Every unplanned production stop, even of individual machines, generates additional effort, higher costs and lower output. Data-driven, continuous and best possible condition monitoring is an essential basis for maximum availability of machines and plants.
Machine monitoring
An essential use case here is continuous machine monitoring. The aim here is to use relevant data such as current, temperature or vibration to detect anomalies in ongoing operation at an early stage, ideally to be able to classify these anomalies and to detect possible errors in advance. This often involves signs of wear and tear, which are detected early and reliably using machine learning-based analytics. This then provides the opportunity to plan service & maintenance interventions in good time, for maximum availability at minimum cost.
Process monitoring
Another essential use case is continuous process monitoring. The aim here is to detect deviations from process parameters at an early stage and to be able to intervene in the process if necessary. Building on the existing control solution and the data that is usually already available, a model-based machine learning solution is again used to detect anomalies and classify them as far as possible. Where rule-based automation reaches its limits, ML enables completely new insights into previously unknown process states. With the result of being able to intervene in the process earlier and in a more targeted way.
On the basis of continuous condition monitoring of machines and plants, various use cases arise that pay attention to the goal of maximum plant availability, which in turn is the key to economic production. Ultimately, the condition of a plant is known continuously, which results in maximum safety for the people responsible for the process. In addition, errors or anomalies in the machine data can be detected at an early stage and the necessary measures can be initiated before the actual malfunction or even failure of the system occurs.
One application example is the automated monitoring of high-speed conveyor belts in intralogistics. One task is to monitor and predict the elongation of the conveyor chain elements, depending on various influencing factors such as speed, load, running time or temperature. This involves the early identification of individual damaged chain areas. This also leads to an illustration or securing of the service technicians' knowledge about the condition of the system, and thus also to a transformation towards continuous data-based monitoring of the system. The specific benefits of automated monitoring are a reduction in service and maintenance costs, as well as higher availability and thus productivity of the system. With a view to new business models, for example, the sale of availability in the form of new or extended service level agreements (SLAs) is made possible. After all, such IIoT services lead to new perceptible features for the end customers and a correspondingly higher customer loyalty.
Another application is the automated monitoring of fans in an electroplating production hall. In electroplating, venting is a production-critical process. For example, oxyhydrogen gas is produced, which poses an explosion hazard at critical concentrations. Acidic substances can also lead to corrosion of the equipment. There is also a responsibility in the area of occupational health and safety and employee health. The task is to continuously monitor the fans using smart sensors and ML-based data analysis. The path to be followed is from preventive, rule-based to condition-based maintenance. This results in a minimisation of unplanned production downtimes and a reduction of maintenance costs. The first benefit of this exemplary end-to-end solution for brownfield applications is continuous, automated condition monitoring. This is the basis for reducing or minimising inspection, maintenance and repair. In this specific case, it was possible to switch from a monthly inspection of the fans with a walk-through on the electroplating roof to a semi-annual inspection. Unplanned downtimes could be reduced and the availability of the plant increased. An IoT use case with specific benefits.
Based on the process data and their evaluation using ML technologies, it is possible in many cases to draw conclusions about the quality of the manufactured products. On the basis of selected process parameters, ML-based patterns can be recognised that can be assigned to a specific process state. Derived from this, tolerance ranges can be monitored, for example, and it is recognised at an early stage when tolerances are or will be out of line in the sense of a prediction.
Smart data analysis helps Grenzebach to achieve real-time quality assurance and predictive machine maintenance for its innovative friction stir welding systems. In this way, the machinery and plant engineering specialists make a contribution to raising 24/7 series manufacturing to a new level.
A rotating friction pin is the central tool with friction stir welding (FSW), the innovative seam welding process that Grenzebach has developed for lightweight metals such as aluminium and its alloys. Through friction and pressure, the pin generates the process heat required to make the metal malleable, which is then stirred along the contact point by the rotational action of the friction pin. Without the need for the addition of welding wire or inert gas, this creates a solid joint that is characterised by its long-term stability and its resistance to distortion. A requirement for this result is that the friction pin behaves as expected. Accurate tensile and pressure forces are critical for achieving the correct degree of deformation of the metal. Until now, quality control was carried out by the machine operator who visually inspected the weld seam after the FSW process - a time-consuming procedure whose success also depended heavily on the personal know-how of the user.
Real-time monitoring during the weld process
Technology developer Dr Carlos Paiz Gatica explains how anomaly detection works: The comparison of the reference model and the current process allows a quality assessment in real time. As a pioneer in the field of Industry 4.0, Grenzebach today makes use of intelligent data analysis processes that enable precise predictions. And for this purpose, they use a tailored Industrial Analytics solution from Weidmüller.
Our analytics software, which has been customised to meet the needs of Grenzebach, compares the forces recorded at the sensors during the weld process with an ideal reference data record. As soon as the system detects a deviation that lies outside the defined parameters, the machine operator is notified and immediately knows that something isn't right with the weld process. Manual inspection of each weld seam is therefore no longer necessary
To determine the reference model, Weidmüller worked together with the engineers at Grenzebach to assess the data sets of several hundred welding seams for their relevance and evaluated them using intelligent data analysis methods. A significant element of the analyses was provided by the know-how coming from Grenzebach. The Weidmüller software may well be able to predict a fault with a certain degree of probability, but to do so it always needs to have been classified beforehand. Only Grenzebach can determine whether an anomaly should actually be classified as a critical error or not.
Product quality and availability on offer
As well as carrying out quality control checks on the weld seams, the analytics software also records the process parameters of each part that is produced, thereby producing complete documentation. This is a significant benefit not only from the legal aspect but also in terms of traceability and reproducibility. The system also delivers a timely warning if a replacement of the welding pin would be advisable. Armed with this information, the machine operator can plan the maintenance schedule in such a way that any downtime is avoided.
“Alongside the minimising of waste that can result from a tool breakage, an important factor, particularly in machinery and plant engineering, is the availability of the machines,” emphasises Kress.
High-tech specialists Grenzebach can see several more advantages in respect of its business model going forward: “Firstly, we can offer our customers very accurate and quantifiable quality control as well as giving them a forecast about possible equipment downtime, enabling them to save on resources and costs. At the same time, we are in a position to implement data-driven services and effectively to use the product quality or the availability of the equipment as selling arguments,” explains Michael Sieren, FSW Sales Manager at Grenzebach.
Do you have questions about our products and solutions or would you like an individual offer? Simply fill out contact form. Our employees will contact you as soon as possible.