Hierarchical Cluster Analysis and Node Centrality Scores for Industrial Symbiosis Network models of Point Lisas Industrial Estate

This dataset contains processed experimental simulation data supporting the results of the unfunded study titled “Filtering ‘3-2’ Industrial Symbiosis Networks at a Carbon-Intensive Cluster in a Small Island Developing State to Reuse CO2 and Water”. The data was processed using the algorithms and software described in the Methods section of that study (https://doi.org/10.1016/j.cherd.2024.10.023).

To investigate the industrial symbiosis (IS) for combined carbon dioxide utilization and water reuse on the Point Lisas Industrial Estate (PLIE), enterprise input-output MILP models of a representative IS network were developed. Discarded water streams with multiple contamination levels and high-purity process CO2 from ammonia processes were selected as materials to be reused in: existing petrochemical plants, a mineral carbonate factory and a propylene carbonate plant. The relative locations of these plants is reported.

Firstly, exploratory hierarchical cluster analysis (HCA) was performed to discover summary knowledge of the chosen cluster model (e.g. the likely number of subclusters) with the Python library, SciPy using agglomerative clustering. The cophenetic correlation coefficient and dendrograms from the HCA with different linkage functions are reported.

Then economic and environmental objectives were set for each material. Combining economic objectives left a tri-objective problem, which was resolved with ε-constraint optimization and multi-criteria decision-making methods, for two different scenarios - with and without the mineral carbonate and propylene carbonate plants. Graph and network algorithm functions in MATLAB® (ver. R2021a) were used to create network graphs from weighted and directed adjacency matrices, which contained the magnitude and direction of inter-plant flows, and then calculate network metrics that characterize structural attributes, in particular five centrality indices: degree, closeness, betweenness, 'hubs' and 'authorities' were calculated using MATLAB’s centrality function. Along with network connectance, the 'in' and 'out' degree, 'in' and 'out' closeness, betweenness, Kleinberg's 'hub' and 'authority' centrality scores are reported for the individual and combined networks for both scenarios.