Digital Process Monitoring of Stationary Processing Equipment—A Step Toward an Optimized Digital Processing Plant

In response to the current trends of digital transformation in the raw materials extraction and processing industry, IFE Aufbereitungstechnik GmbH has set itself the challenging goal of developing robust and versatile sensor systems for monitoring the condition of their predominantly vibrating stationary processing units. It shall be used as a basis for establishing a predictive maintenance scheme and ultimately facilitate the reduction of equipment downtime. In the course of the research project, research activities will be directed in particular toward a better understanding of the operating state of vibrating machines, especially based on their vibration patterns. Namely, a range of experiments with various vibrating screens and vibratory conveyors as well as an already ongoing extended case study on a linear vibrating screen currently operating in a waste treatment plant will be conducted. Furthermore, DEM simulations of vibrating screens as well as a specifically engineered “Laboratory Vibrating Machine” will be part of the research process. The sensors in use to identify and measure the required vibration patterns will be newly developed vibration sensors called “Sapient Edge Sensors”, which, later on, will be combined with other sensors for measuring different parameters.


Introduction
The company IFE Aufbereitungstechnik GmbH, hereafter referred to as IFE, is an Austrian company that has been producing machines for processing bulk material for over 70 years. Its products revolve around three main areas of expertise: vibratory conveyor technology, screening technology, and magnetic technology. A large share of the thusproduced equipment is applying vibration to perform its screening, conveying, or sorting task.
To keep up with the current trends of digitalization in the industry [1], numerous business partners have approached IFE with the need for a digital system for monitoring their vibrating equipment. The company, too, was searching for a way to monitor and assess the parameters of the vibration their machinery produces. A thorough market analysis revealed that a certain demand exists for developing a suitable condition-monitoring system specifically tailored to vibratory equipment. This new system should be used to set up a predictive maintenance scheme as well as optimize equipment efficiency by reducing machine downtime. IFE once again brought the Chair of Mineral Processing on board as a scientific research partner, the research partnership has existed for many years to the benefit of both institutions. The cooperative research work is carried out within the framework of the first author's dissertation project. The start-up company eSENSEial Data Science GmbH, also based in Leoben, has docked on as a development partner for sensor technology.

State of the Art in the Maintenance of Vibrating Machinery
The currently widely used definition for maintenance, according to DIN standards, includes the tasks of maintaining and restoring the target condition of the equipment as well as assessing its current condition state and predicting future ones [2]. Nevertheless, many mineral processing plant operators still see maintenance solely as the task of scheduling and performing repair works. That is primarily due to the lack of readily available real-time machine and process monitoring data. This lack of data in turn is caused by the rough environments for electronic equipment, the mineral and waste processing industries usually entail. These environments make the use of precise sensors to measure vibration or other machine properties a cost-intensive act. However, currently, some predictive maintenance systems for vibrating machinery do already exist. They usually use vibration analysis, although only for predicting failures of some machine parts, such as bearings, engines, or gears [3][4][5]. The present systems available on the market do not involve a comprehensive analysis of the current condition state of a complete vibrating machine.

Development Objectives
The sensor system, as the intended outcome of this dissertation project, should be able to provide the necessary insights into the condition state of the monitored vibrating machines. Thus, it can be used to optimize their operational settings, predict wear part life as well as future failures of parts, and improve the overall efficiency of vibrating equipment. The basis of this tool will be IFE's new vibration sensor, which should be combined with various other sensors for the monitoring of different parameters. The combination of many types of machine monitoring data, gathered by the different sensors in real-time at the vibrating machines, will be used to derive the necessary conclusions. These conclusions will be drawn from data processing, based on machine learning approaches to perform the analysis autonomously.

Vibration Analysis by Using a Sapient Edge Sensor
The analysis conducted in this cooperative research project primarily focuses on vibration data collected at the frame of an industrial vibrating screen as it is in operation in a processing plant. Therefore, the applied type of sensor needs to operate in a lower frequency range than other vibration sensors, which are used for the monitoring of bearings [3,5]. Furthermore, it has to be able to withstand the surrounding environmental conditions at the screen's frame.
To fulfill these and other specifications, the "Sapient Edge Sensor" , in short, "SES" , was developed by eSENSEial in cooperation with IFE. This new sensor can measure vibration at the screen's frame, mathematically calculate its vibration pattern, derive the most important characteristics from that pattern, and send them to a database. By computing the vibration data directly at the sensor and only transmitting the main vibration characteristics instead of the vast amount of measured position data, the amount of stored data is drastically reduced. More details can be found in the contribution "Entwicklung eines Schwingungsüberwachungssystems für Vibrationsförderrinnen und Siebmaschinen" in the section "Praxis und Wissen" (this issue).

Scientific Approach
Vibrating machines usually apply different kinds of vibration for different tasks, thus it is advisable to focus the research on one type of vibration at a time. Hence, the first phases of the project only deal with vibrating screens and among them primarily linear motion vibrating screens. Scientifically assessing the condition state of one linear vibrating screen by using vibration data, provided by SES, already poses some significant challenges, though.
Literature shows that research about the application of vibration monitoring for the same purpose is based primarily on a range of laboratory experiments, simulations, or a limited number of experiments with industrial-size ma- Fig. 1: Overview of first research objectives and planned scientific measures to achieve them chines [6]. The experiments conducted mainly involve the use of high-sophisticated vibration sensors, which are not suitable for permanent use in industrial environments [5,7]. The resulting data is then labeled for analysis, which is a straightforward procedure, since the data is generated in experiments or simulations with pre-designed settings and known boundaries. These settings and boundaries can directly be used as labels for the vibration data points.
However, the main focus of this research project is the interpretation of monitoring data from a vibrating screen that is fully operational in a processing plant. The procedure of interpreting and labeling the thus gathered data to set up a machine learning algorithm poses a significant challenge, since the settings of the screen as well as the boundaries of the system are subject to variations. Primarily, the oscillation pattern of a vibrating machine is defined by the motors that drive the movement as well as the whole construction of the screen. Secondarily, since the machines are used to process bulk material, the feed stream, too, affects the vibration pattern. Additionally, other varying influences can be process liquids, such as water or slurry, that might dampen or intensify the vibration of the screens. Therefore, to comprehensively deduce conclusions from the vibration data, a reduction in variable system boundaries is required and will be accomplished by the use of various approaches to the issue, the first of which are shown in Fig. 1.

Experiments on Various Vibrating Screens
By conducting small ranges of pre-defined experiments on different industrial-sized vibrating screens, with the new SES mounted on their frames, the effect a steady feed stream has on the vibration pattern has already been determined. The goal of these preliminary experiments was to distinguish between the idle state of the screen and its load state only by interpreting the gathered vibration data. The thus gathered data shows a clear correlation between the feed stream and the vibration pattern. In other words, the changes in width and height of the oscillation ellipsis, calculated by the SES, allow a clear statement about the load state of the screen.
After defining the idle and load states of a screen, based on its vibration pattern, later on in the project, additional experiments will be conducted to recreate different condition states of the machine, chosen for the case study (see Sect. 5.2). The data of such experiments can be considered as easy to label, since the settings and system boundaries of the experiments can be tailored precisely to the problem at hand.

First Case Study on Continuously Operational Screen
In the center of the first project phases, one beta installation is required to continuously test the sensor system and obtain the first data points to start the analysis. The machine selected is a linear motion vibrating screen, which is usually in operation for about 16 h on weekdays in a waste processing plant. Its combined use simultaneously encompasses the dewatering of slurry, washing of feed as well as ordinary screening. After equipping the screen with six SES, placed evenly distributed on the frame of the screen, and setting up the required connection to the database, the collection of continuous real-time data could begin. The data gathered from this installation entails all the previously described variable system boundaries and screen settings. To iteratively reduce the influencing factors on the vibration pattern, a series of further project steps will be taken.

Experiments on a Laboratory Vibrating Machine
To draw additional conclusions about the industrial vibration monitoring data, a specific machine was developed at the beginning of the research project, called a Laboratory Vibration Machine (LVM), depicted in Fig. 2. The LVM features spring damper elements and electric motors that are identical in construction to those of state-of-the-art industrial vibrating screens. It offers the possibility of manually adjusting the vibration angle, frequency, and amplitude as well as the option of creating mountings on the construction. However, it does not feature a mesh surface for real screening operations. Hence, it ought to be used to repli-

DEM Simulation
Since many condition states of vibrating screens have to be attributed to fluctuations in the feed stream, this factor needs to be individually investigated as well. First brief studies of the application of DEM simulations to understand the interactions between the feed stream and the vibrating screen show promising results. Therefore, this approach will be part of further in-depth research.

Perspectives on Future Research Steps
Soon the case study (mentioned in Sect. 5.1) will be used to test the application of additional sensors, first of all, an oil sensor. Later on, the research will be expanded to monitoring installations on multiple industrial vibrating screens of different construction types. In a parallel step, the application of machine learning for the interpretation of the gathered data will be investigated. The final step of the project will be the expansion of the monitoring system to other vibrating equipment.
Funding. Open access funding provided by Montanuniversität Leoben.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.