crystAIr - Artificial Intelligence- and sensing-driven combustion burner

crystAIr aims to develop and deploy a digital twin technology for the redesigned hydrogen burner as a key element of the engine. The technology is based on the acquisition of key burner parameters by piezocryst and fiber optic sensors (FOS) and an integrated machine learning (ML) approach to the acquired data to improve the performance of the hydrogen burner for safe and seamless flow operation with minimal environmental impact.

Short Description


The use of sensors in the aerospace industry is mainly for structural health monitoring. However, these huge amounts of data could find many uses in Big Data analytics and especially in AI modeling. The result of this data analysis could be a new design of the engine or aircraft components. Hydrogen combustion faces many challenges: for example, finding an optimal combustion regime and burner geometry. Processing huge amounts of data using AI models could be critical to the strategy of rapid hydrogen adoption.


Sensorize the hydrogen burner to pick up the most important combustion parameters.
Develop the data acquisition strategy and implement a ML-based digital twin for virtual representation and AI control of hydrogen combustion in the new generation of air engines.
An intelligent redesign of the burner geometry, to obtain an optimal air/H2 mixture and flame.


The goal of the crystAir project is to achieve optimal and safe hydrogen combustion by using sensor technology and AI. In the project, the existing sensors are combined with the piezocryst sensors to achieve complete data acquisition. The Piezocryst sensors are able to collect information about vibration, voltage and load at high temperatures to complete the monitoring of the burner. On the one hand, the collected data and its analysis will be useful in the redesign of the burner, offering the possibility to produce new additively manufactured burner geometries that allow a more uniform combustion. On the other hand, the data will be used for training some AI models for intelligent online combustion control.


To achieve this goal, the following challenges must be addressed and overcome:
(a) sensing of the combustion burner: implementation of an appropriate sensor network (1-3 sensors) for SHM data acquisition, based on piezocryst sensors embedded in the burner;
(b) Develop data acquisition strategies, including sensor parameter selection and positioning, as well as data processing and computing facilities, to improve hydrogen combustion burners in terms of flame flashback prevention, improved maintenance, and NOx reduction by testing novel sensor-controlled H2 burners fabricated using AM technologies.
(c) Develop a hybrid digital twin based on an integrated approach to component and combustion process modeling, driven by AI-based solutions (XAL), with the goal of performing unsupervised mixture and flame control in the existing burner design.
The digital architecture will be applied to a newly designed burner, fabricated using mathematical shaping strategies and additive manufacturing, and fully optimized for hydrogen combustion.

Expected results

The key deliverables are a new hydrogen burner design and a digital twin that virtually represents and effectively controls the combustion process of an air engine.

Project Partners

Funding program: Take Off