@inproceedings{FloresLWComparison2025,
abstract = {The efficient estimation of lead times in manufacturing processes is crucial for optimizing production and reducing costs. In wind tower manufacturing, particularly in the longitudinal welding operation, accurate lead time prediction is essential for maintaining smooth workflows and meeting tight delivery schedules. Traditional engineering methods, which rely on analytical models and heuristics, have been widely used to estimate welding times. However, these methods often fail to account for the complexity and variability of real-world conditions, such as equipment wear, environmental factors, and production bottlenecks. In recent years, machine learning (ML) techniques have emerged as powerful tools for predictive modelling, leveraging large datasets and learning from patterns within the data. This study compares traditional engineering methods with ML approaches for lead time prediction in longitudinal welding, using data from a wind tower manufacturing plant. The results demonstrate that ML models, particularly Gradient Boosting, outperform traditional methods in accuracy and flexibility, offering significant potential for improving operational efficiency in the wind tower industry.},
address = {Cham},
author = {Flores-Huam{\'a}n, Kenny-Jes{\'u}s
and Lorenzo-Espejo, Antonio
and Mu{\~{n}}oz-D{\'i}az, Mar{\'i}a-Luisa
and Escudero-Santana, Alejandro},
booktitle = {Data Science, Challenges and Applications in Industrial Operations},
editor = {Barbadilla-Mart{\'i}n, Elena
and Robles-Velasco, Alicia
and Rodr{\'i}guez-Palero, Mar{\'i}a
and Cort{\'e}s, Pablo},
isbn = {978-3-032-10126-6},
pages = {321--326},
publisher = {Springer Nature Switzerland},
title = {Enhancing Lead Time Prediction in Wind Tower Manufacturing: A ML Approach Compared to Traditional Engineering Models},
year = {2025}
}
Abstract
The efficient estimation of lead times in manufacturing processes is crucial for optimizing production and reducing costs. In wind tower manufacturing, particularly in the longitudinal welding operation, accurate lead time prediction is essential for maintaining smooth workflows and meeting tight delivery schedules. Traditional engineering methods, which rely on analytical models and heuristics, have been widely used to estimate welding times. However, these methods often fail to account for the complexity and variability of real-world conditions, such as equipment wear, environmental factors, and production bottlenecks. In recent years, machine learning (ML) techniques have emerged as powerful tools for predictive modelling, leveraging large datasets and learning from patterns within the data. This study compares traditional engineering methods with ML approaches for lead time prediction in longitudinal welding, using data from a wind tower manufacturing plant. The results demonstrate that ML models, particularly Gradient Boosting, outperform traditional methods in accuracy and flexibility, offering significant potential for improving operational efficiency in the wind tower industry.