Big data & the power of analytics

Manufacturing executives from across the UK joined special guest, Roy Haworth (Airbus), Henry Anson (The Manufacturer), as well as Sue Scorer and Jen East from  FICO, at the stately environs of The Gilbert Scott in London, to discuss the capabilities of big data and business analytics, industry’s uptake so far and the implications accompanying that transition.

The topic of big data and the power of analytics produced a varied and insightful discussion around the table. It was highlighted that manufacturers could be at the forefront of advanced analytics adoption due to the significant amount of data produced in their operations.

Diners were keen to understand what their manufacturing peers were looking for in data analytics, as well as, the potential challenges and benefits in this rapidly evolving landscape.

Speaking volumes

Haworth began the conversation by sharing Airbus’ biggest challenges – the management of quality, efficiency and making improvements. Haworth openly admitted that he and his colleagues have an “obsession” with reliability and quality, a concern that stems from – regardless of how Hollywood portrays the industry – the final destination of the firm’s products. For example, the Hubble telescope is 250 miles above the Earth, a little too far to travel if something breaks down.

Unsurprisingly, the level of testing necessary for such products creates monumental levels of data. Airbus’ space crafts have a life expectancy of 15 years within extremely hostile environments, and vast amounts of data are produced when testing in these surroundings.

He said, “Our challenge is that most of that data is gathered independently in our supply chain. We used to be able to afford lots of people to analyse this data, now we need to find ways of analysing it much more efficiently.”

Aside from testing data, information is generated in the products the company builds. Haworth provided Earth Observation Satellites as an example. He said, “In each revolution, the satellite produces roughly one terabyte of data, it completes 16 revolutions a day for 365 days a year and probably lasts between five and seven years.”

Individuals around the table agreed that it is very challenging to decipher the value of data in large volumes, regardless of sector. FICO’s Jen East highlighted the importance of understanding the quality of the data available and taking that into account when modelling a strategy to harness the power of that information.

John Dyson agreed, he said; “The pharmaceutical industry is very similar. The R&D that accompanies the creation of a drug, test after test, how are we going to deal with billions of terabytes that we need to collaborate, with companies that we need to share? There’s a practical problem there.”

Richard Wilkinson, who operates in the same industry, on a smaller scale, concurred, and referred to the struggles of ‘niche manufacturing’. He explained, “Some of our drugs are niche and trying to remain competitive in a highly regulated industry can be challenging.

Mike Osborne weighed in. He said, “We are not mass manufacturers and we are still doing things very slowly and carefully, in much lower volumes.

“The amount of data that has got to sit behind that. The stakes are really high. If you get something in the sequence wrong, there’s a qualification and test regime you have to go through, and in our business that’s a mighty expensive mistake to make if you have to go back and requalify something.”

The supply chain and beyond

Osborne went on to discuss the further data collected within the supply chain and how MBDA keeps track of it. He said, “Our suppliers have got to be in synchronisation with us.

“We employ supplier development engineers that work with suppliers, to prove their capability to meet our requirements.”

Jose Lopes supported this and said, “You are only as robust as your supply chain.” Lopes continued, saying that in the automotive sector it is essential to achieve “repeatability”. To do this he described the data as “critical”.

He elaborated, “For us, it is about compatibility between the systems in the very deep supply chains. In an industry centered around testing and development, we have moved from physical testing through to virtual testing, and that’s becoming more and more difficult. When you are constrained by the supply chain, how do you translate all the data within it?”

Russell Coding agreed and explained that where a company is operating in multiple sectors the levels of testing are increased and in turn, the data, and this is “critical to how we are creating value in that market”.

Cyber secure

From here, the conversation moved to data security, where Nigel Vaughan discussed QinetiQ’s Big Data and Analytical Cyber Department, which evaluates the vulnerability of big data. He said, “On behalf of government, the team hacks into various big data centres, uncovers weak spots and works out how to protect them.”

Vaughan outlined several areas where data can be hacked, he cited environmental systems, smart buildings and smart cars as the most exposed places. He underlined a “serious lack of regulation in regards to big data” and added that QinetiQ was involved with government to address this problem.

Andy Schofield muted the hype when he suggested that big data was not the answer to all obstacles facing the sector. He said, “There are a number of elements in manufacturing strategy: digital technology and big data, skills, culture and training. Big data is just one element, Industry 4.0 is not going to be the answer everything.

Guests agreed that while the advantages of big data were set to have a massive impact on the competitive edge of manufacturers, the collection of such data was only the very beginning. It was underlined that although it is possible to collect masses of data, the true challenge comes when distinguishing and analysing useful data to harness its benefit.

Sue Scorer of FICO said, “Data, big or small is not a solution. The power of that data comes when you are able to take the patterns and relationships discovered in the data and model them to predict events, moving from ‘what has happened’ to ‘what could happen’ – predictive analytics. But where the biggest business potential lies is prescriptive analytics or optimisation – determining the best decision to make, thus moving from ‘What could happen’ to ‘What should we do?’”