Volume 3 Issue 12 December 2021
In this Issue
Welcome to Industree 4.0 for December 2021, exclusively sponsored by SAP.
By James Sullivan and Fawn Fitter, SAP SE
How to Make Sustainable Materials Use Matter -- Part 1, Business Value
A recent SAP Insights survey on sustainability found that companies’ biggest barrier to taking more environmental action is their uncertainty about how to embed sustainability into business processes and systems. Making more sustainable choices about materials use is one way to address that challenge. Part 1 of 2 talks about the business value driving the push to use sustainable materials.

Apple, for example, gives customers the option of returning their used phones, laptops, and other products. Newer items earn trade-in credit, which the company recoups by refurbishing and reselling them. It also disassembles unrepairable and obsolete technology, both for parts and to extract and recycle or reuse materials like aluminum and gold. As one result, the latest MacBook Airs, Mac Minis, and iPads are encased in 100% recycled aluminum from Apple’s own manufacturing scrap as well as postconsumer building and construction waste. This approach recaptures the value of materials that would otherwise be discarded while keeping e-waste out of landfills and incinerators.

Indeed, the SAP survey identified sustainable materials use as one of respondents’ top two priorities for investing in the environment, roughly equal to their commitment to addressing climate change. We also found a minority of companies, which we call the Nows, that believe sustainability is financially material* to their operations today and are already profiting from their investments in it. The Nows are significantly more committed than other survey respondents to investing in sustainability across the board, including sustainable materials use.

And the business case for that investment is incontrovertible. In 2020, the Center for Sustainable Systems at the University of Michigan found that raw materials use has grown at more than three times the speed of population growth in the United States alone. Given an expanding global population that insists on an ever-higher standard of living, combining reuse and recycling with more sustainable raw materials would help ensure companies have enough resources to keep up with demand. Read the research

There are, of course, trade-offs. Companies may have to spend more up front to redesign their products and manufacturing processes to make better use of current materials or substitute different ones. But it’s likely they can build on their existing processes and systems, which will help them see a fairly prompt return on investment. Along the way, it’s possible they may also identify previously unnoticed opportunities to cut costs and create new income streams.

“There’s increased interest in materials choices because they can fit into a larger sustainability narrative,” says Kevin Dooley, chief scientist at The Sustainability Consortium, a nonprofit helping the global consumer goods industry create more transparent and environmentally sound supply chains. “And there are many ways to approach materials use based on what will appeal to customers while meeting sustainability goals.”

The business value in sustainable materials

To some degree, a combination of consumer demand and government regulation is driving many companies toward more sustainable sourcing and use of materials. Consider the current pressure to eliminate single-use plastics, driven by both heartrending photos of animals that starved to death with stomachs full of trash and predictions like that of the World Economic Forum: that without intervention, the oceans in 2050 will contain more plastic waste than fish. Consumers are reacting with demands that companies use less plastic. Meanwhile, governments are responding with increasingly stringent regulations and are ratcheting up taxes, fees, and bans on single-use plastics. At the same time, they’re pushing companies to take responsibility for the disposal costs as a way to encourage them to choose different materials, or at least to design products and packaging that contain less plastic and can be reused multiple times.
There’s also significant evidence that customers are putting their money where their demands are. In 2020, sales of sustainably marketed products grew seven times faster than those of other consumer packaged goods and accounted for more than half of overall growth in the category.
Companies discover savings opportunities when they invest the time and effort into collecting data about their current materials use. The more data they have, and the more visible and accessible it becomes, the more useful it is for generating insights that lead to greater efficiency, increased cost savings, and potential new opportunities.

Calculating the true costs of your materials use – not just to manufacture and package a product, but also the way it’s used and eventually disposed of – may initially seem burdensome or impractical to the point of impossibility. But the more you understand those costs and how you incur them, the more you can identify opportunities to create value, recapture it, or both.

Exploring the full lifecycle costs of materials contributes to a comprehensive view of sustainability in the markets where you do business, which in turn provides metrics for areas where you can set goals and take action. For example, you may identify ways to avoid materials (like plastic) that can’t currently be recycled in a key market. You might also develop new product designs based on more sustainable materials. Sportswear brand Cariuma makes knit sneakers from thread that combines plastic reclaimed from the ocean with highly renewable bamboo – an approach that reduces manufacturing waste and promotes alternative textiles while creating new markets among eco-conscious consumers.

You may even discover new ways to shift or extend your entire business model. Seventh Generation, which makes eco-friendly cleaning products, created an ultraconcentrated version of its flagship laundry detergent that contains 50% less water and requires 60% less plastic to package. By having consumers reconstitute the detergent from their own taps instead of paying to ship water around the world, the company distributed more product at less cost while making that product even more sustainable.

Meanwhile, in places where governments are using regulations to promote recyclable materials and minimize waste, companies willing to pay more for materials that include recycled content will find opportunities to recoup these costs through lower taxes and fees in those markets.

It’s likely one reason the Nows in the SAP Insights survey are so enthusiastic about investing in better materials use is because they see its potential to deliver an impressive ROI, even though that may take some effort to achieve.

Why is AI so dumb?
By Pat Dixon, PE, PMP

Vice President of Automation, Pulmac Systems International (pulmac.com)

One of my favorite publications (besides this one) is IEEE Spectrum. The October issue featured a special report addressing the question “Why is AI so dumb?”
AI (Artificial Intelligence) seems to be synonymous with Industry 4.0, even though it really isn’t. It is all the rage in many articles and presentations on Industry 4.0. Many are making promises that AI cannot currently fulfill.
Artificial Intelligence indeed has field proven applications in automation. I have over 20 years of experience applying AI to automation and have seen it work. When applied properly it can produce remarkable results.
If applied improperly, it can lead to major failures. In the Spectrum issue, one of the articles “7 Revealing Ways AIs Fail” is required reading for anyone intending to invest in AI. An excerpt states:
“Artificial Intelligence systems can perform more quickly, accurately, reliably, and impartially than humans on a wide range of problems, from detecting cancer to deciding who receives an interview for a job. But AIs have also suffered numerous, sometimes deadly, failures. And the increasing ubiquity of AI means that failures can affect not just individuals but millions of people.”
I actually take issue with the ability of AI to decide who is interviewed for a job, but I want to give you a paper industry example from personal experience.
There was a paper mill that invested in AI tools to predict sheet quality. In particular, the predictions used neural networks, which is one of the AI technologies that is commonly used. Neural networks have the remarkable attribute of being able to model just about anything. For the discerning engineer, this is also a caveat. This became obvious to me when I was sent to the facility to help them refine prediction models they were working on. The mill was convinced they had neural network models that accurately predicted product quality. They wanted me to improve the models and apply them to additional quality measurements. As I collected data and tried to train neural networks, I had great difficulty producing models that showed good correlation. I spent days trying to collect additional data, pre-processing data to remove outliers and downtime, and retraining. With no progress being made, I took a closer look at the models they had developed and solved the mystery. They had taken lab samples of sheet quality (which should be outputs of the prediction model) and used them as inputs to train neural networks to predict process data from the historian (which should be inputs to the model). Besides being completely backwards, the dataset was significantly compressed to the point that the neural network could be trained to fit the data without considering all the rest of the process data that should be included as inputs. Their model showed very high R-squared (a measure of model fitness) for their datasets, but when applied generally to their process data it was not nearly as impressive. What was revealed is what is known as overfitting. As I said, neural networks are capable of learning nearly anything, including noise. When I looked at the sensitivities in the model, they defied first principles. They had made the mistake of throwing a bunch of data in, looking at the R-squared, and thinking they were done. In fact, what they produced was at best unusable and at worst misleading.
That is why another article in the Spectrum issue entitled “A Human in the Loop: AI won’t surpass human intelligence anytime soon” is rather pertinent. It contains the following excerpt:
“Just about every successful deployment of AI has either one of two expedients: It has a person somewhere in the loop, or the cost of failure, should the system blunder, is very low.”
I alluded to this in my October 2019 article in this publication entitled “The Singularity”. While Ray Kurzweil foresees machines exceeding human intelligence in about 20 years, this prognostication seems based on a limited focus on processing speed and memory. For industrial automation, human reasoning provides common sense and creativity to solve problems that AI cannot yet approach. That is why successful AI applications, especially in industrial automation, include human design, development, and oversight.
To be clear, in some applications AI is way smarter than us. For example, Big Data exceeds human ability to find patterns in massive datasets. The concern is to assume that brilliance in one domain can be extrapolated to all domains. When compared to the breadth of human understanding and reason, AI is still dumb.

The piece we never talk about
Warning: Long winded story with hopefully a point. My wife and I were in Cincinnati recently and, surprise, surprise, she asked me if I would like to show her any of my old haunts. I took her by the apartment building which I purchased when I was twenty-four. Here it is...
Now, I belong to a group on Face Book called "Cincinnati Nostalgia." So I decided to post this picture there with this caption: "590 Ludlow Ave., Clifton. Built 1925. 6 main apartments plus janitor’s apt. in basement and 6 maids’ quarters plus maids’ bath on 3rd floor in the back. My wife took this picture on 4 Dec 2021..."

It solicited lots of interest. One of the commenters said: "Beautiful. But I'm confused about it being apartments with maids quarters for renters?"

My answer: "Yes, indeed. This was (and still is) a high class building. In the laundry room there were six sinks with washboards built in with a lightbulb over each one marked with the apartment no., so when the maid did the laundry they turned on the light for the apartment they served and the electricity was charged correctly. I took these out and put in coin operated laundry machines. Each apartment had a door bell type button within the apartment and that rang a bell in the appropriate maid's room to summon the maid (who came down the back stairs, the only one that reached the third floor). The janitor in the basement had a back door in their apartment directly to the furnace room so they could manually feed the coal fired boiler 24 x 7. All these things were wiped out with the coming of social security, worker's compensation insurance and unemployment insurance. Hundreds of thousands of maid's jobs around the country were eliminated when these employer taxes came into effect."

In Industry 4.0 and AI (see Pat's article above) we often talk about efficiency and accuracy in repetitive tasks. We don't talk about the simple savings of labor costs with these modern developments. In these homes the maids were eliminated because of (a) costs and (b) the development of modern appliances, like automatic washers and dryers. I recently read that in today's society, the modern aids at our fingertips would likely be the equivalent of 300 personal servants in the Middle Ages.

If this is the situation at home, think about the savings we see in modern industry with Industry 4.0, AI, and all the other tools at our fingertips. We lose perspective on how far we have come so fast.
Key Challenges to Using Automation in Packaging and Processing Operations
By Kim Overstreet
End users are struggling with a lack of common components and industry-wide standards, as well as with automation kits that cannot easily integrate with leading ERP or warehouse management systems (WMSs). As a solution, one CPG called for defining universal standards that could be agreed upon by both CPGs and OEMs.
Manufacturing predictions for 2022
The Manufacturer Magazine
Despite a steadfast, and in many cases, inspiring response from the sector, uncertainty still abounds as we approach the new year. Here, Naveen Poonian, CEO of iBASEt, offers some manufacturing predictions for 2022.
Digital Growth: What Reuters NEXT Says About IoT in 2022
IOT For All
Every year the Reuters NEXT conference brings together CEOs, founders, and other thought leaders from a wide range of industries to discuss what the year ahead holds.
Redefining success using industrial internet of things
Technology Record
Joe Biron shares how the technology is liberating the data in control systems, coupled with advanced analytics and the latest techniques in edge computing to meet today’s smart manufacturing needs.

Industree 4.0 is exclusively sponsored by SAP