Th. doct.-ing.--Génie électrique--Toulouse--I.N.P., 1983. N°: 240.

Th. doct.-ing.--Génie électrique--Nancy--I.N.P.L., 1980.

Nous abordons dans cette thèse deux types de problèmes d’ordonnancement sur une machine ou sur des machines parallèles : 1. les problèmes d’ordonnancement de type juste-à-temps : il s’agit de déterminer un ordonnancement de sorte que les travaux se terminent le plus près possible de leur date de fin souhaitée. On considère le cas où la date de fin souhaitée commune est connue et le cas où elle est à déterminer. De nouveaux algorithmes exacts sont proposés. Des schémas d’approximation sont élaborés. 2. les problèmes d’ordonnancements de travaux interférants : il s’agit de déterminer un ordonnancement qui permet d’optimiser un critère pour tous les travaux, sachant que la solution trouvée doit permettre également l’optimisation d’un autre critère défini uniquement sur un sous-ensemble des travaux. Il s’agit ici d’un nouveau problème d’ordonnancement multicritère, différent de la notion classique. Les approches considérées pour trouver une solution non dominée sont l’approche e-contrainte, la combinaison linéaire de critères et le goal programming. De nouveaux résultats de complexité sont montrés et des algorithmes exacts sont développés. / In this thesis, we consider two kinds of scheduling problems on a single machine or on parallel machines : 1. just-in-time scheduling problems : it aims to determine a schedule so that a job completes as close as possible to its due date.We consider the case where the common due date is known and the case where the common due date has to be fixed. New exact algorithms based on greedy algorithms and dynamic programming are proposed. Approximation schemes are given. 2. scheduling problems with interfering jobs : the aim is to determine a schedule that optimizes a criterion for the whole set of jobs and so that the solution optimizes another objective only for a subset of jobs. It is here a new multi-criteria scheduling problem, different from the classical notion. The approaches considered for finding a non-dominated solution are the e-constraint approach, the linear combination of criteria and the goal programming approach. New complexity results are proposed and exact algorithms are developed.

Dans la présente étude, nous nous sommes efforcés de resituer notre contribution dans un contexte industriel plus large. Abordant tour à tour les problèmes d'identification des systèmes, de la théorie moderne du contrôle et de la commande, nous avons tenté de faire une comparaison des diverses méthodes permettant d'apporter la connaissance du phénomène physique nécessaire pour atteindre avec plus de précision possible les buts que l'on se propose. Cependant, compte tenu des moyens à notre disposition, nous nous sommes cantonnés à des méthodes simples; la voie reste par conséquent ouverte pour des développements plus complexes qui permettraient une approche plus précise et plus systématique. Nous nous sommes placés délibérément dans une optique entièrement déterministe, ce qui conduit à des hypothèses assez peu raffinées. L'analyse dynamique du générateur de vapeur a été faite à partir d'approximations et de linéarisations. Le modèle ainsi obtenu par les équations d'évolution a été simulé sur calculateur analogique; et les résultats comparés à l'expérience. Les équations différentielles d'évolution ont été transformées en équation vectorielle. Après avoir fait le tour des diverses techniques d'optimalisation à notre disposition, nous avons choisi la plus appropriée à notre problème particulier, cette méthode finalement retenue est basée sur la programmation dynamique, grâce à laquelle nous avons pu déterminer quelle devait être la commande optimale à appliquer pour minimiser un critère économique du type: valeur absolue moyenne des variations de pression. Nous nous étendrons plus longuement sur ce choix, pour montrer l'aspect intéressant l'accroissement du rendement global de l'installation.

The efficiency of a 4 inch hydrocyclone was studied using a suspension of fine silica at pulp. densities of up to 507o solids by weight. The variables manipulated were diameter of the vortex finder, flowrate, percent solids in the feed and cyclone length. Slurry temperature was also measured. All measurements were made with the spigot adjusted to give the same degree of roping at the underflow. The product size distributions were analysed using an electronic particle sizing instrument similar to the "Coulter Counter" but interfaced with a digital computer. An advantage of this method is the continuous curves produced (154 data points per run). Equations were developed by means of stepwise regression analyses to predict (I) separating size, (2) inlet pressure, (3) bypass ratio, (4) sharpness of classification, alpha, (5) zero classification size and (6) water recovery in the underflow, as a function of hydrocyclone variables including temperature of the feed slurry. Two forms of a roping constraint equation were obtained. One predicts the underflow percent solids at which roping occurs, whilst the other gives the spigot size required to avoid roping. The separating size was found to depend on several factors including the temperature of the feed slurry. Moreover, the sharpness of classification, alpha, was identified to be a variable parameter. This parameter was found to be dependent on variables such as vortex finder diameter and the volume recovery of slurry to the overflow. The variability of alpha has not been widely recognized. Acceptance of this variability concept offers hope that classification efficiencies may be improved by choosing the proper combination of variables in an operating environment. Interestingly, the zero classification size was judged to be a constant at the 0.05 level of significance. The bypass ratio depended primarily on the water split, as expected, but was also influenced by the cyclone feed percent solids. It is suggested that this particular study advances the understanding of sub-sieve sized particle behaviour in hydrocyclones fed with slurries containing a high proportion of solids. / Applied Science, Faculty of / Mining Engineering, Keevil Institute of / Graduate

Rotary mills are used extensively to reduce the coarse rocks of mined ores down to a product fine enough to facilitate the extraction of the valuable minerals. On the South African gold mines alone about 110 million tons of ore is mined annually, at a power cost of about R90 million, and a liner material cost of about R30 million. The charge motion in mills has been studied both theoretically and experimentally since the beginning of the century. However, a lot of the work has been purely empirical, and some of it illconceived, resulting in there still being a poor understanding of the topic. The influence of lifter-bars upon charge motion has only been considered in the past few decades, and mostly to a very limited extent. This state of affairs has left the field wide open to further research. The charge motion of an isolated rod or ball, and how it is influenced by a flat-faced lifter-bar of any face-angle and of any height, was modelled theoretically. The charge motion of rods in a glass-ended mill was filmed with a high-speed camera. The mill was fitted with a variety of lifter-bars with different face-angles and heights, and was run at a wide range of speeds. The filmed trajectories of the rods were then tracked. The coefficients of friction between the rods and lifter-bar material were measured under vibrating conditions as are found in the mill. The power draw of the mill was also measured at a wide range of mill speeds, and with lifter-bars of different heights in the mill. A good correlation between the theoretical predictions and experimental results was found over a wide range of conditions. The impact point, at which the grinding element strikes the mill shell, was considered to be of primary importance in the analysis. It was found that the height of the lifter-bar has a strong influence on charge trajectories, as the height increases from zero up to just greater than one charge radius. Thereafter the lift increases until a critical lifter-bar height is reached, beyond which the grinding element is projected off the lifter prior to reaching the tip, the height of the impact point increasing slightly and then decreasing to a constant height. However, the change in the height of the impact point is very small, so in practical terms an increase in lifter-bar height, once it is higher than the radius of the grinding element, has a very small effect upon the charge trajectories. An increase in the angle of the lifter-bars was found to have a strong influence upon the height of the charge trajectories. A linear relationship between the mill speed and the impact angle was dicovered, and changes in the mill speed strongly influence the charge trajectories. It was also found that the power draw of a mill depends on the liner configuration, with power draw decreasing as the lifter-bar height is increased. Some new and surprising effects of lifter-bar geometry upon the charge trajectories, which are of great practical importance, have been discovered. The theoretical model is an advance upon all previous models, and agrees well with experimental results over a wide range of conditions.

No description available.

Thèse (M.Sc.)--Université Laval, 2004. / Bibliogr.: f. [161]-165. Publié aussi en version électronique.

Thesis (M. Phil.)--University of Hong Kong, 2006. / Title proper from title frame. Also available in printed format.

Thèse de doctorat : Génie électrique : Nancy 1 : 2009. / Titre provenant de l'écran-titre.