Thesis NURSE ROSTERING PROBLEM: UN CASO APLICADO A UN HOSPITAL EN NUEVA ZELANDA
Loading...
Date
2016
Authors
BARRAZA VARGAS, PATRICIO ALEJANDRO
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
En este trabajo se estudia el problema de asignación de turnos de enfermeras (NurseRostering Problem) en un hospital de Nueva Zelanda. En este hospital, se trabajan tresturnos diarios, todos los días del año. El horizonte de programación se definió en cuatrosemanas. Cada enfermera puede solicitar turnos libres, i.e., no trabajar en determinadosturnos del horizonte de programación. El problema es formulado como un problemalineal entero (ILP) multiobjetivo, donde se requiere minimizar en forma simultánea lainsatisfacción de las enfermeras producto del incumplimiento de sus solicitudes de turnoslibres y la cantidad de enfermeras cuyos requerimientos no se satisfacen completamente.El enfoque de solución está basado en el método de los pesos que pondera cada objetivopor un factor de ponderación (a). Para determinar la calidad de la formulación propuesta ycubrir un amplio rango de datos, se generaron diez instancias. En cada instancia se modificóel factor de ponderación completando un total de cincuenta problemas. La factibilidaddel problema depende del número de enfermeras disponibles que pueden trabajar enturnos de día (2 turnos) y noche (1 turno). Como se esperaba, a medida que se da mayorpeso a la insatisfacción de turnos, la insatisfacción promedio disminuye y el promediode enfermeras afectadas aumenta. Los problemas analizados permiten concluir que laformulación propuesta es tratable porque el tiempo promedio de uso de CPU fue 1.872segundos y el peor caso fue de 24.406 segundos. Para las instancias probadas, se observaque la función objetivo es creciente en a y que el mínimo de la función objetivo se alcanzaen a = 0.
In this work, we study the problem of assignment of nursing shifts (Nurse RosteringProblem) in a hospital in New Zealand. In this hospital, they work three shifts a day,every day of the year. The scheduling horizon was defined in four weeks. Each nursecan request shifts off, i.e., not work on certain shifts in the scheduling horizon. Theproblem is formulated as a multi-objective integer linear problem (ILP) where is requiredto simultaneously minimize the dissatisfaction of the nurses as result of the failure to fulfilthe requirements and the number of nurses whose requirements are not fully satisfied. Thesolution approach is based on weights method which weights each objective by a weightingfactor (a). To determine the quality of the proposed formulation and cover a wide range ofdata, ten instances were generated. In each instance, the weighting factor was changed fora total of fifty problems. The feasibility of the problem depends on the number of nursesavailable that can work in day shifts (2 shifts) and night shifts (1 shift). As expected, asmore weight is given to the dissatisfaction of shifts, the average dissatisfaction decreasesand the average number of nurses affected increases. The problems discussed allow us toconclude that the proposed formulation is treatable because the average CPU time was1.872 seconds and the worst case was of 24.406 seconds. For the tested instances, we notethat the objective function is increasing by a and the minimum of the objective function isattained in a = 0.
In this work, we study the problem of assignment of nursing shifts (Nurse RosteringProblem) in a hospital in New Zealand. In this hospital, they work three shifts a day,every day of the year. The scheduling horizon was defined in four weeks. Each nursecan request shifts off, i.e., not work on certain shifts in the scheduling horizon. Theproblem is formulated as a multi-objective integer linear problem (ILP) where is requiredto simultaneously minimize the dissatisfaction of the nurses as result of the failure to fulfilthe requirements and the number of nurses whose requirements are not fully satisfied. Thesolution approach is based on weights method which weights each objective by a weightingfactor (a). To determine the quality of the proposed formulation and cover a wide range ofdata, ten instances were generated. In each instance, the weighting factor was changed fora total of fifty problems. The feasibility of the problem depends on the number of nursesavailable that can work in day shifts (2 shifts) and night shifts (1 shift). As expected, asmore weight is given to the dissatisfaction of shifts, the average dissatisfaction decreasesand the average number of nurses affected increases. The problems discussed allow us toconclude that the proposed formulation is treatable because the average CPU time was1.872 seconds and the worst case was of 24.406 seconds. For the tested instances, we notethat the objective function is increasing by a and the minimum of the objective function isattained in a = 0.
Description
Catalogado desde la version PDF de la tesis.
Keywords
NURSE ROSTERING , PROBLEMA DE INTEGRAL LINEAL , PROGRAMACION DE PERSONAL DE HOSPITALES