Multi-objective land use optimization for urban areas

Abstract

This research has been carried out with an intension to present an optimization approach for the urban land use allocation problem, encouraging efficient utilization of urban space. In Dhaka, acute scarcity of buildable lands and excessive high population density has resulted in unrestrained accumulation of different land uses in a precinct both horizontally and vertically. In the absence of proper technological advancement, the planning authority finds it very complicated to ensure optimized arrangement of land particularly in the vertical dimension, which engender incompatible land uses. Moreover, the involvement of multiple stakeholders in the land use allocation procedure makes it intricate to reach a decision satisfying their diverging interests. In this background this research aims to provide a decision support system for the land use planners, facilitating generation of Pareto optimum land use allocations considering two objectives - maximizing compatibility among adjacent land uses of a study area without compromising the area’s total land price and to maximize the price of each plot. Considering the mixed use characteristics of Dhanmondi Residential Area, it has been selected as the feasible study area to apply the optimization problem practically. To accomplish the study a massive portion of required data has been collected from primary source. The floor-wise use of every building in the study area, plot size, ground coverage, building storey, adjacency to lake and green space, percentage share of space use in a block, number of school, health facilities per block all these data have been collected through field survey. While data of residential, commercial and office space prices have been collected through questionnaire survey, using stratified sampling method. Apart from these, data regarding distance to important activity centre, adjacency of the plot to major road and data on compatibility indices among different space uses for the study area have been obtained from secondary sources. In order to avoid complicacy of calculation, the land price equation has been developed for only residential, commercial and office space use of the study area, using well accepted hedonic regression method. The hedonic model explains 79% of the natural logarithms of the unit land price of residential, commercial and office plots. Considering the non-linear characteristics of the objective function later a multi-objective evolutionary algorithm approach called Non-Dominated Sorting Genetic Algorithm-II (NSGA-II) has been applied to obtain Pareto optimal solutions of land use allocation subject to different set of constraints. Different model runs of NSGA-II have been explored with different conditions to observe the variation in generated outputs. From those, a model run with 30% of changes in certain space use produces 24 Pareto optimal solutions of land use allocation from a total of 100 solutions. All these solutions in Pareto front allow tradeoff between maximizing compatibility and land price from one solution to other. Of these 24 optimum solutions, nine solutions have been obtained where both compatibility and land price have been maximized simultaneously from the existing condition by incorporating up to 13% of changes in certain space use out of 30% of allowable changes. However maximum increase in compatibility with improved land price was possible through a reduction in commercial space use and addition in office space use. On the other hand the study also reveals that, imposing maximum allowed percentage of changes in certain space use increases the land price of different plots but at the same time reduces the compatibility of the study area. The more the model tries to maximize the land price of each plot in the study area, there is a strong inclination of the plots with single use to be transformed into mixed use plots comprising various share of residential, commercial and office uses. To conclude it is expected that, the generated models of this research will aid the policy makers of development authorities by providing alternative land use plan and thereby predict the consequence of any plan before practical application.