Scaling Enterprise AI at Georgia-Pacific

Challenges

Georgia-Pacific (GP), one of the world’s leading makers of tissue, pulp, packaging, and building products, has over 100 manufacturing facilities across North America, including paper mills. Paper mills are large-scale, complex operations that require efficient and reliable operations to avoid costly unplanned downtime.

In 2018, to dramatically improve operational reliability, GP established the Collaboration and Support Center (CSC), home to over 150 engineers, IT personnel, and data scientists. Since its inception, the CSC delivered remarkable results using rule-based and machine learning (ML)-based models to monitor thousands of individual sensor readings, and generating alerts when a specific tag became anomalous. However, as the CSC looked to scale up its program to monitor more complex assets and processes, they found that existing software platforms were not optimized to meet their data and ML needs.

The CSC needed a new solution that could uplevel their program in two ways: (1) monitor at an asset versus tag level to better contextualize issues and eliminate redundant alerts, and (2) utilize massive data sets from disparate sources and more advanced ML techniques to improve monitoring performance.

Center of Excellence with C3 AI

To meet these needs, in 2020, GP chose to embark on a multi-year partnership with C3 AI to leverage the C3 AI Platform and applications, as well as C3 AI’s expertise in developing and scaling enterprise AI applications. A Center of Excellence (CoE) consisting of engineers, data scientists, and subject matter experts from both companies was formed, and the team quickly got to work on the first use case: leveraging the C3 AI Reliability application to reduce unplanned downtime.

Initial Results and Looking Ahead

By integrating AI into its manufacturing operations, GP is seeing significant improvement in overall monitoring performance, reduced unplanned downtime and maintenance costs, improving overall equipment effectiveness (OEE) by up to 5%. Building on this success, GP plans to expand the application to eight additional critical asset classes.

This deployment is just the start for GP. The CoE continues to identify and scope new AI use cases, while delivering scalable, high value enterprise AI applications across GP.

Company Objectives

• Improve process monitoring performance by leveraging more data and advanced ML techniques
• Rapidly scale across any asset class or process unit
• Effectively deploy and manage ML models at scale

About Georgia-Pacific

• Leading manufacturer of tissue, pulp, packaging, and building products
• HQ in Atlanta, GA
• Over 100 facilities

Project Highlights

• 200+ assets monitored across 13 paper mills
• 1 live ML model per asset, plus continuously deployed challenger models
• 2,816 ML model features
• 6 disparate data sources integrated

Solution Architecture