Less than five per cent of data generated by manufacturing plants is used to bring insights and improve operations, according to a study by Frost & Sullivan. Sensors and other wireless devices in manufacturing plants do not produce sufficient data to make a difference, it finds.
The problems for manufacturers run deep, it seems; Frost & Sullivan points to improper data management strategy, limited application expertise, missing data, and a sheer lack of resource availability. It references process industries in particular, including oil, chemicals, natural gas, power, iron and steel, and pulp and paper.
But the research company reckons half of all companies engaged in process manufacturing will invest two times more in analytics over the next two to three years than they have to now.
They will ‘co-relate’ different types of data using combinational analytics, in a system called ‘synaptic business automation’. This way they will be able to make more of the data at their fingertips, it said.
“The blurring of traditional automation boundaries is steering the development of innovative business models. Edge computing platforms are resulting in the democratisation of analytics and near-real-time interfaces with sensing systems,” said Muthuraman Ramasamy, automation and IIoT industry director at Frost & Sullivan.
“The industry understands the imperatives of digital, but the challenge resides in the ‘how’ of digital. This will require customers to partner with accomplished domain experts who can not only help structure a digital roadmap but also have strong AI application capabilities over plant data and comprehensive expertise over a manufacturing value chain.”
Frost & Sullivan has authored a white paper with Yokogawa, a Japanese electrical engineering and software company, with businesses based on its measurement, control, and information technologies, which sets out a “closed loop from data extraction to value creation”.
More than 90 per cent of plant data contains ‘noise’, said Frost & Sullivan, which obscures patterns in the raw data. The paper recommends manufacturers focus on the most relevant data and understanding the principles and algorithms to interpret data, integrate as far as possible data from both internal and external elements, and build the relationship between business and domain knowledge.
