> Manufacturing and Engineering

Production Planning and Control

Summary

We improved the decision process and automated planning stage by building an AI assisted constraint-based scheduling tool that connects directly with their ERP system. The tool automates order allocation across machines and enforces stable planning windows, resulting in a 10% increase in throughput and a 70% reduction in planning time.

Case study:

Machine-scheduler for orders and maintenance

Sector

Manufacturing and Engineering

Client

A B2B paper manufacturer supplying custom-sized rolls of various weights and qualities to clients in packaging, labeling, printing, and related industries.

Engagement Overview

The client’s production planning relied heavily on manual processes and institutional knowledge. Orders were scheduled reactively, planning windows were unstable, and there was no centralized way to assess which machines could fulfill a given order. Over a 2-month period, we developed and deployed a custom scheduling tool that integrates with their COBOL-based ERP system. The result: improved production stability, faster planning, and increased throughput.

Constraint-Based Scheduler

We built a constraint programming model to automatically generate efficient production plans.

  • Logic: Tasks are scheduled based on machine feasibility, order priorities, and setup minimization.
  • Rules Introduced: Orders must be locked into production 48h ahead. No new orders can be scheduled for the current or next day. Minimum delivery time of 4 days enforced.
  • Outcome: Greater stability and reduced last-minute disruptions in planning.

ERP Integration via API

Rather than interfering directly with the client’s COBOL system, we implemented a lightweight, scalable API-based architecture.

  • Deployment: Optimization engine hosted on a remote Linux server (Ubuntu) with Nginx and SSL.
  • Workflow: A JSON payload with open order and feasibility data is sent via POST request to an endpoint. The server responds with a solution and a visual schedule.
  • Advantage: The standalone API avoided the risk and complexity of modifying the ERP, while still integrating seamlessly with the planning workflow.

Familiar Interface for Fast Adoption

To ensure fast adoption by planners, we embedded the solution directly into Google Sheets and Excel.

  • Input: Users enter open orders and constraints into the spreadsheet.
  • Execution: Add-on compiles the data into a JSON file and sends it to the scheduler.
  • Output: Returns a table with scheduled machine times and a Gantt-style chart showing machine x order allocation.
  • Benefit: No need for planners to learn new systems, just open a spreadsheet, press a button, and get a full production plan.

Quantifiable Results

  • +10% increase in productivity (tons/hour)
  • -70% reduction in planning time
  • More consistent delivery timelines and fewer urgent reworks
  • A scalable process for integrating future planning rules and machine constraints

Technologies Used

Python (Google OR-Tools), Node.js, Nginx, PM2, Ubuntu Server, Google Sheets API, Excel Add-ins, JSON over HTTPS.

Collaboration Style

We worked closely with stakeholders across departments, planning, IT, production, quality, and business intelligence, to ensure the tool aligned with both technical realities and business goals. Our ongoing support includes training, maintenance, and new feature rollouts as the system matures.