The Backbone of Advanced Manufacturing Is Advanced Analytics

Overview

Today, manufacturing generates and accounts for a full third of all data involved in business and industry.  As more industrial companies move to the digital enterprise model, this percentage is likely to increase significantly. The role of data in manufacturing has been typically understated, if not misinterpreted or underutilized. This is changing rapidly, as data and the actionable information derived from data form the backbone of the digital enterprise, which represents the future for advanced manufacturing.

As companies move toward the digital enterprise model, they are often challenged to effectively extract value from the vast amount of manufacturing data -- including both historical data saved in repositories, and data generated by current production systems, machines, and equipment. Predictive and prescriptive analytics and Big Data technologies will be key enablers for extracting value from these data.

The Industrial Internet of Things (IIoT) will connect real-time data from the factory floor to both enterprise systems and the human decision makers that require actionable information. IIoT will provide production line information based on an emerging generation of intelligent sensors, machines, and systems. IIoT provides the connectivity, intelligence, and advanced Big Data analytics to drive not only the next wave of continuous process improvements (CPI), but the digital transformation of today’s business processes.

These analytics-enabled production systems will use operational intelligence, visibility, demand-pull, and a synchronized supply chain to drive the manufacturing process based on an intelligent, event-oriented environment. A pervasively connected and intelligent “system of systems,” combined with analytics engines, will turn the unstructured data of manufacturing into actionable information to help make event-driven manufacturing a reality.

Backbone of Advanced Manufacturing Is Advanced Analytics

Looking at the areas of performance analytics, operational intelligence, and closed-loop product lifecycle management (PLM); it is clear that each discipline supports the idea of performance improvement, in part by helping validate the as-built to as-designed.  Furthermore, as ARC had anticipated, advanced analytics is now progressing from predictive to prescriptive analytics, where we bring together Big Data, statistical sciences, rules-based logic, and machine learning to discover and reveal the origins of the complex problems, and then determine decision-based options to resolve them.

According to the US Bureau of Labor Statistics, manufacturing, both discrete and process, have the most stored data (well over 1,500 petabytes) of any industrial or business sector. One could make the case that this represents a digital repository of manufacturing process history and embodies the primary source of unstructured data that needs to be aggregated, analyzed, and converted into actionable information. Moreover, when this information is placed in the context of a design/build lifecycle, it becomes a closed-loop mechanism connected by a “digital thread” that includes product development, manufacturing, and services in the field.  The data held in this repository becomes a source for operational intelligence, and product performance, enabling production process improvement.

Because a typical large OEM manufacturer can accumulate extremely large volumes of stored data, the only effective way to analyze the sum total of these records is through large-scale pattern matching technology enabled by current machine learning (ML) algorithms.  What ML already does exceedingly well—and will get even better at—is relentlessly processing any amount of data and every combination of variables.  Eventually, pattern matches are made and predictable outcomes form that, in the case of product process records, can result in a set of rules and best practices for improvements in product designs and manufacturing processes.  The process is straightforward and relatively simple in theory, but can involve some rather complex algorithms and require a lot of data.

Analytics for the Factory and Product Development

We have been discussing for some time now how manufacturing and production systems will undergo significant changes with the emergence of IIoT-enabled smart, connected factories and supply chains. There has been a steady progression from simply monitoring machines and production lines to optimizing production processes using predictive and prescriptive analytics.  Eventually, we might even see the autonomous “lights out” factories of the future. Within this progression, two distinct sets of predictive analytics solution are emerging: one for product development and one for factory operations. 

Let us first look at the factory and how analytics is being applied in the manufacturing of products and production operations.  Here, leading PLM solution providers are applying advanced analytics solutions to help optimize manufacturing processes and the overall performance of production operations. These analytics solutions all use various forms of machine learning algorithms and other statistical and rules-based methods to examine and identify patterns in the production process. The goal is to apply an operational intelligence approach to achieve a closed-loop mechanism that first validates the as-built to the as-designed, and then looks for ways to improve both the product itself and the manufacturing process.

The vast repository of production records includes completed operational and executed work records, quality assurance records, work flow histories, operational deviations and variations, engineering changes, machine and tooling metrics, material data, and many other records related to the production process. A good percentage of these data have been maintained and archived to meet specific regulatory requirements in industries such as aerospace & defense, automotive, pharmaceuticals, and medical devices. From this production process-related Big Data, predictive/prescriptive analytics engines that use machine learning, multi-variant statistics, and rules-based logic can empirically reveal the origins of the most complex problems and (in the case of prescriptive analytics) suggest decision options to solve them.

In contrast, on the product development/design side, predictive analytics is used in conjunction with product test simulation tools, and multi-discipline systems engineering methods, generating an emerging branch of product development called predictive engineering analytics (PEA).

One of the most sought after but elusive goals of product design engineering is to validate that you have achieved all the design criteria in the as-built product. That is, closing the loop between the as-built to the as-designed and validating that the physical product will meet all design criteria before the product is actually manufactured. This is where the concept of the digital twin is now being applied to product design.

The digital twin is a virtual representation of the physical product, including its engineering design and operational functions, essentially merging the physical and virtual worlds.  This enables design improvements and continuous process improvements in manufacturing, operations, and maintenance. An important aspect of the digital twin concept is that products can be designed with the intent of being smart and connected while providing operational data for predictive analytics.

To meet the requirements for the IoT and the new generation of smart connected products, product designers are adopting a closed-loop, systems-driven approach. This approach involves a de-composition of all functional elements while considering all intended interactions across the various engineering disciplines involved and controlling the overall system behavior. This approach minimizes risk and avoids late-stage changes by understanding boundary limits for subsystems, and also understanding the global system's behavior at every stage of the development process. Further, PEA enables product designers to model and predict product performance.

Recommendations

Advanced analytics is becoming a permanent and critical element of today’s manufacturing processes and product design and for maintaining and servicing equipment and products in the field. As companies re-examine methods and approaches for improving product design and production systems, analytics will become a requirement and be applied across the entire design/simulate/build/maintain lifecycle. Next-generation analytics applications based on Big Data and machine learning will allow manufacturers to detect and determine best practices for their product design and manufacturing processes.  Clearly, advanced analytics will serve as one of the necessary technologies that will enable advanced manufacturing and the factories of the future.

If you would like to buy this report or obtain information about how to become a client, please Contact Us.

Keywords: Analytics, Descriptive, Predictive, Prescriptive, Big Data, Machine Learning, IIoT, Digital Twin, ARC Advisory Group.