Overview of the MAST3RBoost project

MAST3RBoost project has been up and running for 9 months already. Enough time for our partners to lay down the foundations for the development of the project’s innovations and to substantiate our decision-making processes in this initial stretch. But what exactly has each one of us been up to?

The prof. Mokaya’s group from the University of Nottingham made available an impressive dataset of up to 80 different adsorbent materials from their existing library, that our colleagues from Nanolayers used to train for the first time a supervised Machine Learning tool. As a result, they were able to predict key properties of the material narrowing down the uncertainty to <16 % for key properties such the BET area (a measure of the available surface for H2 adsorption inside the material), and with a little less precision (18-23%) for other key properties such as the volume of micropores and elemental composition. Lessons have been extracted in order to build a much more robust database during the initial 24 months of the project.

Our supervised Machine Learning showing how it can predict some outputs with lower error than expected – a proof of actual learning!

Our experts from INCAR-CSIC, led by Dr. Marta Sevilla, joined forces with Nottingham to agree on a harmonized methodology for the performance of tests and for the analysis of the isotherms. Now we can be confident that independent labs and different researchers will reach the same values for a given material. Furthermore, the team at Nottingham has also been exploring the adaptations needed to measure hydrogen adsorption + desorption at 2 different temperatures? Further updates will be provided in the future.

CSIR and Envirohemp worked together in the definition of the criteria (typically 10-12 independent parameters) to assess the scalability of the different lab-scale recipes for the production of the different adsorbent materials.

They also included threshold values to assess “compliance” with each criterion. Aspects such as yield, input ratios of chemicals/solvents (kg used per kg of product), reusability of secondary streams among others were assessed. This resulted in a priority list of the best “industrial-friendly” methods to be explored in the lab by researchers of the University of Pretoria, the University of Nottingham and INCAR-CSIC. In fact, some of the lab-scale methods will no longer be pursued due to the poor scalability scoring achieved.

Under the supervision of Dr. Rosalina Pérez from the Coatings and Surface Treatments Unit at CIDETEC, a team of experts that included with representatives of partners LKR, TWI and CIDETEC made a comprehensive literature review of all the literature available in relation to the alloys of Aluminum and Magnesium, the 2 candidate families of materials for the hydrogen storage vessel. The review also included the literature on the hydrogen barrier coatings that can be selected to reduce hydrogen exposure. Particular focus was put into assessing the compatibility of all the materials with the use of hydrogen and cryogenic temperatures. Promising candidates include Aluminum alloys as material of construction, and epoxy resins as barrier. The project partners will now work in generating the missing evidence for final qualification of the materials, such as the impact of the uneven microstructure generated by the additive manufacturing techniques when casting the vessel’s walls and the adhesion of epoxy coatings to this type of surface.

Finally, a number of MAST3RBoostspecific tools are currently in the making by other consortium members, including: the adaption of mechanical testing to better recreate cryo-conditions (TWI), the layout of the pilot vessel and test rig (SPIKE) and the definition of the sustainability assessment methodology (Contactica).

Follow us on Twitter and LinkedIn for more MAST3RBoost updates!

Leave a Reply

Your email address will not be published. Required fields are marked *