Identifying Significant Differences between Supply-Cost Modelling Methodologies using GIS Applications for Bioenergy Production in Eastern Ontario
Southeastern Ontario has potential for the production of vast amounts of bioenergy. In order to assess this potential accurately, it is essential to determine accurate distances and costs for the delivery of biofuel residues to bio-refineries for the production of bioenergy. With this in mind, the purpose of this project is to implement GIS-based knowledge in order to compare raster and vector modelling of supply-cost curves for four different biofuel residues at three different cities. The residues include stover, softwood, hardwood and straw at the three cities of Bancroft, Smith's Falls and Cornwall. While raster data exhibits exceptional ability to cover a surface continuously, it must use the 'Euclidean Distance' tool to determine distances. The implementation of Euclidian distances deteriorates the accurate assessment of distances since these do not reflect real-world conditions. Since raster data does not take into consideration road networks, a tortuosity factor is applied to compensate for this limitation as done by previous research. Alternatively, network analysis is also implemented to determine more realistic distances based on a real-world road network. The distance intervals from the cities (bio-refineries) to their supply sources are the same under both models in order to generate a plausible comparison between the two. The distances start at 30 km increasing gradually until a maximum distance of 200 km from each city. Aggregated supply is calculated for each interval distance, giving a cumulative yield for each interval, which is later inputted into spreadsheet formulas to calculate the cost of delivery. After determining delivery cost and distances from supply sources to bio-refineries, supply-costs curves are generated, compared and analyzed to undertsand the differences of raster-based methods to vector-based methods. All cities and residues were analyzed resulting in the variation of supply-cost curves for vector modelling. This resulted in higher yields for vector based analysis compared to raster based analysis. As a result, this small sample of findings demonstrated that there is a significant difference between raster and vector modelling for estimating supply-cost curves.