Selecting cost-effective abatement measures to achieve good water status with the Environmental Costing Model.

Transcription

1 Selecting cost-effective abatement measures to achieve good water status with the Environmental Costing Model. S. Broekx*,a, E. Meynaerts a, P. Vercaemst a, D. Beheydt a a Flemish Institute for Technological Research (VITO), Boeretang 200, B-2400 Mol, Belgium *corresponding author: Tel: Fax: steven.broekx@vito.be Abstract The European Water Framework Directive (WFD) sets ambitious objectives to ensure that all European waters meet good status by 205. Member States have to publish river basin management plans by 2009, including programmes of cost-effective abatement measures. In Flanders, the Environmental Costing Model (Milieukostenmodel or MKM in Dutch) is developed to select least-cost combinations of emission abatement measures to obtain a given surface water quality target. Emission sources incorporated are industry, households and agriculture. By means of mixed integer programming the MKM identifies the least-cost combination of abatement measures to satisfy multi-lutant reduction targets. Reduction targets are determined by comparing VLAREM II standards with actual measurements in the reference period and this at the end of each sub basin. First results for the basin of the Nete show that approximately 60 million per year or 00 per inhabitant per year is needed to achieve the VLAREM II-standards at each sub basin and this for the lutants COD, N and P. The majority of these annual costs result from the connection of households to collective waste water treatment plants and making nitrate targets for agriculture more stringent. However, due to the large amount of non-allocated emissions (e.g. +/- 50% for COD) it is difficult to predict whether surface water quality targets are reached with these measures. Extension of the model with the remaining basins in Flanders and improvement of the methodology by linking the MKM with a water quality model is foreseen in 2007 and Keywords Cost effectiveness analysis, water quality, water framework directive

2 Introduction The European Water Framework Directive (WFD) sets ambitious objectives to ensure that all European waters meet good status by 205. Member States are required to publish river basin management plans by 2009 which outline a programme of measures to achieve a good water status by 205. Withing the WFD economic approaches and tools are seen as important aids to set up management plans. One of these tools is the cost-effectiveness analysis. Article annex III of the WFD states that judgement should be made about the most cost-effective combination of measures in respect of water uses to be included in the programme of measures. Several initiatives have been or are currently being undertaken by the Member States to determine this cost-effective combination of measures. Interwies et al., 2004 for instance developed a multi-step evaluation process for Germany. This evaluation takes into consideration the ecological effectiveness, the probability of reaching the WFD-objectives until 205, the time frame necessary for their implementation and a prioritization with respect to the direct and indirect costs involved. In the Netherlands, the KRW Verkenner was developed to allow decision makers to assess the impact of measures on water quality. (WL Delft, 2006) In Flanders, the Environmental Costing Model (ECM) provides useful economic insights to assist icy-makers in designing a programme of measures to achieve this good water state at least costs. The BAT-Centre of the Flemish Institute for Technological Research (VITO) started in June 200, under the authority of the Flemish Environmental Administration with the development of the Environmental Costing Model. The Flemish Government aspires with this project the development of a tool to (i) determine the costs of environmental icy and (ii) contribute to more efficient environmental icy by indicating how environmental targets can be realised in a cost effective way. Initially, the model was developed for the most important industrial air lution sources (incl. electricity production) of SO 2, NO x and VOCs in Flanders. (Eyckmans et al., 2005; Lodewijks et al., 2007) By analogy with the test case for atmospheric lution, the BAT-Centre started with a test case for lution of surface water which focuses on the lutants P, N, COD, the basin of the Nete and multiple emission sources (industry, households, agriculture).

3 2 Methodology The MKM determines by means of mixed integer programming (in GAMS) the least-cost combination of abatement measures to satisfy multi-lutant reduction targets. The objective function and main constraints are described below: Minimize objective function : TC = TC meas () constrained by: ( TARGET ) LOAD wbody, LOADREFwbody, wbody, with : meas imeas (2) TC = TC + TC (3) TC imeas = (4) c s i impl IC + OC source, source, source imeas = = = n,, source imeas ( + r) r( + r) n TC = c n s r( + r) = = implc IC + OC n source, ( + r) source (5) s source= imeas= s source= imeas= i i impl impl source, source, implc < implc = = 0 (6) c = imeas= i impl (7), source = EAT LOAD,, source Esource, ( EFF ) ( EFF, ) impl, source = (8) c i s [ EAT wbody ],, imeas, source T,,, = (9) wbody = imeas= source=

4 EATREF,, source = Esource, (0) LOADREF wbody ( EFF ) ( EFF, ) implref, source c i s [ EATREF wbody ],, imeas, source T,,, = () = imeas= source= impl, implc = 0 or (2) sets and variables: : collective measure id E: source emission before individual and collective treatment [kg yr - ] EAT: source emissions after individual and collective treatment [kg yr - ] EATREF: source emissions after individual and collective treatment in the reference year [kg yr - ] EFF: efficiency of individual or collective measure IC : investment cost [ yr - ] imeas : individual measure id impl: implementation of individual and collective measures for a specific source (0 or ) implc: implementation of a collective measure (0 or ) implref: implementation of individual and collective measures in the reference situation (0 or ) LOAD: amount of lutants reaching the end of each water body [kg yr - ] LOADREF: amount of lutants reaching the end of each water body in the reference year [kg yr - ] n: lifespan of measure [yr] OC : operational cost [ yr - ] : lutant (COD, TN, TP) r: discount rate source: source id T: transport coefficient or relative amount of the emission after treatment reaching a specific water body TARGET: reduction target for each water body TC : total cost [ yr - ] wbody: water body id The objective of the algorithm as stated in equation is evidently to minimize total costs. The surface water quality targets that have to reached are expressed in annual emission loads (equation 2). Based on a comparison between the observed concentrations and the concentration targets (e.g. VLAREM II-standards), it is derived to which extent the annual loads in the reference year have to be reduced to apply to these targets. Targets are set for each lutant at the end of each VHA-zone. The division into these subbasins is determined in the Flemish Hydrographical Atlas (VHA) and is based on merging points and river characteristics.

5 A distinction is made in equation 3 between individual measures and collective measures. Individual measures can only be applied by one individual source whereas collective measures can be applied by multiple sources at once. This distinction was required as both households and industrial plants can be treated by the same collective treatment plants. Changes made on a collective level have an impact both on households and industrial plants. Costs made to improve collected measures cannot be attributed to individual sources. Therefore it was required to make distinction between costs attributable to a single source (equation 4) and collective costs which have an impact on multiple sources (equation 5). Costs included in equation 4 are for instance individual wastewater treatment for households or industry, but also sewage to connect individual sources to a collective treatment station as these costs can be attributed to individual sources. Costs included in equation 5 mainly consist of construction costs for building new collective treatment plants or improving existing plants. Total costs consider both the discounted investment costs and annual operational costs. For each individual source a single combination of one individual measure and one collective measure is chosen (decision variable impl source, ). If for a single source a collective measure is implemented (for instance the construction of a collective treatment plant), it has to be made sure that all costs for implementing this measure are included on the one hand, but also on the other hand that these costs cannot be accounted for twice (i.e. for other individual sources). Consequently, an additional decision variable implc is required, which is related to impl source, as stated in equation 6. Equation 7 is used to make sure that for each source exactly one combination of individual and collective measures is implemented. Consequently, some artificial measures with zero cost and zero efficiency have to represent discharges in the reference year. If these measures are chosen to be implemented for a specific source in 205, no supplementary measures are selected. This also implies that combinations of individual or collective measures (as for instance a secondary and tertiary treatment for industry) have to be included as a separate measures, as only one measure can be chosen. As these decision variables implc are set as binary variables, it is assumed that measures are used fully whenever they are implemented. This also corresponds with current practice. Most of the measures will be used non stop. Making a distinction between fixed and variable costs, necessitating both a binary and continuous decision variable, and allowing measures to be implemented only partly, was not assessed to be very realistic. Equation 8 is used to estimate the remaining loads after treating emissions, using a combination of an individual and collective measure. Equation 9 indicates the amount of emissions reaching a certain water body by using transport coefficients. These coefficients describe the fraction of lutants emitted by a certain source that reaches the specific location where reduction targets are set. This methodology has been applied for instance in Van der Veeren et al., Transport coefficients can vary between 0, when the discharge point is situated in another tributary or in a sub basin downstream, and, when the discharge point is situated on the target location. These coefficients can be estimated by simulating with a water quality model how concentrations are influenced when loads are reduced at a specific location. This methodology also allows to incorporate impacts in water bodies downstream by reducing emissions further upstream.

6 3 Case study: basin of the Nete 3. Study area The Nete is a tributary of the Scheldt river. The basin is part of the Flanders Region in Belgium. It has a surface of 673 km² and a total length of watercourses of 2224 km. Approximately inhabitants live in the basin and 4.2 companies are situated inside the region. This particular basin is chosen because in the past a lot of investments have been made to achieve water quality targets. Despite these investments quality targets are not reached and further analysis is needed. The MKM could give useful insights and help icy makers to achieve targets at minimal cost. The basin is subdivided into 2 VHA-zones and for each of these zones reduction targets are estimated for 3 lutants. Figure : Nete river basin

7 4.2 Reference scenario The figure below shows for each sector and lutant an overview of the relative share of emissions in the reference situation in the Nete basin. Households are clearly the main contributor to COD and Ptot. Agriculture is the main contributor to Ntot. For each lutant, the contribution of industry is rather limited. Figure : Relative share of emissions of different sectors in reference scenario % 90% 80% Relative share emissions 70% 60% 50% 40% 30% 20% 0% agriculture industry households 0% COD Ntot Ptot 4.2 Reduction targets Reduction targets used in this case study are calculated by comparing the observed 90-percentile concentrations with the 90 percentile quality targets. These targets are based on the VLAREM IIstandards 30 mg/l for COD, 6 mg/l for Ntot and mg/l for Ptot. Reduction targets are equal to the percentage by which the standards are exceeded. If for instance the observed 90-percentile concentration for COD is 60 mg/l, the emissions have to be reduced with at least 50% to apply to the standard of 30 mg/l. This is of course a strong simplification as a one to (?) one relationship between emissions and concentrations is assumed. Previous research on load balances in the General Water Quality Plan 2 for the river basin of the Nete (Paelinckx et al., 200) indicated a large amount of non-allocated emission loads. When comparing emission inventories for phosphorus with the observed emission loads only 42% can be allocated. Due to the absence of a diffuse lution inventory for all sources with the exception of agriculture, the unknown impact of erosion on water quality and measuring errors. Also, the contribution of the different sectors can vary strongly according the season. Agriculture is expected to have a high impact in the winter during heavy rainfall whereas households and industry are expected to have the highest impact in the summer. By means of the ECM alone, it is not possible to know in advance whether or not quality targets will be achieved. Ideally, the ECM is linked to a water quality model as discussed in Cools et al., 2007.

8 The figure below compares for the river basin of the Nete the observed 90-percentile concentrations with the quality targets. The main problems in the Nete basin are caused by COD. For 8 out of 2 VHA-zones, the target is exceeded. For 2 zones (VHA 500 and 540), no concentration measurements are available and as a consequence no reduction targets can be derived. For Ptot the target is exceeded in 7 VHA-zones and for Ntot the target is achieved in all zones. The maximum observed 90-percentile concentration for Ntot is,65 mg/l, namely in zone 54. Figure 2: Comparison 90-percentile concentrations COD, Ntot and Ptot with concentration targets for the VHA zones of the Nete basin in percentile concentration 99-0 (mg/l) COD meas COD target N meas N target P meas P norm VHA zone A comparison of the reduction targets with the maximum reduction potential of all measures considered in the case study shows that all targets are achievable except the target for COD in zone 5. To be able to perform a cost effectiveness analysis, this target is set to the maximum achievable reduction. 4.3 Cost effectiveness analysis The measures considered in the case study are: For households: - Individual treatment (4 possible development scenarios) - Connection to sewerage and collective treatment (households are grouped into clusters according to the existing icy-initiative zoneringsplannen, clusters range according to cost to connect from the cheapest dark green, to light green, to yellow and to red)

9 For industry: - Primary treatment - Secondary treatment - Tertiary treament - Additional treatment (microfiltration) - Closed enterprises (companies which closed down between 2000 and 2006 in the Nete basin) For agriculture: - Setting N-targets more strict (20 kg/ha and 70 kg/ha) - Bufferstrips of 5 and 0 metres - Green fertilizers - Individual treatment in greenhouses and dairy farms At a discount rate of 5% total annual costs are on average 59 million euro to achieve the reduction targets at the end of each VHA-zone. This amount consists of an investment of 442 million euro and an annual operational cost of 3 million euro. Households represent the major share in total annual costs and in reduction of emissions as they are the main contributors to COD emissions. As mentioned before, VLAREM II-standard are exceeded most frequently for this lutant in the Nete basin. Figure 3: Relative share sectors in reducing emissions for each lutant and annual costs according to the cost effectiveness analysis 00% 90% 80% 70% Relative share 60% 50% 40% 30% 20% 0% agriculture industry households 0% Red. COD Red. Ntot Red. Ptot Cost Besides estimating the total cost to reach all quality targets, it is important to determine which measures can be seen as cost effective. To this end, marginal abatement cost curves are set up for

10 all lutants. These curves are derived by including a tax in the objective function (equation ) on the remaining emissions. All measures with a marginal abatement cost lower than this tax will be implemented as it is more cost efficient to implement the measure than to pay the tax. Measures with a higher abatement cost will not be implemented as it is more cost efficient to pay the tax. By gradually increasing this tax in sequential optimization procedures and comparing implemented measures between different procedures abatement cost curves can be derived and a ranking of measures can be established. Figure 5 is the marginal abatement cost curve for reducing COD in the Nete basin. The red line indicates the reduction target at the end of the basin. The marginal cost to reach the target for COD is equal to 9,7 /kg, which means that all measures with a marginal cost lower than 9,7 /kg have to be implemented to achieve the target at the lowest cost. Measures with a lower marginal cost are for example primary and secondary industrial treatment, connection of households (dark green and light green clusters) to collective treatment and individual treatment for the remaining households. Figure 4: Marginal abatement cost curve for COD for the Nete basin with indication of the reduction target at the end of the basin (VHA-zone 56) Marginal abatement cost ( /kg) Marginal abatement cost Target end Nete basin , 0,2 0,3 0,4 0,5 0,6 0,7 Reduction emissions reference year (%) 4 Discussion The Water Framework Directive stipulates that the river basin management plans have to include programmes of cost-effective abatement measures if good water status is not achieved. Consequently, European member states urgently need methodologies and models to perform a cost-effectiveness analysis focusing on surface water quality. Though some work has been done on this subject in the past, De Nocker et al., 2007 indicates however that currently most of the European Member States are still at an early stage in assessing costs and performing a costeffectiveness analysis. In Flanders the environmental costing model is used to assess the cost-effectiveness of abatement measures. Based on the experience in the field of air lution, a case study was developed for

11 surface water lution in the Nete basin. Due to the large differences between air lution and water lution a new methodology had to be developed. An important difference was the methodology to apply individual reduction targets for each subbasin, whereas for air lution only a nationwide ceiling applies. This required the use of transport coefficients to take into account the fact that there is a correlation between water quality in different sub basins, as measures in one sub basin could affect water quality in another. Another consequence of this division into sub basins is the large amount of individual sources which have to be assessed separately. As targets are set at the level of sub basins, sources cannot be aggregated on a scale higher than sub basins. For the case study more than.000 individual sources were included in the model. For the extension to the complete Flemish region the amount of individual sources is expected to be around Some large methodological challenges remain such as linking the ECM with a water quality model to translate emission reduction to water quality. Without this linkage, it is impossible to predict whether or not good water status will be reached with the measures selected by the environmental costing model. This linkage might also influence the cost-efficiency of measures as it enables to incorporate seasonal variations in water quality. A first test described in Cools et al., 2007 suggests that reducing diffuse emissions from agriculture will have a lower impact on concentrations than reducing point emissions from households and industry. This might be explained by the fact that the highest concentrations are observed during dry periods and diffuse emissions mostly take place during rainy periods. However, one could also argue that water quality models are usually calibrated on short term data ( to 3 years) and it is expected that agricultural measures will only have a full impact after at least 5 years as it takes time to reduce nutrient concentrations in groundwater. Another challenge is to establish a ranking of abatement measures based on cost-effectiveness. Although abatement cost curves are specifically set up to establish this kind of ranking, cost curves are very lutant and location specific. Consequently, the ranking of measures alters significantly amongst regions and lutants. How icy makers will translate these differences in ranking to nationwide management plans, is still not clear.

12 References Broekx S., Meynaerts E., Vercaemst P., Ochelen S., Beckers A., Towards a good surface water state in the Flemish Region of Belgium with the Environmental Costing Model, River Basin Management Brouwer R., van den Bergh J., ICES-KIS voorstel: De directe en indirecte waarde van multifunctioneel watergebruik op stroomgebiedniveau Cools, J., Vlieghe, P., Huygens, M., Broekx, S., Bauwens, W., Modelling cost and impact of emission reduction measures. Proceeding of the SWAT 2007 conference, Delft, Netherlands. De Nocker, L., Broekx, S., Liekens, I., Görlach, B., Jantzen, J., Campling, P., Costs and Benefits associated with the implementation of the Water Framework Directive, with a special focus on agriculture, report for DG Environment, 57 pp. Eyckmans J., Meynaerts E., Ochelen S., The Environmental Costing Model: a tool for more efficient environmental icymaking in Flanders, The ICFAI Journal of Environmental Economics, Volume III, nummer 2, mei Lodewijks, P., Meynaerts, E., The Environmental Costing Model: a tool to advise icy makers in Flanders on issues of cost efficiency. Proceeding of the 6 th International Conference on Urban Air Quality, Cyprus, March 2007 Meynaerts E., Ochelen S., Beckers A., Van Biervliet K., Het Vlaams Milieukostenmodel Water: de ontwikkeling van een instrument voor een efficiënter beleid inzake kwaliteit van het oppervlaktewater, Tijdschrift Water, maart/april Meynaerts E., Broekx S., De Nocker L., Finding the appropriate scale and dealing with uncertainty issues in the Environmental Costing Model for the Flemish Region of Belgium, Proceedings International Workshop River Basin Management Interwies E., Borchardt D., Kraemer A., Kranz N., Görlach B., Richter S., Willecke J., Dworak T., Basic principles for selecting the most cost-effective combinations of measures for inclusion in the programme of measures as described in Article of the Water Framework Directive Paelinckx I., Wuyts S., Vannevel R., 200. General Water Quality Plan 2 Nete Van Biervliet K., Gielen B., De Sutter R., Le Roy D., Van Tomme I., Leroy I., Van den Broeke G., Pilootstudie Toepassing Milieukostenmodel op het thema oppervlaktewater voor een prioritair bekken: inventarisatie en relaties, i.o.v. AMINAL

13 Van der Veeren R.J.H.M., Economic analyses of nutrient abatement icies in the Rhine basin. Waterbouwkundig Laboratorium Delft, Handleiding KRW Verkenner,