Gaining insight from data

11 November 2019

Michael Risse believes that there are three main trends driving more rapid insights from manufacturing big data.

Big data is intertwined with analytics because it is the raw material used by analytics tools to extract value or insight. Big data has traditionally led analytics in the sense that data, in particular increasing volumes of data, has been the driving force behind innovation in ways to extract value from information. For example, first there was the web, then big data and tremendous interest in the term then there was a way to make sense of the millions of web pages: Google MapReduce and Page Rank, algorithms wrapped by the Google search bar.

Analytics, therefore, are really nothing new, the concept is as old as data. Merriam Webster’s online dictionary states the first known use of ‘analytics’ was in 1590 when it was defined as ‘the method of logical analysis,’ whereas ‘analysis’ was first used in 1581 and defined as ‘the separation of a whole into its component parts.’ 

Today, the pressure to gain insight from data is so pervasive that ‘analytics’ has become a throw-away term in marketing material for all types of software: visualisation, spreadsheets, business intelligence, and dashboards – and always along with the promise of ‘actionable insights’. And analytics comes in many forms: descriptive, predictive, diagnostic, interactive, prescriptive, basic, real-time, historical, and root cause as examples. 

However, despite all the words and use cases, research on the state of analytics within manufacturing organisations shows a gap between the current and desired state of analytics adoption. Simply put, organisations do not have timely access to the insights they want or need to drive improved business and production outcomes. With data volumes increasing, this gap only promises to widen as the data rich but information poor (DRIP) problem gets worse over time.

There are, however, three trends in analytics with the potential to keep up with the innovations in data creation and storage associated with the Industrial Internet of Things (IIoT) and big data in general. 

Three trends
The first of these is advanced analytics as a new and distinct set of offerings. ‘Advanced’ may seem like an odd adjective to couple with ‘analytics’ as it could refer to anything new or contemporary over the past 50 years, but it’s the term of choice of industry analyst Gartner and management consulting firm McKinsey & Company.  

Their definitions for advance analytics focus on the integration of innovations in machine learning, artificial intelligence, and data science into analytics products. The McKinsey report on advanced analytics, for example – Buried Treasure: Advanced Analytics in Process Industries – has the following explanation of the opportunity from new insights in data. ‘Analytics can improve performance tremendously while reducing costs: the value, spread over thousands of opportunities, can be worth tens or even hundreds of millions of euros across a company’s site network… Manufacturers must therefore embark on an analytics-transformation effort that reaches all the way from the shop-floor operators (who steer processes) to process engineers (who use deep insights that will drive the next wave of improvements) to managers (who constantly oversee performance).’

Applying data science innovation to analytics may also be referred to as ‘augmented’ or ‘accelerated’ in the sense that the user experience is improved through innovations that improve the time to outcomes.

A number of compelling scenarios come into focus with the deployment of advanced analytics solutions. For example, predictive analytics on equipment can warn of impending failure so operators can take action to prevent an unplanned outage. By taking a deeper look into near real-time data, engineers can make small adjustments and shave 10 minutes off a batch cycle time, increasing the annual number of batches by hundreds. Management can receive near real-time alerts about commodity pricing, allowing them to make decisions to boost plant profitability. 

The cloud
The second change is the application of massive amounts of analytics capacity within the manufacturing sector using the cloud, now an option for ever more manufacturing customers. One can see the focus the public cloud platforms—Google, Amazon, and Microsoft—are putting on industry tradeshows. 

For the first time, Google, Amazon, and Microsoft all had a significant presence at IHS’s CERA Week event in March this year. Amazon Web Services (AWS) presently has 50 job openings for employees with oil and gas expertise, to add to a roster of existing employees with experience at GE, AVEVA, and other industrial automation vendors. Microsoft and Google are similarly recruiting from across the automation ecosystem. 

What this means for manufacturing organisations is more cloud-based options to provide faster time to implement analytics solutions via an elastic infrastructure that can grow and shrink according to customer requirements, along with greater agility and the promises of lower computing costs. While cloud deployment is not a requirement for advanced analytics deployments and many customers either choose or are tethered to on-premise deployments, it brings another opportunity to organisations. This will mean a change to big data storage models in process manufacturing, which today are mostly on-premise, historian-based, and proprietary. 

Specific benefits of a cloud-based model for data storage and aggregation includes examples from the pharmaceutical industry where cloud-based data lakes combine data from sensors, batch systems, quality systems, and laboratory information management data sets. This allows these organisations to provide a comprehensive view of the manufacturing process to IT-based data scientist and frontline process engineers seeking insights. 

Another example, this one from the mining industry, is the ability to deploy analytics in the cloud while integrating with on-premise data sources to accelerate deployment time and enable access to any employee with a browser-based analytics application. 

From a management perspective in any industry, the ability to see data collected from across plants to implement consistent best practices and compare KPIs for production metrics is another cloud-enabled benefit.

Empowering experts
Finally, one thing that does not change in an advanced analytics and cloud-enabled organisation is the importance of domain expertise. Subject matter experts – the process engineers and other employees with expertise in the data, the assets, and the processes of the plant or facility – remain a critical component. 

These employees have been analysing data for 30 years with spreadsheets, an entire generation since the prior use of slide rules, pen, and paper. Putting these employees and their expertise together with advanced analytics and current innovations of data science is required for improved outcomes.

An example of reducing the time to insight is an end user company that was able to improve time to insight by their engineers from two weeks to one hour by using advance analytics software. This increased production by 100 barrels a day, resulting in a $2 million-dollar improvement per year. 

Beyond the bottom-line improvements from improved analytics, there is also the opportunity to simply have time for analytics and investigation. As an example, an asset root cause analytics scheduled to take 400 hours and requiring the efforts of five engineers was instead solved by one engineer in less than one hour. 

By enabling the front line of expertise in their organisations, manufacturers can realise a host of yield, availability, productivity, and margin improvements.

The surge in data volumes within manufacturing organisations is increasing the difficulty of finding insights. Yet the pressure to find insights and be ‘analytics driven’ is higher than ever, putting pressure on plant executives and SMEs alike, but changes are afoot to address this issue.

Advanced analytics integrates data science innovations and the cloud model brings new opportunities for data storage, collection, and context. Together, these advancements will enable SMEs to find insights that enable decisions to improve business results and profitability.

Michael Risse is the CMO and Vice President at Seeq Corporation.

Contact Details and Archive...

Related Articles...

Print this page | E-mail this page