Periodic Reporting for period 3 - REACTT (REliable Advanced Diagnostics and Control Tools for increased lifetime of solid oxide cell Technology)
Okres sprawozdawczy: 2024-01-01 do 2025-05-31
The exploitation of renewable energy sources such as wind, water, and solar energy has become one of the priorities of the modern energy sector across the globe. However, renewable sources cannot solely be used due to their intermittent nature, meaning that there are periods when supply exceeds demand and vice versa. Solid Oxide Electrolysis Cell (SOEC) systems are electrochemical energy conversion devices capable of converting electrical energy into carbon-free hydrogen. Hydrogen can be stored and potentially re-used as fuel for fuel cells to be re-converted into electrical energy. SOEC systems operate at high temperatures and have a unique property that the same device can operate in the reverse mode, i.e. as a Solid Oxide Fuel Cell (the so-called rSOC).
Despite their attractive features, SOC systems (a common term for SOEC, SOFC, and rSOC) are still not massively present in in-field applications. To ensure that SOC systems operate reliably, and efficiently and deliver hydrogen or electric power when required, an accurate assessment of their performance, health, and life span is necessary. That was the topic at the heart of the REACTT project (Figure 1).
The overall objectives of the REACTT project were the following:
1. Improve durability, reliability, and maintainability of SOEC and rSOC stacks by developing innovative algorithms for diagnostics and prognostics of lifetime;
2. Develop the advanced control strategy with self-optimizing and fault-tolerant features;
3. Develop the hardware module for implementation of the monitoring, diagnostics, prognostics, and control functions; and
4. Perform characterization of stacks and systems in SOE and rSOC nominal and faulty conditions and validation of the product prototype.
Despite their attractive features, SOC systems (a common term for SOEC, SOFC, and rSOC) are still not massively present in in-field applications. To ensure that SOC systems operate reliably, and efficiently and deliver hydrogen or electric power when required, an accurate assessment of their performance, health, and life span is necessary. That was the topic at the heart of the REACTT project (Figure 1).
The overall objectives of the REACTT project were the following:
1. Improve durability, reliability, and maintainability of SOEC and rSOC stacks by developing innovative algorithms for diagnostics and prognostics of lifetime;
2. Develop the advanced control strategy with self-optimizing and fault-tolerant features;
3. Develop the hardware module for implementation of the monitoring, diagnostics, prognostics, and control functions; and
4. Perform characterization of stacks and systems in SOE and rSOC nominal and faulty conditions and validation of the product prototype.
The main accomplishments of REACTT project:
• Test protocols that include fingerprint tests under non-faulty operation, as well as tests that would emulate faulty operation, all to understand better how stacks would respond to faults and how easily they recover once the normal operation is restored;
• 4 short stacks (6 cells each) and 3 full stacks with 70 repeating elements were delivered by the manufacturer SolydEra (Figure 2);
• An extensive experimental campaign was conducted by EPFL on a segmented cell stack (Figure 3) and by CEA and EPFL on SolydEra SOEC short stacks (Figure 4). Over 11.000h and 2100h of operating hours for the short stacks and the full stack respectively. Several fault regimes were included: (i) high steam conversion, (ii) H2 inlet starvation, (iii) humidity in O2 electrode. Over 1.000 EIS measurements have been performed (sinusoidal and DRBS excitation).
• The second (upgraded) release of the MDPC HW platform, see Figures 5(b) and 5(c). It hosts and executes the monitoring, diagnosis, prognosis, and control algorithms. The overall concept of this HW and of the firmware running on it is shown in Figure 5(a). The platform is low-cost, yet with high computational performance, thanks to the carefully selected components and optimized HW and SW design.
• Firmware for the MDPC platform including the algorithms running on the MDPC board, the database for the data exchange among the MDPC algorithms, and the communication with external control units, c.f. Figure 5(a).
• An innovative excitation module (EM) for stack perturbation with conventional sinusoidal and non-conventional discrete random binary signal (DRBS) has been developed (Figure 6).
• A real-time optimization (RTO) algorithm scaled up for operating the 5kW solid-oxide electrolyzer (SOE) system at optimal efficiency has been designed and excessively tested on the SOEC simulator (Figure 7).
• A consistent diagnostic framework comprising a portfolio of tools supporting two kinds of diagnostic approaches:
- the passive one utilizing conventional process and data-driven models of the stack and system (Figure 8) and
- the active one is based on an additional stack perturbation to get the stack dynamics in terms of the electrochemical impedance spectra (EIS). The diagnosis is pursued by interpreting the spectra by using Equivalent Circuit Models. The results of the diagnostic algorithm are probabilities for a particular fault mode (Figure 9).
• A multiscale model that unravels the complex relationships between the global solid oxide cell response and the reaction mechanisms taking place in the electrodes (c.f. Figure 10).
• A supervisor module for orchestrating the operation of the MDPC functional tasks (c.f. Figure 11).
• Experimental validation of the MDPC tool on 3 SolydEra full stacks at EPFL, CEA and VTT (Figure 12).
• Validated diagnostic algorithms for pump fault, noise and instability in steam flow. An excerpt showing detection and isolation of the fuel pump problems is shown in Figure 13.
• Validated RTO on the SolydEra 70-cell stack at the EPFL. The algorithm automatically enforces feasible operating conditions of the SOE system during real-time operation. Hence there is no need for expert manual tuning. An excerpt from the RTO performance is shown in Figure 14.
• Implemented a data-driven prognostic algorithm that online predicts the trend of the health indicator. An excerpt from the experiment at EPFL is given in Figure 15.
• Test protocols that include fingerprint tests under non-faulty operation, as well as tests that would emulate faulty operation, all to understand better how stacks would respond to faults and how easily they recover once the normal operation is restored;
• 4 short stacks (6 cells each) and 3 full stacks with 70 repeating elements were delivered by the manufacturer SolydEra (Figure 2);
• An extensive experimental campaign was conducted by EPFL on a segmented cell stack (Figure 3) and by CEA and EPFL on SolydEra SOEC short stacks (Figure 4). Over 11.000h and 2100h of operating hours for the short stacks and the full stack respectively. Several fault regimes were included: (i) high steam conversion, (ii) H2 inlet starvation, (iii) humidity in O2 electrode. Over 1.000 EIS measurements have been performed (sinusoidal and DRBS excitation).
• The second (upgraded) release of the MDPC HW platform, see Figures 5(b) and 5(c). It hosts and executes the monitoring, diagnosis, prognosis, and control algorithms. The overall concept of this HW and of the firmware running on it is shown in Figure 5(a). The platform is low-cost, yet with high computational performance, thanks to the carefully selected components and optimized HW and SW design.
• Firmware for the MDPC platform including the algorithms running on the MDPC board, the database for the data exchange among the MDPC algorithms, and the communication with external control units, c.f. Figure 5(a).
• An innovative excitation module (EM) for stack perturbation with conventional sinusoidal and non-conventional discrete random binary signal (DRBS) has been developed (Figure 6).
• A real-time optimization (RTO) algorithm scaled up for operating the 5kW solid-oxide electrolyzer (SOE) system at optimal efficiency has been designed and excessively tested on the SOEC simulator (Figure 7).
• A consistent diagnostic framework comprising a portfolio of tools supporting two kinds of diagnostic approaches:
- the passive one utilizing conventional process and data-driven models of the stack and system (Figure 8) and
- the active one is based on an additional stack perturbation to get the stack dynamics in terms of the electrochemical impedance spectra (EIS). The diagnosis is pursued by interpreting the spectra by using Equivalent Circuit Models. The results of the diagnostic algorithm are probabilities for a particular fault mode (Figure 9).
• A multiscale model that unravels the complex relationships between the global solid oxide cell response and the reaction mechanisms taking place in the electrodes (c.f. Figure 10).
• A supervisor module for orchestrating the operation of the MDPC functional tasks (c.f. Figure 11).
• Experimental validation of the MDPC tool on 3 SolydEra full stacks at EPFL, CEA and VTT (Figure 12).
• Validated diagnostic algorithms for pump fault, noise and instability in steam flow. An excerpt showing detection and isolation of the fuel pump problems is shown in Figure 13.
• Validated RTO on the SolydEra 70-cell stack at the EPFL. The algorithm automatically enforces feasible operating conditions of the SOE system during real-time operation. Hence there is no need for expert manual tuning. An excerpt from the RTO performance is shown in Figure 14.
• Implemented a data-driven prognostic algorithm that online predicts the trend of the health indicator. An excerpt from the experiment at EPFL is given in Figure 15.
Some of the developments in REACTT go beyond the state of the art:
• The MDPC module. It is a low-cost embedded hardware is a new versatile tool for SOC systems operation. It allows the execution of monitoring diagnostics, prognostics, and control techniques to enhance the lifetime while keeping the efficiency at a feasible maximum.
• The excitation module. It can inject practically deliberately chosen waveform on the frequency range up to 18kHz while able to provide 200 A maximum direct current to the stack. The excitation module can be applied directly to a broad range of stacks and on top of the existing power supplies. That discards any need for expensive re-design of the existing power converter and greatly saves efforts in the development stage of the monitoring and control technologies for SOC systems.
• The first successful implementation of the RTO module on the SOEC system for various hydrogen production targets.
• Implementation of a passive diagnostic system with evolving capability to learn from data.
The achievements of the REACTT project indicate their potential towards marketable solutions that ensure more effective and versatile asset monitoring and optimized operation in the context of in-field applications of the SOC technology. REACTT relates to the FCH-2-JU- Multi-Annual Work Plan -2014-2020 - Innovation Pillar 2 (MAWP). The goals are in agreement with those set by the Energy Pillar for the area and, in particular, its strategic objective to “implement an optimal research and innovation program at EU level to develop a portfolio of clean and efficient solutions that exploit the properties of hydrogen as an energy carrier and fuel cells as energy converters”.
• The MDPC module. It is a low-cost embedded hardware is a new versatile tool for SOC systems operation. It allows the execution of monitoring diagnostics, prognostics, and control techniques to enhance the lifetime while keeping the efficiency at a feasible maximum.
• The excitation module. It can inject practically deliberately chosen waveform on the frequency range up to 18kHz while able to provide 200 A maximum direct current to the stack. The excitation module can be applied directly to a broad range of stacks and on top of the existing power supplies. That discards any need for expensive re-design of the existing power converter and greatly saves efforts in the development stage of the monitoring and control technologies for SOC systems.
• The first successful implementation of the RTO module on the SOEC system for various hydrogen production targets.
• Implementation of a passive diagnostic system with evolving capability to learn from data.
The achievements of the REACTT project indicate their potential towards marketable solutions that ensure more effective and versatile asset monitoring and optimized operation in the context of in-field applications of the SOC technology. REACTT relates to the FCH-2-JU- Multi-Annual Work Plan -2014-2020 - Innovation Pillar 2 (MAWP). The goals are in agreement with those set by the Energy Pillar for the area and, in particular, its strategic objective to “implement an optimal research and innovation program at EU level to develop a portfolio of clean and efficient solutions that exploit the properties of hydrogen as an energy carrier and fuel cells as energy converters”.