Tubacex – Reducing Electric Furnace Energy Consumption
Tubacex
Sector: Industry
Business Case
Develop a predictive model based on data collected from processes and PLCs over the past 5 years. This model will recommend the minimum heating time needed to reach the optimal casting temperature.
Objectives
The aim is to reduce energy consumption per casting and to provide operators with tools to optimise the process and avoid casting overheating or casting processes below the optimum temperature.
Use case
Data is shared through Snowflake for dataset preparation. Work is underway to develop a cloud-based predictive model, which is fed by a preceding descriptive model. The real-time results will be presented to the operator via the Grafana platform.
Infrastructure
Cloud
Technology
Automatic or Deep Learning
Data
Private datasets They contain process records reported in the MES system, about 640 tags (PLC) reporting per second. Our “real time” is estimated at 2 minutes.
Resources
A nivel organizativo ha sido necesario 1 jefe de proyecto, 2 técnicos de SW (ETL y ML) y 2 técnicos de procesos internos de ACVA. Desarrollo técnico principalmente por IBERMATICA, y validación técnica, requisitos y viabilidad por parte de TUBACEX.
Difficulties and learning
At the organisational level, 1 project manager, 2 SW technicians (ETL and ML) and 2 ACVA internal process technicians were required. Technical development mainly by IBERMATICA, and technical validation, requirements and feasibility by TUBACEX.
KPIs (business impact and metrics of the model)
RMSE (Root Mean Square Error). RMSE comparing predicted temperatures with real temperatures. The lower the RMSE value, the better the accuracy of the model in predicting the final temperature of the casting and thus the power consumption, our target KPI.
Funding
The project was funded by the European COGNIPLANT programme.
Collaborators, Partners
The project was developed in collaboration with IBERMATICA.
