Air Liquide improves hydrogen service thanks to innovative augmented intelligence project

Darwin: Increased efficiency and more rapid response
Large Industries
24 February 2020

Air Liquide is opting resolutely for a customer-centric approach as it works on digital transformation projects. Darwin - an augmented intelligence project taking customer need as its starting point - is located at the intersection of the two...

“Darwin is a fascinating example of how digital transformation can be deployed to work for our customers,” says Vincent Esbelin, Digital & Data Science Manager at Air Liquide. “In essence, it’s an augmented intelligence project around the collection and processing of large amounts of data allowing us to create added value for our major industrial customers.”

“Darwin helps us deal better with the uncertainty of market demand and all its associated complexity,” he adds. “The point is that many factors and possibilities have to be weighed up over and over again within a short period of time. Darwin helps us by providing statistical indicators and making well-founded predictions. It takes out the guesswork, you might say.”

What do you mean, customer-centric?

The Darwin model starts from the accurately estimated hydrogen demand of the customers. Next, Darwin examines how best to translate this demand into the most appropriate production offer, and then Darwin suggests the best possible configuration to achieve it optimally.

“We use the data flow we receive via the sensors in our network to fully model the network. There are millions of sensors measuring, among other things, the pressure, flow and temperature of the gases.”

“The data are used to ensure we run our production sites as efficiently as possible, both economically and in terms of sustainability. Predictive maintenance periods are also taken into account. All this helps us improve the efficiency of our installations and shorten the response time for when market demand unexpectedly increases.”

Humans versus Darwin

“Given the limited amount of time we have to make decisions, the data flow is too large to be processed efficiently by humans. Experienced specialists can do it, but with Darwin they can free themselves to innovate, such as by improving the tools and identifying value for customers. What’s more, our people now have more time available for one-to-one customer interaction.”

“Darwin is designed to support human decision-making - not replace it - and provide significant added value to our customers.”

Augmented Intelligence rather than Artificial Intelligence

“Since Darwin relies largely on the human knowledge of our experts, I view it as an augmented intelligence model. It also means that less experienced employees can make full use of the experience of their more experienced colleagues. In this way we can work much more efficiently as a team.”

“Our customers are now also passing on the forecasts of their hydrogen demand, and the amount of data is constantly growing, making the models increasingly reliable.”

The added value of Darwin

“Darwin is a kind of single source of truth, which can be safely followed by all internal stakeholders. This way, there is immediate consensus, and the decision-making process can be quicker. The Darwin model also helps to ensure that we can deliver what we promise. In this context, we also carry out stress tests to make sure that any technical problems can be dealt with. The model allows us to switch to plan B if necessary, and even to plan C!”

“The introduction of Darwin has been very successful and has allowed us to win new contracts. This is because Darwin gives us the opportunity to sell any surplus capacity in our production. This could be additional demand from our customers on top of the contractual agreements, or possibly quantities of hydrogen that we can sell proactively.”

The outlook: bigger, more and better

“So far, we’ve only used Darwin for our hydrogen network, because that’s where the added value is greatest. Initially, Darwin will only be operational in the Netherlands and Belgium but, in time, the project will be greatly expanded.”

“In the near future, Darwin will be introduced in other domains and will also be used to support our energy transition plan. The model’s margin of error will also be further reduced and the response time will become shorter.”

“We’ll also investigate the extent to which we can link our systems with our customers’ systems. This could offer additional possibilities, such as automated bidding. In such a mutual value co-creation, we’re looking to share flexibility that will offer added value for both parties.”