Assessing the Robustness of Optimization-Based Decision Support Systems for Sustainable Urban Planning: The Bologna Green Cells Case Study

Optimization-based decision support systems can support sustainable urban planning by making spatial trade-offs explicit, but their recommendations may depend strongly on uncertain input data. This thesis studies that issue in the TALEA Bologna Green Cells case study, where optimization models select candidate locations for modular green interventions over a city grid. The work first reconstructs the baseline data pipeline, preprocessing stages, and implemented MiniZinc models, then evaluates robustness under controlled synthetic perturbation scenarios applied to environmental, demographic, land-use, geometric, and allocation-related inputs. Robustness is assessed by comparing each perturbed solution with the corresponding no-noise baseline. Spatial robustness is measured through Jaccard similarity and centroid-buffer IoU, while quantitative robustness is evaluated through output-ratio metrics for utility, yard allocation, and street allocation. A cross-model comparison further normalizes these metrics to summarize average stability, lower-tail fragility, and degradation trends across increasing perturbation levels. Within the tested perturbation scenarios, many outputs show good average stability, but the results also reveal scenario-specific fragility: some model-source-perturbation combinations produce substantial spatial shifts or unstable quantitative ratios. The analysis is scenario-based and reproducible, but it is not a full Monte Carlo uncertainty quantification. Overall, the thesis shows how robustness evaluation can support more transparent and uncertainty-aware use of optimization-based decision support systems in urban planning.