L'inférence d’un système, par la reconstruction de la structure à partir de l’analyse de son comportement, est reconnue comme un problème critique. Dans la théorie des systèmes, la structure et le comportement se situent aux extrémités de la hiérarchie qui définit la connaissance du système. L'inférence d’un système peut être également considérée comme l’escalade de la hiérarchie depuis la connaissance de bas niveau vers la connaissance de plus haut niveau. Ceci n'est possible que sous des conditions maitrisées et justifiées. Dans cette thèse, une nouvelle méthode d'inférence de système est proposée. La méthode proposée étend la technique Process Mining pour extraire des connaissances depuis les données des événements du système. Les aspects de modularité, de fréquence et de synchronisation peuvent être extraits des données. Ils sont intégrés ensemble pour construire un modèle Fuzzy-Discrete Event System Specification (Fuzzy-DEVS). La méthode proposée, également appelée méthode D2FD (Data to Fuzzy-DEVS), comprend trois étapes: (1) l’extraction depuis des journaux d’évènements (registres) obtenus à partir des données générées par le système en utilisant une approche conceptuelle; (2) la découverte d'un système de transition, en utilisant des techniques de découverte de processus; (3) l'intégration de méthodes Fuzzy pour générer automatiquement un modèle Fuzzy-DEVS à partir du système de transition. La dernière étape est de l’implémenter cette contribution en tant que plugin dans l'environnement Process Mining Framework (ProM). Afin de valider les modèles construits, une approximation de modèle basée sur le morphisme et une méthode prédictive intégrée à Granger Causality sont proposées. Deux études de cas sont présentées dans lesquelles le modèle Fuzzy-DEVS est déduit à partir de données réelles, où l'outil SimStudio est utilisé pour sa simulation. Les modèles ainsi construits et les résultats de simulation sont validés par comparaison à d'autres modèles. / System inference, i.e., the building of system structure from system behavior, is widely recognized as a critical challenging issue. In System Theory, structure and behavior are at the extreme sides of the hierarchy that defines knowledge about the system. System inference is known as climbing the hierarchy from less to more knowledge. In addition, it is possible only under justifying conditions. In this thesis, a new system inference method is proposed. The proposed method extends the process mining technique to extract knowledge from event data and to represent complex systems. The modularity, frequency and timing aspects can be extracted from the data. They are integrated together to construct the Fuzzy Discrete Event System Specification (Fuzzy-DEVS) model. The proposed method is also called D2FD (Data to Fuzzy-DEVS) method, and consists of three stages: (1) extraction of event logs from event data by using the conceptual structure; (2) discovery of a transition system, using process discovery techniques; (3) integration of fuzzy methods to automatically generate a Fuzzy-DEVS model from the transition system. The last stage is implemented as a plugin in the Process Mining Framework (ProM) environment. In order to validate constructed models, morphism-based model approximation and predictive method integrated with Granger Causality are proposed. Two case studies are presented in which Fuzzy-DEVS model is inferred from real life data, and the SimStudio tool is used for its simulation. The constructed models and simulation results are validated by comparing to other models.

A Identificação de Sistemas em Malha Fechada possui considerável apelo prático, uma vez que oferece maior segurança durante a coleta experimental de dados e ao mesmo tempo, em linhas gerais, proporciona a construção de modelos mais adequados para servir de base ao projeto de sistemas de controle. Esta Tese apresenta, como um de seus principais objetivos, a comparação dos Métodos Diretos aplicados à Identificação em Malha Fechada com a classe dos Métodos de Dois-Passos, que se enquadram na abordagem de Identificação Conjunta Entrada/Saída. Complementando esta comparação, propõe-se um novo algoritmo em Dois-Passos, a Dupla Filtragem. As propriedades de convergência deste método são analisadas em detalhe. O desempenho alcançado pelos modelos identificados pelos Métodos Diretos e com o uso dos Métodos de Dois-Passos aqui considerados a saber, Filtragem-u (VAN DEN HOF; SCHRAMA, 1993), Filtragem-y (HUANG; SHAH, 1997) e Dupla Filtragem são comparados em uma abordagem estatística por meio da aplicação de Simulações de Monte Carlo. Também se propõe uma variante ao método da Filtragem-u, proporcionando duas formas distintas de descrever a função de sensibilidade da saída associada ao processo sob estudo (FORSSELL; LJUNG, 1999). Os critérios de comparação de desempenho adotados nesta tese incluem validações dos modelos identificados em simulações livres (operação em malha aberta), em que os objetos de análise são respostas a pulsos retangulares e, com maior ênfase, validações em malha fechada que utilizam o mesmo controlador instalado no sistema sob estudo. Nesta última situação são empregados sinais de excitação de mesma natureza daqueles adotados nos ensaios de identificação, porém com diferentes realizações. Cada uma dessas validações é acompanhada de seu respectivo fit (LJUNG,1999), índice de mérito que mede a proximidade entre as respostas temporais do sistema físico e de seu modelo matemático. Também são consideradas as respostas em frequência do processo, que constituem a base para a determinação do limite máximo para a incerteza associada ao modelo (ZHU, 2001). Tomando como fundamento tais limites máximos de incerteza, em conjunto com as respostas em frequência dos modelos identificados, é possível associar graduações a esses modelos (A, B, C, ou D). Desta forma, esta tese utiliza índices de mérito fundamentados em ambas as respostas temporais e em frequência. Aspectos relativos à influência da amplitude e do tipo de sinal de excitação aplicado à malha, bem como à relação sinal-ruído estabelecida no sistema, são analisados. Também se investiga a relação entre a qualidade do modelo identificado e o ponto de aplicação do sinal de excitação: no valor de referência da malha de controle ou na saída do controlador. Por fim, verifica-se como a sintonia do controlador afeta o modelo identificado. Todas as simulações realizadas utilizam sinais de perturbação do tipo quase não- estacionário, típicos da indústria de processos (ESMAILI et al., 2000). Os resultados indicam que os Métodos Diretos são mais precisos quando a estrutura de modelo e ordem adotadas são idênticas àquelas do processo real. No entanto, os Métodos de Dois-Passos são capazes de fornecer modelos muito confiáveis mesmo quando a estrutura e ordem do modelo diferem daquelas do processo sob estudo. / Closed-loop System Identification has considerable practical appeal, since it provides increased security during the collection of experimental data and, at the same time, provides the construction of suitable models for the design of high performance control systems. This thesis presents, as one of its main objectives, a thorough comparison between Direct Methods (applied to the closed-loop identification) and Two-Step Methods. The latter ones belong to the Joint Input/Output approach. Complementing this comparison, a new two-step algorithm the Double Filtering is proposed. The convergence properties of this method are analyzed in detail. The performance achieved by the models identified by Direct and Two-Step methods is compared in a statistical approach through Monte Carlo simulations. The Two-Step methods considered in this thesis are the u-Filtering (VAN DEN HOF; SCHRAMA, 1993), the y-Filtering (HUANG; SHAH, 1997) and the Double Filtering. A variant of the u-Filtering method is proposed, providing two distinct ways of describing the output sensitivity function associated with the process under study (FORSSELL; LJUNG, 1999). The performance comparison criteria adopted in this thesis include free-run model validations (open-loop operation), in which rectangular pulses responses are analyzed. Greater emphasis is given to closed loop model validation, which uses the same controller installed in the system under study. This type of validation employs excitation signals similar to those adopted in the identification tests, but with different realizations. Each of these validations is accompanied by its corresponding fit (Ljung, 1999), a merit index that measures the proximity between the time responses of the physical system and its mathematical model. Process frequency responses are also considered, since they form the basis for determining the model uncertainty upper-limit or upper-bound error (ZHU, 2001). The upper- bounds, along with the frequency responses of each identified model, provides ranks (A, B, C, or D) for these models. Therefore, this thesis uses merit indexes based on both time and frequency responses. It is analyzed how the type and magnitude (or equivalently, the signal-to-noise ratio) of the excitation signal applied to the loop impacts the accuracy of the identified models. This work also investigates the relationship between the accuracy of the identified models and the point of application of the excitation signal: the reference of the control loop or the controller output. Finally, it is checked how the controller tuning affects the identified models. All simulations employ quasi non-stationary disturbance signals, typical of the process industries (ESMAILI et al., 2000). The results indicate that Direct Methods are more accurate when the model structure and order adopted in the identification are identical to those of the actual process. However, the Two-Step Methods are capable of providing very reliable models even when the adopted structure and order differ from those of the process under study.

A Identificação de Sistemas em Malha Fechada possui considerável apelo prático, uma vez que oferece maior segurança durante a coleta experimental de dados e ao mesmo tempo, em linhas gerais, proporciona a construção de modelos mais adequados para servir de base ao projeto de sistemas de controle. Esta Tese apresenta, como um de seus principais objetivos, a comparação dos Métodos Diretos aplicados à Identificação em Malha Fechada com a classe dos Métodos de Dois-Passos, que se enquadram na abordagem de Identificação Conjunta Entrada/Saída. Complementando esta comparação, propõe-se um novo algoritmo em Dois-Passos, a Dupla Filtragem. As propriedades de convergência deste método são analisadas em detalhe. O desempenho alcançado pelos modelos identificados pelos Métodos Diretos e com o uso dos Métodos de Dois-Passos aqui considerados a saber, Filtragem-u (VAN DEN HOF; SCHRAMA, 1993), Filtragem-y (HUANG; SHAH, 1997) e Dupla Filtragem são comparados em uma abordagem estatística por meio da aplicação de Simulações de Monte Carlo. Também se propõe uma variante ao método da Filtragem-u, proporcionando duas formas distintas de descrever a função de sensibilidade da saída associada ao processo sob estudo (FORSSELL; LJUNG, 1999). Os critérios de comparação de desempenho adotados nesta tese incluem validações dos modelos identificados em simulações livres (operação em malha aberta), em que os objetos de análise são respostas a pulsos retangulares e, com maior ênfase, validações em malha fechada que utilizam o mesmo controlador instalado no sistema sob estudo. Nesta última situação são empregados sinais de excitação de mesma natureza daqueles adotados nos ensaios de identificação, porém com diferentes realizações. Cada uma dessas validações é acompanhada de seu respectivo fit (LJUNG,1999), índice de mérito que mede a proximidade entre as respostas temporais do sistema físico e de seu modelo matemático. Também são consideradas as respostas em frequência do processo, que constituem a base para a determinação do limite máximo para a incerteza associada ao modelo (ZHU, 2001). Tomando como fundamento tais limites máximos de incerteza, em conjunto com as respostas em frequência dos modelos identificados, é possível associar graduações a esses modelos (A, B, C, ou D). Desta forma, esta tese utiliza índices de mérito fundamentados em ambas as respostas temporais e em frequência. Aspectos relativos à influência da amplitude e do tipo de sinal de excitação aplicado à malha, bem como à relação sinal-ruído estabelecida no sistema, são analisados. Também se investiga a relação entre a qualidade do modelo identificado e o ponto de aplicação do sinal de excitação: no valor de referência da malha de controle ou na saída do controlador. Por fim, verifica-se como a sintonia do controlador afeta o modelo identificado. Todas as simulações realizadas utilizam sinais de perturbação do tipo quase não- estacionário, típicos da indústria de processos (ESMAILI et al., 2000). Os resultados indicam que os Métodos Diretos são mais precisos quando a estrutura de modelo e ordem adotadas são idênticas àquelas do processo real. No entanto, os Métodos de Dois-Passos são capazes de fornecer modelos muito confiáveis mesmo quando a estrutura e ordem do modelo diferem daquelas do processo sob estudo. / Closed-loop System Identification has considerable practical appeal, since it provides increased security during the collection of experimental data and, at the same time, provides the construction of suitable models for the design of high performance control systems. This thesis presents, as one of its main objectives, a thorough comparison between Direct Methods (applied to the closed-loop identification) and Two-Step Methods. The latter ones belong to the Joint Input/Output approach. Complementing this comparison, a new two-step algorithm the Double Filtering is proposed. The convergence properties of this method are analyzed in detail. The performance achieved by the models identified by Direct and Two-Step methods is compared in a statistical approach through Monte Carlo simulations. The Two-Step methods considered in this thesis are the u-Filtering (VAN DEN HOF; SCHRAMA, 1993), the y-Filtering (HUANG; SHAH, 1997) and the Double Filtering. A variant of the u-Filtering method is proposed, providing two distinct ways of describing the output sensitivity function associated with the process under study (FORSSELL; LJUNG, 1999). The performance comparison criteria adopted in this thesis include free-run model validations (open-loop operation), in which rectangular pulses responses are analyzed. Greater emphasis is given to closed loop model validation, which uses the same controller installed in the system under study. This type of validation employs excitation signals similar to those adopted in the identification tests, but with different realizations. Each of these validations is accompanied by its corresponding fit (Ljung, 1999), a merit index that measures the proximity between the time responses of the physical system and its mathematical model. Process frequency responses are also considered, since they form the basis for determining the model uncertainty upper-limit or upper-bound error (ZHU, 2001). The upper- bounds, along with the frequency responses of each identified model, provides ranks (A, B, C, or D) for these models. Therefore, this thesis uses merit indexes based on both time and frequency responses. It is analyzed how the type and magnitude (or equivalently, the signal-to-noise ratio) of the excitation signal applied to the loop impacts the accuracy of the identified models. This work also investigates the relationship between the accuracy of the identified models and the point of application of the excitation signal: the reference of the control loop or the controller output. Finally, it is checked how the controller tuning affects the identified models. All simulations employ quasi non-stationary disturbance signals, typical of the process industries (ESMAILI et al., 2000). The results indicate that Direct Methods are more accurate when the model structure and order adopted in the identification are identical to those of the actual process. However, the Two-Step Methods are capable of providing very reliable models even when the adopted structure and order differ from those of the process under study.

The hepatitis C virus (HCV), with a global prevalence of roughly 2%, is among the most serious diseases today. Among the more promising HCV targets is the NS3 protease, for which several drug candidates have entered clinical trials. In this work, computational methods have been developed and applied to the design of inhibitors of the HCV NS3 protease. Comparative molecular field analysis (CoMFA) modeling and molecular docking are the two main computational tools used in this work. CoMFA is currently the most widely used 3D-QSAR method. Methodology for improving its predictive performance by evaluating 6120 combinations of non-default parameters has been developed. This methodology was tested on 9 data sets for various targets and found to consistently provide models of enhanced predictive accuracy. Validation was performed using q2, r2pred and response variable randomization. Molecular docking was used to develop SARs in two series of inhibitors of the HCV NS3 protease. In the first series, preliminary investigations indicated that replacement of P2 proline with phenylglycine would improve potency. Docking suggested that phenylglycine-based inhibitors may participate in two additional interactions but that the larger, more flexible phenylglycine group may result in worse ligand fit, explaining the loss in potency. In the second series, β-amino acids were explored as α-amino acid substitutes. Although β-amino acid substitution may reduce the negative attributes of peptide-like compounds, this study showed that β-amino acid substitution resulted in reduced potency. The P3 position was least sensitive to substitution and the study highlighted the importance of interactions in the oxyanion hole. Finally, docking was used to provide the conformations and alignment necessary for a CoMFA model. This CoMFA model, derived using default settings, had q2 = 0.31 and r2pred = 0.56. Application of the optimization methodology provided a more predictive model with q2 = 0.48 and r2pred = 0.68.

Dans ce travail, nous avons développé un système expert qui intègre les fonctions suivantes :i) l'identification des profils à partir des mesures sur les plasmas de tokamak, par les méthodes bayésiennes; ii) La prédiction des quantités reconstruites, selon les modèles choisis et; iii) une comparaison intelligente des résultats provenant des deux premières étapes. Ce système comprend une vérification systématique de la cohérence interne des quantités reconstruites, permet la validation automatique des modèles. Avec un modèle validé, il peut aider à la détection de nouveaux effets physique dans une expérience. Trois applications de ce système sont présentées dans ce travail. Le système expert peut détecter automatiquement les défauts dans la reconstruction des profils et fournir (lorsque la reconstruction est valide) des statistiques sur l'accord des modèles avec les données expérimentales, c'est à dire des informations sur la validité du modèle. / In this work we developed an expert system that carries out in an integrated and fully automated way i) a reconstruction of plasma profiles from the measurements, using Bayesian analysis ii) a prediction of the reconstructed quantities, according to some models and iii) an intelligent comparison of the first two steps. This system includes systematic checking of the internal consistency of the reconstructed quantities, enables automated model validation and, if a well-validated model is used, can be applied to help detecting interesting new physics in an experiment. The work shows three applications of this quite general system. The expert system can successfully detect failures in the automated plasma reconstruction and provide (on successful reconstruction cases) statistics of agreement of the models with the experimental data, i.e. information on the model validity.

It is widely accepted that correct system modeling and identification are among the most important issues power system operators face when managing instability and post-contingency scenarios. The latter is usually performed involving special computational tools that allow the operator to forecast, prevent system failure and take appropriate actions according to protocols for different contingency cases in the system. To ensure that operators make the correct simulation-based decisions, the power system models have to be validated continuously. This thesis investigates power system modeling, identification and validation problems that are formulated and based on data provided by operators, and offers new methods and deeper insight into stages of an identification cycle considering the specifics of power systems. One of the problems this thesis tackled is the selection of a modeling and simulation environment that provides transparency and possibility for unambiguous model exchange between system operators. Modelica as equation-based language fulfills these requirements. In this thesis Modelica phasor time domain models were developed and software-to-software validated against conventional simulation environments, i.e. SPS/Simulink and PSAT in MATLAB. Parameter estimation tasks for Modelica models require a modular and extensible toolbox. Thus, RaPiD Toolbox, a framework that provides system identification algorithms for Modelica models, was developed in MATLAB. Contributions of this thesis are an implementation of the Particle Filter algorithm and validation metrics for parameter identification. The performance of the proposed algorithm has been compared with Particle Swarm Optimization (PSO) algorithm when combined with simplex search and parallelized to get computational speed up. The Particle Filter outperformed PSO when estimating turbine-governor model parameters in the Greek power plant model relying on real measurements. This thesis also analyses different model structures (Nonlinear AutoRegressive eXogenous (NARX) model, Hammerstein-Wiener model, and high order transfer function) that are selected to reproduce nonlinear dynamics of a Static VAR Compensator (SVC) under incomplete information available for National Grid system operator. The study has shown that standard SVC model poorly reproduces the measured dynamics of the real system. Therefore, black-box mathematical modeling and identification approach has been proposed to solve the problem. Also, the introduced combination of first-principle and black-box approach has shown the best output fit. The methodology following identification cycle together with model order selection and model validation issues was presented in detail. Finally, one of the major contributions is a new method to formulate the uncertainty of parameters estimated in the form of a multimodal Gaussian mixture distribution that is estimated from the Particle Filter output by applying statistical methods to select the standard deviations. The proposed methodology gives additional insight into power system properties when estimating the parameters of the model. This allows power system analysts to decide on the design of validation tests for the chosen model. / Det är allmänt accepterat att korrekt modellering och identifiering av systemet är bland de mest viktiga utmaningarna som kraftsystemoperatörer ställs inför när de hanterar scenarior med instabiliteter och oförutsedda händelser. Det senare är vanligen hanterat med speciella beräkningsverktyg som låter operatören förutse utvecklingen och utföra lämpliga åtgärder enligt de protokoll som finns vid olika systemhändelser. För att försäkra sig om att operatörer tar de korrekta, simuleringsbaseda besluten måste kraftsystemsmodellen kontinuerligt valideras. Denna avhandling undersöker problem inom modellering, identifiering och validering av kraftsystem, formulerade och baserade på data tillhandahållet av operatörer, samt erbjuder nya metoder och fördjupade insikter i delar av en identifieringscykel som beaktar kraftsystemets. Ett av de problem som denna avhandling tar upp är val av en programmiljö för simulering och modellering som ger transparens och möjlighet till otvetydigt modellutbyte mellan systemoperatörer. Modelica är ett ekvationsbaserat programspråk som uppfyller dessa krav. I denna avhandling utvecklades enfasekvivalenter i Modelica som blev validerade mot konventionella program för simulering, såsom SPS/Simulink och PSAT i MATLAB. Parameterestimering i Modelica-modellerna kräver en modulär och utbyggbar verktygslåda. Därför har verktyget RaPiD Toolbox, som tillhandahåller systemidentifieringsalgoritmer för Modelica-modeller, utvecklats i MATLAB. Bidrag från denna avhandling är en implementation av ett partikelfilter (en sekventiell Monte Carlo-metod) och valideringsmetrik för parameteridentifiering. Prestandan i den föreslagna algoritmen har jämförts med partikelsvärmoptimering (PSO) då den är kombinerad med simplexsök och parallellisering. Partikelfiltret överträffade PSO när modellparametrar i turbinregulatorn i ett grekiskt kraftverk skulle estimeras utifrån verklig mätdata.  Avhandling analyserar också olika modellstrukturer (NARX, Hammerstein-Wiener-modeller, och överföringsfunktioner med höga ordningstal) som används för att reproducera den ickelinjära dynamiken hos statiska reaktiv effekt-kompenserare (SVC) vid ofullständig information som är tillgänglig för systemoperatören National Grid. Undersökningen visar att den vanliga SVC-modellen är dålig på att reproducera den verkliga, uppmätta dynamiken. Genom att matematiskt modellera problemet som en svart låda har en identifieringsmetod föreslagits. Vidare, genom att kombinera modelleringen som en svart låda med fysikaliska principer har givit den bästa anpassningen till utdata. Metodologin för identifieringscykeln tillsammans med valet av modellkomplexitet och svårigheter med modellvalidering har utförligt presenterats. Slutligen, ett av de främsta bidragen är en ny metod för att formulera osäkerheten i parameteruppskattningarna i form av en blandning av normalfördelningar med flera typvärden som estimeras med partikelfiltrets utdata genom att använda statistiska metoder för att välja standardavvikelsen. Detta ger kraftsystemanalytiker möjlighet att utforma valideringstest för den valda modellen. / <p>QC 20171121</p> / EU FP7 iTesla project

The increasing use of model-driven software development has renewed emphasis on using domain-specific models during application development. More specifically, there has been emphasis on using domain-specific modeling languages (DSMLs) to capture user-specified requirements when creating applications. The current approach to realizing these applications is to translate DSML models into source code using several model-to-model and model-to-code transformations. This approach is still dependent on the underlying source code representation and only raises the level of abstraction during development. Experience has shown that developers will many times be required to manually modify the generated source code, which can be error-prone and time consuming. An alternative to the aforementioned approach involves using an interpreted domain-specific modeling language (i-DSML) whose models can be directly executed using a Domain Specific Virtual Machine (DSVM). Direct execution of i-DSML models require a semantically rich platform that reduces the gap between the application models and the underlying services required to realize the application. One layer in this platform is the domain-specific middleware that is responsible for the management and delivery of services in the specific domain. In this dissertation, we investigated the problem of designing the domain-specific middleware of the DSVM to facilitate the bifurcation of the semantics of the domain and the model of execution (MoE) while supporting runtime adaptation and validation. We approached our investigation by seeking solutions to the following sub-problems: (1) How can the domain-specific knowledge (DSK) semantics be separated from the MoE for a given domain? (2) How do we define a generic model of execution (GMoE) of the middleware so that it is adaptable and realizes DSK operations to support delivery of services? (3) How do we validate the realization of DSK operations at runtime? Our research into the domain-specific middleware was done using an i-DSML for the user-centric communication domain, Communication Modeling Language (CML), and for microgrid energy management domain, Microgrid Modeling Language (MGridML). We have successfully developed a methodology to separate the DSK and GMoE of the middleware of a DSVM that supports specialization for a given domain, and is able to perform adaptation and validation at runtime.

Many AC grids suffer from decreased frequency stability due to less system inertia. This has increased the risk of large-scale blackouts. This thesis and its papers address the frequency stability problem from a hydropower perspective. Grid frequency stability assessments often require accurate system inertia estimates. One approach is to estimate the inertia of all individual power plants and sum up the results. We implemented three inertia estimation methods on a Kaplan unit and compared their results. The generator contributed with 92-96% of the unit inertia, which verified the results from previous studies. However, the three methods estimated slightly different values for the unit inertia, which raises the question of when to use which method. Hydropower often deliver frequency control, yet we found no studies which validate Kaplan turbine models for large grid frequency disturbances on strong grids. Therefore, we performed frequency control tests on a Kaplan unit, implemented three hydropower models, and compared the simulation results to the measurement data. The models overestimated the change in output power and energy delivered within the first few seconds after a large change in frequency. Thus, it is important to have sufficient stability margin when using these types of hydropower models to assess the grid frequency stability. The Nordic transmission system operators are updating their frequency control requirements. We used measurement data and simulation models to assess whether improved runner blade angle control could help a Kaplan unit fulfill the coming requirements. The results showed that improved runner control does not improve the performance sufficiently for requirements fulfillment. The requirements are based on an assumption on minimum system inertia and became easier to fulfill if they were implemented with more system inertia. Thus, more inertia could allow more participants to deliver frequency control in the Nordic grid.

Thermal energy storage (TES) modules are specifically designed to respond to transient thermal loading. Their dynamic response depends on the overall structure of the module, including module geometry and dimensions, the internal spatial distribution of phase change material (PCM) and conductive heat-spreading elements, and the thermophysical properties of the different materials composing the module. However, due to the complexity of analyzing a system’s dynamic thermal response to transient input signals, optimal design of a TES module for a particular application is challenging. Conventional design approaches are limited by (1) the computational cost associated with high fidelity simulation of heat transfer in nonlinear systems undergoing a phase transition and (2) the lack of model integration with robust optimization tools. To overcome these challenges, I derive reduced-order dynamic models of two different metal-PCM composite TES modules and validate them against a high fidelity CFD model. Through simulation and validation of both turbulent and laminar flow cases, I demonstrate the accuracy of the reduced-order models in predicting, both spatially and temporally, the evolution of the dynamic model states and other system variables of interest, such as PCM melt fraction. The validated models are used to conduct univariate and bivariate parametric studies to understand the effects of various design parameters on different performance metrics. Finally, a case study is presented in which the models are used to conduct detailed design optimization for the two HX geometries.

This thesis aims at formulating and validating models for electrochemical energy storage devices. More specifically, the devices under consideration are lithium ion batteries and polymer electrolyte fuel cells. A model is formulated to describe an experimental cell setup consisting of a LixNi0.8Co0.15Al0.05O2 composite porous electrode with three porous separators and a reference electrode between a current collector and a pure Li planar electrode. The purpose of the study being the identification of possible degradation mechanisms in the cell, the model contains contact resistances between the electronic conductor and the intercalation particles of the porous electrode and between the current collector and the porous electrode. On the basis of this model formulation, an analytical solution is derived for the impedances between each pair of electrodes in the cell. The impedance formulation is used to analyse experimental data obtained for fresh and aged LixNi0.8Co0.15Al0.05O2 composite porous electrodes. Ageing scenarios are formulated based on experimental observations and related published electrochemical and material characterisation studies. A hybrid genetic optimisation technique is used to simultaneously fit the model to the impedance spectra of the fresh, and subsequently also to the aged, electrode at three states of charge. The parameter fitting results in good representations of the experimental impedance spectra by the fitted ones, with the fitted parameter values comparing well to literature values and supporting the assumed ageing scenario. Furthermore, a steady state model for a polymer electrolyte fuel cell is studied under idealised conditions. The cell is assumed to be fed with reactant gases at sufficiently high stoichiometric rates to ensure uniform conditions everywhere in the flow fields such that only the physical phenomena in the porous backings, the porous electrodes and the polymer electrolyte membrane need to be considered. Emphasis is put on how spatially resolved porous electrodes and nonequilibrium water transport across the interface between the gas phase and the ionic conductor affect the model results for the performance of the cell. The future use of the model in higher dimensions and necessary steps towards its validation are briefly discussed.