<p>The aim of this thesis is to discuss and analyse the influenceof data uncertainties with regard to the reliability of materialflow analysis (MFA) studies. MFA, as a part of environmentalsystems analysis, is a method belonging to the research field ofindustrial ecology and more specifically industrial metabolism.As such, the method strives at giving a holistic view of thecomplex world we live in, in order to reduce negativeenvironmental impact. Among other things, MFA studies have beenproposed to be useful for priority setting and following up inmunicipalities.</p><p>Serving as a starting point is a local case study of flows ofnitrogen in a Swedish municipality, Västerås. The casestudy has been performed using the ComBoxmodel. The years studiedare 1995 and 1998. The main sectors in society emitting nitrogento water were identified as the agricultural and householdsectors. The dominating sectors emitting nitrogen to air wereidentified as the agricultural, transport and infrastructuresectors.</p><p>As a basis for discussing data uncertainties qualitatively andquantitatively a literature survey was performed. 50 articles andbooks were identified as in some way or another dealing with datauncertainties in MFA. The literature survey showed that theuncertainties for results from a MFA study might vary between±30 % and a factor 10 depending on what kind of parameter isinvestigated. Only one method was found that dealt with datauncertainties in MFA in a complete way; a model developed byHedbrant and Sörme (HS model).</p><p>When applying the HS model to the case study of nitrogen flowsin Västerås, it was found that when uncertaintyintervals were calculated the possible conclusions changed. Ofthe two pair of flows compared in relation to priority setting,none of the earlier conclusions remained. Of the three flowsanalysed in relation to following up, only the flow from onepoint source supported the same conclusion when uncertainty wasconsidered.</p><p>In all, it is concluded that data uncertainties in MFAanalysis are an important aspect and that further research isneeded in order to improve input data quality estimations andframeworks for determining, calculating and presenting data, datauncertainties and results from MFA studies. However, theunderlying reality remains, e.g. that management of materialflows are important for understanding and reducing the negativeenvironmental impact. Thus, MFA is one useful tool in thiswork.</p><p><b>Keywords:</b>data uncertainties, sensitivity analysis,Material flow analysis, MFA, method to determine datauncertainties, case study, ComBox model, nitrogenflows.</p>

The aim of this thesis is to discuss and analyse the influenceof data uncertainties with regard to the reliability of materialflow analysis (MFA) studies. MFA, as a part of environmentalsystems analysis, is a method belonging to the research field ofindustrial ecology and more specifically industrial metabolism.As such, the method strives at giving a holistic view of thecomplex world we live in, in order to reduce negativeenvironmental impact. Among other things, MFA studies have beenproposed to be useful for priority setting and following up inmunicipalities. Serving as a starting point is a local case study of flows ofnitrogen in a Swedish municipality, Västerås. The casestudy has been performed using the ComBoxmodel. The years studiedare 1995 and 1998. The main sectors in society emitting nitrogento water were identified as the agricultural and householdsectors. The dominating sectors emitting nitrogen to air wereidentified as the agricultural, transport and infrastructuresectors. As a basis for discussing data uncertainties qualitatively andquantitatively a literature survey was performed. 50 articles andbooks were identified as in some way or another dealing with datauncertainties in MFA. The literature survey showed that theuncertainties for results from a MFA study might vary between±30 % and a factor 10 depending on what kind of parameter isinvestigated. Only one method was found that dealt with datauncertainties in MFA in a complete way; a model developed byHedbrant and Sörme (HS model). When applying the HS model to the case study of nitrogen flowsin Västerås, it was found that when uncertaintyintervals were calculated the possible conclusions changed. Ofthe two pair of flows compared in relation to priority setting,none of the earlier conclusions remained. Of the three flowsanalysed in relation to following up, only the flow from onepoint source supported the same conclusion when uncertainty wasconsidered. In all, it is concluded that data uncertainties in MFAanalysis are an important aspect and that further research isneeded in order to improve input data quality estimations andframeworks for determining, calculating and presenting data, datauncertainties and results from MFA studies. However, theunderlying reality remains, e.g. that management of materialflows are important for understanding and reducing the negativeenvironmental impact. Thus, MFA is one useful tool in thiswork. <b>Keywords:</b>data uncertainties, sensitivity analysis,Material flow analysis, MFA, method to determine datauncertainties, case study, ComBox model, nitrogenflows. / NR 20140805

Cette thèse porte sur le calcul d'une sur-approximation conservative pour les solutions d'équations différentielles ordinaires en présence d'incertitudes et sur son application à l'estimation et l'analyse de systèmes dynamiques à temps continu. L'avantage principal des méthodes et des algorithmes de calculs présentés dans cette thèse est qu'ils apportent une preuve numérique de résultats. Cette thèse est organisée en deux parties. La première partie est consacrée aux outils mathématiques et aux méthodes d'intégration numérique garantie des équations diff érentielles incertaines. Ces méthodes permettent de caractériser de manière garantie l'ensemble des trajectoires d'état engendrées par un système dynamique incertain dont les incertitudes sont naturellement représentées par des intervalles bornés. Dans cette optique, nous avons développé une méthode d'intégration hybride qui donne de meilleurs résultats que les méthodes d'intégration basées sur les modèles de Taylor intervalles. La seconde partie aborde les problèmes de l'identification et de l'observation dans un contexte à erreurs bornées ainsi que le problème d'atteignabilité continue pour la véri cation de propriétés des systèmes dynamiques hybrides. / This thesis addresses the computation of conservative over-approximation of the solutions of uncertain ordinary di erential equations and its application to the estimation and the analysis of uncertain continuous-time dynamical systems. The main feature of the methods and algorithms presented in this thesis is the fact that they are numerically veri ed and hence can be used to obtain numerical proof of properties. This thesis is organized in two parts. The first part is devoted to the mathematical tools and the guaranteed numerical integration methods for uncertain ordinary di erential equations. These methods make it possible to characterize in a guaranteed way all the state trajectories generated by an uncertain dynamical system whose uncertainties are in a natural way described by bounded boxes. Accordingly, we have developed a hybrid integration method which gives better results than the integration methods based on interval Taylor models. The second part is dedicated to the resolution of identi cation and observation issues in a bounded error context. It also deals with continuous reachability computation for the veri cation of the properties of hybrid dynamical systems.