Towards greener petrochemical production: Two-stage network data envelopment analysis in a fully fuzzy environment in the presence of undesirable outputs

• Published in: Expert systems with applications Vol. 164; p. 113903
• Main Authors: Mozaffari, Mohammad Reza, Mohammadi, Samira, Wanke, Peter F., Correa, Henrique L.
• Format: Journal Article
• Language: English
• Published: New York Elsevier Ltd 01.02.2021; Elsevier BV
• Subjects: Ammonia; Data analysis; Data envelopment analysis; Decision analysis; Decision making; Fuzzy logic; Fuzzy set; Fuzzy sets; Greenhouse gases; Linear programming; Pollutants; Two stage; Undesirable output; Linear programming; Undesirable output; Two stage; Fuzzy set; Data envelopment analysis
• ISSN: 0957-4174; 1873-6793
• DOI: 10.1016/j.eswa.2020.113903

•We evaluate two-stage production networks in a number of petrochemical companies.•We use a fuzzy linear programming model with the presence of undesirable outputs.•Our model also includes a two-stage network data envelopment analysis model.•We think this is the first study to develop and apply a model with the above features.•Our contribution to practice: better evaluation of efficiency in petrochemical plants. Traditional Data Envelopment Analysis (DEA) models usually consider decision-making units (DMU) as black boxes that consume a set of inputs to produce a set of outputs and ignore the units’ internal structure. In literature, two-stage DEA models have been introduced and developed to deal with the inner workings of such DMUs; however, models in the extant literature can only be used to assess technologies with precise inputs and outputs. In this study we present an innovative two-stage DEA model for DMU evaluation in the presence of imprecise data, which are modeled using triangular fuzzy values in this study. We also innovate by including the treatment of undesirable outputs. We propose some alternative two-stage DEA network models with undesirable outputs in a full-fuzzy mode that are either directional or radial and are embedded within a multi-objective linear planning structure based on fuzzy triangular numbers. We also propose a method to solve fuzzy programming models in the presence of undesirable pollutant outputs from production processes such as greenhouse gas emissions, harmful effluents, among others. Our proposed model is then applied to a real case of 15 ammonia-manufacturing units in the Iranian petrochemical sector. The promising results that we obtained in our research and their implications for theory and practice are then presented and discussed.