CHICAGO — Although many manufacturers are racing to implement automation and artificial intelligence in their factories, experts cite obstacles and urge caution before moving full steam ahead. This could mean considering energy demands, testing physical AI and exploring use cases.
Sven Diedrich, head of digital transformation and business solutions at Pinaxis, pointed to four additional factors to consider before going all-in on automation: people, processes, data and systems. He called these the “four pillars” of automation.
“Are your people, processes, data, and systems actually aligned, or do they improve on their own in silos?” he said at the International Manufacturing Technology Show in Chicago. “That's what we come across a lot: There is a disconnect between the four pillars.”
Diedrich cautioned manufacturers not to be distracted by the latest technology simply because it has theoretical potential.
“When companies start thinking about automation, it's very, very easy to begin with the technology because that's the shiny new object,” he said. “Everybody wants to get involved.”
For example, Diedrich added, companies often turn to robotics for more throughput, software applications for better visibility and AI for better prediction.
“All of those things are driving certain values but do not paint the holistic picture of your organization on their own,” he said. “They have to be connected in order to give you the value, the visibility and the decision-making capabilities that you are expecting these technologies to deliver.”
Diedrich offered a hypothetical scenario of a company that identifies a capacity problem. Engineering, information technology and other departments propose an automated solution, and only then does the company discover that its data is insufficient to implement the solution and the suggested processes are not in line with standard company procedures.
“You can have a high-dollar investment in the latest and greatest technology and still create a mediocre automation environment because you miss the connected pieces,” he said.
Conducting an honest evaluation of a company’s people, processes, data and systems before attempting to implement automation technology can save a lot of time, money and headaches, according to Diedrich.
People
“For the people, we usually ask: Who owns the process? Who operates the solution? Who supports it? Who responds when something goes wrong?” he said. “When we talk about automation, people are sometimes treated almost as an afterthought... The equipment gets the attention, the software gets the attention and the business case and the ROI calculation get the attention, but not the people [who] need to use the automation technology.”
Skills development is another important part of the people equation.
“Automation doesn't simply reduce work,” he said. “It creates different challenges and different skill sets that we need to make automation work and to support it.” A company’s IT and operational technology departments need to work together to “bridge those gaps and kind of define the requirements of these new rules and the changes.”
Organizations also need to take a hard look at whether their existing personnel have the capacity to implement automated systems.
“Organizations sometimes install the new system and assume the current team can simply pick up whatever they install and just smoothly move on with it,” he said. “But somebody has to be the administrator of the application. Someone has to be the one who is investigating reoccurring exceptions. Someone needs to train new employees, and someone needs to improve the system over time. That effort does not disappear simply because that system is automated now.”
Processes
When it comes to processes, Diedrich said it is important to understand the implications of reducing or eliminating human intervention.
“You may have an operator who... knows how to fix certain issues really quick,” he said. However, with automation, “the process is telling the machine what to do, and this machine sometimes is not capable of responding to the changes in the same way as a human can, because the machine actually does exactly what we tell it to do and doesn’t deviate from it. Automation exposes process weaknesses very quickly, because it removes the human being who quietly compensates for all their shortcomings.”
The bottom line is that a company must decide who owns system availability, data, continuous improvement efforts and other key elements of a process.
“If everybody owns just a little bit, something here, a little bit there, nobody owns anything, and we are setting up ourselves for failure and chaos,” he said. “So, clear rules become very important.”
It is also important not to inbed inefficient processes into an automated system.
“Automation can make a good process extremely efficient, but it can also make bad process extremely efficient as well,” he said. “The robot can repeat waste more consistently. And the software... can digitize waste pretty good without you knowing it.” None of these things, he said, address the underlying problem causing the inefficiency.
Data
Trustworthy, fresh data that everyone understands is also key to successful automation.
“We collect data left and right, but do we understand the value? Do we manage the data properly to create value?” Diedrich said. “The more decision-making we give to AI, the more important trustworthy data becomes, and the processes need to be rock solid in order to be combined with this data.”
Companies should pay careful attention to several types of data, he said, including master data such as item numbers, dimensions, weights and build materials; transactional data such as orders, inventory receipts and shipments; and process data such as cycle tollings, quality signals, executions, manual corrections, downtime and rework.
Systems
Finally, manufacturers need to have good systems in place to take advantage of automated processes. In part, this means ensuring there are robust communication channels between departments, as well as between the equipment and its operators.
“We have different IT [and] OT, and everybody's comfortable in their own silo,” he said. “But we don't share the information, so somebody on the shop floor may not get the right information at the right time, and that causes fails.”
Other important system considerations include determining what applications exist and the role of each, how data is inputted and processed, and the value that the company expects to gain from an automated system.
Once the four pillars are firmly in place for a pilot automation project, Diedrich said it’s time to think about scaling it. This doesn’t mean just “cutting and pasting” an operation from one factory or one country to another, where people may operate quite differently.
“A pilot can be successful with their work operators because they have the tribal knowledge to... overcome the shortcomings with a regular workaround,” he said. But what happens... when we're trying to scale what we implemented successfully in a chosen aspect of the process, and now we are trying to scale it towards the entire process? This is where true readiness becomes visible, or where shortcomings become visible.”
