Desafío

The customer, a manufacturer of stoppers for perfume bottles, faced high operational complexity when planning their production:

• Large number of product references, multiple lines and a variety of colors.
  
• Continuous operation with three daily shifts, each with its own planner.
  
• High workload to reorganize production in the face of daily unforeseen events.
  ‍

This generated difficulties in anticipating plant needs and caused an inagile management of available resources.

‍‍

Artificial Intelligence Solution

It developed a operational planning tool tailored, designed to convert existing data into intelligent and operational recommendations. The solution was integrated with the customer's systems and made it possible to automate decision-making in the short and medium term.

‍

Key components of the system:

• Integration with ERP/MES and Microsoft Project from the customer to obtain data on manufacturing orders and plant status.
  
  
• Functional web interface to consult and manage the generated schedule.
  
  
• Planning engine based on heuristic, hybrid or operational research algorithms.
  
  
• Managing manual restrictions, to incorporate unforeseen events not reflected in the system.
  
  
• Plant progress control in real time, with full traceability.
  
  

‍Results

• Capacity of plan 3 to 4 days in advance, with the possibility of extending up to 7 days or more.
  
  
• Improved control over plant execution and the visibility of work orders in progress.
  
  
• Progressive digitization of critical process parameters to power the system.
  
  
• Activation in a real environment with support in the start-up and training of key users.
  
  

‍Lessons Learned

• The joint work between the customer planning team and the Datision engineers was essential to identify and formalize the real operating conditions.
  
  
• The quality of the labeling of historical data was critical to achieving reliable plans.
  
  
• Starting with controlled environment testing and iterative validations allowed us to adjust the planning engine before full deployment.
  
  
• The modular approach facilitated a structured implementation, with continuous improvements based on active use of the tool.

‍