Electrolyser Degradation Estimation via Fleet Bayesian Inference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for quantifying degradation in PEM electrolyser systems are inefficient, as they require sporadic, complex, and time-consuming polarization curve tests, which are difficult to standardize and often impractical for on-site maintenance.
Innovation Solution
An empirical model and degradation reference figure are introduced, utilizing a fleet-based approach that aggregates degradation history from multiple systems to estimate the degradation of a target system, employing statistical inference and Bayesian methods for accurate and continuous degradation quantification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If polarization curve tests are performed to quantify degradation, then degradation measurement is achieved, but the process becomes time-consuming and complex
Solution Approach 1:
The patent uses a fleet-based approach where degradation data from multiple electrolyser systems are aggregated to create a reference degradation model. This model serves as a copy or representation of the degradation process, allowing degradation quantification without performing time-consuming polarization curve tests on each individual system. The reference model captures the essential degradation characteristics that can be applied across the fleet.
Solution Approach 2:
The patent performs preliminary degradation testing on a representative subset of systems to establish the reference degradation model before applying it to the entire fleet. By conducting comprehensive tests in advance on selected systems and using statistical methods to generalize the findings, the approach eliminates the need for repeated time-consuming tests on all systems throughout their operational life.
2Reliability
If polarization curve tests are conducted frequently to track degradation continuously, then continuous degradation monitoring is achieved, but the complexity and resource requirements increase significantly
Solution Approach 1:
The patent merges data from multiple electrolyser systems into a single fleet-based reference model. By combining degradation information across the fleet and using statistical aggregation, the system achieves continuous degradation monitoring capabilities without requiring complex individual system testing. The merged fleet data provides more robust and reliable degradation tracking than any single system could provide alone.
Solution Approach 2:
The patent implements a feedback mechanism where the reference degradation model is continuously updated and refined based on actual measured data from the fleet. This feedback loop allows the system to maintain high reliability in degradation monitoring while keeping the testing procedure relatively simple, as the model self-corrects and improves over time using accumulated operational data.
3Ease of operation
If standardized polarization curve tests are performed on-site at customer sites, then on-site degradation assessment is achieved, but it becomes impractical due to availability requirements and test standardization difficulties
Solution Approach 1:
The patent creates a virtual copy of the degradation assessment process through the fleet-based reference model. Instead of requiring physical presence at customer sites for standardized testing, the system uses the reference model to assess degradation remotely based on operational data. This copying approach maintains assessment accuracy while eliminating the practical difficulties of on-site test standardization and customer availability requirements.
4Adaptability or versatility
If degradation measurements are taken under varying operating conditions, then real-world degradation is captured, but measurement comparability and reliability decrease
Solution Approach 1:
The patent explicitly accounts for parameter changes by incorporating operating conditions (temperature, current, time) as variables in the reference degradation model. The model is designed to normalize and compare degradation across different operating conditions, allowing the system to capture real-world variability while maintaining measurement consistency through statistical standardization and fleet-wide aggregation.
Data Source
Figure 1
Figure 2~3
Figure 4~5
AI summary
A method of determining a degradation of an energy system (1, 1'), such as an electrolyser system, is presented. The method comprises the steps of, (i), providing an empirical model (M) of a degradation reference figure (Uref), wherein the reference figure (Uref) is suitable to estimate the degradation of a given system (1) out of a set of similar energy systems (1, 1'), and wherein at least one unknown model parameter (Cn) is fitted, (ii), providing fleet information (F) from the set of systems (1, 1') to the model (M), wherein fitted model parameters (Cnm) of a plurality of the systems (1, 1'), each parameter (Cn) comprising an evolution over time (t), are aggregated to form a fleet-based parameter (Cn'), and, (iii), matching the model (M) with a target energy system (1) via a statistical inference, particularly a Bayesian inference, by using the fleet-based parameter as prior probability distribution (Pr), wherein a degradation figure (U) of the target system (1) is calculated. Moreover, a related computer program product and a related apparatus, are provided.