Your AI-driven optimization framework for achieving autonomous paper mill operations

In recent years, the paper manufacturing industry has witnessed a significant transformation, with Artificial Intelligence (AI) emerging as a powerful tool for optimization. These technologies have already proven their worth, saving paper manufacturers valuable labor resources and production costs.

If you’re already rolling out autonomous use cases within your facility, you’ve likely established a centralized data repository that includes contextualized up-to-date data. (Clean, contextualized, and high-quality data are vital for AI efforts.)

But what comes next? Even if a plant has had a few successful use cases, there may not be a solid framework in place to scale AI efforts for autonomous operations across the entire mill—or, better yet, at the entire organization. 

Autonomous operations are best visualized as a three-legged stool, with each leg representing a critical component for success:

1. Data preparation, contextualization, and optimization
2. Implementing AI and ML tools in specific use cases
3. Achieving autonomous operations at scale

Many paper manufacturers are still exploring ways to leverage AI’s potential at scale fully. This article outlines a comprehensive framework that paper manufacturers can use to scale AI and achieve autonomous paper mill operations that maximize efficiency and productivity.

Reduce errors with AI-driven detection

One of the most immediate benefits of implementing AI in paper manufacturing is enhanced anomaly detection. Advanced AI algorithms can identify drifts and irregularities in production that often unnoticed by human operators. 

Humans can only monitor, track, and prioritize so many things at once. Even if you’re using technology to give teams insights or visibility into the entire production process, these tools may not be able to parse out the most critical situations. 

Using AI means teams can anticipate drifts or breakdowns earlier. AI can also recommend which situations to address first, reducing decision fatigue for employees while prioritizing mission-critical events. As your organization moves toward autonomous operations, AI systems can provide real-time countermeasures for course correction based on historical precedence or specific objectives.

Arjowiggins, a leading technical and creative paper manufacturer, implemented Braincubeto better utilize asset and process data. Early on, they tested Braincube’s AI on a specific problem. The results were so convincing that they’ve continued using Braincube for over 20 years. 

Bringing AI into their workflows allowed Arjowiggins to analyze processes differently than traditional statistical models, providing operational solutions previously unavailable. For example, Braincube AI-enabled Arjowiggins to optimize processes, improve problem resolution, and enhance performance monitoring, particularly in energy use. 

By implementing AI-driven analysis, Arjowiggins transformed its data utilization paradigm. They now use Braincube as a reference tool for all process-related decisions, leading to more efficient operations, cost savings, and a deeper understanding of their manufacturing processes.

Obtain optimal production parameters for current assets

Even as new tech is brought in, many paper manufacturers still run production on legacy assets. For this reason, one of the biggest challenges that operations leaders face is getting their workforce to utilize (and believe) data that may contradict “the way things have always been done.” 

Your teams may have assumptions on the best setpoint ranges, courses of action, or operating settings based on how things are done when similar situations arise. In most cases, though, these assumptions are simply that: assumptions. They aren’t based on a comprehensive analysis of historical data, different production conditions, or variability trickling down from upstream processes. Moreover, it’s nearly impossible for even the most technical teams to manually analyze the vast amount of data that goes into determining the right settings for different situations.   

From a complex set of 2,450 variables, AI identified which 28 critical variables impacted paper coating cost fluctuations.

AI excels at analyzing vast amounts of data to determine the factors that have the most significant impact on key objectives. Current estimates state that AI can compute 200 quadrillion calculations per second, making it possible to crunch through massive amounts of data in very little time. 

Leveraging machine learning algorithms enables paper manufacturers to identify ideal operating ranges and setpoints for key assets and processes. These gains in time and accuracy mean your teams can put valuable findings into action faster than ever (even in real time!), while also having confidence that these process changes will be safe and effective. 

Oji Paper, a Braincube customer, successfully used our AI tools to minimize fluctuating production costs associated with chemical coatings on their paper products. The team knew that 65% of batches had excessive chemical coatings but they could not hone in on the specific variables that drove up this excess usage. 

This single chemical coating process included 2,450 variables. This made it immensely difficult for their process experts to identify which variables had the greatest impact on cost. 

Braincube’s AI quickly identified 28 pivotal variables that had the greatest impact on chemical coating costs. Armed with this new process knowledge, the Oji team monitored these key variables in the control rooms to ensure they were in compliant operating ranges as much as possible. 

This process change resulted in a drastic 10% reduction in chemical consumption rates in only one week. In addition, Oji allowed their operators to implement new standard settings as they were identified by AI (and validated by process experts). This approach enabled Oji to achieve both process stability and significant cost savings.

Perform root cause analysis with pinpoint accuracy

It’s no secret that papermaking is a highly complex, lengthy process. 

But why is AI the right tool for this instead of traditional analytics tools? Traditional analytics tools—like PowerBI and Excel—have limitations when it comes to comprehensive analyses. These traditional tools (and the human brain) can’t holistically analyze the layered relationships that exist within manufacturing data sets.

One of the most powerful applications of AI in paper manufacturing is its ability to perform detailed root cause analysis. Using AI enables process experts, engineers, and data scientists to understand and optimize multivariate relationships across the entire paper production process. Rather than resolving isolated incidents at the point where a problem occurred, AI can suggest operational changes that collectively enhance production efficiency by highlighting how a misstep upstream caused an issue downstream. 

In other words, AI gives teams the ability to concretely understand why something happened during production, even if an issue originated upstream instead of localized to where the problem occurred.  

One of our global paper manufacturing customers wanted to reduce its production costs, but siloed data prevented engineers from uncovering the root causes of process variability and potential recipe optimizations. Furthermore, The shop floor lacked real-time guidance for recipe control and problem troubleshooting, leading to increased downtime, quality variability, and waste.

The engineering team leveraged Braincube’s CrossRank AI to accelerate root cause analysis when problems occurred. When paired with AI-recommended recipe optimizations, the team was able to improve their process control and troubleshooting efforts. The company experienced a 40% reduction in chemical waste by minimizing the number of paper breaks occurring on the line, driving $750k in savings in just nine months. 

Closing the loop with autonomous operations

As AI systems prove their worth in analysis and decision support, the next step is for paper manufacturers to integrate these models directly using selected systems that trigger automatic actions. This closed-loop integration, operating without human interaction, represents true autonomous operations.

The transition to autonomous operations is typically gradual. Initially, human teams will need to validate AI recommendations and insights before pushing them into production. They will want to verify that the AI is providing safe, viable recommendations and understand the expected outcome of implementing these new findings. 

Remember that AI models learn from actions and outcomes by incorporating feedback from human process experts. Your models—and their recommendations—will only be as good as the humans that train them. This continuous learning allows the AI to refine its decision-making processes and focus on the most critical aspects of operations, as influenced by your teams.

Over time, though, your models will learn what to focus on by incorporating directions and guidance from your process experts. Your AI systems will prove their reliability and effectiveness, enabling you to integrate autonomous operations more deeply into their processes. This integration allows AI to not just analyze and recommend, but also to implement actions directly. This frees up your process experts to work on other high-value tasks that can’t be improved by automation as they move to a primarily supervisory role. 

Keep in mind that Digital Twins also play a crucial role in autonomous operations. Digital Twins ensures models are continuously fed with current, contextualized, and high-quality data for new optimization analyses. They also provide a virtual representation of the physical manufacturing environment, allowing AI models to test scenarios (and humans to validate these scenarios) in a risk-free virtual environment. 

The benefits of autonomous operations for paper manufacturers are truly limitless. Reduced human error, improved productivity and efficiency, and consistent quality output in the midst of changing conditions are just some of the “pipe dream” outcomes that are feasible. 

Of course, autonomous operations are not without their challenges. Your teams will need to continuously ensure the system is operating safely and reliably, but these pulse checks will be few and far between compared to the early stage of autonomous operations. If possible, you can work with a third-party vendor to help you maintain and update the AI systems over time, further alleviating this stressor from your teams. 

Closed-loop systems represent a significant shift in manufacturing paradigms, moving from reactive to proactive operations management, and ultimately to predictive and autonomous systems. It’s a key step towards realizing the full potential of smart manufacturing and Industry 4.0.

Conclusion

The path to autonomous operations in paper manufacturing may seem daunting, but it’s more accessible than many realize. By partnering with the right technology providers and taking a step-by-step approach, paper manufacturers can leverage AI to transform their operations. These autonomous tools free up operators to intervene less while improving compliance with optimal settings. 

Starting with foundational data infrastructure and gradually implementing AI tools for anomaly detection, parameter optimization, and root cause analysis, mills can build toward fully autonomous systems. The benefits are clear: increased efficiency, reduced downtime, improved product quality, and optimized resource utilization.

As we’ve seen from the examples throughout this article, paper manufacturers around the world are already reaping the rewards of AI-driven optimization. Those who embrace this technology and commit to the journey toward autonomous operations will be well-positioned to lead the industry in the coming years.

The paper mill of the future is not just a concept: it’s an achievable reality through the power of AI-driven autonomous operations. By leveraging advanced insights and closed-loop control systems, paper manufacturers can elevate their entire production stream, ensuring competitiveness in an increasingly demanding market.