Technology

When Artificial Intelligence paves the way to "Augmented Labs"

Mar. 4 2019

With over 400 testing laboratories around the world, Bureau Veritas is a global leader in testing, inspection and certification. From hydrocarbons to mineral ores, consumer products and food processing – Bureau Veritas experts control, analyze and test all types of components on the market. Today, Artificial Intelligence (AI) is gradually being installed in Bureau Veritas laboratories. New testing methods and productivity gains are heralding the era of “test 4.0”...

“Charles” – the first steps of AI at Bureau Veritas

Laboratory tests comply with very specific protocols and are mostly repetitive. Three years ago, experts from the Bureau Veritas laboratory in Atlanta, United States specializing in “Oil Condition Monitoring”decided to revolutionize their process by adopting Artificial Intelligence. Named “Charles” in homage to a lab expert who had retired after 40 years of service, the program was designed to free expert analysts from the simplest and most repetitive tasks and allow them to spend more time on high-risk samples in order to issue recommendations with greater value-added.

The Atlanta lab performs approximately 400,000 tests each year. Its mission is to help companies assess the level of potential risks posed by engines within their industrial fleet, for example mechanical diggers, forklifts, trucks, rotating machinery, etc.

The tests carried out on lubricant samples taken from these machines allow a team of experts to assess the risks, on a scale from “normal” to “critical”, in order to support clients in their predictive maintenance approach.

Three million results and 15 years of memory

Operational since mid-2018, Charles first had to analyze a large amount of data to build an efficient calculation model (about three million results corresponding to more than 15 years of testing and analysis data). To do this, Bureau Veritas worked closely with teams from Microsoft and its Azure Cloud platform. The result was a leap in productivity for the teams and the lab. In fact, Charles releases experts from 20% of their repetitive analysis tasks, giving them more time to focus on the most complex cases – particularly the 3% of cases identified as “critical”.

Charles now allows us to optimize our experts’ working capacity!” says Michel-Ange Camhi, Group Chief Data Officer at Bureau Veritas. Eventually, the aim is for Charles to offer new on-demand services, such as the option of online consultations, in order to meet the specific needs of Bureau Veritas’s industrial customers.

Ore samples – a new test process to create a “digital twin” of the mine

On the other side of the globe, in Canning Vale near Perth, Australia, an unprecedented large-scale initiative was launched three years ago in the mining sector, called the “Minerals Machine Learning” project. It is using Artificial Intelligence to accurately estimate the mineralogical composition and chemical and physical properties of ore and mineral samples.

“With a simple infrared spectrometry test, AI can produce analyses for a much lower cost than previously," says Dr. John Carter, Global Technical Manager at Bureau Veritas in charge of the project.  

In the mining sector in particular, costs can quickly prove prohibitive, whether in relation to exploration or exploitation. Since the cost/benefit ratio is sometimes unfavorable, some of these tests are excluded from the process by some industrial companies.

In these cases, AI ​​opens up new perspectives, both in terms of productivity gains and coverage of new risks. For example, it makes it possible to estimate the hardness of a fragment of ore simply by analyzing a powder from it.

The method developed in the Bureau Veritas laboratory in Canning Vale, which is based on AI, will make it possible to create “digital twins” of mines based on historical test data.

This significant first step reveals the extraordinary potential of the "Minerals Machine Learning" project.

Rolling out AI worldwide

Bureau Veritas aims to replicate and roll out this AI expertise to all its laboratories worldwide. This ambition has been promoted by the creation of the Data Lab team, whose mission is to support and develop all projects in the field that use AI. This team will benefit from the technical and commercial partnership established with Microsoft for use of the AI resources offered by the Azure Cloud platform.

Eventually, the aim is for all tests offered by Bureau Veritas to benefit from the contribution and power of AI – in whatever business sector the Group’s clients happen to operate.