Assessing the effect of data granularity on the carbon footprint of an entity within a complex organization : A comparative analysis of data granularity across emission categories, effect on the identification of reduction levers, and a methodological framework transferable to other entities

Details

Supervisors

Agnan, Yannick ; Servais, Marc

Faculty

Faculté des bioingénieurs

Degree label

Master [120] : bioingénieur en sciences et technologies de l'environnement, à finalité spécialisée

Abstract

Organizational carbon accounting frameworks are well established conceptually, yet their practical implementation in decentralized institutions remains constrained by two unresolved challenges: the absence of empirical guidance on where data precision actually matters, and the lack of operational frameworks reproducible by non-expert contributors. This thesis addresses both gaps through a full carbon footprint assessment of the Faculty of Bioscience Engineering at UCLouvain (2024), spanning the three scopes of the GHG Protocol across seven emission categories. Emissions were computed using an activity-based approach with ADEME Base Empreinte® factors. Two objectives are pursued: (1) quantifying, per emission category, the sensitivity of the carbon footprint result to data granularity, and (2) translating these findings into a transferable operational framework for future autonomous entity-level assessments. Granularity sensitivity proves highly heterogeneous: for student mobility the faculty-specific estimate exceeds the scaled university-wide one by 71%, while for purchases the gap is ~2.5%. Where it matters, insufficient resolution renders the dominant reduction lever structurally invisible rather than merely imprecise; for categories that are either minor or institutionally constrained, the residual uncertainty is irreducible by granularity: structural, allocation-based, or sampling-bound, not a matter of finer data. Uniformly high data precision is therefore neither necessary nor efficient: effort concentrated on dominant, granularity-sensitive categories shifts both the total and the identity of the primary reduction lever. A per-category operational guide is proposed as a partial inventory management plan oriented toward preventive data structuring, transferable across UCLouvain entity types. A transfer test on a research institute confirms that the binding constraint on granularity improvement shifts from data quality to emission factor availability for research-intensive procurement profiles. These findings connect to a broader pattern: in complex organizations, the gap between measuring a footprint and acting on it is less a calculation problem than a data-resolution one. Entity-level granularity analysis is thus a prerequisite for moving from carbon reporting to evidence-based reduction.