August 17th, 2023
The RTDT Research Affiliate Programme space - helping to improve data sharing in wind energy
RTDT Laboratories is an ETH Zurich spin-off developing a universal tech stack for Structural Health Monitoring as a service of wind energy assets, for cloud and edge. The RTDT Research Affiliate Programme brings together researchers with real data and WeDoWind Challenges from the industry in order to create state-of-the-art open-source models and analysis tools for wind turbine Structural Health Monitoring. Recently we were able to use the interim results of the first challenge to contribute to a Research Data Alliance grant project aiming to develop recommendations for improved data sharing in wind energy.
Creating and adopting a set of recommendations for improving cross-disciplinary FAIR data sharing in wind energy
The goal of this project, funded by an RDA cross disciplinary science adoption grant that has just been completed, was to create and adopt a set of recommendations for improving cross-disciplinary FAIR data sharing in wind energy. We did this by investigating how existing RDA outputs could be applied to a challenge in the RTDT Research Affiliate Programme WeDoWind space.
The goal of the challenge "Challenge: Field Applications in Structural Health Monitoring for Wind Turbines" is to evaluate and benchmark various analysis methods to solve the following monitoring use cases:
- Operational Modal Analysis to identify the modal properties of the wind turbine
- Identification and forecast of damage/failure of the collective pitch drive
- Identification of aerodynamic imbalance
Participants could engage to solve one or several of the use cases listed above.
FAIR data preparation
For the three use cases, separate data sets were provided by RTDT Laboratories AG. We evaluated the datasets and provided metadata with the RDA FAIR Data Maturity Model.
The results of the evaluation indicated low scores across all four criteria, with lowest overall levels in Findable and Reusable sections. This allowed us to make some improvements, including uploading the data to the Bridge of Knowledge repository and creating our own draft schema to describe the sensor specifications and placement (encoded as JSON and YAML schemas), following and slightly modifying the WindIO wind turbine ontology. Many other indicators could not be tackled due to the lack of community standards, or the absence of public data sharing services catering specifically to the wind energy context.
The data sets can be found here:
As well as this, we provided a Colab notebook to read the HDF5 files in the datasets and create Python dataframes or even export it to .mat files. We also provided guidelines for preparing the code based on the RDA FAIR Principles for Research Software on an open GitHub repository, where the participants were instructed to upload their results.
The interim results
We carried out a launch webinar in March 2023, an interim webinar in April and a final webinar in June, with a total of eight participants. During this time period, we answered questions of the participants and clarified any problems. For the final webinar, we got a total of three solutions submitted on the platform. The challenge is continuing in September!
Some initial data exploration plots from the participants are shown below.
However, the main results so far concern the participant survey, which we designed and implemented in order to assess the effectiveness of the “FAIR Data Maturity Model Excel tool” on data sharing in the WeDoWind environment. Although no quantitative comparison could be undertaken, data sharing was significantly improved compared to previous projects, for which the data had not been prepared using the FAIR Data Maturity Model. In these previous projects, a significant amount of time was spent by the participants in preparing and understanding the provided data.
We found that providing structured metadata was very useful for the participants, allowing them to save significant time. This should be done in future challenges in a similar fashion. As well as this, it was found that the participants preferred to use the YAML metadata rather than the JSON metadata. This should be considered in the future. Furthermore, the ontology schema was not used, showing that adopted ontology was sufficiently intuitive and was easily interpretable by domain experts without the need for aditional information provided by the schema annotations. However, the rich metadata should be further enhanced in the future, by providing more context, for example by describing the environmental and operational variables in more detail so that they are understandable for non-wind turbine experts.
Using the results to improve data sharing
This work allowed us to document the challenges and concerns that arose during the implemention of the FAIR Data Maturity Model in the context of sharing wind energy domain data. We will use these results at the next RDA Plenary event in October, as part of a dedicated wind energy session, to work together as a community to improve data sharing.
More details are available on request.
Continuing with the RTDT Research Affiliate Programme space
The RTDT Research Affiliate Programme space is continuing in September 2023. We are still working on solving the "Challenge: Field Applications in Structural Health Monitoring for Wind Turbines". You can sign up to the space using the button below and selecting the RTDT space in the resulting form. We hope to see you soon!
Join this space
More about RTDT
RTDT is a No-code SaaS for Structural Health Monitoring of wind turbines, founded in 2022 as a spin-off of ETH Zurich (Chair of Structural Mechanics and Monitoring).
Converting data into actionable insights that deliver value can be a challenge. This is where RTDT with an end-to-end intelligence layer comes in, which is both robust and scalable while ensuring domain specificity. RTDT thus provides you with an easy way to access deep structural health assessments on your wind turbines’ components regardless of your skill set and expertise.
Got to RTDT's website