Uncertainty in LCA: An estimation of practitioner-related effects Flavio Scrucca a , Catia Baldassarri b , Giorgio Baldinelli b , Emanuele Bonamente b, c , Sara Rinaldi c , Antonella Rotili c , Marco Barbanera c, d, * a ENEA - Italian National Agency for New Technologies, Energy and Sustainable Economic Development, Department of Sustainability - Rome, Italy b University of Perugia, Department of Engineering - Perugia, Italy c CIRIAF- Biomass Research Centre - Perugia, Italy d University of Tuscia, Department of Economics, Engineering, Society and Business Organization - Viterbo, Italy article info Article history: Received 5 June 2019 Received in revised form 23 April 2020 Accepted 14 May 2020 Available online 20 May 2020 Handling editor: Yutao Wang Keywords: Uncertainty Reproducibility Repeatability Practitioner Carbon footprint Wine abstract The aim of the paper is to quantitatively analyse a main source of uncertainty in LCA practice, i.e. the one due to the LCA practitioner. The same life cycle inventory dataset was used by six practitioners to independently compute six environmental impact categories with a cradle to grave approach, consid- ering a red wine bottle produced by an Italian winery. To obtain the repeatability (r) and reproducibility (R) limits for each impact categories, LCA results were analyzed according to the ASTM E691-05 standard specifications. After a first stage of the study, in which relevant differences in the approach used and results were observed, all the practitioners considered the same system boundaries and processes, and, as a consequence, the results of all the impact categories became comparable. Nevertheless, the choice of different inventory datasets for describing the same process caused variations among the practitioners’ outcomes. This study highlighted how the uncertainties due to the practitioner choices may significantly affect LCA results, especially when lack of information affects the data collection. The practitioner-related uncertainty should be considered in the same way as other uncertainty sources, especially when the Life Cycle impacts of a product are compared to the ones published in other studies. © 2020 Elsevier Ltd. All rights reserved. 1. Introduction Life Cycle Assessment (LCA) is a recognized and widespread tool to evaluate the environmental impact of products, technologies and policies (Igos et al., 2018; Groen and Heijungs, 2017) and it can be considered as a specific method within the environmental impact assessment framework (Bjorklund, 2011). Furthermore, LCA is frequently used as a tool to support decision making processes but several types of uncertainty in all stages of an LCA can sometimes lead to widely varying results, misleading the conclusions in a scenario comparison (Cherubini et al., 2018). In the computation and communication of the result of an experimental activity (including modeling and simulations), in order to allow reliable comparisons with other analyses and discuss the consistency of different outcomes, it is utterly important to provide a properly- computed uncertainty. Uncertainty differs from variability e that is due to the natural heterogeneity of values e and it can be intended as the statistical “difference between a measured or calculated quantity and the true value of that quantity” (Finnveden et al., 2009). Moreover, the definition of uncertainty includes everything that is unknown, comprising both random and sys- tematic errors (during estimating, measuring or collecting data) and epistemic uncertainty (due to lack of scientific knowledge) (Rosenbaum et al., 2018). On the other hand, variability refers to inherent differences within a population due to intrinsic hetero- geneity of values and, unlike uncertainty, cannot be decreased but only better estimated with, for example, a better sampling (Pomponi et al., 2018). In the past decades, several attempts to include uncertainty and variability in LCA were carried out, demonstrating that the aware- ness of the importance and the impact of this topic in the LCA practitioners is increased (Igos et al., 2018). Huijbregts (1998) proposed a first general framework that distinguished different types of uncertainty and variability in LCA, defining, in particular, three types of uncertainty: parameter uncertainty, model uncer- tainty and uncertainty due to choices. In this regard, the choice of functional unit and system boundaries in the goal and scope defi- nition phase, the choice of the allocation procedures in the * Corresponding author. CIRIAF- Biomass Research Centre - Perugia, Italy. E-mail address: m.barbanera@unitus.it (M. Barbanera). Contents lists available at ScienceDirect Journal of Cleaner Production journal homepage: www.elsevier.com/locate/jclepro https://doi.org/10.1016/j.jclepro.2020.122304 0959-6526/© 2020 Elsevier Ltd. All rights reserved. Journal of Cleaner Production 268 (2020) 122304