An Industrial Internet project to optimize tissue paper production through real-time analysis of the production process that allows machine operators to check the parameters thus analyzed and, if necessary, adjust them. All with one purpose: to decrease the variability of the tensile strength of the paper produced. But the system can also go further.
Measuring the quality of the paper product at every stage of its production is the key factor in being able to reduce process variability, thus leading the mill to a decrease in both quality losses and operating costs.
This was the challenge at the heart of a joint project involving Valmet and Sofidel with an Industrial Internet solution. They talked about it at Miac 2023.
The virtual sensor design
The quality of tissue paper is based on certain characteristics: softness, strength, and thickness. Uniformity on the jumbo reel is critical to manage fluctuations-that is, the variations that are created along the papermaking process. “We have always used laboratory measurements to evaluate these characteristics,” explains Matteo Notini, corporate automation engineer at Sofidel, “but these present some problems. First of all, it is necessary to wait for all rolls and reels to be produced, so time is needed; secondly, sample evaluations are made, which, as a result, cannot be representative of the totality of products.” The way to overcome these problems is to rely on digital.
The project undertaken with Valmet was aimed precisely at the possibility of using virtual sensors capable of obtaining in-line parameters of paper characteristics. “Our target was first of all to focus on the paper characteristics that are important for our paper mill. By using the data collected by the sensors, which are then processed and stored at the cloud level, it is possible to analyze the predefined parameters of the paper without having to wait any longer for the production time of an entire reel to know its characteristics.” In this way, one is able to achieve continuous monitoring of the paper web and certainty that it meets the set specifications, particularly with regard to the tensile strength parameter.
Improved line management
The expected benefits, says Notini, are several. “First, as anticipated, there is less waiting time for laboratory values. We then have-and this is a key aspect-a greater adherence of characteristics to the finished product. Also, another very important aspect, we can also have a greater understanding of the process and papermaking, because we have the possibility of using a tool that helps us immediately understand what impacts the changes made have on the finished product.”
The idea of the project, explains Jonas Johansson, head of Tissue Industrial Internet at Valmet Tissue, is to use a set of data collected in the systems and, with the use of special algorithms, make calculations and estimates, i.e., make predictions of values that, otherwise, would not be possible to measure. “The advantages we get from virtual sensors are a greater understanding on the part of the operator of what is happening within the process, then the ability to have immediate feedback on the values entered. This also allows for real-time analysis of production and having greater responsiveness precisely for the purpose of improving it. Thus, by receiving the necessary information from us, operators can be faster and make the right corrections; we can even help implement corrections on the process that are the same for all operators.” Real-time prediction of quality parameters also enables the property to optimize raw material utilization as well.
Project preconditions
Before starting the project, Sofidel determined and verified that all the necessary preconditions were in place to start it. “A transformation and digitalization project like this is quite complex,” Notini emphasizes. Different parts of the production process are involved and interconnected, which it is not always easy to make work together. Not to mention the mill’s need to integrate the new systems with those already in the plant. “Our previous systems were updated and modified so that we could integrate them in the right way with the digital solution proposed by Valmet. In addition to the integration, the customization aspect was also worked on,” such a digital solution, in order for it to meet the organizational needs of the plant, had to be customized to the peculiarities of the paper mill. Another critical point, Sofidel’s engineer recalls, were the unforeseen contingencies or design changes that came up during the course of the project.
“When we started this innovation journey, we defined selection criteria both in terms of the project itself and the personnel involved. We wanted a state-of-the-art update of the systems and excellent management of our data repository. In addition, it was necessary to have a dynamic staff that was very attentive to the management of this data, so we needed someone to help us use it in the right way.”
The need then was to have the assurance of relying on a company that had the know-how of production processes, with skills and knowledge in data management and systems engineering. “But most of all, we wanted the presence of a dedicated team that would develop the necessary algorithms and then set up an operational structure to help us not only implement the system but also to use it later over time.”
Project phases
The project kicked off in 2021. “We worked together with both the global team and Sofidel’s paper mill team,” Johansson says, “agreeing on a data collection framework to identify key ones, identify potential for improvement, and then build a model that we implemented and delivered between late 2022 and early 2023.”
The first step then involved identifying the data. “We needed to ensure that the quality of the data was good and also that the quantity of the data available was appropriate for the target.” The intent, the Valmet manager explains, was to understand what the mill’s use of this data would look like, in order to make a preliminary model for accurately predicting the characteristics of the paper produced.
“With Sofidel’s data, we were able to build a reliable model from the very beginning, from the early stages of the project. The measurements were adequate and met the requirements.” Another interesting aspect then emerged, “The operators were active and dedicated to adjusting the parameters according to the needs of the site, and the quality of the paper was also good.”
In implementing the system, Valmet also had to take into account an explicit request from Sofidel, which became a key aspect of the project, “ownership did not want additional interfaces. The new analytics had to be available to operators already in the existing systems. So we implemented the planned improvement in the cloud, making it accessible to operators in their daily work.” The choice was to build simple solutions to enable them to be more aware of the situation and the possible need to intervene when critical issues arise. “We made sure that the information was provided directly to the operators.” These, thanks to the new system, always have a real-time view of the situation: from product quality validation to expected quality. “But not only that, we also wanted to help operators understand what to do in case an improvement is needed. So to manage the calculations, to understand what parameters are adjustable and how to manage adjustments to achieve an improvement in the production process.”
Since the ultimate goal of the project is improvement in reducing variability, Johansson continues, “we focused on the values that, over a given period of time, were outside the expected range and verified how those results were also confirmed by our prediction data.”
The model developed through detailed process, experimental, and validation analysis thus aims to provide operators with a tool that decreases the variability of tensile strength with real-time information that is available on the Valmet DNA DCS interface.
Future Applications
The potential offered by this digital technology for the future is also interesting, “there is good room for improvement starting with process variables such as softness and creping,” Notini declares. “Virtual sensors can also allow us to monitor in real time parameters such as the energy consumption and balance that we are getting from a machine, the use of resources and the consequent management of the attached costs; or even sustainability, trying to understand what the impact of our process is, for example, from an environmental point of view, so that we can limit it and also predict what the impact of our production will be.”
Going further, Notini points out, it is also possible to consider a number of other KPIs – key performance indicators – giving the production process a completely new perspective. “We are trying to imagine a new vision of processes that has to do with a forecasting mode capable of opening up completely new scenarios.”
This technology, Johansson adds, “is flexible and also opens up new opportunities for other types of virtual sensors that can be useful within paper mills and within the paper industry.”
Virtual sensors thus provide a look inside the process and offer solutions to significantly improve processing. “They can best support operations teams by helping them choose the right settings to achieve the highest possible efficiency. Therefore,” Notini concludes, “we are convinced that these sensors will be a game changer in the paper industry.”