Skip to main content

Real operation pem fuel cells HEALTH-state monitoring and diagnosis based on dc-dc COnverter embeddeD Eis

Periodic Reporting for period 1 - HEALTH-CODE (Real operation pem fuel cells HEALTH-state monitoring and diagnosis based on dc-dc COnverter embeddeD Eis)

Reporting period: 2015-09-01 to 2017-04-30

HEALTH-CODE project aims at improving and validating an advanced Monitoring and Diagnostic Tool (MDT) capable of evaluating the State-of-Health (SoH) of Polymer Electrolyte Membrane Fuel Cell (PEMFC). The focus is mainly related to micro-Combined Heat and Power (µ-CHP) and backup applications equipped with air- and oxygen-fed stacks, respectively. The MDT is based on the use of suitable data derived from Electrochemical Impedance Spectroscopy (EIS) measurements performed during systems operation. In the available literature, EIS has been widely recognized as rather effective in extracting valuable information by means of just one measurement process. However, such method is applied mainly in laboratory environment due to its costs and complexity of execution. HEALTH-CODE leverages EIS advantages while ensuring the design of proper hardware for on-board uses, bridging laboratory and on-field applications.

FC technologies are currently spreading in the public market due to the great technological improvements achieved in the last decades. Nevertheless, their reliability, availability and durability are still below competitive values with respect to conventional power systems. The on-board implementation of suitable control, diagnostic and prognostic algorithms can improve such features, making fuel cell more efficient and competitive.

HEALTH-CODE has set the following three main technical objectives:
- the enhancement of EIS-based diagnosis for residential μ-CHP and backup applications;
- the development of a Monitoring and Diagnostic Tool (MDT) for State-of-Health (SoH) assessment, fault detection and isolation as well as degradation level analysis for lifetime extrapolation;
- the reduction of experimental campaign time and costs by means of innovative scaling-up methodology.

The three main objectives are being achieved by the project, whose framework is structured with three main pillars, that are i) hardware and power electronics, ii) monitoring and diagnostic algorithms and iii) experimental analysis. Figure 1 gives a schematic representation of the main elements of HEALTH-CODE.
The achievements of the HEALTH-CODE project reached at M20 (April 2017) are related to:
- EIS board design, manufacturing and preliminary validation;
- High voltage (HV) DC/DC converter design and commissioning;
- Low voltage (LV) DC/DC converter modifications and peliminary testing;
- Experimental test protocol and FC characterization;
- EIS measurements performance, in nominal and faulty conditions, for both FC technologies;
- Monitoring and diagnostic algorithms design and preliminary testing;
- Scaling-up algorithm design and validation on literature data.

More than 50% of the expected work has been completed at Mid Term.

For the design and manufacturing of the EIS board and converters (Fig. 1) required for the two investigated systems, the starting point was the HW available from the D-CODE project. The final assembly of the EIS board and its box is reported in Fig. 2, which shows a set of available connection ports for the purpose of testing and validation activities. From the figure, it is remarkable the possibility to further reduce the size for better on-board installation.

The HV converter for air-fed system - integrated with a DC/AC inverter - has been designed and commissioned. The HW has been conceived for the maximum efficiency (expected to be 94%) and features grid connection and DC loads. The low voltage converter of the backup system has been modified for the right interfacing with the EIS board. A preliminary test of the communication protocols has been also performed. A picture of the tests for the communication between the EIS board and the low voltage converter is presented in Fig. 3. Further advancements will entail the final coupling of the EIS board with the two converters and the verification of EIS measurements quality during on-board operations.

The first step towards experimental characterization of the both FC stacks entailed the definition of proper test protocols to increase the information, retrieved with the experimental activity, and reduce the amount of required experiments. Then, tests have been considered for both standard and faulty conditions. More than 1200 EIS spectra have been collected so far, of which about 25% are in nominal conditions and the other 75% in faulty states. Examples of EIS spectra acquired during the first test campaign are shown in Fig. 4 in nominal operation (a, e) and under fuel starvation (b), air starvation (c), anode flooding (d) and faulty states at 120 A (f).

For the description of the work done on scaling-up methodology, a complete description is available on Journal of Power Sources as open source publication at the following link: .

The diagnostic algorithms proposed by HEALTH-CODE are listed below and classified per involved partner:
- active diagnosis based on FC response evaluation to random stimuli;
- signal-based diagnosis through Adaptive Neural Fuzzy Interference System (ANFIS) approach;
- data-based diagnosis through Fuzzy Clustering approach;
- model-based diagnosis through Equivalent Circuit Modelling (ECM) approach, combined with waveform analysis for preliminary faults identification.
The status of the activities performed by HEALTH-CODE at the present stage are in line with the scheduling and the overall achievements are in good agreement with the purpose of the project and the defined objectives. The advancements with respect to the EIS board development and DC/DC converter design confirm the possibility to fulfill proper HW configuration to apply EIS measurements on-board of real systems in relevant industrial applications. Furthermore, the approach and methodology chosen for Monitoring Diagnostics Tool development proved their applicability for any fault/undesired operation due to their high generalizability. The robustness of the approach is also represented by the proposal of several techniques, which can merge the advantages of physical knowledge, retained for instance by the model-based approach, with the applicability and customization of the signal-based and data-based ones.

As further remark, the identification of a proper scaling-up technique confirmed the feasibility of transposing experimental evidences achieved in specific operating conditions and technology types to other applications, allowing the reduction of costs and time for dedicated experimental activities. Nevertheless, the knowledge of the involved physical phenomena and the use of experts’ findings represent a strategic requirement for the correct design and implementation of such algorithm. In such prospect, the HEALTH-CODE project become the exemplary paradigm of matching between industrial needs and advanced research.

The results achieved so far have confirmed the potential of building effective monitoring and diagnostic approaches along with related applications (power electronics) within affordable cost limits. These results will have an impact on stationary FCS with a direct increase in electrical efficiency, availability and durability, leading to a reduction in maintenance and warranty costs, thus increasing the customers’ satisfaction. Therefore, HEALTH-CODE contributes to the enhancement of FC competitiveness towards a wider market deployment.
Fig. 1 – Schematic representation of grid connected fuel cell system
Fig. 2 – Box with EIS board and analog fronte end communication board
Fig. 3 – Low voltage converter and EIS board at test bench being tested for EIS
Fig. 4 - Examples of EIS spectra in nominal and faulty operations