27 February 2026
AI enables accurate real-time flow measurements
Artificial intelligence can make advanced flow measurements both faster and more reliable. New research shows how deep learning can be used to analyse flows in a detailed manner, that is how liquids and gases move in real time, helping to improve energy systems, reduce wear in machinery and prevent unexpected operational shutdowns.
Measuring the motion of liquids and gases is central to many areas of engineering, including energy, automotive, aerospace and process industries. In laboratory environments, Particle Image Velocimetry (PIV) is commonly used to map such flows, but the large and detailed datasets generated by the method are difficult to analyse using conventional tools.
In his doctoral thesis, Yuvarajendra Anjaneya Reddy, Doctoral Student in Experimental Mechanics at Luleå University of Technology, has investigated how artificial intelligence can be used to interpret these data in a more robust way. A key focus has been to develop end-to-end methods that are not only fast, but also consistent with the physical laws governing fluid flow.
“When artificial intelligence is built with a clear physical foundation, it can function as a reliable measurement tool rather than a black box,” says Yuvarajendra Anjaneya Reddy.
More temporally coherent flow fields
The thesis shows how new AI-based models can reconstruct flow fields with high precision even under demanding measurement conditions, such as strong shear, flow separation and low signal levels. By combining spatial and temporal information from several consecutive image frames, the results become smoother and more physically consistent and have high resolution than those produced by conventional methods.
Taking into account how a flow evolves over time is crucial for mitigating noise and measurement errors, particularly when the results are intended for monitoring or control in production sectors.
“Temporally coherent measurement data are essential if the results are to be used in real time,” says Yuvarajendra Anjaneya Reddy.
Yuvarajendra Anjaneya Reddy, doctoral student in experimental mechanics at Luleå University of Technology.
Applications in industry and digital twins
The methods developed in the thesis could have significant impact on industrial applications where flows need to be monitored continuously. Examples include cooling in energy systems, aerodynamic testing and early detection of deviations that might otherwise lead to inefficient operation or system failure.
The research is also relevant for the development of digital twins, where virtual models are continuously updated with measurement data from physical systems in real-time. With more reliable real-time flow measurements, such models can become both more accurate and more autonomous.
“In the long term, this could contribute to smarter sensors and systems that adapt directly to how flows actually behave in real-world conditions,” says Yuvarajendra Anjaneya Reddy.
Contact
Yuvarajendra Anjaneya Reddy
- Doctoral student
- 0920-49
- yuvarajendra.anjaneya.reddy@ltu.se
- Yuvarajendra Anjaneya Reddy
Published:
Updated: