As organisations look to derive more value from all their assets -- including their data -- advanced analytics techniques, the ability to handle "big data" and improved reporting and business intelligence (BI) capabilities are becoming increasingly critical to business success.
For more on advanced analytics in manufacturing
Tips on manufacturing BI and supply chain analytics from SearchManufacturingERP.com
A quick reminder on how data analytics is best defined
Alas, manufacturing analytics often comes up short
For example, Denmark-based wind turbine manufacturer Vestas Wind Systems is using analytics tools in an effort to give its business operations an edge over rivals.
Situating wind turbines is about more than raising finance and winning governmental planning permission. The location of wind farms is an increasingly science-based activity, with developers modeling not just where to put the farms but also how to arrange each individual turbine to maximum effect.
From working out how to maximise electricity generation and minimise maintenance costs, power companies need to look at wind patterns as well as rainfall, humidity and atmospheric pressure. And they need to model the airflow across turbine blades to calculate the best way to position each turbine tower.
"People buy turbines not because they are beautiful, but because they want a return on investment," said Lars Christian Christensen, a vice president at Vestas. "Our job is to make sure the buyer has the right site, but also the optimum layout for the turbines."
Using advanced analytics software and other tools to optimise wind turbine locations allows Christensen's team to "improve the output of the power plant by several percent," he said. "You obtain a better output for the same cost. We use software modeling and tools to provide a better business case assessment and more certainty."
‘Hindcasts’ -- weather models with hindsight
The scope of data investigated by the Vestas team is vast. The company has maps of the world, down to 3 km by 3 km squares on a grid, and can call up wind, temperature, solar radiation and humidity data from the present back to the year 2000 on an hourly basis. "That allows us to create a global, historical view," Christensen said.
The data is used to generate "hindcasts," or models of what weather conditions would have been at prospective turbine sites, and the electricity that would have been generated by different combinations of turbine locations.
Methods and tools employed as part of the process include computational fluid dynamics, database flow models, and computational and analysis tools such as IBM's InfoSphere BigInsights, a Hadoop-based technology for analyzing large volumes of both structured and unstructured data.
"Our big data challenge is that we have a library that runs to 3 petabytes," Christensen said. "If you can't retrieve that information from storage, it serves no purpose." Maintaining such a large database and the tools required to analyse it is essential for a global operation, like Vestas: not everywhere that wants green energy has a large flat space in which to build a wind power plant.
Advanced analytics enables Vestas to tell if a particular location would be a good site for wind turbines, Christensen explained. "Sometimes we help customers find undiscovered resources," he said. "The data we have is the digital equivalent of a map of gold mines."
Blinds maker gains visibility into sales data
Hillarys Blinds is a UK manufacturer with 27% of the UK market for made-to-measure blinds and other windows coverings. The company operates a direct sales model, with 1,000 self-employed sales advisers who visit customers in their homes or businesses and measure and order the goods. They are supported by the 100-person manufacturing and head office team in Nottinghamshire.
The challenge for Hillarys was to ensure a speedier flow of information to and from the sales advisers. But, as IT director Julian Bond points out, since they are self-employed, the head office could not impose IT systems on them. "We can't take a sheep dip approach," he said. "We have to understand the effects of regional variations, competition and the economy, so we can fine-tune our offerings county by county."
But the company also wanted to shorten the cycle between sales, manufacturing and finance. "This is a retail business, so we have a weekly cycle of trading meetings and the finance team would spend all Monday trying to form the trading picture," Bond said. "And if someone wanted some different data, it would take a week to find the answer."
Hillarys also wanted to split out data from its rapidly-growing custom products business. This unit mostly takes on commercial contracts, which are far larger than domestic orders; as a result, it skewed the weekly figures.
The problem, Bond said, is that there was no easy way to pull together all the data: "We had whole families of spreadsheets, and it was leading to frustrations in the business as they didn't have access to the totals they wanted."
After evaluating a number of BI and analytics products, Hillarys opted to roll out SAP's Business Information Warehouse (BW), primarily because of its tight integration with the company's SAP-based manufacturing systems.
Other BI tools might be richer in functionality than BW is, Bond said, but the integration was the most important factor for the business. "We are not making stock items, such as 30 different types of blinds," he explained. "Each order is unique and so needs its own bill of materials."
And the results have justified the choice, according to Bond: the company has been able to bring its weekly trading meeting forward, and the quality of the data available for analysis has improved. "We are able to look at questions [in our sales data] we would not previously have been able to answer," he said. "We have a much clearer view."