Periodic Reporting for period 2 - RUBY (Robust and reliable general management tool for performance and dUraBility improvement of fuel cell stationarY units)
Reporting period: 2022-02-01 to 2023-07-31
Fuel Fells (FCs) are becoming more and more attractive for real world applications, their performance meet the strong constraints imposed by regulations and international policies aiming at reducing the carbon footprint of power systems. Despite the noticeable improvements attained in the last decade, still the competitiveness with respect to conventional solutions is yet to be achieved for FCs efficiency, reliability, availability and durability. The implementation on FC Systems of suitable control, diagnostic and prognostic algorithms can improve those performance, making FC more efficient and competitive. The project covers the areas of Proton Exchange Membrane FC (PEMFC) backup and Solid Oxide FC (SOFC) µ-CHP systems that are among the most mature stationary FC solutions available on the market today.
The project aims at developing, integrating, engineering and testing a comprehensive and generalized Monitoring, Diagnostic, Prognostic and Control (MDPC) tool capable of improving efficiency, reliability and durability of SOFC and PEM systems for stationary applications. The MDPC tool features Electrochemical Impedance Spectroscopy-based stack monitoring for fault diagnosis and lifetime estimation, Balance of Plant diagnostics, supervisory RTO control and mitigation strategies. The tool relies on advanced dedicated HW and will be embedded in the Fuel Cell Systems (FCSs) for on-line validation in relevant operational environment.
Therefore, the RUBY tool will reach at the end of the project the TRL 7. It is foreseen that the tool will be ready for engineering scaling up of production, together with certification for embedding within commercial FCS.
To successfully fulfil the aims of RUBY, the following four main technical objectives were set:
1. Improve FCS performance and durability by implementing an advanced and integrated algorithm that combines monitoring, diagnosis, prognosis, control and mitigation actions for both SOFC and PEM systems.
2. Design and engineer the HW required for MDPC algorithms application, with attention to sensors reduction issues and the specific constraints imposed by stack technologies and systems applications towards industrial scalability.
3. Perform dedicated experimental campaigns for stacks and system characterization and MDPC tool prototype validation embedded on FCSs running in operational environment.
4. Develop an advanced FCS management strategy (supervisory level), with functionalities integrated with remote monitoring, for future smart-grid interaction and predictive maintenance application.
The project aims at developing, integrating, engineering and testing a comprehensive and generalized Monitoring, Diagnostic, Prognostic and Control (MDPC) tool capable of improving efficiency, reliability and durability of SOFC and PEM systems for stationary applications. The MDPC tool features Electrochemical Impedance Spectroscopy-based stack monitoring for fault diagnosis and lifetime estimation, Balance of Plant diagnostics, supervisory RTO control and mitigation strategies. The tool relies on advanced dedicated HW and will be embedded in the Fuel Cell Systems (FCSs) for on-line validation in relevant operational environment.
Therefore, the RUBY tool will reach at the end of the project the TRL 7. It is foreseen that the tool will be ready for engineering scaling up of production, together with certification for embedding within commercial FCS.
To successfully fulfil the aims of RUBY, the following four main technical objectives were set:
1. Improve FCS performance and durability by implementing an advanced and integrated algorithm that combines monitoring, diagnosis, prognosis, control and mitigation actions for both SOFC and PEM systems.
2. Design and engineer the HW required for MDPC algorithms application, with attention to sensors reduction issues and the specific constraints imposed by stack technologies and systems applications towards industrial scalability.
3. Perform dedicated experimental campaigns for stacks and system characterization and MDPC tool prototype validation embedded on FCSs running in operational environment.
4. Develop an advanced FCS management strategy (supervisory level), with functionalities integrated with remote monitoring, for future smart-grid interaction and predictive maintenance application.
The main achievements at end of 2nd Period are:
- PEM stack for Backup System characterized during 100+ h of nominal and 3700+ h of faulty conditions
- 100+ EIS spectra measured on PEM stack for Backup operated in nominal & faulty conditions
- PEM Backup System characterized during 1000+ h of operation
- PEM Backup installed ready for MDPC tool implementation and final testing
- SOFC stack for µ-CHP System under testing (700+ h) for EIS characterization in nominal & faulty operations
- 74 EIS spectra available for the SOFC stack module, of which 20 EIS spectra in faulty conditions
- SOFC system tested during 10000 h of nominal & faulty operations
- 32 EIS spectra measured on SOFC stack fitted in the µ-CHP in faulty operations
- Database of features for monitoring & diagnosis extracted from EIS spectra
- Supervisory control for MDPC tool functions activation
- Ethernet communication protocol developed for interaction among all HW
- Firmware for MDPC board (RUBY-Box)
- MDPC board developed & tested
- Specifications of converter for on-board EIS implementation ready for third party contracting
- Systems (Backup & µ-CHP) installed and operational, ready for MDPC tool implementation and final testing
An analysis was conducted to develop MDPC tools for PEM and SOFC technologies. Various algorithms were designed using data from FCSs manufacturers and WP2/WP3 measurements. It was developed an innovative fault isolation method for SOFC technology, machine learning and deep learning techniques. Conventional voltage monitoring and classical approaches for predicting RUL have been implemented together with multiple models for PEM stack prognosis. EIS data were employed for model-based diagnostics. A multiscale model was used to assess the impact of Ni coarsening on SOFC degradation. Enhanced sensor-based diagnostics and soft sensors for adaptive control was exploited and a real-time optimization (RTO) algorithm for fuel cell stacks and systems was implemented as well.
The focus of WP6 was on designing and manufacturing HW components for MDPC algorithm implementation and onboard EIS spectra acquisition. The design of the HW for EIS stimuli to be injected by a DC/DC converter was completed for both SOFC µ-CHP and PEM backup. The RUBY-Box was built incorporating MDPC, Analog Front End board, Ethernet network connection and cost-effective Raspberry Pi boards. Six new RUBY-Box prototypes are currently undergoing electrical testing in laboratory. The Figure 1 shows the schemes of both µ-CHP and Backup systems integrated with the DC/DC converter embedding the EIS perturbation functions (p) and the RUBY-Box (red boxes) featuring control and measurement functions (f).
In WP2, extensive experimental activities were conducted for PEM FCSs. The testing included simulations of various faults, such as fuel and air starvation and cathode contamination, and utilized EIS with sinusoidal perturbations for performance characterization. A modified PEFC stack system was installed on a dedicated test rig, assessing EIS changes with respect to nominal and faulty operations. For SOFC technology within WP3, stack and system were tested in nominal and faulty operations as well as long-term and EIS measurements. Two µ-CHP systems are installed in a laboratory and in a gas utility company (Figure 2).
As far as smart grid concerns, advanced functions are being developed to implement energy management systems that would help achieving the optimal use of the energy within the grid. The partners are developing a supervisory strategy for an energy grid accounting for the three main energy vectors (namely electricity, heat, gas), along with final uses and production sites, with PEM and SOFC (including reversible technology) as the main energy conversion devices. To build the supervisory level the main functions have been identified. The main challenges being solved refers to communication, data transfer, code integration and parallel execution.
- PEM stack for Backup System characterized during 100+ h of nominal and 3700+ h of faulty conditions
- 100+ EIS spectra measured on PEM stack for Backup operated in nominal & faulty conditions
- PEM Backup System characterized during 1000+ h of operation
- PEM Backup installed ready for MDPC tool implementation and final testing
- SOFC stack for µ-CHP System under testing (700+ h) for EIS characterization in nominal & faulty operations
- 74 EIS spectra available for the SOFC stack module, of which 20 EIS spectra in faulty conditions
- SOFC system tested during 10000 h of nominal & faulty operations
- 32 EIS spectra measured on SOFC stack fitted in the µ-CHP in faulty operations
- Database of features for monitoring & diagnosis extracted from EIS spectra
- Supervisory control for MDPC tool functions activation
- Ethernet communication protocol developed for interaction among all HW
- Firmware for MDPC board (RUBY-Box)
- MDPC board developed & tested
- Specifications of converter for on-board EIS implementation ready for third party contracting
- Systems (Backup & µ-CHP) installed and operational, ready for MDPC tool implementation and final testing
An analysis was conducted to develop MDPC tools for PEM and SOFC technologies. Various algorithms were designed using data from FCSs manufacturers and WP2/WP3 measurements. It was developed an innovative fault isolation method for SOFC technology, machine learning and deep learning techniques. Conventional voltage monitoring and classical approaches for predicting RUL have been implemented together with multiple models for PEM stack prognosis. EIS data were employed for model-based diagnostics. A multiscale model was used to assess the impact of Ni coarsening on SOFC degradation. Enhanced sensor-based diagnostics and soft sensors for adaptive control was exploited and a real-time optimization (RTO) algorithm for fuel cell stacks and systems was implemented as well.
The focus of WP6 was on designing and manufacturing HW components for MDPC algorithm implementation and onboard EIS spectra acquisition. The design of the HW for EIS stimuli to be injected by a DC/DC converter was completed for both SOFC µ-CHP and PEM backup. The RUBY-Box was built incorporating MDPC, Analog Front End board, Ethernet network connection and cost-effective Raspberry Pi boards. Six new RUBY-Box prototypes are currently undergoing electrical testing in laboratory. The Figure 1 shows the schemes of both µ-CHP and Backup systems integrated with the DC/DC converter embedding the EIS perturbation functions (p) and the RUBY-Box (red boxes) featuring control and measurement functions (f).
In WP2, extensive experimental activities were conducted for PEM FCSs. The testing included simulations of various faults, such as fuel and air starvation and cathode contamination, and utilized EIS with sinusoidal perturbations for performance characterization. A modified PEFC stack system was installed on a dedicated test rig, assessing EIS changes with respect to nominal and faulty operations. For SOFC technology within WP3, stack and system were tested in nominal and faulty operations as well as long-term and EIS measurements. Two µ-CHP systems are installed in a laboratory and in a gas utility company (Figure 2).
As far as smart grid concerns, advanced functions are being developed to implement energy management systems that would help achieving the optimal use of the energy within the grid. The partners are developing a supervisory strategy for an energy grid accounting for the three main energy vectors (namely electricity, heat, gas), along with final uses and production sites, with PEM and SOFC (including reversible technology) as the main energy conversion devices. To build the supervisory level the main functions have been identified. The main challenges being solved refers to communication, data transfer, code integration and parallel execution.
The project preliminary outcomes are confirming the potential in generating positive impacts on both technology and market fields, thus contributing to the progress of FCSs products beyond the state-of-the-art. The results achieved so far confirm the expected advancements though a complete assessment will be performed in the second part of the action. One interesting result deals with the design of a HW configurations for EIS perturbation. In both cases the flexibility of the solutions can guarantee the implementation of EIS-based monitoring on any FC system. Furthermore, with respect to the project proposal, additional algorithms have been developed; this result will enlarge the possibility to find more appropriate and effective techniques for diagnostics, prognostics and control to enhance the lifetime while keeping the efficiency closer to that at beginning of life (BOL).