Skip to content

Floor Show: To Optimize A Factory Floor, Visibility Is Key

Michael Anderson
By Michael C. Anderson Contributing Lead Editor

Recent advances have made factory optimization better, easier and more cost-effective, according to leadership at Advanced Manufacturing International Inc. (AMI).

The first goal of optimizing a factory floor is visibility, said Larry Megan, vice president of Advanced Manufacturing International Inc. (AMI), Clearwater, Fla. Manufacturers need to know how things are currently working and identify opportunities to improve quality, reliability and productivity—ultimately leading to increased throughput and operating efficiency.

Collecting and analyzing data is key to gaining insight into overall operations, compared with making guesses based on anecdotal information, Megan said.

“Manufacturers need to know what’s happening on the factory floor in real time as well as retrospectively,” Megan said. “You’ve got raw material coming in, you’re making products with multiple steps, you’re analyzing for quality, and you’re shipping it out to customers. To maximize the production of the entire factory, you need to look at all of those things holistically.”

Fixing a bottleneck in one area without considering total operation could simply move the bottleneck elsewhere without solving the overall problem, Megan said. 

Data is critical

The missing ingredient is often data—the relevant data.

“You can’t really optimize the factory until you integrate the data,” said Stephan Biller, CEO and founder of AMI.

Advances in the past decade have made it easier and more cost-effective to get better data, Biller said, citing advances in 5G, improvements in artificial intelligence and machine learning, advances in the Internet of Things, and cheaper storage in the cloud. Improved machine learning and artificial intelligence have improved modeling so manufacturers can better predict outcomes, Biller said.

“We need to look at key performance indicators—KPIs—and use mathematical models to figure out: if we act upon something, what is the impact of these key performance indicators ahead of time before we flip the switch and put something into action,” Biller said. “Now you can develop models that predict what is going to happen in that factory if you take certain actions. These kinds of simulations are really helpful in testing results of action on the factory floor of key performance indicators such as throughput, quality, cost, on-time delivery, safety, sustainability and resiliency.”

The five steps to the factory optimization process, Biller said, are:

  1. Data collection: Collecting and syncing relevant data.
  2. Data visualization: Looking at simple charts to understand where problems exist and if the factory had a good or bad day. Determining if the data makes sense.
  3. Decision support: What would happen if I do X? Should I do X, Y or Z?
  4. Optimization: What would be the best decision, but keeping the human in the loop to ultimately make that decision.
  5. Automation: Based on 100 or more accurate recommendations, automate the decision process and take the human out of the loop.

Some manufacturers are at the optimization stage already, Biller said, while others, especially small and medium manufacturers, are still at the data collection stage.

“When I was at GM, GM had gotten to the point where we would recommend optimal decisions in some parts of the plant, but people still wanted to be involved in the implementation of the decisions,” Biller said. “But small manufacturers are 10 to 15 years behind the large manufacturers, because the tools were so expensive, and because of the resources they have available.” 

Smaller manufacturers lag in factory optimization

Optimization tools have been more cost-effective and, thus justifiable, for larger manufacturers because their factories are more complex and such tools are more valuable in more complex situations, Biller said.

The challenges also vary with the size of the manufacturer, Biller said. Large manufacturers may have 15 to 20 IT systems and equipment generating data, with each system running on its own clock, he said. Since variances of even a few seconds can make a difference, results are skewed unless the systems are exactly in sync, he said. The challenge for large manufacturers is to sync all those data sources.

“You want to detect the patterns among those different data sources,” Biller said. “How do they correlate? As you are trying to get a complete picture of the factory, you want to integrate those data sources so you have one picture, one database, where everything is stored and time synced.”

Meanwhile, small to medium manufacturers may not have any infrastructure at all, let alone analytic capability, he said. The challenge is helping smaller manufacturers get tools that are neither too expensive nor too complex to collect the necessary data themselves, Biller said.

“If it’s pretty simple, you can do it on a white board,” Biller said. “If you have a thousand machines, you cannot.” 

Helping smaller manufacturers modernize

AMI, a not-for-profit organization, is focused on helping small and medium manufacturers cross the digital divide and implement their first Industry 4.0 capability.

“Most software providers have focused on the large companies because that is where they think they can make more money,” Biller said. “That is a market failure. We have to think about how we can help small and medium companies get to that stage. We are focused on low cost, low complexity and high security.”

Advances in the past decade have made such optimization better, easier and more cost-effective, Biller said, citing advances in 5G, improvements in artificial intelligence and machine learning, advances in the Internet of Things, and cheaper storage in the cloud.

Quick wins help sell the importance of optimizing the factory. “You need to know what is the company’s biggest problem. Delivering orders on time? Throughput? Unable to make enough product to meet demand? Then, it’s critical to have leadership drive the process and identify the problems that are most pressing. Typically, they say, ‘If I could only see X, that would be so helpful.’ In general, they first want to visualize some data. Data analytics will follow.”

In addition to buy-in from leadership, front-line workers also must be part of the process, Biller said. “The people on the floor who are actually using the tools must be an integral part of the integration and innovation,” he said. “They know a lot more than the providers of the tools. Only then do you get ownership and solutions that are truly integrated into the process and will make a lasting difference.” 

Legacy equipment poses (not impossible) challenge

Optimizing a factory floor with mostly legacy equipment and machines—where replacing old but expensive equipment is not a consideration—is challenging but not impossible, Megan said.

“Nobody ever replaces equipment until they can no longer find parts for it,” Megan said.

The good news: Equipment made within the past 15 years usually communicates via current standard data protocols, and thus can be readily connected to a data collection and analytics system, Megan said. For older equipment, often low-cost sensors can be wired to the machine to collect data, he said.

More good news: The cost of these sensors has dropped dramatically, making them more an affordable, realistic choice for small and medium manufacturers, which are AMI’s target customer group.

“In many cases, the sensor technology has been around for years, even decades,” Megan said. “As that technology continues to evolve, it becomes cheaper and cheaper. What was a $100 sensor 10 years ago has become a $10 sensor.”

Such tools also have moved beyond the sole domain of the IT department and become accessible to citizen data scientists­—those without specialized expertise, Megan said.

“Those tools are democratized,” he said. “You don’t have to be a programmer or an IT specialist to be able to use them.”

As one of its offerings, AMI provides a foundational platform to help manufacturers collect, control and manage data, Megan said.

In one example, a furniture company making high-end plastic outdoor furniture used this system, known as the Low Investment Manufacturing System (LIMS) to collect data that had been previously locked inside its CNC machines. The company is facing increasing demand and needed to ensure that the productivity of its existing assets was maximized before considering more costly solutions such as adding another shift or buying new equipment, he said.

The AMI team updated the data protocols on the machines to make it easier to collect data and create real-time visibility on cell utilization. This is helping the manufacturer remove production bottlenecks. The manufacturer may add more sensors from the CNC machines and other parts of the process to gain additional insight to maximize throughput.

With this added insight enabled by real-time data, the company has moved to a quantitative analysis of the cell and can identify targeted projects to improve things such as change over time and shift-to-shift production variability—ultimately improving profitability.

During the pandemic, manufacturers of all sizes have learned that optimizing manufacturing for just-in-time delivery can cause supply chains to become brittle, he said.

“We need to look at the supply chain from a multi-objective point of view and compare different decisions in terms of costs, resiliency, sustainability and on-time delivery,” Biller said.

  • View All Articles
  • Connect With Us

Always Stay Informed

Receive the latest manufacturing news and technical information by subscribing to our monthly and quarterly magazines, weekly and monthly eNewsletters, and podcast channel.