Optimal Placement of Solar Photovoltaic Farms Using Analytical Hierarchical Process and Geographic Information System in Mekkah, Saudi Arabia

摘要

This paper presents a combined approach of Multi Criteria Analysis (MCA) and Geographic Information system for the optimal placement of solar photovoltaic large farms in Makkah region in western of Saudi Arabia. This region is characterized by its heavily electric power consumption that may reach up to 16GW per year. To perform the multi criteria analysis, a set of geographic criteria was used including solar radiation, topography, land use, accessibility and proximity to electric transmission lines. The Analytic Hierarchy Process (AHP) method was adopted to determine the corresponding weight of each criterion. Suitable lands are then identified through a linear overlay of spatially distributed weighted criteria using a GIS solution. To implement the Analytic Hierarchy Process, an arrangement tree was constructed. The first level corresponds to the MCA objective to be satisfied. The second level of that tree corresponds to the criteria mentioned above. The third level contains the sub-criteria of each criterion of the first level (e.g. Local roads, Primary roads and Express roads are sub-criteria of the main criterion Road). Then, pairwise comparison matrices were constructed to evaluate the relevance of each criterion or sub criterion. These comparison matrices were compiled from questionnaires filled by experts. The calculated relative weights have shown that the solar radiation criteria acquire the most important weight of 0.42, followed by the slope criteria with a weight equal to 0.35. The obtained satisfactory Consistency Ratio of the AHP was 0.08. In order to carry out the second phase of this study, the geographic distribution of each spatial criterion was represented by raster data. Then, it was reclassified into a new grid that reflects the relevance of each cell location regarding its suitability to Solar PV installation. For example, cells that are near roads will take more important score than those which are distant. In addition, a Slope raster (genera