According to the EU Council in 2007, a target of 20% Renewable Energy Sources (RES) energy share was determined by the year 2020. Maximizing RES penetration, whilst simultaneously ensuring grid stability and security of electric supply, has become a major challenge for the grid operators. The aggregated effect of Distributed Generation (DG) units will affect increasingly the transmission grid operation and planning. More and more, the High Voltage (HV) grid has to export the excess of power produced at the Medium Voltage (MV) level, where DG units are connected. The energy flows become variable both in value and direction in substations at the interface with distribution networks, which is a complete change for the grid operator. Power flow congestions and voltage problems are particularly more likely to arise. Systematically reinforcing the network in order to absorb the last MWh produced by DG units located in unfavorable areas, while maintaining the traditional operation of the grid, is not efficient, i.e. neither economically viable for the community nor acceptable from the point of view of environmental impact. The intermittency of DG units makes it irrelevant to define the amount of connectable units on the basis of their installed power and the N-1 criterion. New paradigms to increase the grid capacity of accepting DG units before reinforcement are to be considered. And new methodologies for long-term and operational grid planning, giving allowance to this inherent variability in the generation, are therefore necessary.Active Network Management (ANM) allows to moving away from conventional grid operation towards a new approach, comprising (almost) real-time supervision and control of the DG units and network elements. Thanks to this new management of the system and accounting for the intermittent (i.e. weather-dependent) RES production, more DG units can be connected to an existing grid: the power produced by some DG units can be curtailed to eliminate possible congestions encountered for specific combinations of loads, generations and weather conditions. In others words, the use of an ANM scheme makes possible to maximize the grid utilization in enhancing the required flexibility of system operation to maintain power system security margins.A reasonable level of security in applying ANM is however required and it must be assessed before any possible application to the grid. This assessment can be performed based on a probabilistic approach: the uncertain parameters, i.e. each load and power produced by a DG unit, are modeled with probability density functions (pdf’s); the latter are then randomly sampled, to create so-called variants. These variants serve as input data for an Optimal Power Flow (OPF) module to find the possible redispatching or curtailment that could be necessary in each case. The state space is extremely vast, however, due to combinatorial explosion. Creating a sufficiently large sample of variants to cover all significant situations the grid can face appears intractable, and alternative approaches, combining a systematic search in the state space with an acceptable computation time, are to be developed.This research proposes a pragmatic methodology to handle the high dimensionality of the problem and estimate the impact of connecting a new DG unit, via the computation of several risk indices. A systematic approach guarantees searching all over the plausible congestion zones of the state space, while an on-target sampling drives the computational effort towards the direction of interest. This combined approach allows managing the computation time without falling into oversimplification or losing too much accuracy. / Doctorat en Sciences de l'ingénieur et technologie / info:eu-repo/semantics/nonPublished

As an alternative and renewable energy source, the shallow geothermal energy evolving as one of the most popular energy source due to its easy accessibility and availability worldwide, and the ground source heat pump (GSHP) systems are the most frequent applications for extracting the energy from the shallow subsurface. As the heat extraction capacity of the GSHP system applications arises, the design of the borehole heat exchangers (BHE), which is the connected part of the system in the ground, become more important. The backfilling materials of BHEs, particularly, the grout material must provide a suitable thermal contact between the ground and the heat carrier fluid in the high density polyethylene (HDPE) pipes and ensure durability to the induced thermal stresses due to the heat loading. In addition, for the heating purposes of buildings, BHEs immerged in groundwater may be operated below the freezing point of water with anti-freeze mixture in the pipe, leading to freezing-induced ice pressure which may damage the grout.In order to propose a proper grouting for BHEs, the thermo-hydro-mechanical behavior of the grout and its interferences with the adjacent ground conditions must be evaluated in the near field, and the thermal interactions of each BHE in a multi-BHEs field in the long-term operations must also be considered at a further field.Primarily, we have evaluated the performance of various grouting materials, through thermal, hydraulic and mechanical laboratory characterizations. In particular, we have proposed a homemade grout material, with the addition of graphite powder to improve the thermal properties of grout material. In parallel, the characteristics of two different widely used commercial grouting materials (i.e. calcite-based and silica-sand based materials) have been also investigated. In the subsequent study, the heat flow rate per meter of a BHE and the borehole resistance of borehole heat exchangers are assessed experimentally in a 1×1×1 m3 sandbox under, successively, dry sand and fully water-saturated sand conditions. During the operations, the monitored temperatures in the sandbox are in good agreement with analytical predictions. This study demonstrated that the homemade admixture prepared with 5 % natural flake graphite can be considered as an appropriate grout for BHEs regarding to its rheological and thermo-physical properties. Thermally-enhanced grouting can be of significant interest in a high thermal conductivity ground (such as saturated sand) because it minimizes the thermal resistance of the BHE.After characterizing and testing the efficiency of various grout materials, the thermal stresses occurred in BHEs due to heat injection or extraction has been investigated with the analytical solution of hollow cylinder model that is adapted for time-dependent heat loading, the geometry of a BHE, and the thermo-mechanical properties of surrounding ground conditions. Firstly, the hollow cylinder model has been solved for the considered boundary conditions in 2D plane stress. Secondly, the temperature differences at the inner and outer circles of the cylinder is evaluated with the heat line source models for continuous and discontinuous loadings to observe the impact of the heat loading schedule. The developed analytical solution for thermal stress investigation is validated with numerical models. It is demonstrated that the analytical solutions agree well with numerical results for two types of BHE configurations (co-axial and single U-shaped pipes). Furthermore, the calculated maximum stresses are compared with the tensile strength of grout materials obtained from Brazilian tests. It is predicted that thermal contraction of the grout, partially constrained by the surrounding rock, generates tensile stresses that may lead to cracking in the BHE. According to the results, the stiffness of rock has primary role on the developed tensile stresses, and the relationship between the thermal conductivity of the ground and of the grout induces a proportional impact on the magnitude of thermal stresses.Another major concern is the freeze-resistance of the grout materials, when the system is operated for heating purposes. Firstly, we conducted an experimental setup in a small-scale sandbox to understand the behavior of the grout material by evaluating the permeability change during freeze-thaw cycles of a BHE. According to the results, the permeability of grout materials did not change after 10 freeze-thaw cycles due to the thermal transfer with the adjacent soil partially reducing the impact of freezing in the grout material. Therefore, in order to test the freeze-resistance of a BHE, we have investigated the freezing impact of pore water pressure and thermal stress with analytical models and experimental setups on BHEs. For the theoretical approach, an analytical solution has been developed by using the hollow cylinder model that accounts for both the HDPE pipe and the grout material. Firstly, the freezing pore water pressure is adapted to the generalized Hooke’s law equations in 2D plane stress, and secondly the model is solved for the considered boundary conditions. In order to validate the developed model, the experimental setup is conducted in agreement with the geometry of the considered analytical model and the BHE probes are prepared with three different grout materials having large difference in the thermal and hydraulic characteristics (i.e. silica-sand based, calcite based and the homemade enhanced thermally with natural flake graphite powder). According to the experiments for 50 h of freezing operation, the calcite based grout and the homemade grout, having lower permeability and relatively higher porosity, are fractured. In contrast, the silica-sand based grout having higher permeability did not exhibit any damage. Compared with the theoretically obtained results, the observations from the experiments are consistent with the calculated stress results. The effective tangential stress induced by the freezing pore water pressure causes the crack development and agrees with the crack patterns. As a conclusion, the porosity and the permeability play a significant role on the grout failure.In a multi-BHEs field, the thermal interaction between each BHE may have a significant influence on the near-field investigation results in long-term operations. Therefore, in order to complete the near-field investigation, a far-field long-term operation study is required. However, existing analytical solutions for thermal analysis of ground source heat pump (GSHP) systems evaluate temperature change in the carrier-fluid and the surrounding ground in the production period of a single BHE only if a continuous heat load is assigned. In this study, we modified the Green’s function, which is the solution of heat conduction/advection/dispersion equation in porous media, for discontinuous heat extraction by analytically convoluting rectangular function or pulses in time domain both for single and multi-BHEs field. The adapted analytical models for discontinuous heat extraction are verified with numerical finite element code. The comparison results agree well with numerical results both for conduction and advection dominated heat transfer systems, and analytical solutions provide significantly shorter runtime compared to numerical simulations (approx. 1500 times shorter). Furthermore, we investigated the sustainability and recovery aspects of GSHP systems by using proposed analytical models under different hydro-geological conditions. According to the engineering guideline VDI 4640, a linear relationship between thermal conductivity of the ground and the sustainable heat extraction rate is demonstrated for multi-BHEs. In addition, we developed an MATLAB interface for users in which the analytical model can be used easily and more efficiently.In addition, in order to extend the case studies for a ground including several layers, we proposed a finite line source model for BHEs that takes into account conduction/advection/dispersion mechanism in multilayer porous media. Firstly, the anisotropy is added to the moving finite line source model, and we used an existing composite model approach for conductive multilayer ground. The comparison with the numerical model results demonstrates the suitability of the approach. The proposed model can provide a faster solution than classical numerical approaches and help to optimize the heat extraction rate in multilayer media. However, further investigations are required to validate the model with the field measurements. / Doctorat en Sciences de l'ingénieur et technologie / info:eu-repo/semantics/nonPublished

It is estimated that concrete represents 5% of the anthropogenic CO2 emissions, mainly originating from the production of cement, the most essential component of concrete. The recent awareness to the environmental challenges facing our civilization has led the cement industry to consider substituting cement by mineral additions, by-products of existing industries. In this work, a combination of limestone filler and blast furnace slag is used to design an “eco-concrete”, defined as a concrete combining a massive decrease of the cement content as well as mechanical properties and potential durability equivalent to a normal strength concrete. This research reviews the main concrete mix design parameters and their importance when designing an eco-concrete, resulting in five compositions containing various amounts of water, limestone filler and slag. The independent and combined effects of these parameters are observed on the development of the main concrete properties through the use of specific experimental equipment. Continuous non-destructive methodologies to assess efficiently and accurately the properties of any cement-based materials are developed by combining adapted traditional measurements and ultrasonic testing. The question of the early age cracking risk of these concretes is introduced by studying the effect of temperature on the main concrete properties, and a unified model is proposed for taking its effect of the early age autogenous deformation. Finally, new experimental observations are made regarding the behaviour of concrete under fully restrained deformations. The proposed eco-concrete meets many of the requirements of a sustainable construction material. / Doctorat en Sciences de l'ingénieur et technologie / info:eu-repo/semantics/nonPublished

The current work concerns the study and the implementation of a modern algorithm for a posteriori error estimation in Computational Fluid Dynamics (CFD) simulations based on partial differential equations (PDEs). The estimate involves the use of duality argument and proper consistent discretisation of primal and dual problem.A key element is the construction of the adjoint form of the primal differential operators where the data term is a quantity of interest depending on the application. In engineering, this is typically a physical functional of the solution. So, by solving this adjoint problem, it is possible to obtain important information about local sensitivity of the error with respect to the current target quantity and thereby, we are able to perform an a posteriori error representation based on adjoint data. Through this, we provide local error indicators which can drive an adaptive meshing algorithm in order to optimally reduce the target error. Therefore, we first derive and solve the discrete primal problem in agreementwith the chosen numerical method. According to consistency and compatibility conditions, we can use the same discretisation for solving the adjoint problem, simply by swapping the position of the unknowns and the test functions in the linearised variational operator. Remembering that the corresponding adjoint problem always remains linear, the computational cost for obtaining these data is limited compared to the effort needed to solve the primal nonlinear problem.This procedure, fully developed for Discontinuous Galerkin (DG) and Finite Volume (FV) methods, is here for the first time applied in a fully consistent way for Petrov-Galerkin (PG) discretisations. Differently from the latter, the biggest issue for the PG method becomes the need to handle two different functional spaces in the discretisation, one of which is often not even continuous. Stabilized finite element schemes such as Streamline Upwind (SUPG), bubble stabilized (BUBBLE) Petrov-Galerkin and stabilized Residual Distribution (RD) have been selected for implementation and testing. Indeed, based on local advection information, these schemes are naturally more suitable for solving hyperbolic problems and therefore, interesting alternatives for fluid dynamics applications.A scalar linear advection equation is used as a model problem for convergence rate of both primal and adjoint solutions and target quantity. In addition, it is also applied in order to verify the accuracy of the adjoint-based a posteriori error estimate. Next, we apply the methods to a complete collection of numerical examples, starting from scalar Burgers’ problem till 2D compressible Euler equations. Through suited quantities of interest, we illustrate aspects of the adjoint mesh refinement by comparing its efficiency with respect to the standard a posteriori error estimation. / Doctorat en Sciences de l'ingénieur et technologie / info:eu-repo/semantics/nonPublished

Knee joint kinematics is the result of a complex roto-translation movementcharacteristic of the tibio-femoral (TF) and patello-femoral (PF) articulations.This movement depends on the shape of the femur, the tibial plateau andthe patella. Moreover, it depends also on the morphological and mechanicalproperties of the soft tissues of the knee joint. In fact, the knee is characterizedby an extrinsic stability due to the active constraints (muscles and tendons)and passive soft tissues (menisci, retinaculum and ligaments) that surround it.As a result, knee kinematics and kinetics are different in each human being, andsometimes, even in the same person, with the right knee behaving differentlycompared to the left one.The ideal total knee arthroplasty TKA, used to correct pathologies that couldaffect the knee joint, should enable the restoration of the patient’s functionalknee kinematics and kinetics, so that the patient does not normally notice theTKA implant.Nowadays, TKA surgery is a well-established procedure and surgeons maychoose from among the broad range of TKA solutions available on the marketto meet the patient’s request. Prostheses may differ because of shape, materials,and mechanical constraints of their components. Consequently, the restorationof the knee joint biomechanics is limited by the degrees of freedom guaranteedby the adopted design solution.Despite the success of TKAs, pain and limited motor skills are reportedto still affect the clinical outcomes and not all patients are shown to be happyafter a TKA.Current complaints regarding post-TKA surgery might be related to the absenceof a proven tool that enables predicting patient-specific outcomes based ondifferent TKA solutions and providing guidelines to surgeons. In fact, surgicalpre-planning is usually based on a patient’s evaluation that the clinician canmake also based on medical images, and clinical experience. Data reported inthe literature can help in guiding the surgeon to a final decision regarding thebest subject-specific solution.Numerical methods, able to simulate knee biomechanics for various configurations,can be fundamental for the development of the appropriate reliableand effective tools to support clinically-tailored responses to a question.In particular, they can be used for subject-specific analyses on the intact kneeand for supporting the surgical pre-planning phase by comparing the effect ofdifferent solutions.When developing a subject-specific knee model, different kinds of datainputsare needed, such as the knee shapes and alignment information, softtissuesbehavior and boundary conditions describing the investigated motortasks. Often, most of this requested data are unlikely to be available (e.g.subject-specific soft-tissues material properties). Consequently, it is a commonoperating procedure to integrate literature data with subject-specific informationin order to develop knee models for collecting personalized outputsthat could be used to address research and clinical questions.However, up to now, the resulting effect of different generalized sources, asa mix of subject-specific and literature data, still needs to be evaluated for itsimpact on personalized outputs concerning knee behaviour.Furthermore, clinical questions are often focused on specific requests thatpartially use features of more complex knee models that could require too muchtime to be efficiently incorporated into daily clinical evaluations.For these reasons, the principal aims of this research have been to assess,first, the impact of differently derived generalized sources on the developmentof an intact subject-specific knee model or after a TKA; second,to provide guidelines to identify efficient clinically-tailored data sourcesused in and for knee modeling.To accomplish these tasks, a numerical knee model of an intact knee wasdeveloped based on both subject-specific and literature data sources. Theinfluence of different approaches to deal with a subject’s information, such asthe reconstruction of the knee geometries from different imaging sources, hasiiibeen evaluated. Moreover, a sensitivity analysis was performed to understandthe potential changes on kinetics and kinematics outcomes due to differentlyderived literature inputs, such as models and the properties that characterizethe joint materials and ligaments description. The outputs collected after finiteelement analyses were analyzed and compared with already published experimentaloutcomes for the same analyzed specimen and replicated boundaryconditions.Additionally, the effects on knee joint contact forces and kinematics afterTKA surgery and due to the mis-alignment of implant components or misidentificationsof ligament insertions were evaluated in another sensitivityanalysis performed with a rigid body analysis for four different TKA designsimplanted in a subject-specific knee model. As for the intact knee model, theanalyzed configurations were compared against already published experimentaloutputs or literature data replicating similar boundary conditions.Moreover, several dedicated knee models were developed to address specificclinical questions, such as the lack of biomechanical explanations for certainbehaviours of TKA designs.Once compared to already published experimental or literature data, the resultsof the developed models agree.The main results from the numerical simulations performed show that, changingthe values of some of the parameters used as inputs, the knee model kinematicsis less influenced than the contact forces and stresses outputs.In particular, in developing an intact knee model, the main effecting parameteris the material properties selection for the knee cartilage layers. Among theconfigurations analyzed using subject-specific knee models with TKAs, theposition of the tibial component and the height of the patellar buttonare the most effecting inputs.Exploring the different chapters of this research thesis, several specific resultsare shown regarding each main step followed in developing a knee numericalmodel. For example, new approaches based on MRIs have been suggested andtested proving that they are suitable for collecting subject-specific informationregarding geometrical shapes and landmark definitions. Moreover, a newgraphical method was proposed resulting more effective and immediate thanconventional representations in reporting huge amount of data. In particular,the method is the favourite to show complex biomechanical analyses especiallyfor the clinical audience that replied to a survey. Furthermore, the differentmodels tailored to address specific clinical questions collected useful biomechanicalresults, to provide clinical advice or industrial guidelines, and can beconsidered as examples of what should be included in a knee model for similarscenarios.The results of this thesis offer several contributions. Generally, these findingscould provide useful guidelines for knee-model developers to achievea more balanced approach to subject-specific intact knee models based upongeneral sources in order to improve the understanding of personalized kneebiomechanics.To address a general comment to the title of this thesis, there is no singleanswer. In fact, the selection of data sources is case-dependent using, forexample, the subject’s or literature available data to describe material’s behavioror the boundary conditions of a specific motor task. Moreover, differentclinical questions can be addressed with different numerical approaches, e.g.finite element analysis is necessary especially in the case that stress outputs arerequested, but can be too time-consuming for addressing complex sensitivityanalyses.Once the knee model developer has identified the necessary data sources andthe approaches to be implemented, the question-tailored knee models can thusbe used for several applications such as predicting subject-specific abnormalknee kinematics and kinetics for different TKA designs, polyethylene wear,patellofemoral dislocation and bone remodeling, choosing the best fitting TKAdesign for a specific patient, and developing a procedure to optimize TKAimplant designs. / Doctorat en Sciences de l'ingénieur et technologie / info:eu-repo/semantics/nonPublished

In recent years, obesity has literally reached epidemic proportions throughout the world. Gastric electrical stimulation is a recent technique that uses an implanted device to stimulate the stomach and produce a feeling of satiety for overweight and obese patients. However, to place the implant, the patient currently needs to undergo an invasive surgical procedure. Endoscopic implantation could be used to place the gastric stimulator in the stomach. This would help moderately and morbidly obese patients that are ineligible for surgery, but such a solution brings new challenges for the development of the gastrostimulator. This thesis is a first step toward a commercial implant, endoscopically implantable and treating obesity. This work mainly focuses on the design and implementation of a first endoscopically implantable prototype, to be validated on animals. This will allow us to prove the feasibility of the project, through a pre-clinical phase on animals. Compared to previous devices, the prototype developed is specifically designed to stimulate the stomach using embedded electrodes. Its design allows endoscopic implantation. Its packaging resists long-term implantation in an acidic environment, hence overcoming the shortcomings of previous gastrostimulators. A flexible and low-cost manufacturing chain is implemented. The prototype is built with discrete components, commercially available, and encapsulated with silicone rubber. This thesis also provides a method to select a silicone rubber for the encapsulation of an electronics circuit. The validation of the prototype is performed on a bench-test, in ex-vivo and in-vivo on three dogs. A new method using surface cutaneous electrodes is introduced to validate the functioning of the stimulator. / Doctorat en Sciences de l'ingénieur et technologie / info:eu-repo/semantics/nonPublished

Nowadays, in the field of energy production, particular attention must be paid to improving efficiency, reducing pollutants and fuel flexibility. To reach those goals, cogenerative systems represent an appealing solution. One of the most promising cogenerative systems available nowadays is the micro turbine, which provides reasonable electrical efficiency of about 30%, multi-fuel capability, low emission levels and heat recovery potential, and need minimum maintenance. Among the several options, micro gas turbines (mGT) are particularly interesting. Beside theuse of natural gas, other fuels like landfill gas, ethanol, industrial waste off-gases and other bio-based gases can be used. Moreover, it is possible to further improve the efficiencies and reduce the emissions for mGTs by paying particular attention at the design of the combustion chamber. To this goal, flameless combustion could be an interesting solution. Flameless combustion is able to provide high combustion efficiency with low NOx and soot emissions. The increasing interest in flameless combustion is motivated by its large fuel flexibility, representing a promising technology for low-calorific value fuels, high-calorific industrial wastes as well as in presence of hydrogen. Moreover, flameless combustion is very stableand noiseless, so it is suited for gas turbine applications where conventional operations may lead to significant thermo-acoustic instabilities (“humming”) and stresses. Flameless combustion needs the reactants to be preheated above their self-ignition temperature and enough inert combustion products to be entrained in the reaction region, in order to dilute the flame. As a result, the temperature field is more uniform than in traditional combustion systems, and it does not show high temperature peaks. Hence, NOx formation is suppressed as well as soot formation,due to the lean conditions, low temperatures and the large CO2 concentration in the exhausts.mGTs operating in flameless combustion regime represent a promising technology for the combined production of heat and power with increased efficiency, reduced pollutants emission and high fuel flexibility. The objective of the present Thesis is the design of a combustion chamber for amGT for residential applications. The design is performed employing CFD-tools. Thus, it is necessary to develop a reliable numerical model to use in the design process. Therefore, the first step of the Thesis consists in a series of validation studies, with the goal of selecting the most appropriate and reliable models to describe flameless combustion. The validation will be carried on three differenttest cases, which have different nominal powers and employ different gaseous fuels. The second part of the Thesis focuses on the design and optimization of the combustion chamber. Finally, the third part shows the experimental investigation of the aforementioned chamber.The study of those three cases shows that, to correctly predict the behavior of those systems, it is necessary to take into account both mixing and chemical kinetics. The best results have been obtained with the Eddy Dissipation Concept model, coupled with detailed kinetic schemes. As far as the NOx emissions are concerned, it is fundamental to include all the formation routes, i.e. thermal, prompt, via N2O and NNH route, to estimate properly the NOx production in flameless conditions.The aforementioned models have been used for the design and optimization of a combustion chamber for a mGT operating in flameless combustion regime. Both the design and the optimization have been carried out by means of CFD simulations and both are goal-oriented, meaning that they are carried out with the purpose of improving one or more performance indicators of the chamber, such as pollutants emissions, efficiency or pressure losses. The configuration that satisfies the criteria on the performance indicators has been built and investigated experimentally. The combustion chamber is stable and performs well in terms of emissions for a wide range of air inlet temperature and air-fuel equivalence ratio, lambda, values. Except for the condition closer to the stoichiometric one, both CO and NOx emissions are extremely low for all !and air inlet temperatures. Thechamber performs the best at its nominal operating condition, i.e. lambda = 3.5 and air inlet temperature 730 °C, In this case CO is 0 ppm and NOx is 5.6 ppm. The numerical model employed to describe the combustor performs quite well, except for the CO prediction, for all the conditions investigated. The final step of the present work is the application of a different kind of fuel, namely biogas. First the feasibility of such application has been evaluated using CFD calculations, and then the experimental evidence has been discussed. Due to a calibration error on the gas flow meter, it has not been possible to investigate the conditions of the design point (lambda = 3.5). Three other conditions have been examined,characterized by lower values of !closer to the stoichiometric conditions. Despite the relatively high values of NOx emissions due to the lower air excess and to the consequently higher temperatures, the combustion chamber has proven to be fuel flexible. Both ignition and stable combustion can be achieved also when biogas is burnt. Numerical simulations have also been performed; the results are in good agreement with the experimental evidence. / Doctorat en Sciences de l'ingénieur et technologie / info:eu-repo/semantics/nonPublished

Technology has a major role in today’s world. The development and massive access to information technology has enabled the use of computers to provide assistance on a wide range of tasks, from the most trivial daily ones to the most complex challenges we face as human kind. In particular, optimisation algorithms assist us in taking decisions, improving processes, designing solutions and they are successfully applied in several contexts such as industry, health, entertainment, and so on. The design and development of effective and efficient computational algorithms is, thus, a need in modern society.Developing effective and efficient optimisation algorithms is an arduous task that includes designing and testing of several algorithmic components and schemes, and requires considerable expertise. During the design of an algorithm, the developer defines parameters, that can be used to further adjust the algorithm behaviour depending on the particular application. Setting appropriate values for the parameters of an algorithm can greatly improve its performance. This way, most high-performing algorithms define parameter settings that are “finely tuned”, typically by experts, for a particular problem or execution condition.The process of finding high-performing parameter settings, called algorithm configuration, is commonly a challenging, tedious, time consuming and computationally expensive task that hinders the application and design of algorithms. Nevertheless, the algorithm configuration process can be modelled as an optimisation problem itself and optimisation techniques can be applied to provide high-performing configurations. The use of automated algorithm configuration procedures, called configurators, allows obtaining high-performing algorithms without requiring expert knowledge and it enables the design of more flexible algorithms by easing the definition of design choices as parameters to be set. Ultimately, automated algorithm configuration could be used to fully automatise the algorithm development process, providing algorithms tailored to the problem to be solved.The aim of the work presented in this thesis is to study the automated configuration of algorithms. To do so, we formally define the algorithm configuration problem and analyse its characteristics. We study the most prominent algorithm configuration procedures and identify relevant configuration techniques and their applicability. We contribute to the field by proposing and analysing several configuration procedures, being the most prominent of these the irace configurator. This work presents and studies several modifications of the configuration process implemented by irace, which considerably improve the performance of irace and broaden its applicability. In a general context, we provide insights about the characteristics of the algorithm configuration process and techniques by performing several analyses configuring different types of algorithms under varied situations. And, finally, we provide practical examples of the usage of automated configuration techniques showing its benefits and further uses for the application and design of efficient and effective algorithms. / Doctorat en Sciences de l'ingénieur et technologie / info:eu-repo/semantics/nonPublished

Doctorat en sciences appliquées / info:eu-repo/semantics/nonPublished

Doctorat en sciences appliquées / info:eu-repo/semantics/nonPublished