Manufacturers looking to take their automation strategy to the next level need to be thinking about data analytics. This key component of Industry 4.0 systematically analyzes production data to help users understand which areas of their operation hold the most optimization potential.
Beyond these core insights, the right analytics tools enable workers to monitor their processes in real time, providing opportunities for continuous improvement. The software helps drive continuous improvement by alerting workers to developing problems before they have a chance to fester and impede productivity, diminish product quality or increase operating costs.
To understand more about how data analytics are being used in industry today, we spoke with Morgan Bowling, director of industry at Seeq, a supplier of analytics software for the process industries.
AW: What are the most critical challenges that analytics software can address in the manufacturing industries?
Bowling: I would say these challenges fall into two, somewhat-related categories: profit and planet. In the profit category, we’re talking about ways that manufacturing companies can increase their revenue or their profit margins. [To increase revenue] companies are using analytics technologies to help them make more products, either by reducing downtime or by pushing other capacity constraints to create a higher average throughput over the course of a year.
These levers really only matter in markets where companies can sell more if they can make more. A lot of markets are in a state of oversupply right now, so those companies are working on increasing their revenue not only by making more product, but also by trying to differentiate their product by increasing quality so they can sell it at a higher margin.
They’re increasing profit margins by also attacking their fixed and variable costs. They’re doing things like optimizing the energy inputs so they’re only paying for the energy that they need when they need it.
In the sustainability category, while there is usually a financial benefit associated with energy, water and waste optimization, the bigger picture is that many companies—particularly Fortune 500 companies—have pledged to reduce emissions by, say, 50% or to be net zero or near zero over a specified time horizon. So, they’re using analytics to monitor emissions and identify periods of suboptimal energy consumption, waste generation and water consumption. All of this is helping them to understand what their carbon footprint is now and to look at different ways to reduce it.
AW: How does analytics software handle real-time data analysis and decision-making in a manufacturing setting?
Bowling: Our advanced analytics software, for example, does this by connecting live to the data source. Having a real-time connection as process data is being scanned and transmitted to the centralized historian ensures that you’re able to run your calculations and get outputs in real time. Seeq does not provide closed-loop feedback to the control system. Rather, it is running calculations and providing them to the subject matter experts in real time. That person still has to look at the insight provided by that calculation and then make the decision to take action.
AW: How are you seeing analytics software used most to help optimize production processes and improve operational efficiency?
Bowling: In a batch process, your levers for making more product are to shorten the duration of the batch, shorten the time between batches, and increase the size of the batch. We’re seeing the first two being used quite heavily. [The third lever is not used as much because,] often, increasing the size of the batch means increasing the size of the vessel, which is a capital plant modification. But if you’re able to slice and dice your data and carve out all of the individual procedural steps, as well as dead time or lag time, then you can easily find optimization opportunities to shorten batch durations.
For example, a particular step may take anywhere between eight and 32 minutes, and you don’t know why it’s sometimes long and sometimes short. So, you could look at that step with analytics. You would figure out what the best case scenario is, build a monitoring profile for that, and then monitor every batch in near real time to see when it’s going longer than the prescribed time.
In the continuous [processing] world, analytics are used a lot to figure out ways to increase the average production rate over time. That might mean reducing downtime, or it might mean taking opportune downtime to do maintenance or cleaning to restore production rates.
AW: What are the potential cost savings or ROI that manufacturers typically experience when implementing analytics software?
Bowling: Users typically realize between 10 to 100 times return in the first year of using our software. These returns are comprised of a lot of individual use cases, both the ones that users knew about before investing in the software and cases they hadn’t even thought of when they started using the software. Let’s take, for example, a batch cycle-time optimization use case. Analytics may allow a user to make $1.5 million in additional batch revenue per year. Maybe another use case is optimizing a heat-exchanger maintenance program, which may save $300,000 on maintenance resources over the course of the year.
Now, in most cases, manufacturers conduct a use case once at a plant and then scale it out. So, that value is quickly amplified when you apply it to 30 heat exchangers across a big refinery, for example.
AW: What level of technical expertise is required to implement and operate analytics software effectively?
Bowling: There’s a full spectrum of potential users of analytics software. Some software has straight up coding interfaces, and that’s typically used by a data scientist.
Seeq, on the other hand, falls into the category of self-service advanced analytics, meaning the software is accessible to anybody regardless of technical encoding prowess. Operators and process engineers with no coding knowledge whatsoever can use the software. With its point-and-click interface, they can locate the time periods they want to focus their analysis on, run some calculations during those time periods, or build a model or a golden profile without ever typing any code.
On the flip side, Seeq also offers a data science interface for leveraging the calculation engine. It resides in a coding interface for folks who are really interested in coding and want to work with it that way.
[To help users with the software] we offer an introductory boot camp that lasts two hours. It helps users get a feel for some of the things you can do with the software. We also do a foundations training course that covers all the capabilities of the point-and-click format of the software. That course runs three hours a day over three days—so a nine-hour commitment total. Beyond that, as users play around in software and take advanced training courses, they can realize the full potential of what they’re able to do.
