L’optimisation multi-objectif concerne la résolution de problèmes pour lesquels plusieurs objectifs (ou critères) contradictoires sont pris en compte. Contrairement aux problèmes d’optimisation ayant un seul objectif, un problème multi-objectif ne possède pas une valeur optimale unique mais plutôt un ensemble de points appelés “ensemble non dominé”. Les bornes inférieures et supérieures d’un problème multi-objectif peuvent être également décrites par des ensembles. Dans la pratique, les variables utilisées en optimisation multi-objectif représentent souvent des objets non fractionnables et on parle alors de problèmes multi-objectif en nombres entiers. Afin d’obtenir de meilleures bornes qui peuvent être utilisées dans la conception de méthodes exactes, certains problèmes sont formulés avec un nombre exponentiel de variables de décision et ces problèmes sont résolus par la méthode de génération de colonnes. Les travaux de cette thèse visent à contribuer à l’étude de l’utilisation de la génération de colonnes en programmation linéaires en nombres entiers multi-objectif. Pour cela nous étudions un problème de tournées de véhicules bi-objectif qui peut être considéré comme une généralisation de plusieurs autres problèmes de tournées de véhicules. Nous proposons des formulations mathématiques pour ce problème et des techniques pour accélérer le calcul des bornes inférieures par génération de colonnes. Les sous-problèmes qui doivent être résolus pour le calcul des bornes inférieures ont une structure similaire. Nous exploitons cette caractéristique pour traiter simultanément certains sous-problèmes plutôt qu’indépendamment / Multi-objective optimization deals with finding solutions to problems for which several objectives (or criteria) are considered. Unlike in single objective optimization, the optimal value of a multi-objective problem is a set of points called “the non dominated set”. Lowerand upper bounds of a multi-objective problem can also be described using sets. For most practical problems, the variables considered in multi-objective optimization represent non fractionable items and thus we talk of multi-objective integer programs. In order to obtain good lower and upper bounds that can be used in the design of exact methods, some problems are usually formulated with an exponential number of decision variables and these problems are solved by column generation. The work of this thesis seeks to contribute to the study of the use of column generation in multi-objective integer linear programming. We do this by studying a bi-objective vehicle routing problem which may be seen as a generalization of several other vehicle routing problems. We propose mathematical formulations for this problem and also find ways to quickly compute lower bounds by column generation. Since the subproblems solved when computing lower bounds have similar structures, we propose intelligent ways of treating some of these subproblems simultaneously rather than independently

In the aftermath of the liberalization of European Energy Markets in the 2000s, Power Exchange platforms have constantly evolved towards more integrated and competitive designs, where quality forecasts and effective optimization strategies play decisive roles. This study presents the development of a hydropower scheduling optimization algorithm for the Day-Ahead spot market using Mixed Integer Linear Programming (MILP). This work was supported by the hydro asset management team of ENGIE Global Energy Markets (GEM) located in Brussels.  The model developed is focusing on the optimization of Coindre Hydraulic Power Plant (HPP), located in the highlands of Massif Central in France. With the combined water discharge of its two interconnected reservoirs, Grande-Rhue and Petite-Rhue, the powerhouse can reach up to 36 MW of power output capacity. The two reservoirs are located kilometres apart from each other and have different storage capacities and catchment areas. The reservoirs naturally exchange water due to the level difference along an interconnection pipe. Maximum power output is limited by water level differences in both reservoirs, which makes modelling complicated. These operational constraints are a limiting factor in terms of operability, as a result the scheduling process is a non-trivial task and is time-consuming.  A framing study of the power plant was conducted over a hydraulic year to identify the governing parameters of the model. The multi-reservoir nature of the optimization problem oriented the model development towards a Mixed Integer Linear Formulation. After experimenting with different solvers, Gurobi 28.1.0 was chosen for its performance in the Branch and Cut Algorithm for the power scheduling task.  The performance of the new model has been validated by re-running the model on past production plans, results show that reservoir volume errors are less than 5% of their respective capacities on a 5 days’ time-horizon. After backtesting it was found that the new optimization strategy results in higher revenue for the plant due to the optimized operation at higher average energy prices. The results also bring out the importance of proper valve actuation in the optimization strategy, as well as the need for future studies. / Till följd av liberaliseringen av de europeiska energimarknaderna under 2000-talet har energiföretagen och elbörserna ständigt utvecklats mot mer integrerade och konkurrenskraftiga lösningar, där kvalitetsprognoser och effektiva optimeringsstrategier spelar avgörande roller. Detta examensarbete presenterar utvecklingen av en algoritm för optimering av vattenkraftplaneringen på Day-Ahead elmarknaden med hjälp av en matematisk modell av typen Mixed Integer Linear Programming (MILP). Arbetet initierades av och utfördes hos ENGIE Global Energy Markets (GEM) i Bryssel.  Modellen som utvecklats är tänkt att optimera Coindre vattenkraftverk, som ligger på höglandet inom Massif Central i Frankrike. Med det kombinerade vattenutsläppet från dess två fördämningar, Grande-Rhue och Petite-Rhue, kan kraftverket leverera upp till 36 MW el netto till elnätet. Vattenreservoarerna ligger flertalet kilometer ifrån varandra och har mycket olika kapacitet och upptagningsområden. Båda reservoarerna är kopplade till varandra genom det gemensamma tilloppsröret till kraftverket, där en reglerventil finns endast vid Petite-Rhue. Vatten kan växlas naturligt mellan de två dammarna när ventilen är öppen på grund av skillnaden i varderas vattennivå. Den maximala effekten från kraftverket är begränsad av vattennivåerna i båda reservoarerna vilket gör optimeringsmodelleringen komplicerad. Dessa operationella begränsningar är mycket hindrande vad gäller valet av driftsregim, eftersom kalkylering av driftsplaneringen blir en svår och tidskrävande uppgift.  En ramstudie av vattenkraftverket genomfördes under ett typiskt hydrauliskt år för att identifiera modellens styrparametrar. Den möjliga vattenöverföringen mellan de två dammarna orienterade modellutvecklingen mot en Mixed Integer Linear Programming (MILP) formulering. Efter att ha experimenterat med olika kalkylverktyg valdes Gurobi 28.1.0 för sin bra prestation i lösningen av Branch and Cut-algoritmen.  Systemets hydraulik har validerats genom att injicera realiserade produktionsplaner som input till modellen. Resultaten visar att volymfelet är mindre än 5% av deras respektive kapacitet under en 5-dagars tidshorisont. Efter tvärstester mot historiska data konstaterades det att den nya optimeringsstrategin resulterar i bättre genomsnittliga elpriser på varje kWh inmatad till nätet och högre intäkter för kraftverket. Resultaten visar också på vikten av korrekt ventilmanövrering i optimeringsstrategin.  Modellen körs i rimliga beräkningstider och redan används i den dagliga optimeringen av Coindre kraftverket, vilket sparar mycket tid. Specifika exempel på den optimerade prestandan och framtida förbättringar hittas i slutet av denna rapport.

[pt] O Problema de Roteirização de Veículos (Vehicle Routing Problem - VRP) é um problema clássico combinatório bem conhecido. Este trabalho apresenta um novo fator no modelo de otimização matemática de otimização do VRP, considerando restrições que abordam a probabilidade de roubo de cargas nas regiões visitadas, além das restrições tradicionais, tais como o número de veículos, janelas de tempo, a capacidade do veículo e tempo de ciclo dos veículos. O modelo desenvolvido é testado em um estudo de caso real, considerando uma empresa de distribuição de produtos farmacêuticos do Rio de Janeiro. As soluções de rota com e sem risco de roubo de carga são comparadas. O modelo é resolvido usando o software AIMMS, para análises com instância pequenas, e resolvidas executando a Metaheurística Simulated Annealing, para o estudo de caso, onde se utiliza de duas instâncias. / [en] The Vehicle Routing Problem (VRP) is a classic well-known combinatorial problem. This paper introduces a new factor in the VRP mathematical optimization model, considering restrictions that address the probability of cargo theft in the regions visited, beyond the traditional constraints such as the number of vehicles, time windows, the capacity of the vehicle and the vehicle s cycle time. The paper proposes a mixed integer linear model that minimizes total transportation costs and cargo theft costs. The model is tested in a real-life case study, a company that distributes pharmaceutical products in Rio de Janeiro. The route solutions with and without cargo theft risk are compared. The model is solved using AIMMS software for analysis with small instance, and resolved by running the Simulated Annealing Metaheuristic, for the case study, which uses two instances.

A characteristic challenge of software development is the management of omnipresent change. Classically, this constant change is driven by customers changing their requirements. The wish to optimally leverage available resources opens another source of change: the software systems environment. Software is tailored to specific platforms (e.g., hardware architectures) resulting in many variants of the same software optimized for different environments. If the environment changes, a different variant is to be used, i.e., the system has to reconfigure to the variant optimized for the arisen situation. The automation of such adjustments is subject to the research community of self-adaptive systems. The basic principle is a control loop, as known from control theory. The system (and environment) is continuously monitored, the collected data is analyzed and decisions for or against a reconfiguration are computed and realized. Central problems in this field, which are addressed in this thesis, are the management of interdependencies between non-functional properties of the system, the handling of multiple criteria subject to decision making and the scalability. In this thesis, a novel approach to self-adaptive software--Multi-Quality Auto-Tuning (MQuAT)--is presented, which provides design and operation principles for software systems which automatically provide the best possible utility to the user while producing the least possible cost. For this purpose, a component model has been developed, enabling the software developer to design and implement self-optimizing software systems in a model-driven way. This component model allows for the specification of the structure as well as the behavior of the system and is capable of covering the runtime state of the system. The notion of quality contracts is utilized to cover the non-functional behavior and, especially, the dependencies between non-functional properties of the system. At runtime the component model covers the runtime state of the system. This runtime model is used in combination with the contracts to generate optimization problems in different formalisms (Integer Linear Programming (ILP), Pseudo-Boolean Optimization (PBO), Ant Colony Optimization (ACO) and Multi-Objective Integer Linear Programming (MOILP)). Standard solvers are applied to derive solutions to these problems, which represent reconfiguration decisions, if the identified configuration differs from the current. Each approach is empirically evaluated in terms of its scalability showing the feasibility of all approaches, except for ACO, the superiority of ILP over PBO and the limits of all approaches: 100 component types for ILP, 30 for PBO, 10 for ACO and 30 for 2-objective MOILP. In presence of more than two objective functions the MOILP approach is shown to be infeasible.

[pt] Este trabalho trata do problema real de avaliar a capacidade de processamento de pátios ferroviários planos de classificação. Nesses pátios, os vagões são recebidos em trens e movimentam respeitando a disposição dos trilhos e a formação sequencial do trem de saída. Movimentações ineficientes implicam em uma capacidade de processamento inferior à potencial do pátio dado seu layout. O objetivo desta pesquisa é descrever o problema e incitar um método capaz de calcular a capacidade de processamento de pátios ferroviários planos de classificação no horizonte estratégico, indicando se existe ou não a necessidade de um projeto de expansão para garantir atendimento à demanda prevista. O problema foi modelado através de programação linear inteira mista (MILP) baseado na teoria de sequenciamento de produção. O modelo foi aplicado em instâncias de teste, reproduzindo movimentações reais de vagões, e provou avaliar diferentes layouts adequadamente, porém com elevado tempo de execução. A inicialização de algumas variáveis binárias do modelo permitiu um incremento de tamanho nas instâncias, porém ainda inviável para aplicação na prática. / [en] This work deals with the real problem of evaluating the processing capacity of flat rail classification yards. In these yards, the railway cars are received on trains and move respecting the car sequence of the outgoing train. Inefficient movements imply a lower processing capacity than the yard s potential given its layout. The objective of this research is to describe the problem and to incite a method capable of calculating the processing capacity of flat rail classification yards in the strategic horizon, indicating whether or not there is a need for an expansion project to ensure meeting the expected demand. The problem was modeled using mixed integer linear programming (MILP) based on production scheduling theory. The model was applied to test instances, reproducing real railway car movements, and proved to evaluate different layouts properly, but with a high execution time. The initialization of some binary variables of the model allowed an increase in the size of the instances, however it is still unfeasible for practical application.

Swedish home care industry has been facing both external and internal problems, such as ageing population, varying quality and unsatisfactory continuity. Accordingly, workforce scheduling system, as one of the most common and useful software within home care planning nowadays, is in need of constant improvement and upgrading. This master’s thesis aimed to explore and analyze improvement potential of an established workforce scheduling system for an IT-company. The thesis was divided into two phases, of which a pre-study in Phase I tried to understand the planning process for planners and identify the perceived problems and shortcomings of the current system from a planner’s perspective. Based on the analysis from the pre-study, the caregiver continuity was chosen as the research area for Phase II. The current system was re-implemented and was modelled as an optimization problem. Furthermore, the system mainly consisted of two key parts, mixed integer linear programming (MILP) and heuristics. Different approaches in terms of modifications in both MILP and heuristics were applied to the re-implemented system. The performance of the modifications was measured by multiple evaluation indicators. The test results showed that there was a potential to improve caregiver continuity with 1.2% to almost 13% depending on the modification type. The modifications were lastly suggested for further examination regarding their practical appropriateness by applying them to the current running algorithm. / Den svenska hemtjänsten möter både yttre och inre problem såsom åldrande befolkning, varierande kvalitet och bristande kontinuitet. Schemaläggningssystemet som är en av de vanligaste och användbaraste programvarorna inom hemtjänsten behöver därmed en ständig förbättring och uppgradering som bemöter de existerande utmaningarna. Detta examensarbete hade som syfte att utforska och analysera förbättringspotentialen av ett etablerat schemaläggningssystem för ett ITföretag. Arbetet var indelat i två faser, varav förstudien i Fas I försökte förstå planerarnas planeringsprocesser och identifiera upplevda problem och brister i det nuvarande systemet utifrån ett planerares perspektiv. Baserat på analysen från förstudien, personalkontinuitet valdes som ett forskningsområde för Fas II. Nuvarande systemet implementerades om och det modellerades som ett optimeringsproblem. Systemet bestod huvudsakligen av två nyckeldelar, blandat heltalslinjärprogrammering (MILP) och heuristik. Olika metoder i form av modifieringar i både MILP och heuristik tillämpades på det omimplementerade systemet. Modifieringarnas prestanda mättes sedan med flera utvärderingsindikatorer. Testresultaten visade att, beroende på vilken modifiering det gäller, fanns det en potential att förbättra personalkontinuiteten med 1,2% till nästan 13%. Det föreslogs slutligen att modifieringarnas praktiska lämplighet behövs undersökas ytterligare genom att applicera det på det nuvarande systemet som är i drift.

Scheduling problem and shift design problems are well known NP-hard problems within the optimization area. Often time, the two problems are studied individually. In this thesis however, we are looking at the combination of both problems. More specifically, the aim of this thesis is to suggest an optimal scheduling policy given that there are no predefined shifts to begin with. The duration of a shift, along with the start and end time may vary. Thus we have proposed to split the problem into two sub-problems: weekly scheduling problem and daily scheduling problem. As there are no exact solution methods that are feasible, two meta-heuristics method has been employed to solve the sub-problems: Simulated Annealing (SA) and Genetic Algorithm (GA). We have provided proofs of concepts for both methods as well as explored the scalability. This is especially important as the number of employee is expected to grow significantly throughout the year. The results obtained has shown to be promising and can be built upon for further capabilities. / Schemaläggning och skiftdesignsproblem är välkända och välstuderade NP-svåra beslutsproblem inom optimeringsområdet. Oftast så studeras dessa problem enskilt, men i detta arbete så studeras en kombination av båda problemen. Mer specifikt är målet med detta arbete att föreslå ett förnuftigt handlingsätt till att skapa ett veckoschema där skift inte är predefinierade för alla veckor. Starttiden, sluttiden och varaktigheten av ett skift kan förändras från vecka till vecka. Därför har problemet delats upp till två delar: Veckoschemaläggnings- och dagsschemaläggningsproblem. Trots uppdelningen så är båda delproblem för komplexa för att lösas exakt. Därför har två metaheuristiska metoder använts som lösningsmetoder: Simulerad Glödgning och Genetisk Algoritm. I detta arbete bevisas båda lösningsmetoderna till att vara bra nog, och dessutom studeras även skalbarheten av modellen. Detta senare är särskilt viktigt eftersom antal anställda som ska schemaläggas förväntas att öka genomåren. De erhållna resultaten har visat sig vara lovande och bevisligen så kan modellen expanderas med er villkor

This master thesis in optimization and systems theory is a development of two different optimization models formulated to schedule multi-skilled agents for contact centers depending on the forecasted demand, assigned by Teleopti. Four mixed integer linear programming models are created with the optimization programming language GAMS and solved by the internet based solver NEOS. Two of the models are formulated to perform an optimal scheduling that matches a forecasted demand per skill and day and the remaining two models are formulated to perform an optimal scheduling that matches a forecasted demand per skill, day and half hour. The first two models are referred to as the Basic Models and the second two are referred to as the Complex Models. The Basic Models includes seven constraints and the Complex Model includes nine constraints, describing regulations at the contact center. The main goal of the project is to find an optimal solution that results in an as even distribution of under or over scheduling. The scheduling optimization covers a period of 28 days, starting on a Monday which results in four weeks. The optimization models are based on two sets of data, there are 104 assigned agents that possesses one, two or three of the skills Channel, Direct and Product. All agents are bound to work according to a contract specified through the constraints. In the Basic Model the forecasted demand is given in amount of hours per day and skill, the demand is non-cyclical. In the Complex model the forecasted demand is given in amount of half hours per day, skill and half hour. Each day is scheduled from 7 a.m. to 11 p.m. resulting in 32 available half hours. All optimization models are developed to correctly mathematically formulate the constraints specified by Teleopti. Any non-linear equation that arises are linearized to maintain linearity, this is favourable in the sense of computational time solving the models. The objective functions in this thesis are formulated to describe the main goal of even distribution as correctly as possible. The result for the Basic Model shows that an optimal solution is achieved after 34 seconds. This model contains 169,080 variables and 39,913 equations. In the Complex Models integer solutions are achieved, but no optimal solution is found in 8 hours of computational time. The larger Complex Model contains 9,385,984 variables and 1,052,253 equations and the smaller Complex Model contains 5,596,952 variables and 210,685 equations. Teleopti’s scheduler produces an integer solution matching the Complex Model in 4 minutes. / Detta examensarbete i optimering och systemteori är framtagningen av två olika optimeringsmodeller formulerade för att schemalägga multikompetenta agenter för kontaktcenters beroende av den förväntade efterfrågan, tilldelad av Teleopti. Fyra blandade heltals linjära programmeringsmodeller skapas med optimeringsprogrammeringsspråket GAMS och löses av den internetbaserade lösaren NEOS. Två av modellerna är formulerade för att utföra en optimal schemaläggning som matchar en prognostiserad efterfrågan per skicklighet och dag och de återstående två modellerna är formulerade för att utföra en optimal schemaläggning som matchar en prognostiserad efterfrågan per färdighet, dag och en halvtimme. De två första modellerna i detta arbete benämns de Grundläggande Modellerna och de resterande två benämns de Komplexa Modellerna. Grundmodellerna inkluderar sju bivillkor och de Komplexa modellerna innehåller nio bivillkor, vilka beskriver arbetsvillkoren på kontaktcentret. Projektets huvudmål är att hitta en optimal lösning som resulterar i en jämn fördelning av under- eller överschemaläggning. Den schemalagda optimeringen täcker en period av 28 dagar, vilken börjar på en måndag vilket resulterar i fyra veckor. Optimeringsmodellerna är baserade på två uppsättningar data, det finns 104 tillgängliga agenter vilka har en, två eller tre av kompetenserna Channel, Direct och Product. Alla agenter är bundna att arbeta enligt det kontrakt som specificeras genom bivillkoren. I grundmodellen anges den prognostiserade efterfrågan i timmar per dygn och kompetens, efterfrågan är icke-cyklisk. I den komplexa modellen anges den beräknade efterfrågan i mängd halvtimmar per dag, kompetens och halvtimme. Varje dag är schemalagd från kl. 07.00 till 23.00 vilket resulterar i 32 tillgängliga halvtimmar. Alla optimeringsmodeller är utvecklade för att matematiskt beskriva de begränsningar som Teleopti specificerar. Alla icke-linjära ekvationer som uppstår linjäriseras för att upprätthålla linjäritet, detta är gynnsamt i avseendet mängd tid beräkningen av modellerna tar. Målfunktionerna i detta arbete är formulerade för att beskriva huvudmålet för jämn distribution så korrekt som möjligt. Resultatet för grundmodellen visar att en optimal lösning uppnås efter 34 sekunder. Denna modell innehåller 169,080 variabler och 39,913 ekvationer. I de komplexa modellerna uppnås heltalslösningar, men ingen optimal lösning hittas på 8 timmars beräkningstid. Den större komplexa modellen innehåller 9,385,984 variabler och 1,052,253 ekvationer och den mindre komplexa modellen innehåller 5,596,952 variabler och 210,665 ekvationer. Teleoptis schemaläggare producerar en heltalslösning som matchar den komplexa modellen på 4 minuter.

Yes / This article proposes a novel technique for operation of distribution networks with considering active network management (ANM) schemes and demand response (DR) within a joint active and reactive distribution market environment. The objective of the proposed model is to maximize social welfare using market-based joint active and reactive optimal power flow. First, the intermittent behavior of renewable sources (solar irradiance, wind speed) and load demands is modeled through scenario-tree technique. Then, a network frame is recast using mixed-integer linear programming, which is solvable using efficient off-the-shelf branch-and cut solvers. Additionaly, this article explores the impact of wind and solar power penetration on the active and reactive distribution locational prices within the distribution market environment with integration of ANM schemes and DR. A realistic case study (16-bus UK generic medium voltage distribution system) is used to demonstrate the effectiveness of the proposed method. / This work was supported in part by the Ministry of Higher Education Scientific Research in Iraq and in part by British Academy under Grant GCRFNGR3\1541.

With increasing shares of variable renewable energy sources in the power mix, the need for energy storage solutions is projected to increase as well. Storage can in such combined systems help mitigate the issues with relying on intermittent sources by time-shifting the supply and smoothing out frequency fluctuations, to name some examples. This thesis has focused on Azelio ABs flagship product, the TES.POD, which is a long-duration thermal energy storage technology. When integrated with, for example, solar PV power, the TES.POD can store excess energy and dispatch it during times of low supply or when during the evening/night. The aim of the thesis has been the development of a day-ahead dispatch optimization tool for systems that include multiple TES.PODs, combined into a Cluster, and solar PV. The model was to be built using the Python programming language and based on Mixed-Integer-Linear-Programming (MILP) methods. The PV+storage system was then allowed to be connected to supplementary power sources such as a larger electric grid, or diesel generators in off-grid locations. The purpose of the optimization model is to find the most economic way to operate the individual TES.PODs while also keeping track of other system components, using a cost-based objective function (minimize costs). A focus has been on using high time resolution (small time step) in order to investigate the impact that the TES.PODs dynamic constraints has on operation. Another strength compared to pre-existing models was the ability to operate individual units indifferent to each other, as opposed to having them all operated in unison. Final results from benchmarking tests and two case studies indicated that using the optimization tool with smaller time steps had an effect on key indicators, and could lead to improved economy in the system. It was observed in both cases that the cost of electricity was reduced by running the optimization tool with time steps of either two or three minutes when compared with using an hourly resolution. Furthermore, several usage parameters for the TES.PODs, notably the total amount of operated hours and energy output per cycle, saw improvements which could lead to reduced cost of operation and maintenance. While not the main intent, testing different Cluster sizes and amount of installed PV capacity with the model, it could also be used in strategic decisions for system sizing. However, due to rapidly growing computational times in systems with large TES.POD clusters and using smaller time steps, the possibility of adding more complexity to the model in future work must be done with caution. To combat this issue, either improvements to the model formulation could be attempted, or by using more powerful hardware or optimizer (imported software algorithm that handles solving the model).