Can data and artificial intelligence benefit sustainable development?

Data is fuel for artificial intelligence (AI). It’s a precious tool that can increase business productivity and turnover... but it doesn’t stop there! Because companies are now using data to accelerate their environmental transition. Yes, AI can make businesses sustainable AND profitable! Let’s find out how.

Data and AI to improve sustainability
Using data to find contact points between profitability and environmental impact.

In 2015, the United Nations (UN) set 17 sustainable development goals to achieve before 2030, in the hope of “meeting the needs of current generations without compromising the needs of future generations”. Reducing hunger in the world, fighting climate change, consuming and producing responsibly: these objectives obviously apply to us all, but they mostly challenge large organizations, including businesses. Again, through the United Nations initiative, the organization AI for Good was founded. This platform unites researchers, associations and governments around one idea: artificial intelligence (AI) is a key asset for achieving the 17 UN sustainable development goals.

But how exactly can artificial intelligence be applied to develop a more sustainable – yet still profitable – business environment?

Mesurer son empreinte carbone

Measuring carbon footprints

The number one beneficial and money-saving way of using data is to measure a company’s energy consumption, which is the first step toward tangibly reducing its carbon footprint. “The goal is to see just how much energy is consumed, and where that energy is purchased. Once we have this data, then we can take action”, explains Benoît Lepetit, Group Chief Data & Analytics Officer at Saint-Gobain. At Saint-Gobain, this very concrete process involves setting quantified CO2 reduction targets for each plant in the Group.

Data is used for this purpose. It involves analyzing past data, or by “feeding” a predictive management system which can be used to regulate the temperature and/or lighting at the sites depending on the time of day, as well as identifying possible energy leaks.

Companies that manage their energy consumption with data analysis all end up reducing their energy spending. And those savings can really add up. Energy consumption has been reduced by 20% in a United States-based living-lab loaded with sensors and other connected objects.

Energy-saving strategies can also be applied to different machines at an industrial production site. In plasterboard manufacturing at Placo®, a subsidiary of the Saint-Gobain Group, the gas combustion process to heat the boards during the drying stage requires a lot of energy. But by fitting the machines with sensors and carrying out a detailed, real-time data analysis, energy-saving techniques were found: “The combustion can be fine-tuned using an algorithm which adapts the intensity of the dryers using gas consumption sensors”, explains Benoît Lepetit. The process is currently being tested at around ten sites within the Group and could lead to a significant reduction in CO2 consumption on every production line.


Detailed raw materials management

Smart use of data is also a useful way to achieve “decoupling”. De-what? “Decoupling” means separating the use of resources (natural, non-renewable) and energy sources from the economic performance of a company or industry. For example, applying artificial intelligence to agriculture can help improve crop yield monitoring and therefore cut down on chemicals, streamline water consumption and decrease food waste by anticipating demand and identifying expired products.


Designing sustainable products

Once the resource management process has been streamlined, companies still need to design products and solutions that are intrinsically sustainable. Here again, data and artificial intelligence prove to be valuable allies, especially in terms of taking into account the entire value chain and gaining insight on certain blind spots. “For example, if I want to spend less on transport and therefore decide to manufacture lighter products, their lifespan could be affected. In that case, my initial approach is counter-productive because the production volume will outweigh the carbon savings I made with a lighter product,” explains Benoît Lepetit.

And that’s where data analysis comes in. With this approach, a product’s entire life cycle can be visualized even before it has been produced. Therefore, the consequences of certain decisions can be understood and anticipated (such as in our example, weight reduction), up to its end of life and/or when it is recycled.


In with the right crowd

Designing beneficial products and solutions is great. Joining forces with the most “sustainable” producers, partners and service providers is even better.

To do that, companies can draw on machine learning and specific software that identifies the best social and environmental partners; those who can help render the company’s products more recyclable, make sustainable investments or monitor the environmental performance of the entire supply chain, throughout the collaboration.

Accenture confirms this finding in its study “Supply chain analytics and AI in driving relevance, resilience and responsibility” claiming that collecting and analyzing large volumes of data can improve the overall supply chain. More specifically, the consulting firm states that CO2 and energy consumption can be reduced by putting data to use in all interactions between the company, its suppliers and its customers. This allows for a more reactive and interactive value chain. Enough is enough.




Smart transport drivers

Obviously, transport is an important issue for companies trying to become more sustainable: in 2018 the sector accounted for 25% of CO2 emissions, according to the International Energy Agency. Why? Fossil fuel combustion. Here too, artificial intelligence has a role to play. By allowing more precise traffic forecasts and optimizing transport routes according to carried volume and delivery routes.

“Saint-Gobain has a large number of suppliers around the world, which means there are many possible routes,” says Benoît Lepetit, explaining that Greennav, an experimental GPS project still in development, should ultimately be able to offer the least fuel-intensive route, in real time, depending on vehicles and perhaps in the future, on loads.

Tech giant Google has already begun recommending the most “sustainable” route for each trip on its “Maps” tool. The feature is only available in the United States at the moment. For Sundar Pichai, Google CEO, the initiative is set to save 1 million metric tons of carbon dioxide every year and would have the same effect as removing 200,000 cars from the road. Smart transport using data




Improving waste management

Manufacture, transport, use, and then what? In a product’s lifecycle the highly sensitive issue of waste is inevitable. The waste we produce, the waste we would like to stop producing and – fortunately – the waste we recycle and reuse. According to Benoît Lepetit, “the risk of the industrial process is to produce waste that cannot be reused”. Machine learning, fed by data and “equipped” with robotic arms, can be used to visually evaluate a chain of waste to identify recyclable elements and work out their volume before removing them directly*.

At Saint-Gobain, through the recent takeover of construction chemical specialist Chryso, some waste can now be reused to produce “green” cement with innovative additives. What’s the link between artificial intelligence and data? Real-time monitoring of the product’s reaction, to make it as efficient as possible. This applies to the Maturix sensors by Chryso. When inserted into this low-carbon concrete, they measure its temperature and identify the ideal moment to unmold it.




Prioritize accessibility & data governance

Data helps find contact points between profitability and environmental impact, which are not necessarily easy for everyone to access.

“That is why a policy for collecting, aggregating and making data available must be put in place within the company”, explains Benoît Lepetit. Another obstacle to the use of artificial intelligence: governance. The data is only reliable if it has been entered and tracked (where, when and how the figures were collected). And finally, last but not least: the figures must be understandable. “As human beings, we have to connect these numbers to the truth they tell, what they mean”, points out Benoît Lepetit.

It’s an important issue, because artificial intelligence allows us to consider many more options that are beyond the scope of the human brain. “AI is able to find solutions that fit with physical realities, stakeholder requirements and sustainability goals while respecting the financial constraints of a project”, said Janne Liuttu, Chief Data Scientist at the Norwegian manufacturer Rambøll, during an interview with Hyperight magazine.

Having said that, Benoît Lepetit emphasizes that data is not and should not be the only way to move toward a cleaner planet. “The challenge is so great that other tools should not be overlooked, we shouldn’t be blinded by technology. So, we need to go making progress on this subject, while optimizing our knowledge and our chances of success.”


* How Machine Learning and Robotics are Solving the Plastic Sorting Crisis / Plug and Play


Photo credits : Bigone/Shutterstock and NicoElNino / Shutterstock

The 7 pillars of an efficient supply chain

Materials transport: on the road to sustainability