Raw materials and energy have been the foundations of every manufactured product since the industrial revolution. With Industry 4.0, data is becoming more important than ever before, not only to minimize operational costs and increase efficiencies but to wring every bit of product and profit from raw materials and energy.
Decades ago, factories pioneered the Internet of Things (IoT) with the use of technologies such as supervisory control and data acquisition (SCADA) and programmable logic controllers (PLC). Valuable as they were, and still are, those traditional IoT applications are mainly about controlling robots and other equipment, monitoring production, and identifying problems such as overheating pumps.
Industry 4.0 builds on that foundation to enable additional, deeper business insights. For example, the trend toward just-in-time manufacturing means suppliers are under more pressure than ever to minimize downtime and maximize yields. A new generation of “smarter” sensors and instrumentation, such as a pyrometer, not only measures the product under processing but can also measure other critical parameters about itself, and ensure that they’re within the ideal range. This data helps maintain product consistency across shifts, lines, and plants.
If temperatures are trending too high or too low, it’s also a sign that something might be wrong upstream in the production process. In the early stage of failure, the temperature shift may be so subtle that employees won’t notice. Same thing with a robotic arm that’s gradually slowing because of a failing bearing — just not yet to a degree that its production rate suffers. When used correctly, the data provided by an integrated sensor can begin to detect this failure before it occurs.
These kinds of insights enable the manufacturer to adjust production and still meet deadlines. For example, if the product temperature shift is still within range, the manufacturer now can move production to another line when it reaches the point that maintenance is required. With enough advance warning, the manufacturer also could schedule that downtime on a weekend.
This data also enables insights when it’s time to replace equipment. For example, a tool such as Cisco Kinetic can ingest and analyze the data so the manufacturer knows which brands and models of equipment and components have higher or lower failure rates and maintenance costs. Now, it knows which ones to buy and which ones to avoid. As a result, it has tighter control over operating expenses such as maintenance.
More IoT sensors yield more insights and more benefits. Take the example of the product temperatures trending out of range. Upstream sensors on equipment and unfinished products provide additional data that industrial mechanics can use to quickly pinpoint the problem. Every hour, they don’t have to spend tracking down a problem is an hour they can spend fixing it. The more productive each mechanic is, the fewer that each factory needs to have on staff. All of these savings flow to the bottom line.
Waste Not, Want Not
IoT data also helps manufacturers meet their goals for energy efficiency, pollution, and product yields. In the case of greenfield plants, the energy efficiency goals frequently include achieving LEED certifications.
Raw materials often are heavily regulated, including expensive licenses based on the amount used. Take the example of a leading European manufacturer of advanced raw materials used in the production of fiber optic cables, whose production requires potentially hazardous materials. Their production was limited by the number of precursor materials that could safely be stored on-site. As their product demand increased, the maximum allowable precursor materials became a limiting factor; their only choice for increasing production was to improve the efficiency of the process itself. This required deep learning and insight into their process in order to understand how to gain this improvement.
IoT data enabled this company to wring more meters of cable out of its existing gas allotment. It began using pyrometers to tightly monitor and control temperature, which is a key measure of the effectivity and efficiency of their process. These insights enabled the company to optimize its processes in ways that maximized product quality and yield, thus reducing the use of their precursor materials to achieve the same yield and product quality.
The business benefits of these and other Industry 4.0 applications are the main reasons why manufacturers are investing in IoT, but they’re not the only ones. The rapidly declining cost of digital transformation initiatives is another driver of change. Not only is the cost of cloud-based data storage and computing reducing, but at the same time, the tools available for analyzing, understanding, and converting this data into actionable information are also increasing. Off-the-shelf solutions, such as Microsoft BI, are ideally suited for analyzing vast amounts of collected data, and they don’t require a small army of Ph.D. data scientists to run. The cost of storing and analyzing terabytes of data has plummeted from thousands of dollars to as little $5. What was once hopelessly aspirational is now practical and affordable. But even though the tools are more affordable and accessible, the trick remains how to increase learning into a process in order to drive a specific action.
But that’s not the same as saying that digital transformation is as simple as buying software and sensors. Manufacturers typically seek expert guidance in identifying where to deploy IoT devices, connecting legacy factory equipment, and picking the right network technology to knit everything together securely and reliably.
What to Do With All That Data
Many manufacturers also want help understanding how to make sense of the flood of data that IoT provides, such as determining which data is worthy of immediate alerts and which can be stored for analysis. Vertical-market experience is particularly helpful because, for example, the IoT partner can recommend specific parts of the production process to monitor, which metrics to track, and what data trends indicate. Today’s manufacturing processes are so complex that it can be difficult to quickly and easily understand the complexities for operating all its equipment, and in many cases, expertise can be gained from suppliers themselves. Critical suppliers will already have years of learning about their unique products and can help define critical parameters to ensure optimal performance in a given application.
Another example is measuring a pump’s oil viscosity to determine when it needs to be changed rather than simply relying on the vendor’s recommended but overly frequent maintenance schedule. Machine learning can play a role by teaching artificial intelligence when viscosity, vibration, and other attributes indicate a need for maintenance. AI and ML also are examples of technologies that were too immature and expensive just a few years ago.
Like virtually every other type of business, manufacturers see digital transformation as a necessary process for maximizing efficiency, productivity, profitability, and competitiveness. IoT, artificial intelligence, and machine learning lay the foundation for those digital transformations.
Road to Transformation
Lastly, I’d like to end this post talking a bit about the evolutionary steps associated with digital transformation. The first step is acknowledging that something has failed (such as a sensor no longer functioning correctly. The second step is being able to understand why this sensor/process failed. These are the simplest steps in the IoT transformation and the factory of the future. Many companies are at this level today.
The third step becomes more predictive: this means being able to determine when something is going to fail before it fails, with as much lead time as possible. At a minimum, producers don’t want to affect/damage their final product or equipment as a result of the failure. But even better, they want advanced notice to schedule the necessary planned maintenance (as you’ve already indicated above).
The final and most transformative step in this evolution is migrating from predictive to prescriptive, meaning that you can see in your data that something is about to fail, but a producer is able to tune and adapt the process in order to control exactly when the downtime can be incurred. This requires significant knowledge and insight into a process, usually through the aggregation of huge amounts of data and precisely understand which parameters affect the process. In this phase, all instruments and sensors begin to act as a symphony.
This last step is also a major transition point when companies, armed with so much information and learning about their process, can then begin to more precisely tune and control their process to achieve desired product uniformity, quality, manufacturing cost, and throughput. This learning can be deployed on a single site, but even more powerfully at multiple sites across an entire enterprise. It can be very difficult for companies to achieve identical product consistency at different manufacturing locations, which can be so easily affected by facilities, geography, personal and supply. This IoT revolution can be to harmonize these global operations and drive product and manufacturing synchronization and improvement across an entire organization.