Grey-box modelling and virtual sensors enabling continuous commissioning of hydronic floor heating

Lucile Sarran, Kevin M. Smith*, Christian A. Hviid, Carsten Rode

*Corresponding author for this work

Research output: Contribution to journalJournal articleResearchpeer-review

116 Downloads (Pure)


Operating district heating systems with low supply and return temperatures improves heat production and distribution efficiency, permitting greater integration of renewable heat sources. Low-temperature district heating is viable without compromising comfort, but faults in end-users’ heating systems constrain temperature reductions. Such faults include malfunctioning valves, improper hydronic balancing, and excessive supply temperature setpoints. Occupants lack the resources to detect and diagnose these faults, so there is a need for automated solutions without requiring additional hardware. This paper proposes a method for improving the operation of an apartment's hydronic floor heating system using data from room thermostats, a heat meter and a circulation pump to identify a grey-box model of the system. The resulting model virtually senses each room loop's heat flux, flow, return temperature, and flow coefficient. The authors tested the model on a low-energy apartment in Denmark, using it to diagnose causes of high return temperatures, including poor hydronic balancing and an excessive supply temperature setpoint and pump setting. The authors also used the model to predict the minimum permissible supply temperature maintaining comfort, yielding a reduction in the energy-weighted supply and return temperatures of 8.6 °C and 6.5 °C, respectively.
Original languageEnglish
Article number125282
Number of pages14
Publication statusPublished - 2022


  • Fault detection and diagnosis
  • Virtual sensor
  • Grey-box modelling
  • Floor heating
  • District heating


Dive into the research topics of 'Grey-box modelling and virtual sensors enabling continuous commissioning of hydronic floor heating'. Together they form a unique fingerprint.

Cite this