Logistics for Energy Crops' Biomass. Deliverable 4.2. Feedstock supply scenarios
Résumé
This work aims at determining the geographical location, under different biomass consumption scenarios, of potential volumes for different feekstocks used in the two project case studies : miscanthus for the French case study (BP) and a mix of agricultural and forestry residues for the Spanish case study (MBP). These scenarios will be use to optimize the logistic chain and to assess the sustainability of the energy production process. This study has been conducted following two steps : first a spatially explicit assessment of the biomass quantities then the proposition of biomass demand scenarios. Different approaches were for the two case studies. For the French case study a spatially explicit location model was deleveloped. This model was developed at the local scale based on the detailed data provided by Bourgogne Pellets on their supply agricutlural fields'locations. This model takes into account one feedstock, miscanthus, which is consistent with the current situation of the BP case plant. After having described all the land parcels of th study area with potential driving factors of miscanthus location (slope, distance from the farm, size of the field, etc.) a BRT (Boosted Regression Tree) data mining method was applied to relate the location of the miscanthus fields to the selected driving factors at the scale of the supply area. This method yielded a spatially explicit model that provided miscanthus location probability for each land parcel of the study area. The model was then applied to a wider area to assess a potential widening of the BP plant supply area. Miscanthus location probabilities were then obtained for each land parcel of four French districts (departements). We have chosen to propose three miscanthus location scenarios additionally to the baseline scenario : a slight increase in miscanthus volume,doubling the current volume and maximum capacity of the BP plant. Miscanthus yields for all land parcels of the study area were derived form the CERES-EGC model estimations and missing values were filled with the mean estimate for the study region. Regarding the Spanish case study, the Miajadas plant currently uses more than 20 diffrent feedstock species. Most of these species are not the pirmary focus of the LogitEC project or the demonstrations of WP5. The species studied in the LogitEC project for the Spanish case are sorghum, poplar SRC and triticale. While triticale is currently part of feedstock mix of the Miajadas plant and cropped in the Extramadura region, sorghum and poplar SRC are still experimental crops for this case study. To add realism to the modelling, analysis and assessment in WP4, we decided to go beyond the three focused species and include agricultural residues and wood as these are major feedstocks for the plant. We then first assess the location of several feedstocks currently used in the Miajadas plant. Based on the relative proportion of the different feedstocks used in the plant and the availability of census data, we selected 9 species (7 annual species and 2 woody species - triticale, sorghum and poplar were not taken into account) for the baseline description. We then estimated an average yield for each selected species based on the available census data. The first output is a potential feedstock quantity available for each municipality of the Extramadura region for 9 feedstocks. Then, the location of triticale, sorghum and poplar SRC was assessed for prospective purposes. Three demand based scenarios were developed. We assumed that those three feedstocks met the entire biomass demand of the plant. The difference between the three scenarios lies in the relative proportions of those feedstocks. As we previously based the potential feedstock quantities on average yields for the Extramadura region, for the scenarios we preferred to propose a range of yields for each feddstock to explore a variety of situations. We developed the baseline maps and prospective scenarios at the regional level (Extramadura). The modelling of spatial feedstock potential for the French case could be based on detailed mapped information about miscanthus fields (location and yields). Such information is not available for the Spanish case and potential feedstock location modelling was based on census data as it proved the best source of information available. Moreover, for the three species : triticale, sorghum and poplar, even census data was mission or inadequate for modelling purposes. Spatial location modelling and yield estimations for these three species had then to be based on heterogeneous information partly issued from census data at the municipality, province and regional level, Miajadas plant practices and expert knowledge.