Outcomes
Scientific advances and impacts
The research has developed and demonstrated
- (i) a comprehensive PEM fuel model, using a bond graph approach, encapsulated within a reliability assessment framework of petri nets
- (ii) a strategy for diagnostic evaluation using data based and model based methods and
- (iii) a knowledge based model offering greater insight into diagnostics.
These scientific advances have made a significant contribution to the academic literature, including, journal publications and multiple national and international conference presentations (as listed in the Resources section).
Industrial impact
This research is supported by Intelligent Energy and with these enhanced modelling capabilities, including quantification of the likelihood of PEM fuel cell stack failure degradation, alongside the diagnostic capability, will enable them and related industry to better evaluate the performance of their systems. The development of the reusable knowledge based model has potential for use across industry.
Public & social impact
Society and the public will benefit from the capability of longer life of fuel cell products through increased diagnostics and mitigation strategies, reducing overall costs and increasing likelihood for utilisation in differing fields.
Economic impact
The steps forward in this project will pave the way for reduced costs in supporting the operation of PEM fuel cells, with longer lifetimes.
Economic impact
The steps forward in this project will pave the way for reduced costs in supporting the operation of PEM fuel cells, with longer lifetimes.
Policy impact
Advanced capabilities in current fuel cell applications could enable policy makers to make more informed decisions regarding use of this techniques in society for other applications.
International impact
Collaborative meeting planned with Brno University of Defence, in Czech Republic, in March 2018, regarding uses and advances of this research.
Other impact
Two PhDs aligned to this research proposal are both in the writing up phase.
Work Packages
Activity: Establish the fuel cell functional description
This was achieved by constructing an initial reference model.
Activity: Fuel cell failure model generation
This was covered by the Petri-nets modelling incorporating a Bond graph model.
Activity: Fuel cell design optimisation
This work is ongoing.
Activity: Component degradation model development
This work is ongoing.
Activity: Asset management strategy development
This work is ongoing.
Activity: Prognostic model development
The approach taken in this work package included, a neural network, an adaptive neuro-fuzzy inference systems (ANFIS) and particle filtering.
Activity: Sensor type and location selection
Used the largest gap method, sensitivity analysis considering both sensor sensitivity and noise resistance.
Activity: Fault diagnosis
This was achieved by integrating two methods. Firstly a data-driven approach using various fuel cell information, like sensor measurements during the fuel cell operation, polarisation curve and EIS. Secondly by adopting a knowledge driven approach using an ontology.
Activity: Data hub infrastructure
Ontology constructed using flooding and dehydration rules.
Activity: Optimisation analysis
This work is ongoing.
Activity: Dynamic Health Monitoring Visualiser
Java & JavaFX client server system built. This simulates the streaming of data to allow the ‘real-time’ diagnostics of the sensory data
Activity: Demonstration and Validation of the Fuel Cell Design and Health Monitoring Decision Support Tool
Initial evaluation of diagnosing fuel cell flooding carried out using one and two fuel cell stacks.