DiagRAMS Technologies has been working for a year and a half now with Storengy on an industrial maintenance project for natural gas storage. The device set up concerns the RKs, equipment that regenerates the TEG (Tri Ethylene Glycol) necessary for the dehydration of the stored gas.

The aim is to detect any drift or anomaly in order to anticipate a breakdown and monitor energy consumption. Antoine Boudehent, digital and industrial project manager for Storengy, gives an initial assessment.

What were the challenges for Storengy in setting up this system?

Antoine Boudehent –  We wanted to be able to understand and explain failures on very sensitive equipment such as compression and RKs in more detail. Better still, it was a question of being able to anticipate them thanks to predictive maintenance, which warns of any change in the equipment’s operating mode based on weak signals. There was also an ecological issue which is extremely important for Storengy. We have a strong presence in renewable energy and biogas and we try to ensure that our facilities have the lowest possible carbon footprint by reducing methane emissions. The system set up with DiagRAMS responds to this logic in order to optimise the performance of equipment, extend its life, and reduce and avoid malfunctions that cause excessive energy consumption. 

How did the project and AI address this?

A.B. – It was not necessary to install sensors that were already present on the equipment. The approach taken with DiagRAMS was to complement our process knowledge (linked to physics) with a new approach based on data collected in real time. We then determined together the parameters that could anticipate any malfunction, and then created a model with artificial intelligence to predict the opening of a fundamental valve in the process. The slightest deviation between the model’s prediction and reality alerts us to a change in operating mode.

How would you assess this first for you?

A.B. – This first season was successful as the model worked very well and will allow us to think calmly about a possible change in the strategy for using RKs according to these parameters. DiagRAMS was particularly attentive to our specificities and enabled us to bring experts and data scientists together around the table. Together, we tried out ways, ruled out dead ends and moved forward to achieve such a result. It is on this basis of exchange between experts from two different worlds that we have succeeded in identifying the influential parameters and thus elaborating the relevant models and then refining them. When we presented this success story to our management and maintenance teams, they were amazed at the results and more confident in the future knowing that they could now anticipate breakdowns.

In view of these convincing results, how do you see the next steps?

A.B. – We want to go further by working on DH, gas dehydration, which takes place upstream of RK. We want to understand the most influential parameters so that our settings are a little less conservative and thus gain in efficiency, and therefore in industrial, energy and environmental performance.

Do you see artificial intelligence as being of increasing importance in the future in the industry?

A.B. – I think we have to start now to move forward with the technologies and build the future. Above all, artificial intelligence is really complementary to a process approach. It is particularly powerful because it allows correlations to be found in a very short time outside of physical equations!