La gestion des risques est un enjeu majeur, mais difficile pour les projets de construction. La difficulté à gérer les risques dans les projets de construction vient de leur complexité. Ils sont composés de nombreuses entités (activités, acteurs, contrats, ressources, etc.) dont le comportement collectif influencent les comportements individuels. Afin de mieux appréhender et comprendre la complexité du système dans son ensemble, il est nécessaire de capitaliser et structurer la connaissance dans le but de proposer un modèle capable de décrire et simuler le comportement du système étudié. Cependant, la formalisation de tels modèles se confronte à de nombreuses difficultés : présence de facteurs humains, raretés de modèles, connaissances souvent expertes et qualitatives difficiles à formaliser, méconnaissance des mécanismes régissant certains processus, données parcellaires, hétérogènes et souvent imparfaites, échelles multiples, etc. L’objectif est de proposer des approches conceptuelles permettant d’assembler des morceaux de connaissances hétérogènes multi-sources et multi-échelles dans le but de proposer un modèle capable de réduire les incertitudes liées au fonctionnement, au devenir, à la conception et au pilotage des projets de construction.Différentes approches et outils ont été proposés pour modéliser et simuler les projets de construction : structure de répartition des risques, réseaux bayésiens, théorie des réseaux, simulation de Monte Carlo, réseau analytique, etc. Ces outils et méthodes sont utilisés pour simuler le comportement de systèmes, mais inadéquats pour représenter des systèmes complexes dynamiques à grandes échelles. Ils sont pour la plupart parcellaires et ne présentent pas ou peu de généricités. Dans ce contexte, les modèles relationnels probabilistes (MRPs) fourniront un formalisme mathématique pratique permettant de représenter et de simuler des systèmes dynamiques complexes entachés d’incertitudes. Les MRPs étendent le formalisme des réseaux bayésiens en ajoutant la notion de paradigme objet où l'incertitude attachée au système est alors prise en compte en quantifiant la dépendance probabiliste entre les propriétés des objets.Pour ce faire, une ontologie du domaine a été développée pour (a) fournir un vocabulaire commun capable de représenter les connaissances sur les projets de construction, (b) identifier les interconnections entre les différentes entités techniques, humaines, économiques à différents niveaux de description. Guidé par cette ontologie unMRP a été élaboré et utilisé pour simuler le comportement des projets de construction tout en prenant en compte les incertitudes. On montrera comment il peut être utilisé pour prédire la réponse incertaine du système ainsi que pour étudier comment la réponse globale du système est sensible aux valeurs ou hypothèses locales. Enfin, le MRP sera utilisé pour deux études de cas (la construction de routes et de ponts à Hue-Vietnam et d’un bâtiment en France). Les résultats montrent que le formalisme des MRPs permet (1) d’instancier tout type de projets de construction, (2) de prendre en compte l'incertitude, (3) de simuler et prédire le comportement du système et (4) d’extraire de la connaissance à partir d’informations partielles. / The difficulty to manage risks in construction projects comes from their complexity. They are composed of many entities (activities, actors, contracts, resources, etc.) among which interactions exist at many levels and influence the system response. In turn, this response can influence the behaviour of some entities. In order to capture the complexity of the system, it is necessary to structure, model and share cross-disciplinary and interdisciplinary knowledge flows in a common and unifying framework. Because of this high complexity, the system response may appear as unpredictable. Uncertainties at all scales are source of risk for the construction project itself. Tackling this complexity could improve our grasp of the whole system, in order to provide more robust and efficient decision alternatives in risk management. It is then essential to propose conceptual approaches able to represent the behaviour and the interactions of system entities over the time.Different approaches and tools have been proposed to model and simulate risk of construction project as Risk Breakdown Structure, Bayesian networks, Network Theory, Monte Carlo Simulation, Analytical Network Process, etc. These tools and methods can be used to simulate the behaviour of the system, but they are inadequate for representing large and complex dynamical system because they are based on case-dependant model (i.e. a specific model has to be built for each studied construction project), the fragmented representation of knowledge, the lack of common vocabulary, the lack of generic character. Hence, an ontology paradigm is developed in order (a) to provide a common vocabulary able to represent the knowledge about construction projects and its risks, (b) to shape the structure (interrelations) between those identified database and (c) to represent construction project integrating as well technical, human, sustainability dimensions at different detailed levels of uncertainty.In this context, by coupling the advantages of ontology and Bayesian network, the framework of probabilistic relational model (PRM) will provide a practical mathematical formalism allowing to represent and simulate complex stochastic dynamical systems. PRMs extend the formalism of Bayesian networks by adding the notion of object paradigm where uncertainty attached to the system is then taken into account by quantifying probabilistic dependence between the properties of objects and other properties of related objects. To the best of our knowledge, this thesis report will be the first application in which PRM have been proposed to model and simulate construction project while accounting uncertainties.Therefore PRM is used to simulate the propagation of uncertainties existing in this complexdynamic and multi-scale system, which lead to construction project risk. A prototypal software framework has been developed to check the consistency and the viability of the concept. It will be shown how it can be used in order to predict the uncertain response of the system as well as to study how the overall response of the system is sensitive to local values or assumptions. Lastly, PRM will be applied for two case-studies (a road and bridge construction in Hue-Vietnam and another building project in France). Results show that the formalism of PRMs allows to (1) implement any kind of construction project, (2) to take uncertainty into account, (3) to simulate and predict the behaviour of system and (4) to derive information from partial knowledge.

Probabilistic methods have been widely used in structural engineering to model uncertainties in loads and structural properties. The subjects of structural reliability analysis, random vibrations, and structural system identification have been extensively developed and provide the basic framework for developing rational design and maintenance procedures for engineering structures. One of the crucial requirements for successful application of probabilistic methods in these contexts is that one must have access to adequate amount of empirical data to form acceptable probabilistic models for the uncertain variables. When this requirement is not met, it becomes necessary to explore alternative methods for uncertainty modeling. Such efforts have indeed been made in structural engineering, albeit to a much lesser extent as compared to efforts expended in developing probabilistic methods. The alternative frameworks for uncertainty modeling include methods based on the use of interval analysis, convex function representations, theory of fuzzy variables, polymorphic models for uncertainties, and hybrid models which combine two or more of alternative modeling frameworks within the context of a given problem. The work reported in this thesis lies in the broad area of research of modeling uncertainties using non-probabilistic and combined non-probabilistic and probabilistic methods. The thesis document is organized into 5 chapters and 6 annexures. A brief overview of alternative frameworks for uncertainty modeling and their mathematical basis are provided in chapter 1. This includes discussion on modeling of uncertainties using intervals and issues related to uncertainty propagation using interval algebra; details of convex function models and relevance of optimization tools in characterizing uncertainty propagation; discussion on fuzzy variables and their relation to intervals and convex functions; and, issues arising out of treating uncertainties using combined probabilistic and non-probabilistic methods. The notion of aleatoric and epistemic uncertainties is also introduced and a brief mention of polymorphic models for uncertainty, which aim to accommodate alternative forms of uncertainty within a single mathematical model, is made. A review of literature pertaining to applications of non-probabilistic and combined probabilistic and non-probabilistic methods for uncertainty modeling in structural engineering applications is presented in chapter 2. The topics covered include: (a) solutions of simultaneous algebraic equations, eigenvalue problems, ordinary differential equations, and the extension of finite element models to include non-probabilistic uncertainties, (b) issues related to methods for arriving at uncertainty models based on empirical data, and (c) applications to problems of structural safety and structural optimization. The review identifies scope for further research into the following aspects: (a) development of methods for arriving at optimal convex function models for uncertain variables based on limited data and embedding the models thus developed into problems of structural safety assessment, and (b) treatment of inverse problems arising in structural safety based design and optimization which takes into account possible use of combined probabilistic and non-probabilistic modeling frameworks. Chapter 3 considers situations when adequate empirical data on uncertain variables is lacking thereby necessitating the use of non-probabilistic approaches to quantify uncertainties. The study discusses such situations in the context of structural safety assessment. The problem of developing convex function and fuzzy set models for uncertain variables based on limited data and subsequent application in structural safety assessment is considered. Strategies to develop convex set models for limited data based on super-ellipsoids with minimum volume and Nataf’s transformation based method are proposed. These models are shown to be fairly general (for instance, approximations to interval based models emerge as special cases). Furthermore, the proposed convex functions are mapped to a unit multi-dimensional sphere. This enables the evaluation of a unified measure of safety, defined as the shortest distance from the origin to the limit surface in the transformed standard space, akin to the notion used in defining the Hasofer- Lind reliability index. Also discussed are issues related to safety assessment when mixed uncertainty modeling approach is used. Illustrative examples include safety assessment of an inelastic frame with uncertain properties. The study reported in chapter 4 considers a few inverse problems of structural safety analysis aimed at the determination of system parameters to ensure a target level of safety and (or) to minimize a cost function for problems involving combined probabilistic and non-probabilistic uncertainty modeling. Development of load and resistance factor design format, in problems with combined uncertainty models, is also presented. We employ super-ellipsoid based convex function/fuzzy variable models for representing non-probabilistic uncertainties. The target safety levels are taken to be specified in terms of indices defined in standard space of uncertain variables involving standard normal random variables and (or) unit hyper-spheres. A class of problems amenable for exact solutions is identified and a general procedure for dealing with more general problems involving nonlinear performance functions is developed. Illustrations include studies on inelastic frame with uncertain properties. A summary of contributions made in the thesis, along with a few suggestions for future research, are presented in chapter 5. Annexure A-F contain the details of derivation of alternative forms of safety measures, Newton Raphson’s based methods for optimization used in solutions to inverse problems, and details of combining Matlab based programs for uncertainty modeling with Abaqus based models for structural analysis.

Distributed generators (DGs) are proposed as a possible solution to supply economic and reliable electricity to customers. It is adapted to overcome the challenges that are characterized by centralized generation such as transmission and distribution losses, high cost of fossil fuels and environmental damage. This work presents the basic principles of integrating renewable DGs in low voltage distribution networks and particularly focuses on the operation of DG installations and their impacts on active and reactive power. In this thesis, a novel technique that applies the stochastic approach for the operation of distribution networks with considering active network management (ANM) schemes and demand response (DR) within a joint active and reactive distribution market environment is proposed. The projected model is maximized based on social welfare (SW) using market-based joint active and reactive optimal power flow (OPF). The intermittent behaviour of renewable sources (such as solar irradiance and wind speed) and the load demands are modelled through Scenario-Tree technique. The distributed network frame is recast using mixed-integer linear programming (MILP) that is solved by using the GAMS software and then the obtained results are being analysed and discussed. In addition, the impact of wind and solar power penetration on the active and reactive distribution locational prices (D-LMPs) within the distribution market environment is explored in terms of the maximization of SW considering the uncertainty related to solar irradiance, wind speed and load demands. Finally, a realistic case study (16-bus UK generic medium voltage distribution system) is used to demonstrate the effectiveness of the proposed method. Results show that ANM schemes and DR integration lead to an increase in the social welfare and total dispatched active and reactive power and consequently decrease in active and reactive D-LMPs. / Ministry of Higher Education and Scientific Research - Iraq / The selected author's publications, the published versions of which were attached at the end of the thesis, have been removed due to copyright.