Planification de la maintenance d'un parc de turbines-alternateurs par programmation mathématique
Bibliographic record
Abstract
RESUME Une population grandissante de groupes turbines-alternateurs dans les centrales hydroelectriques arrive a sa fin de vie utile et les gestionnaires apprehendent la concurrence des arrets pour des renovations majeures avec un nombre tel que les ressources disponibles dans une meme periode ne seraient pas suffisantes. Ces retraits du reseau peuvent durer jusqu'a une annee entiere et mobilisent des ressources importantes en plus de la perte de production electrique. Les previsions a la hausse des ventes a l'export et des rythmes de production severes font craindre la mise a l'arret de beaucoup de groupes en meme temps. Actuellement, le jugement des experts est au coeur des decisions des retraits qui se basent essentiellement sur des inspections periodiques et des mesures effectuees in-situ et dont les resultats sont centralises chez l'equipe de planification des retraits. La nature aleatoire des phenomenes de degradations qui ont lieu, font en sorte que la capacite de prevision de l'usure par l'inspection a un caractere de court-terme. Une vision des renovations majeures sur le long terme est activement recherchee par les gestionnaires dans un souci de justification et de rationalisation des ressources budgetaires allouees aux renovations. Les gestionnaires disposent d’une quantite impressionnante de donnees. Parmi elles, figurent la production horaire de chaque groupe depuis plusieurs annees, l'historique des reparations sur chaque organe ainsi que les retraits majeurs effectues depuis les annees 1950. Dans ce projet de recherche, nous nous proposons de resoudre le probleme de planification de la maintenance d’un parc de 90 groupes turbines-alternateurs du reseau de production d'Hydro- Quebec sur un horizon de 50 ans. Nous developpons une methodologie scientifique et rationnelle de preparation des plans des retraits qui serviront de support a la prise de decision en exploitant les donnees de mesures et les historiques disponibles tout en respectant un ensemble de contraintes techniques et economiques. Pour respecter la confidentialite de certaines donnees, toutes les denominations originales ont ete modifiees pour les rendre anonymes. Ce probleme de planification est traite comme un probleme d’optimisation avec contraintes. D'abord, un groupe est etudie pour ressortir les organes les plus influents. Un modele de taux de defaillance est developpe pour prendre en compte les caracteristiques technologiques de l'organe et d'utilisation du groupe. Ensuite, des strategies de remplacements et de reparations sont----------ABSTRACT A growing number of Hydro-Quebec's hydro generators are at the end of their useful life and maintenance managers fear to face a number of overhauls exceeding what can be handled. Maintenance crews and budgets are limited and these withdrawals may take up to a full year and mobilize significant resources in addition to the loss of electricity production. In addition, increased export sales forecasts and severe production patterns are expected to speed up wear that can lead to halting many units at the same time. Currently, expert judgment is at the heart of withdrawals which rely primarily on periodic inspections and in-situ measurements and the results are sent to the maintenance planning team who coordinate all the withdrawals decisions. The degradations phenomena taking place is random in nature and the prediction capability of wear using only inspections is limited to shortterm at best. A long term planning of major overhauls is sought by managers for the sake of justifying and rationalizing budgets and resources. The maintenance managers are able to provide a huge amount of data. Among them, is the hourly production of each unit for several years, the repairs history on each part of a unit as well as major withdrawals since the 1950's. In this research, we tackle the problem of long term maintenance planning for a fleet of 90 hydro generators at Hydro-Quebec over a 50 years planning horizon period. We lay a scientific and rational framework to support withdrawals decisions by using part of the available data and maintenance history while fulfilling a set of technical and economic constraints. We propose a planning approach based on a constrained optimization framework. We begin by decomposing and sorting hydro generator components to highlight the most influential parts. A failure rate model is developed to take into account the technical characteristics and unit utilization. Then, replacement and repair policies are evaluated for each of the components then strategies are derived for the whole unit. Traditional univariate policies such as the age replacement policy and the minimal repair policy are calculated. These policies are extended to build alternative bivariate maintenance policy as well as a repair strategy where the state of a component after a repair is rejuvenated by a constant coefficient.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".