SIXTH FRAMEWORK PROGRAMME Project no: 502687 NEEDS New Energy Externalities Developments for Sustainability INTEGRATED PROJECT Priority 6.1: Sustainable Energy Systems and, more specifically, Sub-priority 6.1.3.2.5: Socio-economic tools and concepts for energy strategy. Deliverable n° 8.5 - RS 1b “Integration of External cost data and LCA data in TIMES and application to two policy issues” Denise Van Regemorter (CES K.U.Leuven) and F. Pietrapertosa, S. Di Leo and C. Cosmi (CNR-IMAA) I. INTRODUCTION The objective of this WP was to develop an approach for the integration of external cost and LCA data, as derived in stream RS1a and RS1b, into the TIMES modelling framework and to evaluate the expanded modelling framework with policy cases. The integration of external cost allows taking into account in the analysis of policy options the benefits/costs coming from the mitigation of emissions of pollutants. In section 2 the integration approach is briefly described1. In section 3 two groups of policy scenarios are examined: one with the Italian TIMES model on the impact of internalization of the cost of local pollutant and of greenhouse gas emissions (IMAA) and one with the Pan European TIMES model on the impact of the internalization of external cost of local pollutant (CES K.U.Leuven). These scenarios are meant to test the modelling approach and to illustrate what can be done with the fully integrated Pan European TIMES model but there is still a need for the further development of the emission data and the abatement technologies. II. INTEGRATION OF EXTERNAL COST AND LCA DATA IN TIMES A. External cost integration 1. TIMES modelling framework Within TIMES two approaches are modelled for the integration of damages from pollutants: 1. the environmental damage are computed ex-post, without feedback into the optimisation process 2. the environmental damages are part of the objective function and therefore taken into account in the optimisation process. The damage per pollutant is modelled as follows: DAM (env) EV coef (env) * EM (env) EV ( env ) where EV (env) reflects the relation between the damage cost and the level of emissions; EVcoef(env) is computed through calibration from the marginal cost, EVcost(env) , i.e. the derivative of the damage function w.r.t. emissions, and EVREFLEV(env), the emissions in the reference year, for which the marginal cost has been evaluated.: EV coef (env) EV cos t (env) /( EV (env) * EVREFLEV (env) ( EV ( env ) 1) ) In the first approach the damage function is used to compute ex-post the damage associated with the model solution. In the second approach, a term is added to the objective function, which contains the sum of the damage-functions per pollutant. The link with stream RS1b is through the marginal cost EVcost(env) derived from RS1b data. It is the marginal damage generated by the emission of a pollutant. There are several complex intermediate steps between the emission of a pollutant and the damages caused by it. These steps are not explicitly and separately modelled in TIMES, but the different steps are aggregated in the marginal damage figure used. This raises the question of the constancy of the marginal damage cost. From discussion with RS1b it seems that the linear approximation of the dose response function could be justified, but the dispersion/transformation function cannot be considered as linear. The linear approximation can be used when the scenario does not imply drastic changes compared to the reference case. However when the changes are important (e.g. when a 50% reduction in CO2 emissions are imposed) this is not justified and therefore this should be integrated in the data from RS1b or could be modelled through a specific value 1 See Integration report of Stream Integration for a more detailed overview 2 for EV (env), reflecting the non linearity between emission and air quality. At this stage, this was not possible within the timeframe of the project. As the computation are based on dose response functions which give the incremental damage from air pollution, the results should also be interpreted in these terms, i.e. in terms of the change in total damage compared to a reference year (the base year). The damages considered for TIMES are those linked to air quality, acid deposition and climate change generated from the use of energy; other categories linked to the use of a fuel or a technology within the energy system can also be considered, e.g. damage linked to the nuclear technology, if data are available. Ideally, it should be possible to link the region causing the damage and the region suffering the damage through dispersion/transformation matrices, though this is only needed for the ex-post evaluation of the damage per country. Within the optimization, it is sufficient to associate a damage figure in monetary terms to the emissions or to the activity/technology generating the damage within a country (the damage can be national, within the EU or outside the EU). The damage generated abroad because of the import of a fuel within the EU can also be taken into account through the association of a damage figure to the import category considered. At this stage the possibility of internalisation of external cost is modelled as a YES/NO option for all pollutant for which an external cost is given in the database. In a next version of the model code it will be possible to include or not each pollutant separately for internalisation, the ex-post computation covering the pollutant not included. 2. The external cost data from RS1b a) External cost of local air pollutant The local air pollutants considered for integration in TIMES were NMVOC, NO x, PPMcoarse (PM10), PPM2.5 and SO2. The damage factors for the local and regional pollutants derived by RS1b for TIMES are distinguished by region, by height of release (average and high) and by period (2010 and 2020) to allow for a certain variation in the damage costs caused by source characteristics and background emissions. At this stage, the background emissions in the RS1b data correspond to a BAU scenario. The damage categories considered by RS1b are impacts on human health, crops yield loss, damage to building materials and loss of biodiversity. The data from RS1b represent the damage generated by a country emission to Europe as a whole. RS1b could not yet evaluate the impact on the damage cost of a drastic reduction of the background emissions as would occur with a climate policy. Thus in the scenarios with TIMES, the same damage figures were used in all scenarios. The ‘high release’ figures were applied to the electricity sector, while the ‘average height’ for the other sectors. The table below report the data for the damage generated in EU27 as illustration, for the implementation in TIMES the country data were used. For each emission, a damage figure equal to the sum of damage for the different categories was associated to each pollutant emission, with equal weight for each category. Table 1: Damage generated in the EU (€2000/ton) Average Height 2010 2020 Health+Biodiversity NMVOC 516 190 NOX 6493 7487 PPMcoarse 1325 1381 PPM25 24412 24103 3 High Height 2010 2020 516 4945 490 12256 190 6186 481 11929 SO2 Crops+Materials NMVOC NOX SO2 b) 6247 6866 5708 5952 189 398 219 103 506 205 189 251 204 103 389 192 External cost of greenhouse gasses The evaluation of the marginal cost of greenhouse gases emissions and climate change is a very contentious issue, with high inherent uncertainties. Since the EU-funded “External costs of EnergyExternE” project began in 1990 (Externalities of Energy, 2005), the discussion on CO2 damage costs evaluation has been intense because of the very high variability of cost estimations related to assumptions behind the data. It was also an important point of discussion in the NEEDS project. An overview of CO2 external cost estimations done by different experts within and outside the project is reported in Table 4 (R. Friedrich, 2008). Table 2: Overview of the existing values of CO2 external cost, with and without equity weighting (EW). Source/ Explanation Kuik, Avoidance Costs GHG; in €2005 Target MNP, EU-targets until 2020; Marginal Abatement Costs; abatement compared to 1990; prices 2000 target -20% unilateral without CDM target -20% unilateral with CDM target -30% multilateral regimes for emission target 500ppm CO2e NEEDS, RS1B; Marginal Abatement Costs; converted with Methodex Conversion Tool in €2005 Watkiss: existing SCC central Watkiss: SCC average (all) Watkiss: Marginal Costs, until 2050 60% (only UK), prices 2005, converted with Methodex Conversion Tool in €2005 4 year 2010 2020 2025 2030 2040 2050 2020 2020 2020 2005 2020 2030 2050 2000 2010 2020 2030 2040 2050 2060 2000 2010 2020 2030 2040 2050 2030 2040 2050 Euro/t CO2 19 19 23 30 46 61 96 23 74 22 46 74 198 32 37 41 46 51 55 60 26 31 37 45 64 98 0-67 6-105 1-247 PAGE (in Watkiss); in year of emission; only mean; SCC; converted with Methodex Conversion Tool in €2005 with EW without EW FUND (in Watkiss); in year of emission; SCC; trimmed mean (trimmed 5%); converted with Methodex Conversion Tool in €2005 with EW no EW, average 1% trimmed Anthoff: Marginal costs, €2000 external damage WeuEW, average 1% trimmed Nordhaus: optimal carbon price, €2005 globally harmonized 550ppm CO2-eq. IPCC AR4 550ppm CO2-eq., technology integrated 2000 2010 2020 2040 2060 2000 2010 2020 2030 2040 2060 2000 2010 2020 2030 2040 2060 2005 2015 2025 2035 2045 2055 2005 2015 2025 2035 2045 2055 2005 2015 2025 2035 2045 2055 2065 2075 2085 2095 2105 2030 2050 2030 2050 21 28 35 58 86 11 13 17 20 21 26 21 26 31 34 37 46 7 11 14 15 17 27 97 122 148 137 143 196 6 10 12 15 17 23 27 32 37 43 50 15-62 23-119 4-50 12-100 After an exhaustive literature review and a lively internal debate, two set of reference values (scenarios) were identified (Table 3): 5 Ambitious scenario, a combination of the values proposed by RS1b (more ambitious scenario”) and RS1a ( “preferred scenario”) [Ricci A., NEEDS Internal Communication], taking in the recent policy decisions. Realistic scenario, which represent an average between the values proposed by Anthoff in the World EW scenario and Watkiss P. Table 3: GHG external costs (Euro/t of CO2). Ricci proposal (D 1.1 - RS3a, 2008). Scenarios Ambitious CO2 CH4 N2O Realistic CO2 CH4 N2O B. 2010 2015 2020 2025 2030 2035 2040 2045 2050 23.5 493.5 7285 31 651 9610 46 966 14260 51 1071 15810 74 1554 22940 87 1827 26970 110 2310 34100 146 3066 45260 198 4158 61380 23.5 493.5 7285 27 493.5 7285 29 567 8370 32 609 8990 34 672 9920 37 714 10540 50 777 11470 66 1050 15500 77 1386 20460 LCA data integration The TIMES model has the capability to represent energy consumed/produced and emissions caused by the operation of technologies, but also those attached to the construction of technologies, as well as energy and emission releases at dismantling time. It is also possible to model materials sunk at construction and released at dismantling, but only if these materials are represented in the TIMES model. The key problem in integrating LCA and energy modelling is to avoid double counting of energy and emissions. The LCA data distinguishes different phases for a technology, construction, operation and dismantling and can generate for each phase the associated energy consumption and emissions release. The energy consumed and produced at phases 2 and 3 are fully modelled in TIMES (including fuel supply), and require no additional transfer of data from RS1a to TIMES. There remains the issue of accounting for energy, materials and emissions attached to phases 1 and 4 (construction and dismantling). The general approach followed for the integration of LCA data into TIMES was to exclude the material and energy flows included in the LCA datasets for construction and dismantling from the material and energy flows modelled in TIMES. LCA data would then provide the cumulative emissions (and other externalities such as radioactive emissions, land use, etc) attached to phase 1 (respectively phase 4). By cumulative emissions, we mean the emissions actually released during construction (respectively dismantling) plus the emissions produced during the upstream production of fuels and materials required for construction (respectively dismantling), including fuels produced outside EU25. These emissions are provided by RS1a per unit of capacity of the plant. It should however be realized that it is very difficult to isolate the flows of materials and fuels linked to construction and dismantling of the technologies. Those flows are not explicitly represented in TIMES, but included in the total flows modelled. Moreover, part of the construction of a technology does not necessarily occur in the country where it is implemented. Therefore to take into account the difference in external cost linked to the technologies and at the same time avoid possible double counting, the implementation in TIMES was simplified. Only the difference compared to a reference was implemented in TIMES, i.e. the difference from the emissions derived from an average of existing technologies, this average being implicitly reproduced in the reference scenario. It is expected that the approximation will be acceptable, since the amounts of materials and fuels involved are small relative to the overall 6 consumption of fuels and materials by the economy. Then TIMES attach this coefficient to the technologies and combine them with external cost coefficients provided by RS1b to endogenize the external cost for the construction and dismantling phases. The data implemented in TIMES for some typical power technologies are given in the table below; Table 4: Emission coefficient for typical technologies (difference compared to average existing technologies) Coal PP CO2 NOx NMVOC PM2.5-10 PM2.5 SO2 ton/MW kg/MW kg/MW kg/MW kg/MW kg/MW 70 22 27 119 197 213 Coal PP with CC 362 554 186 425 490 745 Nuclear 100 83 29 22 72 114 Wind offshore 0 0 0 6 0 101 Fuel cell on gas 294 482 120 172 136 3357 PV 162 281 625 0 0 1060 At this stage, only data for power generation technologies are provided by RS1a and this might introduce a bias in the choice of technologies in the TIMES model. The additional emissions remaining rather small, the bias will also remain small. III. THE POLICY SCENARIOS A. The Italian policy scenario 1. Basic modelling enhancements The NEEDS Italy model, carried out in the framework of RS2a activities (D. 5.14, RS2a, 2006) was improved to enable a detailed representation of renewable energy cycles and their potential development as well as to allow a full characterisation of the pollutants associated to the different fuels and energyrelated technologies. This mainly led to an in depth technical review of the model in order to associate to energy flows by sector additional emission factors for local air pollutants (SO2, NOx, PM10, PM2.5, VOC) as well as for GHGs (CO2, CH4, N2O). A calibration of the model on UNFCCC data (APAT, 2007) was also made, to verify the consistency of the model’s outcomes with the official national inventories. The improved representation of pollutants emissions allows to account for 94% of SO2, 84% of NOx, 95% of VOC, and for 92% of the total GHG emissions for year 2000 (about 448 Mton of CO2eq, where the remaining 8% could be imputed to a different fuel aggregation in the supply sector). The external costs for local pollutant as derived by RS1b for Italy were implemented and are given in the table below. Table 5: External costs of LAP – Italy (€2000/ton) NMVOC 2001 2005 2010 2015 2020 639 670 712 343 364 High height of release NOX PPM10 PPM25 6602 7034 7617 9175 9933 650 696 757 827 900 13404 14339 15600 16977 18470 SO2 NMVOC 6566 7019 7629 8773 9540 639 670 712 343 364 7 NOX Unknown height of release PPM10 PPM25 8362 8907 9640 11411 12353 1730 1851 2014 2190 2382 29303 31347 34104 35305 38409 SO2 7434 7947 8640 9938 10808 2025 2030 2035 2040 2045 2050 386 411 424 438 452 467 10757 11653 12133 12633 13155 13699 979 1065 1111 1159 1209 1261 20094 21861 22806 23792 24821 25894 10375 11284 11770 12277 12805 13357 386 411 424 438 452 467 13377 14492 15089 15711 16359 17036 2592 2820 2941 3069 3201 3340 41787 45462 47427 49477 51616 53847 11755 12785 13335 13910 14510 15135 For GHG cost, the higher values of the ambitious scenario values were used to perform the Italy test runs in order to highlight their influence on the model’s choice in terms of technology and fuel mix. 2. The Scenario analysis a) Scenario assumptions The scenario analysis was mainly addressed to emphasise the policy implications of external costs in terms of model’s choices. Therefore, two reference scenarios and several case variants were defined, whose basic assumptions are coherent with the ones adopted for the policy scenarios selected at Pan European level (Kypreos & Van Regemorter, 2006; Kypreos et al., 2006; Cosmi et Al.,2007): BAU (Business as Usual): the baseline scenario, whose macroeconomic and energy price background assumptions are in line with DG TREN 2005 projections (Mantzos et al., 2005) and all the exogenous assumptions around drivers, energy prices and policies follow a rather business as usual trend. No climate policy is considered. KYOTO_FOREVER: a climate policy scenario aimed to achieve the national Kyoto Protocol’s target (-6.5% of GHGs in the period 2008-2012 compared to the 1990 values2). Thus, starting from the values of the reference scenario, a reduction of 448 Mton of CO2eq was imposed from 2010 to 2050 to model the GHGs stabilisation on the full time horizon. No tradable permits or flexible mechanisms are allowed to achieve the prefixed -6.5% target. Starting from these reference scenarios, the following case variants were defined to analyse the behaviour of the model due to the integration of the external costs: b) BAU_GHG: Baseline scenario assumptions with the internalisation of the externalities related to CO2, CH4 and N2O only (the “Ambitious scenario” values are considered) on BAU. BAU_LAP: Baseline scenario assumptions with the internalisation of the externalities related to local air pollutants only (SO2, NOx, NMVOC, PM10 and PM2.5). BAU_LAP-GHG: Baseline scenario assumptions with the internalisation of the externalities on both local and global air pollutants, including CO2 (the “Ambitious scenario” values are considered). Kyoto_LAP: KYOTO_FOREVER scenario assumption with the internalisation of the externalities on local air pollutants (SO2, NOx, NMVOC, PM10 and PM2.5). Scenarios comparison In order to compare the effects of the internalisation of damage costs on the total system cost, all the case variants were analysed with both the damage included in the optimization (ex-ante) and damage computed ex-post (Table 6). Table 6: Total discounted system cost with and without the internalization of external costs 2 Total emission WITHOUT land-use, land-use change and forestry 8 Total discounted system cost (MEuro) Case variants ex post energy system cost ex ante damage total energy system cost damage total BAU_GHG 5159 GHG dam 608 5767 5196 531 5728 BAU_LAP 5159 LAP dam 584 5744 5224 454 5679 BAU_LAP_GHG 5159 GHG and LAPdam 1192 6351 5246 999 6245 Kyoto_LAP 5244 LAP dam 5774 5273 475 5748 531 The internalization of the external cost (ex-ante approach) induces a general increase of energy system costs (excluding damage cost) due to new investments in technologies more efficient and with lower environmental impacts. On the total discounted system cost, including damage cost, the effect is a reduction in all cases, up to 1.7% in BAU_LAP_GHG case. In terms of avoided damage costs, the combined policy turns out to be less beneficial: the reduction is 193 MEuro in the BAU_LAP_GHG case, against 206 MEuro summing up the reduction of BAU_LAP and BAU_GHG cases (respectively 130 and 77 MEuro). This is due to the lower LAP emissions (that have the higher external costs per unit of pollutants) in BAU_LAP case compared to BAU_LAP_GHG (Figure 3) with a reduction of respectively -26.5% and -21.4% over entire time horizon. Nevertheless, the effectiveness of combined policies on LAP and GHGs, is clear in terms of GHG emissions trends (Figure 1, Figure 2). The BAU_LAP case variant shows a strong increase of GHGs (+25%), whereas in the BAU_LAP_GHG case (-17%) the internalization of both externalities fosters a reduction of GHGs stronger than in the BAU_GHG case (-15%). Figure 1: Total GHG emissions (Mton CO2eq). BAU BAU_GHG BAU_LAP BAU_LAP_GHG KYOTO KYOTO_LAP 650 GHG emissions (MtonCO2eq) 600 550 500 450 400 350 300 2000 2010 2020 2030 9 2040 2050 Concerning CO2 emissions the internalisation of externalities (BAU_GHG scenario) is effective on long term but not in the short term, highlighting that only the higher values of CO 2 external costs have an influence on the energy system configuration changes. As shown in Figure 2, a higher CO2 reduction is obtained in the conversion sector, through carbon capture processes that give their maximum contribution (about 23%) in BAU_GHG case, showing the lowest CO2 emissions in 2050. In fact, in this case, the highest contribution to CO2 emissions reduction is due to Conversion and Transport (respectively about -78% and -37% respect to the reference emissions of year 2000), obtained besides the use of carbon capture processes by fuel switching to less carbon intensive fuels (in particular, biofuels use in Transport). Households, Commercial and Agriculture that shows a 4% reduction of CO2 emissions in the BAU scenario and about 13% in presence of a Kyoto target, reduce their CO2 emissions about 17% in the BAU_LAP_GHG case. On the other hand, Industry increases its CO2 emissions in every scenario and case variant (from 95% of BAU scenario to 43% of BAU_GHG_case variant) because of a larger use of natural gas consequent to an increase of demand. In particular, natural gas consumption in industry sector doubled in all the case variants, reaching the higher value in Kyoto_LAP case (from 906 PJ in 2000 to 2412 PJ in 2050) compared with a 72% increase of energy intensive industry’s demand and a 32% demand increase for non energy intensive industry. Figure 2: Carbon emissions per sector (Mton/year). 600 CO2 sequestration Emissions of CO2 [Mio t] 500 Transport 400 300 Households, commercial, AGR 200 Industry 100 Conversion, production 2010 2030 Kyoto_LAP Kyoto_forever BAU_LAP_GHG BAU_LAP BAU_GHG BAU Kyoto_LAP Kyoto_forever BAU_LAP_GHG BAU_LAP BAU_GHG BAU Kyoto_LAP Kyoto_forever BAU_LAP BAU_LAP_GHG BAU BAU_GHG 2000 0 2050 The strong increase of CO2 emissions that can be noticed in the BAU_LAP case, is mainly caused by an increase in natural gas consumption used in a Fischer-Tropsch process to produce synthetic diesel fuel (Fischer-Tropsch diesel) characterised by a very low sulphur and aromatic content (Norton, 1998) and by the absence of the carbon capture processes that, without specific constraints on GHGs emissions, are not activated. Table 7 reports the CO2 marginal abatement costs in Kyoto_forever scenario and BAU_GHG case and the percentages CO2 emission reductions obtained compared to base year. As can be seen, on the long 10 term in the BAU_GHG the CO2 damage is much higher than in the KYOTO scenario and induces thus a higher CO2 reduction respect to Kyoto cap scenario. In the short term, the CO2 damage figure used here is not sufficient to achieve the Italian Kyoto target. Table 7: CO2 costs and reduction. 2010 2030 2050 marginal cost (Euro/ton) CO2 reduction 230 12% 230 14% 100 14% marginal cost (Euro/ton) CO2 reduction 20 2% 70 3% 200 21% Kyoto_forever BAU_GHG An in-depth examination of the local air emission pollutants (NOx, SO2, NMVOC and particulates (Figure 3) shows that apart from the decreasing trend due to technical improvement already in the BAU, the internalization of external costs is effective to obtain a further reduction with or without an explicit constraint on GHGs. In particular, the reduction percentage is the highest (about -42%) in the BAU_LAP_GHG case, with a remarkable reduction of NMVOC (-55%). As no specific abatement technologies were inserted in the model (e.g. end of pipe technologies), the abatement of LAP emissions is achieved by fuel switching and technology substitution. The cost of the policy or the percentage reduction could therefore change with the model improvements. Figure 3: Local air pollutants Nox NMVOC Particulates SO2 Emissions of LAP [Mio t] 3500 3000 2500 2000 1500 1000 500 2010 2030 Kyoto_LAP Kyoto_forever BAU_LAP_GHG BAU_LAP BAU_GHG BAU Kyoto_LAP Kyoto_forever BAU_LAP_GHG BAU_LAP BAU_GHG BAU Kyoto_LAP Kyoto_forever BAU_LAP_GHG BAU_LAP BAU_GHG BAU 2000 0 2050 Table 8 summarizes the situation by scenario relatively to the emissions of the main pollutants on the overall time horizon. Table 8: Pollutant emissions by scenario from 2000 to 2050. 11 Scenario Total emissions (Mt) Case variants CO2 BAU BAU_GHG BAU_LAP BAU_LAP_GHG VOC Particulates 24.70 68.40 40.93 14.60 21934 24.17 63.15 35.52 13.42 (-10.6%) (-3.3%) (-8.8%) (-14.2%) (-7.1%) 27041 22.97 (+10.3%) 21039 25.98 13.27 (-8.1%) (-30.7%) (-37.2%) (-8.1%) 11.66 (-14.2%) (-10.7%) (-23.3%) (-25.3%) (-19.2) 21101 22.32 47.97 30.93 (-14.0%) Kyoto_LAP NOx 24523 21101 KYOTO_FOREVER SO2 23.41 53.09 33.40 13.14 (-6.3%) (-12.2%) (-19.3%) (-9.0%) 22.33 60.75 51.43 28.85 11.50 (-14.0%) (-10.6%) (-25.7%) (-30.3%) (-20.4%) To assess the effects of internalization of external costs, it is also worth investigating the behaviour of the energy system in terms of energy supply and consumption. According to demand projections, primary energy consumption increases in all scenarios (Figure 4), from the initial 6799 PJ of the reference year to 7329 PJ in 2050 for the BAU_LAP_GHG case and 8063 PJ for Kyoto_LAP. The increased energy demand is fulfilled mainly by renewable and natural gas. In particular, there is a huge increase of renewable in the Kyoto_LAP case (about 21% in 2050), whereas natural gas consumption is more or less doubled in the BAU_LAP case, to compensate a drastic decrease of oil consumption (quite halved as expected). Figure 4: Primary energy consumption. 12 Electricity import 8000 Waste 7000 6000 Other renewables 5000 4000 3000 Hydro, wind, photovoltaic 2000 Natural gas 1000 0 2010 2030 Kyoto_LAP Kyoto_forever BAU_LAP_GHG BAU_LAP BAU_GHG BAU Kyoto_LAP Kyoto_forever BAU_LAP_GHG BAU_LAP BAU_GHG BAU Kyoto_LAP Kyoto_forever BAU_LAP_GHG BAU_LAP BAU_GHG BAU Oil 2000 Primary Energy Consumption [PJ] 9000 Lignite Coal 2050 This behaviour is emphasized by the electricity production modes (Figure 5). In particular, the endogenous production increases from 254 TWh of year 2000 to 343.6 TWh in 2050 for the Kyoto_LAP case. Electricity from fossil fuels is produced mainly by coal, enabling the use of technologies provided with sequestration processes to diminish the CO2 emissions, and natural gas (that increases about 74% from 2000 to 2050 without any constraint whereas internalizing external costs of GHGs and LAP emissions this percentage decrease up to 49% in the BAU_LAP_GHG case). Among renewables, it can be noted, a remarkable increase of wind energy (from 0.6 TWh in 2000 to a maximum of 27.8 TWh without the internalization of externalities on LAP and 21.5 TWh with the internalization of externalities on LAP). Photovoltaic also increases (+68% comparing the share in 2050 of all the constrained cases with BAU) even if its contribution is still very low (about 2.2 TWh). Figure 5: Net electricity generation (TWh). 13 Others Net electricity production [TWh] 400 350 300 Solar photovoltaic 250 Wind 200 Hydro 150 100 Natural gas 50 Oil 2010 2030 Kyoto_LAP Kyoto_forever BAU_LAP_GHG BAU_LAP BAU_GHG BAU Kyoto_LAP Kyoto_forever BAU_LAP_GHG BAU_LAP BAU_GHG BAU Kyoto_LAP Kyoto_forever BAU_LAP_GHG BAU_LAP BAU_GHG BAU 2000 0 Lignite Coal 2050 Similarly to primary energy consumption, the final energy consumption increases in all scenarios from 2000 to 2050 (Figure 6), with a maximum in the BAU scenario (+25%) a minimum in the BAU_GHG and BAU_LAP_GHG cases (+21%). The most used fossil fuels on long term in all scenarios are obviously natural gas and oil products, but there are significant differences in their share along the examined cases. In particular, oil products share in 2050, which is 36% in the BAU scenario, up to 18% in the BAU_LAP case and about 24% in the Kyoto_LAP case. On the contrary, natural gas share is about 30% in the BAU scenario and increases up to 38% in the Kyoto_LAP case, with an average increase of about 52% respect to base year. There is also a remarkable increase of renewable use in all the scenarios, whose share in 2050 is more than 14% in BAU_LAP_GHG and Kyoto_LAP cases, their consumption increasing from 81 PJ of the base year to respectively 989 and 1028 PJ. The strong increase of other fuels (“Others”) in the BAU_LAP case (about 938 PJ) is fostered by a large use of FT-diesel in Transport sector. Figure 6: Total final energy consumption. 14 Total final energy consumption [PJ] 8000 Others (Methanol, Hydrogen) 7000 Waste 6000 5000 Renewables 4000 Heat 3000 2000 Electricity 1000 Gas 2010 2030 Kyoto_LAP Kyoto_forever BAU_LAP BAU_LAP_GHG BAU BAU_GHG Kyoto_LAP Kyoto_forever BAU_LAP_GHG BAU_LAP BAU_GHG BAU Kyoto_LAP Kyoto_forever BAU_LAP_GHG BAU_LAP BAU BAU_GHG 2000 0 Petroleum products Coal 2050 The effects of internalization of externalities in terms of renewable share in final energy consumption are summarized in Table 9. The 2020 objective of 17% share of energy from renewable sources stated for Italy by the European Directive on the promotion of the use of energy from renewable sources is nearly reached in the cases with GHG cap (Kyoto_forever, Kyoto_LAP), whereas the externalities alone are not effective to give a boost to use of renewables. In 2050, all the cases increase the RES share until 24% for BAU_LAP_GHG case. Table 9: Impact of internalization on renewable use – Final energy consumption. Scenario Case variants BAU BAU_GHG BAU_LAP BAU_LAP_GHG KYOTO_FOREVER Kyoto_LAP B. % of renewable in final energy consumption 2000 2010 2020 2050 4% 5% 7% 12% 5% 11% 22% 5% 6% 13% 5% 15% 24% 4% 10% 16% 20% 9% 16% 21% The European policy scenario The European policy3 evaluated with TIMES in this WP aims at internalizing the external cost linked to local pollutant emissions from the energy system. This policy could be implemented through an emission tax differentiated by country and equal to the external cost generated by the country emissions. The results for this scenario are compared to the reference scenario of RS2a WP4, in which a small CO2 tax is 3 The Pan European TIMES model includes, besides the EU 27 countries, Norway, Switzerland and Iceland 15 implemented from 2010 onwards (5€2000/ton CO2)4. The internalisation is implemented from 2010 onwards. This scenario is meant to illustrate the integration of external cost in TIMES, other scenarios with the European model combining different environmental targets are done in RS2a, WP4. 1. System cost The cost of the policy (without including the damage reduction) represents an increase of 0.6% compared to the reference. When the damage reduction is included, the policy generates a benefit, 1.6% compared to the reference, as can be seen in Table 10. The reduction in damage is already effective from the start of the policy and discounted, it is around 35% over the entire horizon compared to the reference. It should be mentioned that, except for transport and the power sector, no specific local air pollution is assumed in the reference scenario. Also, the perfect foresight and optimisation approach in TIMES limit the adjustment cost of such a policy, which would be observed in real life. Table 10: Total EU discounted cost 2000-2050 (% difference compared to reference) incl. damage -1.6% -10.2% -0.5% total discounted cost difference as % GDP2000 annualized cost as % GDP2000 2. excl. damage 0.6% 3.8% 0.2% Emissions The reduction in local pollutant emissions is rather drastic. The local air pollution policy has however only a small impact on the CO2 emissions, except at the beginning and the end of the horizon. Table 11: Air pollutant emissions in the EU (% difference compared to reference) NOX PM SO2 NMVOC CO2 3. 2010 -29% -70% -58% -41% -9% 2020 -40% -74% -47% -42% 1% 2030 -38% -72% -44% -39% -1% 2040 -30% -70% -38% -34% -5% 2050 -23% -68% -36% -29% -8% Energy The reductions in emissions occur through four mechanisms: reduction in demand for energy services energy saving through improved efficiency fuel switching end of pipe abatement technologies The last option is however rather limited because the technology database of TIMES does not contain yet many end of pipe technologies. 4 The reference is slightly different from the reference used for the Italian model and coresponds to an update of the NEEDS reference in 2008. 16 a) Impact on demand The reductions in demand are concentrated in the industry and commercial sector. In the transport sector the decrease is smaller because already in the reference more drastic emission controls are implemented through the euro-norm and this limits the price increase through the internalisation. The high decrease in agriculture can be explained by the lack of other reduction options in that sector, which is modelled until now in a very generic way with no fuel or technology switch possibilities. Table 12: Energy services/Material demand in the EU (% difference compared to reference) 2010 -10.5% -3.4% -1.6% -0.5% -0.2% -3.4% -0.8% -2.7% -4.1% -2.6% -3.0% -2.5% -0.8% 0.0% -1.2% Agriculture (PJ) Commercial (PJ) Residential (PJ) Freight transport (tkm) Passenger transport (pkm) Industry (PJ) Industry Aluminium demand in ton Industry Ammonia demand in ton Industry Cement and Lime demand in ton Industry Copper demand in ton Industry Glass demand in ton Industry Iron and Steel demand in ton Industry Paper demand in ton Aviation transport (PJ) Navigation transport (PJ) b) 2020 -11.4% -2.2% -0.6% -0.7% -0.2% -3.4% -0.3% -2.3% -4.3% -2.8% -1.8% -2.7% -0.5% 0.0% -1.4% 2030 -12.0% -2.0% -0.7% -0.6% -0.3% -3.4% -0.3% -2.6% -5.1% -1.8% -2.5% -2.8% -0.6% 0.0% -1.4% 2040 -12.5% -1.5% -0.6% -0.6% -0.2% -3.9% -0.2% -2.3% -6.2% -3.5% -2.3% -2.7% -0.9% 0.0% -1.4% 2050 -12.0% -1.5% -0.5% -0.6% -0.1% -3.9% -0.3% -2.4% -6.4% -2.6% -2.0% -2.6% -0.9% 0.0% -1.4% Impact on final energy consumption The decrease in final energy consumption lies between 5 and 7%, the higher decreases are observed between 2020 and 2030. Table 13: Final energy consumption by fuel and by sector in the EU (difference compared to reference in PJ and in %) By fuel in PJ Bioenergy Solid fuels Electricity Gas Heat Oil Renewables Synthetic fuels Total By fuel in % Bioenergy Solid fuels Electricity Gas 2010 -109 -1034 -23 -1595 28 -1593 0 1724 -2602 2020 -128 -1403 190 -1768 82 -5342 52 4958 -3358 2030 -327 -1761 238 -1527 -49 -4709 87 4617 -3431 2040 -304 -1763 198 -1532 -83 -1787 142 2272 -2857 2050 -341 -1699 223 -1344 -67 -298 42 711 -2772 -12% -27% 0% -12% -8% -33% 2% -13% -11% -38% 2% -13% -8% -35% 2% -13% -8% -33% 2% -11% 17 Heat Oil Renewables Synthetic fuels Total By sector in PJ Agriculture Commercial Industry Residential Transport By sector in % Agriculture Commercial Industry Residential Transport 1% -8% -1% 46236% -5% 4% -26% 249% 16804% -6% -2% -24% 374% 369% -6% -4% -11% 439% 50% -5% -3% -2% 22% 9% -5% 2010 -125 -1026 -824 -577 -50 2020 -140 -1350 -905 -862 -102 2030 -157 -1222 -1081 -828 -144 2040 -172 -1031 -1246 -285 -122 2050 -172 -1046 -1180 -226 -149 -11% -17% -5% -4% 0% -11% -20% -5% -7% -1% -12% -18% -5% -8% -1% -13% -15% -6% -3% -1% -12% -14% -5% -2% -1% The highest decrease is observed in the commercial sector, mainly through a shift towards heat pump on electricity in place of bio and gas heating system. The same is observed in the residential sector but to a lesser extent. There is also a shift towards LPG for coking and hot water. In the transport sector synthetic fuels, here DME from coal, is penetrating more and much faster than in the reference scenario, from 2020 onwards instead of 2040 and mostly for freight transport by truck. In the industry, there is mainly a shift from solid fuels to bioenergy, triggered by their relative emission coefficients. The shifts between fuels in the different sectors are mainly driven by the differences in their emission coefficients for local pollutant and this must clearly be further examined. c) Impact on electricity production and other supply The overall production of electricity is approximately 2% higher over the entire horizon, except for the first period (2010). This reflects the shift towards electricity in the final energy consumption. Within the electricity sector there is a shift away from solid fuels towards gas, biomass (in CHP) and renewables. Nuclear is penetrating more (+5%) at the end of horizon in the countries where the option is available. Table 14: Net electricity production by fuel type in the EU (difference compared to reference in PJ) Solids Gas Oil Nuclear DME Biogas Biomass Renewables 2010 -1030 757 -29 0 0 1 17 144 2020 -411 611 -31 -4 0 -1 47 -54 18 2030 -649 616 -13 -3 0 -1 71 168 2040 -687 639 0 0 0 -5 54 145 2050 -673 568 -1 131 0 -3 99 58 -139 -1.2% Total Total (%) 156 1.3% 189 1.4% 147 1.0% 179 1.2% Coal and lignite steam power plants are replaced by IGCC power plants, CHP on coal are replaced by CHP on gas and biomass. Renewables: wind and hydro are penetrating more rapidly in 2010 but then less in 2020 to remain stable thereafter. Solarthermal and wave powerplants see their share slightly increasing with 1% up to 5.5% at the end of the horizon. Combining the fuel shifts in final consumption and in the electricity production, the overall share of renewables and bioenergy as computed for the EU renewable target5 increases with 1.3% reaching 14.1% in 2020. The carbon tax imposed in both scenarios induces carbon sequestration mostly through afforestation in the reference scenario because it is the cheapest option but with a limited capacity in TIMES. With internalisation of local pollutant and the production of DME from coal, the cost minimisation induces more carbon sequestration. As the technology for DME is modelled without carbon sequestration, the model chooses technologies in the electricity sector with carbon sequestration from 2020 onwards while the supply sector sequesters carbon through afforestation. The electricity produced with these technologies doubles compared to the reference but represent still only 6% of the total. d) Impact on primary energy consumption The impact of the policy in the different energy consumption sectors is reflected in the impact on primary energy consumption. There is a shift from oil to coal because of the fuel switch in the transport sector. The increase in renewables reflects their penetration in the electricity sector and in the final energy consumption. However overall, the reduction in the primary energy is not higher than 3%. Table 15: Primary energy consumption in the EU (difference compared to reference in PJ and in %) in PJ Bioenergy Gas Nuclear Oil products Solid fuels Renewable Total in % Bioenergy Gas Nuclear Oil products Solid fuels Renewable Total 5 2010 -107 69 0 -1642 -2149 144 -3685 2010 -8% 0% 0% -5% -17% 5% -5% 2020 -105 -499 -11 -5400 4935 7 -1073 2020 -5% -3% 0% -17% 40% 0% -1% 2030 -443 -267 -7 -4750 3700 255 -1513 2030 -14% -2% 0% -15% 23% 5% -2% 2040 -525 -368 -1 -1817 105 289 -2316 2040 -13% -3% 0% -6% 0% 5% -3% 2050 -395 -554 364 -320 -2227 105 -3027 2050 -9% -4% 4% -1% -8% 2% -3% It is computed as the share of electricity from renewables and renewable final energy consumption in total final energy consumption (incl. electricity losses). 19 The figure hereafter gives another view of the impact of the internalisation on the composition of the primary energy. Figure 7: Primary energy consumption in the reference and the internalisation scenario (PJ) 100000 90000 80000 Renewable 70000 Solid fuels 60000 Oil products 50000 Nuclear 40000 Gas 30000 Bioenergy 20000 10000 2000 2010 2020 2030 2040 REF_decinterlocn REF_decn REF_decinterlocn REF_decn REF_decinterlocn REF_decn REF_decinterlocn REF_decn REF_decinterlocn REF_decn REF_decn 0 2050 IV. CONCLUSION The analysis above is mainly meant to evaluate the development of the TIMES model regarding externalities linked to energy consumption. From a methodological point of view it was important to set up an integrated modeling framework covering all environmental dimensions linked to energy that allows to exploit synergies and trade-offs between the various targets imposed on the energy system (climate, air quality, energy security, import dependence). In particular, this modeling framework is particularly suitable to explore the medium to long term effects of energy, environmental and resources use policies on the energy system structure and on its cost. With the Italian model, the focus was on the assessment of combining climate and air quality policies. It showed that the internalisation of externalities of both GHGs and LAP simultaneously reduces the total discounted system cost compared to the case where the internalization is done separately for the two environmental issues because of the exploitation of the synergies. The results obtained so far emphasize the role of renewable and carbon capture in the mix of options to reduce emissions. With the European model the test was on the internalization of the external cost of local air pollution alone. It shows that a policy internalizing the external cost of local pollution could be very effective for local pollution but has only a small effect on CO2 emissions. 20 Both scenario exercises show the importance of having an integrated model covering the total energy system because of the interaction between its different components and the synergies and trade-offs between policies adressing issues faced by the energy system. But further improvements of the model technology database are clearly needed to draw final conclusions from this scenario analysis. First, the emission coefficients associated with the fuels and technologies have to be further examined because they play a crucial role in the choice of technologies and fuel switching (e.g. in transport) for reducing emissions. Secondly more end-of pipe abatement technologies for local pollutant have to be included for a better trade-off between the different options. V. REFERENCES The Italy NEEDS TIMES model – D5.14 RS2a NEEDS Project Italian Greenhouse Gas Inventory 1990-2005. National Inventory Report 2007. Annual Report for submission under the UN Framework Convention on Climate Change and the European Union’s Greenhouse Gas Monitoring Mechanism APAT - Agency for Environmental Protection and Technical Services. available at http://unfccc.int/national_reports/annex_i_ghg_inventories/national_inventories_submissions/ite ms/3929.php D 1.1 - RS3a “Report on the procedure and data to generate averaged/aggregated data”. P. Preiss, R. Friedrich and V. Klotz. August 2008 Externalities of Energy. Methodology 2005 Update. Edited by Peter Bickel and Rainer Friedrich. Institut für Energiewirtschaft und Rationelle Energieanwendung — IER. Universität Stuttgart, Germany. 2005 R. Friedrich. Internal Communication “Note on the choice of values of marginal external costs of greenhouse gas emission” June 30, 2008 D3.13 - RS 2a Technical Report. "Interim Report on draft Pan European integrated model" D. Van Regemorter, C. Cosmi, M. Salvia, K. Smekens, A. Kanudia, R. Loulou, S. Kypreos, M. Blesl, D. Bruchof, T. Kober. November 2007 Antti Lehtila, Richard Loulou. TIMES Damage functions. Energy Technology Systems Analysis Programme. TIMES Version 2.0 User Note. November, 11 2005 Kypreos, S., and Van Regemorter, D. Scenarios to be generated with the TIMES model for NEEDS. NEEDS Internal working paper RS2a WP2.3, 27 February 2006. Kypreos, S., Van Regemorter, D., and Guel, T. Key Drivers for Energy Trends in EU; Specification of the Baseline and Policy Scenarios. NEEDS Internal working paper RS2a WP2.3, 12 January 2006. Cosmi C., Blesl M., Cuomo V., Kypreos S., Van Regemorter D. The NEEDS scenarios: Final configuration. II NEEDS Policy Workshop, Ljubljana, March 9, 2007. http://www.needsproject.org. Mantzos, L. et al. (2005): European energy and transport scenarios on key drivers, published by DG TREN, Brussels Norton P., Vertin K. D., Bailey B. K., Clark N., Lyons D. W., Goguen S. J., Eberhardt J. J. Emissions From Trucks Using Fischer-Tropsch Diesel Fuel. SAE Technical paper series. ISBN 982526. 1998. COMMISSION OF THE EUROPEAN COMMUNITIES Proposal for a Directive of the European Parliament and of the Council on the promotion of the use of energy from renewable sources. COM(2008) 19 final. 21