Introduction

The continuously increase of electric vehicles is well known and manufacturers are adapting their shopfloors. The shift towards e-mobility has consequences throughout the whole supply chain since it has brought a fundamental change in both - the design and quality requirements of a car’s drivetrain. The number of gears in the gearbox has been significantly reduced, while the noise behavior of the gearbox has gained even greater importance as a decisive quality characteristic. This has given rise to the requirement for a robust generating grinding process as well as 100% quality testing of the gears before they are installed in the gearbox. The minimization of the number of complaints during end-of-line testing is the primary goal. With the Speed Viper and R 300 machines by Klingelnberg there are profound hardware solutions for generating grinding and roll testing. This is an important success factor for quiet gears.

But when it comes to an early identification of noise problems in the drivetrain one has to take data analytics and its integration in the manufacturing process into account. The big vision here in particular is preventive quality. By evaluating sensor data of the machining process, it promises to predict whether a gear is ok or not ok. And if not, this approach comes with the right tools to have a quick result when searching the root causes of part issues. The goal is to increase the output of the grinding machines as well to reduce cost per part significantly. Additionally, the number of parts, which go into the roll testing process, is reduced.

On the way to realize the vision of predictive quality for gear production this contribution lists some important milestones. As we all know, many challenges must be overcome as we move toward such an advanced Industry 4.0 applications. One of the more challenging tasks is to continuously acquire and analyze data taken from the machining process of the machine tool. Did you know that a Klingelnberg generating grinding machine generates up to 1.5 GB of data per hour? This is roughly equivalent to two tons of printed paper. In addition to these IT challenges, detailed knowledge of the design and control-related conditions of the machine tool and the specific toothed gear machining process is required. Otherwise, the data analysis will only reveal banal correlations and no valuable causations. Indeed, added value can only be generated through a combination of IT technology, control engineering, and grinding process knowledge. We cannot expect out-of-the box solutions here and we need joint ventures between machine tool providers and their customers to get there.

Closed Loop 2.0 

A common concept in the gear industry is the Closed Loop between measurement machines and grinding machines. Measurement results of parts are used to calculate corrections of the movement of the grinding machine. By doing so, geometrical imperfections of ground parts caused by differences in temperatures or material inputs in a mass production process can be handled. This concept is depicted in Figure 1 by the example of the Closed Loop concept between Klingelnberg machines for generating grinding of cylindrical gears.

When it comes to phenomena on a microscale like waviness, it gets more interesting and out of the box solutions like the one mentioned before do not exist. Surely a roll tester identifies these effects. You will get suspicious amplitudes in the spectral analysis. Of course, the software of the Klingelnberg R 300 helps you very elegant to identify potential problems. The problem here is to identify the root causes of amplitudes since the grinding machine’s capacity is very limited in avoiding the creation of a certain waviness. The central question here is which dynamic system in the grinding machine has caused a certain micro scale effect on the flank. The quicker you find the answer the sooner you return to a stable production process.

Until today, the appearance of new amplitudes in the spectral analysis leads to many activities in order to identify the root cause. These activities are done – although software supported – in a manual way. Furthermore, they incorporate machine users and machine builders. A better approach to this process would be a data driven Closed Loop between roll testing machines and grinding machines. The challenge for such a data driven Closed Loop is the data acquisition on the grinding side, the feedback integration on the testing side and the data processing in between. The good news is that solutions exist for all of them. The bad news is they are not out of the box. What is more they can be implemented by collaborative projects with machine tool builders and their customers.

The problem for data acquisitions straight from the grinding process is enabled by edge devices. These devices are usually a combined hard- and software solution for machine tools. They enable data analysis directly at the point of origin, the machining process of the machine tool. They collect continuous data and subsequently processes these data with integrated software modules. Special applications on the Edge hardware can then preprocess and analyze this data. This reduces the network load, the data transfer from the machine to the local network, to a necessary minimum. Once high frequency data of the grinding process is available, this data needs to be related to the feedback of roll testing. Simply said every snapshot of a grinding process is correlated to a certain test result. The challenge here is to reduce the large amount of data to the very minimum. Furthermore, a deep knowledge on software systems for machine tools is required to process these data straight forward.

The advantages of the edge devices can only be fully realized, however, if their software is ideally integrated into the generating grinding production system. This is where data platforms like Klingelnberg’s GearEngine comes into play as a platform for toothing data. GearEngine brings the two worlds of machine data and process knowledge together. The extremely precise, but non-process-specific, Edge data are integrated into the specific toothing data, giving the user a comprehensive overall picture for improved decision-making processes.

Machine tool with Edge Device. 

 

Of course, there are many open questions to move towards predictive quality and quiet gears. But the hard- and software equipment for this vision is not a problem. If you go with competent machine tool builders and machine of the current state of the art, you get solutions quickly. The biggest challenge starts once the data acquisition is running. There you need to collaborate strongly with the machine tool builder in order to define the right activities data relations.

Conclusion

The next big concept after common Closed Loops is the data driven Closed Loop which enabled predictions based on correlations of tester and process data. Integrating IoT equipment like edge devices is an important step towards this vision. Step by step Klingelnberg is moving towards this vision of providing new, digital solutions that are easily integrated into customers’ production processes. The goal is to ensure that no one will have to deal with complex commissioning procedures and cloud connections in the future. Klingelnberg is a one-stop source for it all.

klingelnberg.com