KI-Pirinha: intelligent protection versus H2-detonation

Intelligent protection against hydrogen detonation

KI-PIRINHA stands for Key enabling technologIes on Performance, Efficiency and Resilience by Artificial INtelligence for Hydrogen Applications. 

Contrary to its name, the aim of this research project is the avoidance of a detonation cocktail at all costs.

It aims to develop viable technical solutions to improve the safety and operational efficiency of H2-powered applications by combining the knowledge of experts in combustion research (CBOne) and experts in the simulation and control of complex dynamic systems using AI (FH JOANNEUM, Institute of Aviation). FH JOANNEUM is the project leader. This Austrian national research project is funded by BMK and FFG. 

Recent incidents involving or testing hydrogen show that SoA-type monitoring and control systems are not capable of quickly correcting an event leading to a catastrophic outcome (e.g. recently in the region of Styria, the rupture of a hydrogen pipe under very high pressure as a result of human error, leading to the closure of a motorway section for several hours and the detour of aircraft for a similar period), nor of assessing the severity of a situation (was there or was there not a risk of H2 ignition/detonation on the motorway or in the air corridor?)

KI-PIRINHA's goal is not only to respond quickly and well to ensure the safety of personnel and infrastructure, but also to anticipate and prevent such events. Intelligent online health monitoring systems that allow rapid response to abnormal operation of H2 combustors are not yet available on the market. Such systems are based on conventional statistical monitoring of the live process when deviations exceed a given operating range.

The methodology of the KI-PIRINHA project is to implement an AI-based combustion health monitoring system using an intelligently arranged array of multi-channel opto-acoustic sensor probes in the combustion chamber, and then to feed the real-time data into an AI-based algorithm (not yet available) to prevent detonations. The same project will allow us to continue our work on flame monitoring in the emotion and emootion projects, as we have found that the combined opto-acoustic probe is perfectly adapted to the combustion of hydrogen (see links below).

In addition, the project will investigate the transfer of the trained system to combustors of different layout or size to enable transfer learning. The use of AI in this application area contributes significantly to the safe and reliable operation of H2 combustion and thus to the achievement of Austria's climate goals.

In detail, the tasks of KI-PIRINHA are:

  1. Create a data acquisition pipeline that equally combines simulation and experimental data to enable reduced-order, data-driven modelling of green H2 combustion (digital twin).
  2. Creation of a health monitoring/early anomaly detection pipeline based on a minimal configuration of sensors (smart sensor placement) that can safely prevent abnormal operation, thus increasing the safety and reliability of green H2 combustor operation and thus increasing the resilience of the energy supply.
  3. Leverage the trustworthiness of AI for anomaly detection through increased explainability.
  4. Promote the philosophy of open science by following the ideas of open data and open methodology.



The Austrian Research Promotion Agency (FFG) is entrusted by the Austrian Ministry for Climate Protection and Energy with the operational management of funding awards.  The FFG therefore manages the contract "KI-PIRINHA" Nr. FO999910254. Hereby we would like to thank all grantors.


Related links: