Thesis Diseño de un modelo de segmentación y control de inventarios para almacenes de mantenimiento
Loading...
Date
2025-08
Journal Title
Journal ISSN
Volume Title
Program
Ingeniería Civil Industrial
Departament
Campus
Campus Santiago Vitacura
Abstract
La gestión eficiente de inventarios en almacenes mecánicos es fundamental para garantizar la continuidad operativa y optimizar el uso de recursos.  Este trabajo presenta un modelo de segmentación y control de inventarios que combina técnicas de clasificación multivariable y análisis de patrones de consumo, con el propósito de fortalecer la toma de decisiones y reducir costos operativos.  La metodología propuesta integra el análisis histórico del consumo de repuestos con la clasificación de materiales según su valor, criticidad y cantidad demandada, junto con la segmentación de almacenes considerando su ubicación, valor consumido y diversidad de negocios atendidos.  Las clasificaciones multivariables se desarrollan mediante el algoritmo K-Means, implementado en Python, lo que posibilita un procesamiento eficiente, replicable y escalable de grandes volúmenes de datos.  A partir de esta segmentación, se definen límites de control (s,S)1 específicos para cada combinación material–almacén, aplicando criterios estadísticos y juicio experto, y posteriormente se evalúa su impacto frente a la situación actual y en un escenario conservador.  Los resultados evidencian que la aplicación del modelo permite reducir en un 91,18% el valor total de inventarios, equivalente a un ahorro estimado de 1.702 millones de CLP, manteniendo niveles de cobertura adecuados para la operación.  El escenario conservador, que incrementa los límites de inventario, alcanza un ahorro de 1.557 millones de CLP y asegura una cobertura más  holgada.  Se concluye que la estrategia propuesta resulta altamente efectiva para optimizar el capi tal inmovilizado sin comprometer la continuidad del servicio.  Se recomienda su implementación progresiva, estableciendo un nivel de servicio objetivo y fortaleciendo la clasificación con métodos adicionales, como series de tiempo o análisis de tasas de falla, para mejorar la precisión en la estimación de la demanda.
Efficient inventory management in mechanical warehouses is essential to ensure operational continuity and optimize resource utilization. This work presents a segmentation and inventory control model that combines multivariable classification techniques with consumption pattern analysis, aiming to enhance decision-making and reduce operating costs. The proposed methodology integrates the historical analysis of spare parts consumption with the classification of materials according to their value, criticality, and demand volume, along with the segmentation of warehouses based on location, consumed value, and diversity of business operations served. Multivariable classifications are performed using the K-Means algorithm, implemented in Python, enabling efficient, replicable, and scalable processing of large datasets. Based on this segmentation, (s, S) control limits are defined for each material–warehouse combination, applying both statistical criteria and expert judgment, and subsequently assessing their impact compared to the current situation and under a conservative scenario. The results show that the implementation of the model can reduce the total inventory value by 91.18 %, equivalent to an estimated saving of CLP 1.702 MM, while maintaining adequate coverage levels for operations. The conservative scenario, which increases inventory limits for critical cases, achieves savings of CLP 1.557 MM and ensures more robust coverage. It is concluded that the proposed strategy is highly effective in optimizing tied-up capital without compromising service continuity. Progressive implementation is recommended, establishing a target service level and strengthening classification with additional methods such as time series analysis or failure rate analysis to improve demand estimation accuracy.
Efficient inventory management in mechanical warehouses is essential to ensure operational continuity and optimize resource utilization. This work presents a segmentation and inventory control model that combines multivariable classification techniques with consumption pattern analysis, aiming to enhance decision-making and reduce operating costs. The proposed methodology integrates the historical analysis of spare parts consumption with the classification of materials according to their value, criticality, and demand volume, along with the segmentation of warehouses based on location, consumed value, and diversity of business operations served. Multivariable classifications are performed using the K-Means algorithm, implemented in Python, enabling efficient, replicable, and scalable processing of large datasets. Based on this segmentation, (s, S) control limits are defined for each material–warehouse combination, applying both statistical criteria and expert judgment, and subsequently assessing their impact compared to the current situation and under a conservative scenario. The results show that the implementation of the model can reduce the total inventory value by 91.18 %, equivalent to an estimated saving of CLP 1.702 MM, while maintaining adequate coverage levels for operations. The conservative scenario, which increases inventory limits for critical cases, achieves savings of CLP 1.557 MM and ensures more robust coverage. It is concluded that the proposed strategy is highly effective in optimizing tied-up capital without compromising service continuity. Progressive implementation is recommended, establishing a target service level and strengthening classification with additional methods such as time series analysis or failure rate analysis to improve demand estimation accuracy.
Description
Keywords
Gestión de inventario, K-Means, Clasificación de repuestos, Python, Inventory management, Spare parts classification

 Política de Privacidad
 Política de Privacidad