review of fuel input parameters

Transcription

1 European Commission assessment of the impact of ethanolblended petrol on total NMVOC emissions from road transport in the EU: review of fuel input parameters Report to the European Commission DG JRC Institute for Environment and sustainability September, 2005

2 1 DG Joint Research Centre study: approach and results On 5th April at the European Commission (EC) Fuel Quality Review stakeholders meeting in Brussels, the DG Joint Research Centre s Institute for Environment and Sustainability (IES) presented the results of a modelling study into the VOC emission impact of the widespread introduction of 10% ethanol blended in petrol (so called E10 ) in the EU15, projected to the years 2010 and The results were produced from the Copert III model which generates vehicle emissions scenarios by, inter alia, varying fuel input data. Two temporally distinct E10 scenarios were presented by the IES. In the scenario named ETH3 the ethanol is blended directly into petrol (so-called splash-blend) for 12 months of the year. In another scenario, ETH1, splash blending only occurs during the summer months specified in the Fuel Quality Directive. For the remainder of the year, the ethanol substitutes other unnamed components of the petrol in order to maintain the Reid Vapour Pressure of the base petrol (so-called match blending ). For both ETH3 and ETH1, the VOC emissions were compared with those produced by a base petrol called BL1, which had the same specification for both scenarios. For this purpose the IES used reported averaged annual Member State fuel data from 2002 provided by AEA Technology and published in the EC s website. To generate ETH1, the IES made just two changes to the BL1 fuel dataset: o summer RVP was set at 70 kpa for the EU11 that adopted the 60 kpa Fuel Quality Directive limit, and at 80 kpa for the EU4 that adopted the 70 kpa Directive limit; and o oxygen content was set at 3.5% by mass (m/m). Scenario ETH3 included these changes, but also increased winter RVP by 5 kpa above the existing national maximum limits that were used in BL1. A third E10 scenario, ETH2, was also carried out that repeated the blend scenario of BL1/ETH1. The difference between it and BL1/ETH1 was that it used the Copert III model s default fuel data for BL2 and ETH2 instead of the AEA Technology reported data. Table 1 below highlights the fuel parameters that were varied between BL1 and ETH1 and ETH3, and between BL2 and ETH2. Table 1: Fuel parameters varied by IES Fuel parameter BL1 ETH1 ETH3 BL2 ETH2 Summer RVP varies by country kpa kpa kpa kpa Winter RVP maximum permitted maximum permitted maximum permitted Maximum permitted Oxygen content varies by country 3.5% m/m 3.5% m/m 1.5% m/m 3.5% m/m 1

3 At the European level, numbered results (as opposed to bar charts) for the years 2010 and 2020 were presented for the BL1/ETH1 scenario. In 2010, the difference in VOC emissions between BL1 and ETH1 is forecast to be 0 with an expected range of -0.3% to +0.3% for the road fleet (and 0.06% to +0.06% for all VOC sources in the European Union). By 2020 the model predicts a net increase of 1.3 to 1.4% in road fleet VOC emissions from ETH1, which represents an increase of 0.21 to 0.22% from all sources. At the national level the IES presented results for both BL1/ETH1 and for BL2/ETH2 for the year The results are extraordinarily consistent across Member States. For ETH1 the highest achieved reduction in emissions is only 1.9% (in Austria), but the highest increase in emissions is just 2.2% (in Luxembourg). Compared with BL1/ETH1 the BL2/ETH2 results converge towards 0 (i.e., neutrality) in 10 of the 15 Member States and only diverge in 3 cases. In other words, the overall result is even more consistent. The IES concluded that these results present a worst case scenario for E10 petrol, inter alia, because they discounted the VOC benefits of oxygen content from post Euro II cars. 2 Scope of this study Abengoa Bioenergía, a member of ebio (the European Bioethanol Fuel Association) asked the IES to have access to the activity data that were used to run the model for two Member States for the year 2010, with the aim of reviewing the fuel input parameters. In return, Abengoa Bioenergía was asked, on the basis of the findings, to recommend two changes to the E10 input datasets that IES might then review for the EU15. The two countries selected were Germany and Spain because they represent two major markets with diverging VOC emission results. For both reported scenarios (BL1/ETH1 and BL2/ETH2) Germany was forecast to obtain VOC emissions reductions with E10 by 1.2 and 0.6% respectively, whilst Spain s VOC emissions with E10 were forecast to increase by +1.6% and +1.3% respectively. This divergence was thought to principally reflect differences in the input data for: o the vehicle fleet (both age of fleet and petrol/diesel share); and o ambient temperatures 2

4 3 Review of fuel input data 3.1 Fuel parameters in Copert III Abengoa Bioenergía identified the following petrol parameters in the Copert III model as having an impact on VOC emissions: o summer RVP o winter RVP o E100 evaporation o E150 evaporation o aromatics content o olefins content o sulphur content o oxygen content Abengoa Bioenergía believes that all these fuel parameters change when splash blending bioethanol in petrol with the exception of E150 evaporation. Tables 2 and 3 below present the changes made by Abengoa Bioenergía to the BL1 and ETH1 fuel data for the analysis. (Similar changes for BL2 and ETH2 are not presented here). This is followed by an explanation of the recommended changes. Table 2: Recommended changes to BL1 Fuel Parameter Germany data Spain data BL1 Abengoa Bioenergía BL1 Abengoa Bioenergía Sulphur (mg/kg) Table 3: Recommended changes to ETH1 Fuel Parameter Germany Spain ETH1 Abengoa Bioenergía ETH1 Abengoa Bioenergía Summer RVP (kpa) E100 evaporation (% v/v) * * Aromatics (% v/v) * * Olefins (% v/v) * * Sulphur (mg/kg) 0 6.3* 0 6.3* Oxygen content (% m/m) * The recommended changes may (or may not) be appropriate only for the summer period described in the Fuel Quality Directive. 3

5 Summer RVP (change to ETH1): The IES fixed the summer RVP for ETH1 for all Member States at 70 kpa no matter the average RVP for BL1. This significantly overestimates the increase in RVP brought about by splash blending. It is unknown what the average increase of RVP would be in European summer E10 splash blend, but we do know that in the USA the maximum increase is 6.9 kpa. Since European and American petrol are closely matched products (increasingly American petrol is imported from European refineries), Abengoa Bioenergía expects that the average increase in Europe would be between 5.0 and 6.5 kpa, and has chosen 6.0 kpa as a best estimate. E100 evaporation (change to ETH1): This variable was not adjusted by the IES but is well-known to increase with E10 splash blends. Again, it is unknown what the scale of the increase would be with European commercial-grade petrols. On this occasion US experience is not directly applicable since they do not report E100 evaporation. The recommended increase of 6% is no more than a best guess based on the average of the IES s EVAP study and data reported by the French Environment Agency from an emissions test (ADEME, 2003). Whether it will vary or not from BL1 as a match blend entirely depends upon the fuel components that the ethanol substitutes. This issue was not addressed by the IES although the E10 match blends in its EVAP study are affected. Abengoa Bioenergía chose conservatively to assume no effect. Aromatics and olefins (change to ETH1): This variable was not adjusted by the IES. However, aromatics and olefins are petrol components, so splash blending of ethanol dilutes them by the amount of ethanol added. Thus, for ETH1 in summer time Abengoa Bioenergía reduced these components by 10%. Whether they are diluted or not in a match blend entirely depends upon the fuel components that the ethanol substitutes. This issue was not addressed by the IES although the E10 match blends in its EVAP study are affected. Abengoa Bioenergía chose conservatively to assume no effect. Ethanol can not only dilute the aromatics content, like most oxygenates, it is an octane substitute for aromatics. The Caldecott tunnel study by Kirchstteter et al (1996) observed that all aromatics were scaled back by at least 20% when the oxygen fuel programme went into effect in California using only 2% m/m oxygen. Abengoa Bioenergía chose to make the conservative assumption only to measure the dilution effect, but only for lack of European data that would demonstrate what appears to be the stronger octane substitution effect. 4

6 Sulphur (changes to BL1 and ETH1): The IES did not include sulphur in the fuel data. From 2009 the sulphur content in EU petrol is required to be no greater than 10 ppm. Abengoa Bioenergía was advised by Repsol-YPF that a likely average sulphur content for BL1 (in both Spain and Germany) will be around 7ppm. Since ethanol does not generally contain sulphur, for splash blending Abengoa Bioenergía diluted BL1 by 10%. Whether sulphur will be diluted or not in a match blend entirely depends upon the fuel components that the ethanol substitutes. This issue was not addressed by the IES. The E10 match blends in its EVAP study were had reduced sulphur compared to one of the base petrols but not for the other. Abengoa Bioenergía chose conservatively to assume no effect. Oxygen content (change to ETH1): The IES limited the oxygen content of E10 petrol at 3.5% mass. In reality this would only be true under two extreme conditions: o the ETH1 petrol would have no other oxygenates in the petrol; and o the BL1 petrol would all be sold at the highest density permitted by the European petrol standard. However, European petrol is generally sold at a much lower density than the highest density permitted by the European petrol standard. If E10 is splash blended into a petrol without other oxygenates that has the average density of European petrol, (the average between the minimum permitted and the maximum permitted density), the oxygen content is 3.7% m/m. The removal of the oxygenates reported in BL1 from ETH1 creates a data inconsistency in the IES analysis, since their removal would inevitably have a knock-on effect on all the other petrol characteristics. Most petrol in Europe today has oxygen in it. Therefore, E10 splash blending will inevitably lead to more oxygen in the petrol than that which is in the ethanol, unless an overall oxygen limit is set that precludes the other oxygenates. Is such a limit justified? The European Auto-Oil II programme technical study stated that: blends of gasoline with ethanol up to 22% (E22G) can be used in spark ignition engines without any material or other operating problems (Arcoumanis, 2000). The European Commission subsequently stated that: Most vehicles registered in the EU can technically run on a blend of fuel of up to 15% bioethanol. (EC, 2001). This statement covers both diesel 5

7 and petrol engines. This year, a study published by the Dutch Government agreed with the Commission commenting that: there seems to be a consensus that low ethanol blends in fuel (for example, less than 20% in petrol and less than 15% in diesel) can generally be used in unmodified spark ignition and compression ignition engines without any material problems or operating problems (Smokers and Smit, 2005). Bearing in mind that E10 has on average 3.7% m/m, then E20 has double that amount of oxygen. Abengoa Bioenergía therefore estimated oxygen content based on two different assumptions from the IES: o the average oxygen content of the 10% ethanol alone is 3.7% mass; and o that E10 can be splash blended to already oxygenated petrol. 3.2 RVP and VOC exhaust emissions The Copert III model does not include an equation that reflects the relationship between Reid Vapour Pressure (RVP) and VOC exhaust emissions as reported in the technical literature. Abengoa Bioenergía therefore examined this as a separate variable. The most recent exhaustive analysis of the European fleet that demonstrated this relationship that Abengoa Bioenergía knows of was reported by CONCAWE in Their comparative study of two petrols (one with an RVP of 61 kpa and the other with an RVP of 96 kpa) on 8 cars reported lower VOC exhaust emissions from the petrol with the higher volatility. The general result held true across the entire range of five tested temperatures from +25ºC to 15ºC. For the 4 oldest cars without catalytic converters and canisters the reduction in emissions was on average a non-significant 3.3%, but for the more modern cars with canisters and catalytic reductions the VOC exhaust was reduced by more than 6% on average. Abengoa Bioenergía assumed that the average increase in RVP from E10 splash blend will be 6 kpa. By interpolation from the CONCAWE report we estimated a reduction of VOC exhaust emissions of BL1 of 1.1% for each of the five months of splash blending. This approach can be criticised. First, the CONCAWE study on which the relationship between RVP and VOC exhaust emissions is based is now 12 years old. Secondly, a linear interpolation of those results can be no more than a best estimate. Moreover, it is not clear what effect if any the bioethanol might have on the result. Nevertheless, the inclusion of this effect can be justified from the observation that the best available scientific knowledge is being used and that many of the equations found in the Copert III model may be based on empirical CONCAWE data of the same vintage, so a criticism based on the historical nature of the relationship described by CONCAWE is perhaps really to criticise the exercise as a whole. 6

8 4 Analysis As a first step Abengoa Bioenergía was helped by the IES to calibrate the Copert III model and the data inputs to generate the results that the IES presented for Germany and for Spain at the April stakeholders meeting. Having generated BL1 and ETH1, Abengoa Bioenergía decided to make a preliminary assessment of the direction of the change in VOC emissions of each individual fuel parameter (6 within the Copert III model + CONCAWE s inverse relationship between RVP and VOC exhaust emissions), see Table 4, and the likely scale of the impact. Table 4: Fuel parameter impact on VOC emissions Fuel parameter Summer RVP E100 evaporation Aromatics Olefins Sulphur Oxygen content RVP & exhaust emissions Likely reduced VOC emissions from E10 Likely increased VOC emissions From E10 Two conclusions were drawn from the exercise. First, 6 of the 7 recommended changes in the fuel parameters would favour ETH1. Secondly, the E100 evaporation, aromatics, olefins and sulphur changes were likely to be of a lower order of impact than the other parameters so, to save time, Abengoa Bioenergía grouped these 4 parameters together. Consequently, 5 analyses were undertaken with respect to both the IES and the Copert default fuel input data for BL1 and ETH1: 1. Change in summer RVP, all other parameters held constant. 2. Change in oxygen content, all other parameters held constant. 3. Change in E100, aromatics, olefins and sulphur, all other parameters held constant. 4. Extension to IES and Copert default results by use of inverse relationship between RVP and VOC exhaust emissions. 5. Combination of changes to the model (1-3) and extension to the model (4). Abengoa Bioenergía decided to err on the side of conservatism for the E100, aromatics, olefins and sulphur, by assuming that these parameters would not be affected by match blending, despite the evidence to the contrary in the IES EVAP programme. This raised a further 7

9 problem, because the Copert III model is unable to disaggregate these fuel inputs at a seasonal or monthly level. The best accuracy obtained was by running the model with the changed inputs for 12 months and then to divide by five-twelfths in accordance with the duration of the summer period. Exactly the same procedure was necessary in order to obtain a first order estimate of the increased RVP on VOC exhaust emissions during the summer period. 5 Results Abengoa Bioenergía compared its modelled results with both those of ETH1 (the IES dataset based on the AEAT reported fuel data) and ETH2 (the Copert model default fuel data), and concluded that the results were consistent. We present below the comparison between the results of ETH1 and Abengoa Bioenergía s recommended changes to ETH1. Please note that for all tables a + sign indicates that the E10 (ETH1) data has increased VOC emissions compared to BL1, whilst a - sign means that the E10 has reduced VOC emissions. Summer RVP Tables 5 and 6 compare respective VOC emissions from IES ETH1 scenario with the changes made by Abengoa Bioenergía for summer RVP, all other parameters held constant. Table 5: Germany VOC emissions: varying summer RVP data IES ETH1 Abengoa Bioenergía ETH1 Evaporative emissions % +5.38% Total emissions -1.17% -1.73% Change in total emissions -0.56% Table 6: Spain VOC emissions: varying summer RVP data IES ETH1 Abengoa Bioenergía ETH1 Evaporative emissions % +5.61% Total emissions +1.57% -0.60% Change in total emissions -2.17% 8

10 Oxygen content Tables 7 and 8 compare respective VOC emissions from IES ETH1 scenario with the changes made by Abengoa Bioenergía for oxygen content, all other parameters held constant. Table 7: Germany VOC emissions: varying oxygen content IES ETH1 Abengoa Bioenergía ETH1 Exhaust emissions -2.44% -2.95% Total emissions -1.17% -1.64% Change in total emissions -0.47% Table 8: Spain VOC emissions: varying oxygen content IES ETH1 Abengoa Bioenergía ETH1 Exhaust emissions -2.84% -3.76% Total emissions +1.57% +0.90% Change in total emissions -0.67% E100 evapoation, aromatics, olefins and sulphur Tables 9 and 10 compare respective VOC emissions from IES ETH1 scenario with the changes made by Abengoa Bioenergía for E100 evaporation, aromatics, olefins and sulphur content, all other parameters held constant. Table 9: Germany VOC emissions: varying E100 evaporation, aromatics, olefins and sulphur content IES ETH1/BL1 Abengoa Bioenergía ETH1/BL1 Exhaust emissions -2.44% -2.59% Total emissions -1.17% -1.31% Change in total emissions -0.14% Table 10: Spain VOC emissions: varying E100 evaporation, aromatics, olefins and sulphur content IES ETH1/BL1 Abengoa Bioenergía ETH1/BL1 Exhaust emissions -2.84% -3.03% Total emissions +1.57% +1.43% Change in total emissions -0.14% 9

11 RVP and VOC exhaust emissions Tables 11 and 12 compare respective VOC emissions from IES ETH1 scenario with the changes made by Abengoa Bioenergía for the RVP relationship with exhaust emissions, all other parameters held constant. Table 11: Germany VOC emissions: RVP relationship with exhaust emissions IES ETH1/BL1 Abengoa Bioenergía ETH1/BL1 Exhaust emissions -2.44% -3.34% Total emissions -1.17% -2.00% Change in total emissions -0.83% Table 12: Spain VOC emissions: varying summer RVP data IES ETH1/BL1 Abengoa Bioenergía ETH1/BL1 Exhaust emissions -2.84% -3.86% Total emissions +1.57% +0.83% Change in total emissions -0.74% Combined results Tables 13 and 14 compare respective VOC emissions from IES ETH1 scenario with the changes made by Abengoa Bioenergía for all the parameters described above, all other parameters held constant. Table 11: Germany VOC emissions: combined revised parameters IES ETH1/BL1 Abengoa Bioenergía ETH1/BL1 Exhaust emissions -2.44% -3.56% Evaporative emissions % Total emissions -1.17% -2.75% Change in total emissions -1.58% Table 12: Spain VOC emissions: combined revised parameters IES ETH1/BL1 Abengoa Bioenergía ETH1/BL1 Exhaust emissions -2.84% -4.75% Evaporative emissions % +5.61% Total emissions +1.57% -2.00% Change in total emissions -3.57% 10

12 6 Findings We have two key findings to report. 1 The RVP fuel parameter is highly significant The difference in VOC emissions between ETH1 and BL1 in the IES study is precisely 0, with Germany having slightly reduced emissions and Spain having slightly increased emissions. Abengoa Bioenergía s RVP fuel parameter means that now both countries have reduced VOC emissions. In fact Spain has swung from having one of the highest increases in VOC emissions in the EU15 to having a greater reduction in VOC emissions than any of the other Member States reported by IES. A sensitivity analysis was undertaken on the Spanish data to determine just how high average RVP would need to increase in order for VOC emissions to balance (all other IES data held constant) and found it to be around 8 kpa. This is an average increase above and beyond Abengoa Bioenergía s predicted range of 5.0 to 6.5 kpa average increase. 2 The combined fuel parameters are highly significant. As 6 of the 7 adjusted parameters favour E10, the combined changes to the IES datset and the Copert dataset are greater than any single adjustment. The extension of the model to include the relationship between RVP and exhaust emissions only adds to the impact of the result. The significance of these results becomes apparent when comparing them with the worst result of all Member States in the IES BL1/ETH1 scenario. The IES analysis for ETH1 suggests that 6 Member States reduce VOC emissions with E10 petrol. In order that all 15 Member States have VOC emissions reductions with E10, the results need to swing by at least 2.2%. In the case of Germany, Abengoa Bioenergía s revised fuel parameters reduce VOC emissions by 1.6% and for Spain by 3.6%. Hence Abengoa Bioenergy s analysis indicates that if the revised parameters were applied to the EU15 net reductions in VOC emissions would be observed for most if not all 15 Member States. 11

13 7 Recommendations The IES has suggested that Abengoa Bioenergía propose to them to make two changes to the fuel parameters already found in the Copert III model. Abengoa Bioenergía observed that the combined recommended changes in the fuel parameters have a much greater impact than any individual parameter and that this combination is highly significant. It was also noted that the IES suggestion rules out consideration of the relationship between RVP and VOC exhaust emissions, because it is not found within the Copert III model. Believing these parameter changes to be fair given the best available scientific analysis, ebio would advise the IES to think again about before making what seem to us to be perhaps more of a political choice than a scientific one. However, following the IES suggestion, of the 6 parameters changed in the model the most significant are the summer RVP and the oxygen content. These are the two that ebio would suggest the IES to reconsider above all others. In the case of the RVP increase with E10 splash blend, if the IES is uncomfortable with Abengoa Bioenergía s proposed increase of 6 kpa, which is recognised to be no more than a best estimate, we would recommend communication with the US Environment Protection Agency who use fuel models with and without E10 based on their extensive experience of splash blending. In the case of the oxygen content, ebio proposes to add existing oxygenates to E10 petrol in line with the European Commission s observation that: Most vehicles registered in the EU can technically run on a blend of fuel of up to 15% bioethanol. (EC, 2001), and the fuel experts of the European Auto-Oil II programme and more recently of TNO who claim that E20 or more can be used in European petrol without problems (Arcoumanis, 2000 and Smokers & Smit, 2005), since this constitutes approximately 7.5% oxygen in petrol. Finally, we wish to raise a number of research issues: o Abengoa Bioenergía assumed that for match blending the ethanol will entirely substitute highly volatile petrol components such that E100 and other parameters such as aromatics and sulphur remain unaffected. The JRC s EVAP programme would suggest that we have been highly conservative. o Abengoa Bioenergía assumed that the only potential effect of E10 on aromatics is dilution, whereas the literature indicates a much more substantial octane substitution effect. o The CONCAWE data demonstrating the relationship of RVP on VOC exhaust emissions is now ageing and should be re-examined. 12

14 o Whatever the accuracy of the Copert III model in predicting the evaporative emissions of different petrols without ethanol, the inclusion of ethanol may have impacts not measured in the model. On the one hand, the literature suggests that the auto industry may have chosen fuel system materials that permit greater permeation with ethanol. On the other hand, the RVP indicator used as a variable in the model is believed to overstate the volatile emissions impact of E10 under real-life European temperatures. It is unclear to what degree these counteract one another. References Ademe, Mesures d émissions pollutants sur véhicules légers à allumage commandé, alimentés en carburants contenant de l éthanol et de l ETBE. Arcoumanis, C., A technical study on fuels technology related to the Auto-Oil II Programme - volume II: alternative fuels. Prepared for European Commission Directorate- General for Energy. CONCAWE, The effect of gasoline volatility on vehicle exhaust emissions at low ambient temperatures. Report no. 93/51. (Available at EC, Effect of fuel qualities and related vehicle technologies on European vehicle emissions: an evaluation of existing literature and proprietary data: Working Group, European Commission Industry Technical Group 1, Rev. 1. Kirchstetter, T.W. et al, Impact of oxygenated gasoline use in California light-duty vehicle emissions. Environmental science and technology, Vol. 30, pp , cited in Whitten, G.Z., (2004). Air quality and ethanol in gasoline. (Available at Smokers R., and R. Smit, Compatibility of pure and blended biofuels with respect to engine performance, durability and emissions: a literature review. SenterNovem report for the Dutch Government. 13