has to increase by 40,000 units to stabilize at the same ratio. But the latter seems fairly unrealistic. Of course, this can be seen as policy building the network before fostering the fleet. But, so far, that has been only one-sided since nothing has changed in terms of incentives to buy or to retrofit cars. As a proof of that, Government still had not run out of incentives for OEM cars16 belonging to 2003, according to Consorzio Ecogas in April 2004. These results suggest that maybe the present policy is not, in overall terms, the most effective. The network issue is undoubtly a big one, but it can not be only about that. 16 Original Equipment Manufacturer.

The following graph depicts the total number of plants to be built and how the two plant stocks evolve.

15.25 v en 6 ago 2004

Untitled

Page 12004.00 2008.00 2012.00 2016.00 2020.00

Years

1:

1:

1:

2:

2:

2:

3:

3:

3:

0

1500

3000

0

250

500

1: Monof uel plants stock 2: Multif uel wCh4 plants stock 3: Plants to build

1

1

1

1

22

2

2

3

3

3

3

Graph no. (5): Splitting the stock of filling plants into mono- and multi-fuel and the number of new constructions. This leads the modeler to define what flow of investments is required in scenario A. As anticipated, in the year 2005 there is a trough because no additional plants are required for the following year.

15.55 v en 6 ago 2004

Untitled

Page 12004.00 2008.00 2012.00 2016.00 2020.00

Years

1:

1:

1:

2:

2:

2:

3:

3:

3:

0

35000000

70000000

20

60

100

1: Construction Mono 2: Construction Multi 3: Monof uel f illng stations to build

1

11

1

2

2

2

23

3

3

3

Graph no. (6): Investment requirements in nominal terms sorted by year and plant type; line 3 depicts the mono-fuel share of new plants. Even if the number of mono-fuel plants seems to decrease deeply17, its weight on total investments remains significant. This, of course, is due to the fact that more than double the amount of money is necessary to start up a mono-fuel filling plant.

17 It monotonically dwindles till 2020 to become 20% of the share of new plants in construction.

Then it would be interesting to put side by side CO2 savings and investments required for the infrastructure. As one can see from graph no. (7) (below), CO2 hypothetical savings18 are always lower than investments, but the second year.

12.26 mer 11 ago 2004

Untitled

Page 12004.00 2008.00 2012.00 2016.00 2020.00

Years

1:

1:

1:

2:

2:

2:

0

60000000

1200000001: CO2Sav ings 2: Total Construction

11

1

1

2

2

2

2

Graph no. (7): Comparing infrastructure costs and CO2 monetary savings (values expressed in Euros are adjusted for inflation). 4.4 Externalities and Net Present Value (NPV) comparison Infrastructure investments are put side by side with avoided externalities and an evaluation of the avoided CO2. CO2 is considered only to give a wider view of what might be the determining factors for choosing automotive NG. But, of course, it deals only with global emissions. Avoided externalities, conversely, tend to provide a comprehensive assessment of all emissions. That is why the researcher chose to use only global emissions in the following NPV comparison. In the very first years, the relatively small additional fleet is not able to offset the investments required to expand the infrastructure in order to avoid externalities. But soon it gains foothold and from 2010 until 2020 it widens the gap, becoming at times even bigger than the investments of those years. When assessing this using Net Present Value technique, it becomes quite clear how much this last years “overshooting” can weigh. Indeed, even if they are farther apart in time, and therefore should count less in financial terms, avoided externalities are still more than twice as big as infrastructure. The first reason for this, as already mentioned, is the widening scissor effect at the end. The other reason is the low interest rate assumed. Interests rates are now at their lowest historically since WW2. It’s likely that they will grow, but a 3% average could nevertheless be a not far away from the assumption. This means that even having with, say, a 7-8% internal rate of return19, results will not changed very much. A sensitive variation would come from assuming much higher interest rates. But this might be even less realistic than the initial assumption. The last consideration derives from assessing the different nature of these cash flow streams. Those related to infrastructure could involve real money allocation. At the same time, avoided externalities quantify the beneficial effects. So, we should bear in mind, while comparing them, that avoided externalities work mainly as a reference while the infrastructure cash flow could represent real expenditures. 18 Estimated CO2 value per ton is 20 €. 19 Interest rate used as input in the NPV.

Year Infrastructure costs

Externalities avoided

CO2 evaluation

2004 17.083.900 € 0 € 0 €2005 0 € 2.016.570 € 311.092 €2006 52.399.037 € 14.523.526 € 2.157.840 €2007 60.284.475 € 29.521.897 € 4.421.383 €2008 66.343.686 € 47.728.442 € 7.066.383 €2009 72.112.398 € 69.099.195 € 10.168.340 €2010 77.519.862 € 94.043.124 € 13.864.866 €2011 82.508.189 € 121.748.219 € 17.997.526 €2012 87.035.339 € 155.460.266 € 22.553.946 €2013 91.077.236 € 197.930.420 € 27.819.903 €2014 94.628.649 € 245.655.749 € 33.653.226 €2015 97.702.546 € 307.272.947 € 40.163.461 €2016 100.327.821 € 370.945.493 € 47.274.755 €2017 102.545.446 € 430.840.844 € 55.012.342 €2018 104.403.385 € 514.293.842 € 63.269.107 €2019 105.950.771 € 599.683.848 € 72.006.624 €2020 107.232.014 € 683.961.282 € 81.168.335 €

Infrastructure costs

Externalities avoided

Gap

969.277.341 € 2.627.494.187 € 1.658.216.846 €

Nominal values

NPV comparison

Table no. (3): Scenario A comparisons between additional infrastructure costs, externalities avoided and CO2 evaluation; NPVs summarizes them all.

5. Final remarks Although CH4 has the highest H/C ratio of all hydrocarbons used in transport, it still has LCA CO2 emissions that are not significantly lower than that of gasoline cars. Its positive contribution is undeniable, but its effects on overall fleet global emission tends to be limited. Conversely, natural gas employed for automotive purposes seems to have a competitive advantage with some local pollutants. In particular, it over-performs gasolinewith regard to CO, HC and O3. The close-to-zero emissions in this case have a relevant effect, taking into account that only 10% of the fleet by 2020 is assumed to be fuelled by methane. When substituting gasoline there are problems to be solved dealing with NOX and soot. Here the major contributor is diesel that, for a variety of reasons, has become the most competitive fuel. First of all, new diesel car prices are almost the same as gasoline car prices. Moreover, the range of brand new diesel-fuelled car models can not be compared with what that of OEM natural gas cars. In terms of fuel price, NG is more competitive but the gap is much narrower than with gasoline. High diesel engine efficiency and diesel price itself contribute to this. On top of that, there still is the network issue. Diesel car drivers can count on an existing and extensive network when NG drivers can not, so far. This is why it is not likely that CH4 will take away market share from diesel. Consequently, there will be problems dealing with NOX and soot. The only way to deal with them would be to work on both sides. Maybe exacerbating fiscal revenues from diesel and easing those for NG. Or making automotive CH4 price being sticky benchmarked to oil prices. This way, raw material price fluctuations will not impact the end users. This would have a great appeal, especially in days like these when we are experiencing roller coasting oil prices. NPV analysis gave us a hint that CH4 is a enterprise worth undertaking, even as a share of the total. The ultimate problem would be that the Italian Government should need to adopt a set of measures to back all of the required financing. This remains something to disentangle. The evidence of NG benefits should support the decision. Ultimately, NG can be a determining contributor to improved air quality, especially in highly trafficked urban areas. Part of the local pollution issue, NOX and soot emissions, has to be addressed differently. This problem will be dealing with the increasing number of diesel-fuelled car sales and how to face their competitiveness. Building a NG filling station infrastructure is within reach if one looks at the financing required each year. One way to make it more efficient would be to add NG refuelling modules to the existing plants, making them multi-fuel. This is motivated by the costs to build them which is the less than half and the fact that the Italian filling plant infrastructure is already oversized.

6. References

“Environmental external costs of transport” (2001), R. Friedrich – P. Bickel, Ed. Springer, Ch. 15th, par. 5th, “Germany: Benefits of introducing CNG vehicles in the Federal State Baden-Wuerttemberg”.

“Market development of alternative fuels” (2003), Alternative Fuels Contact Group.

“Energy systems aspects of natural gas as an alternative fuel in transport” (2003), Ramhesohl-Merten-Fischedick-vor der Brueggen, Wuppertal Institute for Climate Environmental Energy.

“L’uso del metano nei trasporti: aspetti ambientali locali e globali” (2003), Giuseppe Onufrio, Istituto Sviluppo Sostenibile Italia, presented by the conference “Verso una mobilità sostenibile: il ruolo del gas”, Rome, Italy, 5th November 2003.

“Il Metano per i Trasporti: le ragioni di una scelta” (2003), Poli Vettori, Fedemetano, presented by the conference “Verso una mobilità sostenibile: il ruolo del gas”, Rome, Italy, 5th November 2003.

“Assessing Strategies for the Market Introduction of Natural Gas Vehicles in Switzerland”(2004), Arthur Janssen, Stephan F., Lienin, Fritz Gassmann and Alexander Wokaun, presented by the “International System Dynamics Conference”, Oxford, England, 25th – 29th July 2004.

“1996 Position Paper”, International Association for Natural Gas Vehicle.

“The history and the evolution of CNG engine technology: from the 1930’s until today” (2004), Rinolfi and Cornetti, ATA - Associazione Tecnica dell'Automobile, vol. 57 n.1/2.