Transportation is one of the most vital services in modern society. It makes most of the other functions of society possible. Real transportation systems are so large and complex that in order to build the science of transportation systems it will be necessary to work in many areas, such as: Modeling, Optimization and Simulation. We are interested in solutions for the so-called fleet-sizing-and-allocation problem (FSAP). Fleet sizing and allocation problems are one of the most interesting and hard to solve logistic problems. A fleet sizing and allocation problem consists of two interdependent parts. The fleet sizing problem is to determine a number of transportation units that optimally balances service requirements against the cost of purchasing and maintaining the transportation units. The allocation problem is dealing with the repositioning of transportation units to serve future transportation demand. To make the fleet sizing and allocation problem a little bit more tractable we concentrate on logistic systems with a special hub-and-spoke structure. We start with a very simple fleet sizing of one-to-one case. This case will cause us to focus attention on several key issues in fleet sizing. Afterwards, the generalization of the one-to-one system is the one-to-many system. As a simple example can serve the continuous time situation where a single origin delivers items to many destinations. For the case that items are produced in a deterministic production cycle and transportation times are stochastic. We also studied a hub-and-spoke problem with continuous time and stochastic demand. To solve this problem, based on Marginal Analysis, we applied queueing theory methods. The investigation of the fleet-sizing-and-allocation problem for hub-and-spoke systems is started for a single-period, deterministic-demand model. In that the model hub has to decide how to use a given number of TU’s to satisfy a known (deterministic) demand in the spokes. We consider two cases: 1. Renting of additional TU’s from outside the system is not possible, 2. Renting of additional TU’s from outside the system is possible. For each case, based on Marginal Analysis, we developed a simple algorithm, which gives us the cost-minimal allocation. Since the multi-period, deterministic demand problem is NP-hard we suggest to use Genetic Algorithms. Some building elements for these are described. For the most general situation we also suggest to use simulation optimization. To realize the simulation optimization approach we could use the software tool “Calculation Assessment Optimization System” (CAOS). The idea of CAOS is to provide a software system, which separates the optimization process from the optimization problem. To solve an optimization problem the user of CAOS has to build up a model of the system to which the problem is related. Furthermore he has to define the decision parameters and their domain. Finally, we used CAOS for two classes of hub-and-spoke system: 1. A single hub with four spokes, 2. A single hub with fifty spokes. We applied four optimizers – a Genetic Algorithm, Tabu Search, Hybrid Parallel and Hybrid Serial with two distributions (Normal Distribution and Exponential Distribution) for a customer interarrival times and their demand.

For cost and reliability efficiency, optimal design and operation of pressurized water distribution networks is highly important. However, optimizing such networks is still a challenge since it requires an appropriate determination of: (1) dimension of pipe / pump / tank - decision variables (2) cost / network reliability - objective functions and (3) limits or restrictions within which the network must operate - a given set of constraints. The costs mentioned here consist in general of capital, construction, and operation costs. The reliability of a network mainly refers to the intrinsic capability of providing water with adequate volume and a certain pressure to consumers under normal and extreme conditions. These contradicting objective functions are functions of network configuration regarding component sizes and network layout. Because considerable uncertainties finally render the overall task to a highly complex problem, most recent approaches mainly focus only on finding a trade-off between minimizing cost and maximizing network reliability. To overcome these limitations, a novel model system that simultaneously considers network configuration, its operation and the relevant uncertainties is proposed in this study. For solving this multi-objective design problem, a simulation-based optimization approach has been developed and applied. The approach couples a hydraulic model (Epanet) with the covariance matrix adaptation evolution strategy (CMA-ES) and can be operated in two different modes. These modes are (1) simulation–based Single-objective optimization and (2) simulation-based multi-objective optimization. Single-objective optimization yields the single best solution with respect to cost or network reliability, whereas multi-objective optimization produces a set of non-dominated solutions called Pareto optimal solutions which are trade-offs between cost and reliability. In addition, to prevent a seriously under-designed network, demand uncertainties was also taken into account through a so called “robustness probability” of the network. This consideration may become useful for a more reliable water distribution network. In order to verify the performance of the proposed approach, it was systematically tested on a number of different benchmark water distribution networks ranging from simple to complex. These benchmark networks are either gravity-fed or pumped networks which need to be optimally designed to supply urban or irrigation water demand under specific constraints. The results show that the new approach is able: • to solve optimization problems of pressurized water distribution network design and operation regarding cost and network reliability; • to directly determine the pumping discharge and head, thus allowing to select pumps more adequately; • to simulate time series of tank water level; • to eliminate redundant pipes and pumps to generate an optimal network layout; • to respond well to complex networks other than only to simple networks; • to perform with multiple demand loading; • to produce reliable Pareto optimal solutions regarding multi-objective optimization. In conclusion, the new technique can be successfully applied for optimization problems in pressurized water distribution network design and operation. The new approach has been demonstrated to be a powerful tool for optimal network design not only for irrigation but also for an urban water supply.

A robust simulation-based optimization approach is proposed for truck-shovel systems in surface coal mines to maximize the expected value of revenue obtained from loading customer trains. To this end, a large surface coal mine in North America is considered as case study. A data-driven modeling framework is developed and then applied to automatically generate a highly detailed simulation model of the mine in Arena. The framework comprises a formal information model based on Unified Modeling Language (UML), which is used to input mine structural as well as production information. Petri net-based model generation procedures are applied to automatically generate the simulation model based on the whole set of simulation inputs. Then, factors encountered in material handling operations that may affect the robustness of revenue are then classified into 1) controllable; and 2) uncontrollable categories. While controllable factors are trucks locked to routes, uncontrollable factors are inverses of summation over truck haul, and shovel loading and truck-dumping times for each route. Historical production data of the mine contained in a data warehouse is used to derive probability distributions for the uncontrollable factors. The data warehouse is implemented in Microsoft SQL, and contains snapshots of historical equipment statuses and production outputs taken at regular intervals in each shift of the mine. Response Surface Methodology is applied to derive an expression for the variance of revenue as a function of controllable and uncontrollable factors. More specifically, 1) first order and second order effects for controllable factors, 2) first order effects for uncontrollable factors, and 3) two factor interactions for controllable and uncontrollable factors are considered. Latin Hypercube Sampling method is applied for setting controllable factors and the means of uncontrollable factors. Also, Common Random Numbers method is applied to generate the sequence of pseudo-random numbers for uncontrollable factors in simulation experiments for variance reduction between different design points of the metamodel. The variance of the metamodel is validated using leave-one-out cross validation. It is later applied as an additional constraint to the mathematical formulation to maximize revenue in the simulation model using OptQuest. The decision variables in this formulation are truck locks only. Revenue is a function of the actual quality of coal delivered to each customer and their corresponding quality specifications for premiums and penalties. OptQuest is an optimization add-on for Arena that uses Tabu search and Scatter search algorithms to arrive at the optimal solution. The upper bound on the variance as a constraint is varied to obtain different sets of expected value as well as variance of optimal revenue. After comparison with results using OptQuest with random sampling and without variance expression of metamodel, it has been shown that the proposed approach can be applied to obtain the decision variable set that not only results in a higher expected value but also a narrower confidence interval for optimum revenue. According to the best of our knowledge, there are two major contributions from this research: 1) It is theoretically demonstrated using 2-point and orthonormal k-point response surfaces that Common Random Numbers reduces the error in estimation of variance of metamodel of simulation model. 2) A data-driven modeling and simulation framework has been proposed for automatically generating discrete-event simulation model of large surface coal mines to reduce modeling time, expenditure, as well as human errors associated with manual development.

Health scientists often rely on simulation models to reconstruct groundwater contaminant exposure data for retrospective epidemiologic studies. Due to the nature of historical reconstruction process, there are inevitably uncertainties associated with the input data and, therefore, with the final results of the simulation models, potentially adversely impacting related epidemiologic investigations. This study examines the uncertainties associated with the historically reconstructed contaminant fate and transport simulations for an epidemiologic study conducted at U.S. Marine Corps Base Camp Lejeune, North Carolina. To achieve an efficient uncertainty analysis, sensitivity analysis was first conducted to identify the critical uncertain variables, which were then adopted in the uncertainty analysis using an improved Monte Carlo simulation (MCS) method. Particularly, uncertainties associated with the historical contaminant arrival time were evaluated. To quantify the uncertainties in an efficient manner, a procedure identified as Pumping Schedule Optimization System (PSOpS) was developed to obtain the extreme (i.e., earliest and latest) contaminant arrival times caused by pumping schedule variations. Two improved nonlinear programming methods Rank-and-Assign (RAA) and Improved Gradient (IG) are used in PSOpS to provide computational efficiency. Furthermore, a quantitative procedure named Pareto Dominance based Critical Realization Identification (PDCRI) was developed to screen out critical realizations for contaminant transport in subsurface system, so that the extreme contaminant arrival times under multi-parameter uncertainties could be evaluated efficiently.

Tooling design is crucial for the production of cost-effective and durable composite products. As part of the current search for cost reduction (by reducing capital investment, energy use and cycle time), integrally-heated tooling is one of the technologies available for ‘out-of-autoclave’ processing of advanced thermoset polymer composites. Despite their advantages, integrally-heated tools can suffer from uneven distribution of temperature, variability in heat flow rate and inconsistency in heating/cooling time. This research, therefore, investigates a number of design variables such as shape and layout of heating channels in order to improve the heating performance of an integrally-heated tool. Design of Experiments (DoE) has been carried out using Taguchi’s Orthogonal Array (OA) method to set several combinations of design parameters. Each of these design combinations has been evaluated through numerical simulation to investigate heating time and mould surface temperature variation. The simulation results suggest that the layout of the channels and their separation play a vital role in the heating performance. Signal-to-Noise (S/N) ratio and analysis of variance (ANOVA) have been applied to the results obtained to identify the optimal design combination of the integrally-heated tool. Statistical analysis reveals that the heating performance of an integrally-heated tool can be significantly improved when the channels’ layout is parallel. The shape of the channels has negligible effect and the distance between the channels should be determined based on the production requirement. According to the predicted optimal design, a developed integrally water-heated tool is manufactured. The actual thermal properties of the constituent materials of the produced tool are also measured. Then a numerical model of the experimental tool model is simulated in ANSYS software, with setting the actual material properties and boundary condition to define the temperature uniformity and heating rate of the experimental tool. Comparison of the experimental and numerical results of the experimental tool confirmed the well assigning of the boundary conditions and material properties during simulation the heated tool. The experimental results also confirmed the predicted optimal design of the integrally heated tool. Finally, in order to define its thermomechanical behaviour under the effective (in service) thermal loads, a tool model is simulated. Numerical results presented that the produced extremes of thermal deformation, elastic strain, normal and plane shear stresses, under the effective thermal loading, are within the allowable elastic limits of the participated materials.

For cost and reliability efficiency, optimal design and operation of pressurized water distribution networks is highly important. However, optimizing such networks is still a challenge since it requires an appropriate determination of: (1) dimension of pipe / pump / tank - decision variables (2) cost / network reliability - objective functions and (3) limits or restrictions within which the network must operate - a given set of constraints. The costs mentioned here consist in general of capital, construction, and operation costs. The reliability of a network mainly refers to the intrinsic capability of providing water with adequate volume and a certain pressure to consumers under normal and extreme conditions. These contradicting objective functions are functions of network configuration regarding component sizes and network layout. Because considerable uncertainties finally render the overall task to a highly complex problem, most recent approaches mainly focus only on finding a trade-off between minimizing cost and maximizing network reliability. To overcome these limitations, a novel model system that simultaneously considers network configuration, its operation and the relevant uncertainties is proposed in this study. For solving this multi-objective design problem, a simulation-based optimization approach has been developed and applied. The approach couples a hydraulic model (Epanet) with the covariance matrix adaptation evolution strategy (CMA-ES) and can be operated in two different modes. These modes are (1) simulation–based Single-objective optimization and (2) simulation-based multi-objective optimization. Single-objective optimization yields the single best solution with respect to cost or network reliability, whereas multi-objective optimization produces a set of non-dominated solutions called Pareto optimal solutions which are trade-offs between cost and reliability. In addition, to prevent a seriously under-designed network, demand uncertainties was also taken into account through a so called “robustness probability” of the network. This consideration may become useful for a more reliable water distribution network. In order to verify the performance of the proposed approach, it was systematically tested on a number of different benchmark water distribution networks ranging from simple to complex. These benchmark networks are either gravity-fed or pumped networks which need to be optimally designed to supply urban or irrigation water demand under specific constraints. The results show that the new approach is able: • to solve optimization problems of pressurized water distribution network design and operation regarding cost and network reliability; • to directly determine the pumping discharge and head, thus allowing to select pumps more adequately; • to simulate time series of tank water level; • to eliminate redundant pipes and pumps to generate an optimal network layout; • to respond well to complex networks other than only to simple networks; • to perform with multiple demand loading; • to produce reliable Pareto optimal solutions regarding multi-objective optimization. In conclusion, the new technique can be successfully applied for optimization problems in pressurized water distribution network design and operation. The new approach has been demonstrated to be a powerful tool for optimal network design not only for irrigation but also for an urban water supply.

No description available.

In this thesis, we present and analyze three algorithms that are designed to make computer simulation more efficient, valid, and/or applicable. The first algorithm uses simulation cloning to enhance efficiency in transient simulation. Traditional simulation cloning is a technique that shares some parts of the simulation results when simulating different scenarios. We apply this idea to transient simulation, where multiple replications are required to achieve statistical validity. Computational savings are achieved by sharing some parts of the simulation results among several replications. We improve the algorithm by inducing negative correlation to compensate for the (undesirable) positive correlation introduced by sharing some parts of the simulation. Then we identify how many replications should share the same data, and provide numerical results to analyze the performance of our approach. The second algorithm chooses a set of best systems when there are multiple candidate systems and multiple objectives. We provide three different formulations of correct selection of the Pareto optimal set, where a system is Pareto optimal if it is not inferior in all objectives compared to other competing systems. Then we present our Pareto selection algorithm and prove its validity for all three formulations. Finally, we provide numerical results aimed at understanding how well our algorithm performs in various settings. Finally, we discuss the estimation of input distributions when theoretical distributions do not provide a good fit to existing data. Our approach is to use a quasi-empirical distribution, which is a mixture of an empirical distribution and a distribution for the right tail. We describe an existing approach that involves an exponential tail distribution, and adapt the approach to incorporate a Pareto tail distribution and to use a different cutoff point between the empirical and tail distributions. Then, to measure the impact, we simulate a stable M/G/1 queue with a known inter-arrival and unknown service time distributions, and estimate the mean and tail probabilities of the waiting time in queue using the different approaches. The results suggest that if we know that the system is stable, and suspect that the tail of the service time distribution is not exponential, then a quasi-empirical distribution with a Pareto tail works well, but with a lower bound imposed on the tail index.

Malgré la prédominance actuelle du mode routier, pour le transport de marchandises en milieu urbain, une alternative durable est nécessaire, au vu des enjeux environnementaux et sociétaux. Dans cette thèse, nous proposons l’étude d’une des perspectives possibles, pour absorber une partie de ce flux de marchandises toujours plus dense, en utilisant le réseau ferroviaire urbain, initialement dédié aux voyageurs. Une méthodologie intégrant le fret dans ce dernier est proposée, avec comme première étape, l'identification et la classification de tous les niveaux de mixité fret / voyageurs possibles. Le niveau le plus contraint est retenu, car sa faisabilité induira celle des autres. Notre seconde contribution est relative à une approche par décomposition du problème d’insertion du flux de fret en plusieurs sous-problèmes interdépendants, selon les trois horizons temporels (long, moyen et court). Dans le but d’évaluer la capacité du système global, à absorber un flux supplémentaire de nature différente, le problème de détermination du meilleur plan de transport des marchandises est identifié comme central et critique. La troisième contribution concerne la simulation du système de transport, puis sa formalisation par un PL en variables mixtes, pour affecter chaque commande à un train, en déterminant le moment auquel elle sera chargée et en minimisant les temps d’attente cumulés des commandes. Plusieurs variantes de colonies de fourmis sont développées, pour la résolution d’instances de grande taille. La quatrième contribution concerne le couplage du modèle de simulation, qui permet l’évaluation des performances de cette nouvelle solution de transport, avec les différents algorithmes optimisant le plan de transport. Enfin, nous proposons une approche de replanification par horizon glissant, pour absorber les perturbations de la demande, en minimisant les changements du plan de transport / Urban freight transport is almost exclusively carried out by truck. Beyond the drawbacks caused in the city, this transport mode is nearly saturated. This study discusses an alternative way of transporting freight by using urban rail infrastructure. The first contribution deals with the identification and classification of all different sharing possibilities of mixing freight with passenger’s traffic using rail network. The second contribution is the definition of global freight/passenger transport problem, which is decomposed into several optimization interdependent sub-problems with different temporal decision horizon. In order to show the capacity of the global system to absorb an additional flow with different nature, the Freight Rail Transport Schedule Problem “FRTSP” is identified as the bottleneck of transportation system and is formalized with MIP model. As third contribution, this problem determines train and loading time for each demand to be assigned respecting several constraints while minimizing total waiting time. The fourth contribution deals with a discrete event simulation approach, which studies this alternative and validates several proposed decision algorithms. Finally, the fifth contribution consists in a dynamic approach based on a rolling horizon, which is proposed in order to update the initial plan. The updated plan allows to determine a new assignment regarding new demand such as the modifications from the previous plan are minimized

O estudo de sistemas logísticos envolve a concatenação de elementos estratégicos e operacionais, comumente compondo sistemas com múltiplas facetas, objetivos antagônicos e grande número de alternativas. Nesse contexto, o presente trabalho discute a utilização de análise de decisão multicritério (MCDA), simulação de eventos discretos (SED) e otimização para simulação. A metodologia MCDA captura, mensura e pondera os objetivos e valores dos tomadores de decisão. Por sua vez, a SED representa o sistema estudado com alto nível de detalhamento, permitindo a avaliação de diversas configurações do sistema. Por fim, métodos de otimização para simulação possibilitam a busca e comparação de alternativas mais eficientes. As três metodologias são avaliadas, identificando suas vantagens, desvantagens e complementaridades quando aplicadas a sistemas logísticos. Através da aplicação de um estudo de caso sobre o dimensionamento de um sistema de transporte, constatou-se que: a) a SED incorporou detalhes importantes para a avaliação mais precisa de vários indicadores de desempenho b) a metodologia MCDA possibilitou a captura de vários objetivos e valores, propiciando a realização de tradeoffs robustos; c) um método de busca exaustiva e técnicas de redução de variância permitiram a comparação das alternativas em tempos computacionais reduzidos. Por fim, conclui-se que a metodologia híbrida apresentada expande o potencial de aplicação da SED em sistemas logísticos complexos. / A logistic system study involves strategic and operational elements, commonly composing multi-faceted systems with antagonistic goals and large number of alternatives. In this context, this thesis discusses the use of multi-criteria decision analysis (MCDA), discrete event simulation (DES) and optimization for simulation. The MCDA methodology captures, measures and weighs the goals and values of decision makers. DES is useful for representing systems with high level of detail, allowing the evaluation of several system configurations. Finally, optimization for simulation procedures are useful for searching and comparing more efficient alternatives. These three methodologies are assessed and their advantages, disadvantages, and complementarities are identified for logistics systems applications. Through a case study of a transportation system, we conclude that: a) the SED incorporated important details for more precise evaluation of various performance indicators b) the MCDA methodology was useful to capture several goals and values, so that robust tradeoffs could be carried out c) an exhaustive search routine and variance reduction techniques allowed the comparison of several alternatives in feasible computational times. Finally, we conclude that the presented hybrid methodology expands the application of DES to complex logistics systems.