How GEA turns its fleet into a data foundation for AI services
Around 150 separators connected in a few weeks, without downtime and on existing hardware. Today, condition monitoring and the AI-based service GEA PerformancePlus run on this foundation.
- Customer
- GEA Westfalia Separator Group GmbH
- Industry
- Machine building
- Services
- Edge connectivity, AI services, engineering automation
- Technology
- Edge connectivity, RAG, reinforcement learning
Starting point
GEA is one of the largest machine builders for the food, beverage and pharmaceutical industries. GEA Westfalia Separator builds industrial separators for food, pharma and chemicals.
For the global rollout of an IoT solution, GEA needed an IoT gateway that could be installed on all existing and new machines. Time-consuming and costly downtime was not an option.
- Flexible and cost-efficient to deploy
- Easy to commission
- Integration into existing cloud and IoT platforms
Implementation
The foundation is codestryke’s edge and middleware software. On existing machines, it runs via Docker on the IPC already in place; new machines get a dedicated edge box. The device wizard scans the network and detects manufacturer and model. Presets and the integration with TIA Portal shorten setup. Software and configuration reach the devices over the air.
On this data foundation, we build the AI services together with GEA. A neural network on the edge gateway analyzes vibration data from the ifm VSE sensors, detects bearing damage early and classifies it. The AI Knowledge Hub brings together knowledge about the machine and the process for chatbots and agentic workflows. An RL agent controls the bowl and the paring disc in real time.
In engineering, we generate PLC code and HMI screens automatically from modular hardware variants – via TIA Portal Openness and the Rockwell Logix Designer SDK.
Result
Within a few weeks, data from 150 machines worldwide was available for GEA PerformancePlus, a new service model from GEA. Because existing hardware stays in use and travel is no longer needed, GEA saved an amount in the upper six-figure range.
Analyzing high-frequency vibration data reduces how often the separators need maintenance. Chatbots help with questions about the machine and reduce the training effort.
- Machines worldwide connected in a few weeks
- 150
- Savings from using existing hardware and eliminating travel
- 6-figure
- Performance from the RL agent on the separator fleet
- +9 %
- Damage to the bowl thanks to drift detection
- −40 %
- Training effort thanks to chatbots
- −80 %
Testimonials
“With codestryke, we found the perfect match for turning the data from our machines into AI services for our customers.”
Senior Director Automation & Controls
“With codestryke, we found the optimal edge solution.”
Project Manager
More references
All references
Let’s talk about your project.
Tell us which machine, which line or which platform you want to make better use of. In a first call you get an honest assessment.