Skip to main content

ImplementatioN in real SOFC Systems of monItoring and diaGnostic tools using signal analysis to increase tHeir lifeTime

Periodic Reporting for period 2 - INSIGHT (ImplementatioN in real SOFC Systems of monItoring and diaGnostic tools using signal analysis to increase tHeir lifeTime)

Reporting period: 2018-08-01 to 2019-12-31

The INSIGHT project aims at developing a Monitoring, Diagnostic and Lifetime Tool (MDLT) for Solid Oxide Fuel Cell (SOFC) stacks. The project implemented prediction methodologies and demonstrated their effectiveness by on-field tests on a real micro-Combined Heat and Power system for residential applications. To achieve a cost-efficient and robust solution, 2 advanced complementary techniques have been exploited, Total Harmonic Distortion (THD) and Electrochemical Impedance Spectroscopy (EIS, both sine and PRBS excitation), in addition to conventional Stack Signals Analysis.
These methodologies already developped for PEM fuel cells have been adapted to SOFC in the project.
INSIGHT outcomes opens the perspective to decrease the costs of service and SOFC stack replacement by 50%, which would correspond to a reduction of TCO by 10%/kWh by prolonging SOFC lifetime by 5%, increasing availability by 1%. These, within the limit of adding less than 2% to the total SOFC system’s cost.
Obj 1: To implement an advanced Monitoring, Diagnostic and Lifetime Tool to prolong SOFC lifetime and increase availability.
The most critical faults have been defined: fuel starvation, carbon deposition and gas leakage, as well as a test protocol and the testing matrix. The test matrix has been completed with short stack testing campaigns. Finally, a 32-cell stackbox has been tested, with the fuel starvation protocol.
EIS was found to be an adequate technique to identify the fuel starvation and carbon deposition fault, at stack level. On the contrary, for gas leakage, it was found to be efficient at the RU scale. However, the signature of the defect was unfortunately lost through averaging at stack level. EIS diagrams extracted from PRBS excitation are well superimposed to those obtained from sine excitation, meaning that PRBS is a valuable technique in order to obtain faster EIS results that do not disturb too long the stack/system from its setpoint.
THDA method is sensitive to highlight high FU and to identify a weak component, the method is therefore a suitable indicator to enhance reliability.
Classical signals remain a useful tool to see a deviation of a stack/system from its standard operating conditions. It was particularly useful for the case of a leakage, in association with some specific deviation trials as compared to the nominal operating conditions.
An identification algorithm for EIS measurements metrics extraction based on Equivalent Circuit Model approach has been developed (UNISA patent) and validated.

Obj 2: To develop the hardware for the implementation of advanced Monitoring, Diagnostic and Lifetime (MDLT) algorithms on real SOFC system with low cost (less than 3% of system cost).
For the implementation of the diagnostic methods on board, the development of the firmware architecture and parameters configuration of the BitronBox device have been completed. Both BitronBox hardware and software have been completed. Each diagnostic method operates in two steps, the first one is devoted to the stimulation and data acquisition. The second one is dedicated to the data analysis for extracting the SOFC features. The functions for the data analysis have been developed for running on a high performance linux boards.
A 2nd version of the BitronBox embedding the MDLT tools has been released. 7 were produced.
In order to permit THD, EIS and PRBS measurements, the DC/DC converter integrated in SP commercial SOFC system has been modified and installed in the system used for the on-field test in WP6.
According to SP and BITRON data on system and board costs, the share of the MDLT tool is below 2% thus reaching the targets set.

Obj 3: To identify control actions able to mitigate the impact of both degradation mechanisms and faults on performance and durability of SOFC.
A diagnostic algorithm for Detection and Isolation of faults based on EIS was developed.
It is fast (<5 s). When one of the features is different for that expected under nominal conditions, an alarm is generated, and a detection and fault isolation process starts.
This method has been validated upon experimental data from WP2 but also during on-field tests.
In addition, fast models capable of predicting cell/stack degradation have been developed for different degradation phenomena. Several lifetime prediction algorithms were developed.
CEA Expected results: Local current monitoring. Implementation of EIS/PRBS of a short stack test in order to use it as a diagnostic and lifetime evaluation tool. Perform THD analysis
Results: Local current measurements done on a classical cell + identification of limit conditions
Fuel starvation test with EIS/PRBS measurements implemented every 6 h during 5000 h.
Benefit of THD analysis to detect fuel starvation fault and to identify the weak component of the stack.
Test of 32-cell stackbox

DTU Expected results: Implemented set of diagnostic tools for stacks/systems into a practical device.
Results: Carbon deposition and leakage detection measurements conducted on a 6-cell short stack.

UNISA Expected results: Monitoring, Diagnostic, Lifetime, Mitigation tools and HW & firmware engineering for on-board use.
Results: Algorithm for fault diagnosis developed and validated. Mitigation strategies defined. estimation of remaining useful lifetime done.
New firmware for installation of the MDLT tools on the Bitron box and for interaction with DC/DC converter.

EPFL Expected results: Addition of new diagnostic tools on existing instrumented advanced stack test benches for validation and Improved lifetime prediction tool calibrated with spatially resolved measurement data, e.g. signal response from a SRU with a known local defect introduced on purpose.
Results: EIS/DRT/resistances quantification.

IJS Expected results: PRBS-based diagnosis and prognosis
Results: PRBS probing entirely validated.
Prognostic algorithms built.

BITRON Expected results: MDLT HW device.
Results: new release of the BitronBox was produced in 7 samples. Bitronbox installed on the real system and validated with the in-field test.

SolidPower & HTC Expected results: Small size high efficiency micro-cogenerator SOFC based systems, for small commercial and residential markets.
Results: Modified DC/DC converter for the project done and validated.
MDLT hardware and the excitation and diagnostic routines validated during tests where fuel shortage was simulated, indicating out-of-spec operation beyond 68% fuel utilization. Both PRBS and sine excitation found to give good quality spectra.

VTT Expected results: Procedure for developing MDLT algorithms. Computational hardware for SOFC systems
Results:VTT’s ability for reliable and good-quality high-frequency measurements improved significantly. This includes better understanding and know-how in all levels of measurements, meaning from enhanced wiring to better usage of measurement sensors, devices and equipment

AVL expected results: Algorithm Validation, HW Validated, SOFC-test-station designed
Results: two patents. THDA-Algorithm implemented and tested on real system.
Test station designed and currently commissioned
Bitron Box
SOFC System