Dominik Neugebauer

180km Steam Network: Wireless Condition Monitoring with LoRaWAN

CHEMPARK operates around 1,000 kilometers of pipelines, roughly 180 kilometers of which is steam network. The network was historically designed for different load profiles – demand and offtake points have changed significantly since then. Conventional temperature measurement is barely economically viable out in the field without extensive cabling. Steam traps – components that drain liquid condensate from the steam system – were previously inspected manually; as a result, defects went unnoticed for months. Currenta Conneqtive has built a LoRaWAN network that covers the entire CHEMPARK with just a few outdoor antennas. Battery-powered sensors capture temperatures at network nodes and monitor the status of the steam traps via Condition Monitoring. The goal is to reduce response times for defective traps from months to hours – and to gradually optimize the network design based on better data. In parallel, an AI-powered dynamic simulation model is being developed that maps the network's condition even between measurement points. Because external service providers couldn't deliver the necessary combination of ML expertise and thermodynamic process knowledge, the model is being developed in-house. CHEMPARK serves as a stress test here – what runs reliably here is intended to be offered as a standardized SaaS product for external industrial customers. Your key takeaways LoRaWAN covers expansive industrial sites cost-effectively with just a few antennas – without extensive cabling infrastructure. Condition Monitoring of the steam traps reduces response times for defects from months to hours. Live sensor data from the field improve network simulations and enable better-informed investment decisions in steam network operation. AI-powered simulation in a process environment requires ML expertise and thermodynamic domain knowledge in equal measure – without combining both, you will fail at the model. A LoRaWAN network as shared infrastructure can be used for multiple use cases simultaneously and gradually expanded with new applications.