Why analytics in continuous flow manufacturing is failing, and how to fix it

Continuous flow Manufacturing

Table of Contents

In continuous manufacturing, there is a revolution underway in how data is being used to drive operations. Corporate leadership is looking to better understand what is happening across the organization and driving manufacturing analytics initiatives to the plant floor. I continue to talk with manufacturers of all sizes who have embraced this mandate and yet are drowning in pools of disjointed data. Many continuous manufacturers have disparate tools all handling data from different sources; the challenges associated with weaving the data from these sources together into something useful have led too many companies off the ledge into data doom.

The push towards factory-wide digital technology adoption has resulted in changes in the volume and utility of sensors on the plant floor:

  • The number of sensors providing data on equipment is increasing, allowing a more in-depth view of conditions within a specific machine.
  • The breadth of sensor data is also increasing, as manufacturers now have the ability to tie together sensor data across machines to capture a view of the entire production process.
  • Process flow manufacturers are also increasing the amount of data they capture on production batches in an effort to support quality and compliance initiatives.

Unfortunately, at most manufacturers, the tools and resources in place for analyzing this growing volume of data have not advanced significantly over the past decade.

Continuous flow Manufacturing

The traditional model for production analysis is unsustainable

In the traditional view of data-driven problem solving, data scientists and process engineers retrieve and analyze data captured in a particular segment of production that is being investigated. This analysis is single-use. The analysis is backward facing and time intensive, making it difficult to affect production decisions in real-time. Any change, alteration, or adaptation for future use requires redoing the full analysis with new data. Because of these limitations, analysis tends to happen infrequently if at all.

Past optimization efforts also typically focus on analysis associated with individual machines or assets. This narrow focus made it difficult to account for the interdependencies between different elements of the continuous production process.

What makes this end-to-end view even more challenging for continuous flow manufacturers, is that most have multiple process types. Typically, production involves a mixture of continuous, batch, and discrete processing (for areas like packaging). Most software tools capture and work with data from only one of these processes.

Single-use software packages will never deliver a holistic view of your factory lines

Many software and integration firms deliver targeted solutions that address specific issues such as predictive maintenance or asset performance management. This specialization comes at the expense of thousands of other variables that need to be considered in the depiction of the entire operation for holistic optimization and improvement.

I have encountered multiple manufacturers who have endured a painfully detailed configuration process for niche software that supported one aspect of their line. Within weeks of go-live, the software’s inability to integrate with another essential part of their process was discovered.

So how can continuous flow manufacturers use this data more effectively?

Manufacturers need to rethink the way data is used to drive decision making. To truly deliver transformation, continuous flow manufacturers must take a holistic view of end-to-end production.

At Sight Machine, we have worked with leading continuous flow manufacturers to improve product quality and efficiency by integrating their process and batch data into a holistic view of production. This holistic model is a digital twin of the relationships between machines, lines, processes, and production batches.  

Continuous flow Manufacturing

Creating a digital twin that models the production environment in real-time, changes the paradigm for how data is used to drive improvements. The availability of usable data in a digital twin allows the manufacturer to overlay analytics with streaming data to proactively address quality and efficiency issues.

A digital twin allows the manufacturer to understand the relationships between product quality and the machines touching the product, and to perform complex analysis such as identifying root causes of quality issues. It also allows you to analyze the interrelationships and dependencies between machines involved in continuous, batch, and discrete production so you can see what is happening as product moves through production.

Some of our customers are taking this a step further – not just looking at process and batch data, but also integrating energy or materials used data into their digital twin  – enabling new levels of insight to drive even higher levels of efficiency.

The ability to analyze data from end-to-end processes is fundamental to digital transformation. Visit the Sight Machine use case page to learn how leading continuous flow manufacturers are transforming their operations with digital twins.

Knowledge is Power: Build a Competition-beating Manufacturing Strategy Are you realizing the full value of your data? While big data analytics is designed to unlock the value of data, the reality is that horizontal solutions aren’t well-suited to the manufacturing industry. That’s where manufacturing analytics comes in. Download Now

Picture of Kurt DeMaagd

Kurt DeMaagd

Chief AI Officer and Co-Founder – Kurt co-founded Slashdot.org and has served as a professor at Michigan State University in information management, economics, and policy. Kurt is an accomplished analytics programmer.

Curious about how we can help? 
Schedule a chat about your data and transformation needs.