Industry 4.0 and digitization in durability

Pioneers of big data analyses

Since the founding of Fraunhofer LBF as a laboratory for structural durability and before its inclusion in the Fraunhofer Society, Ernst Gaßner published on the importance of taking account of the complex, variable operating stresses as a precondition in the construction of lightweight structures. He noted that the key success factor is the correct coordination of material, design, production, and load. Right from the beginning, it is pointed out that these influences do not have a serial effect, but rather influence each other. Recognising this correlative complexity, for the last 80 years Fraunhofer LBF has worked successfully to identify and quantify the key influencing factors.

Ernst Gaßner’s main idea – i.e. jointly considering the load and load capacity or stress and stress capacity under a variable operating load – led to the development of the “8-stage block test”. This made it possible to use the testing machines available at that time to demonstrate stress factors on the test bench and the method remained the structural durability test standard up until the advent of servo-hydraulic testing systems. For the “Gaßner test”, knowledge of the operational load is required – a prerequisite that, in the period between 1940 and 1975, led to the development of the corresponding measuring devices needed to detect the operating loads. Such devices included the Svenson’s contact strain gauge and the classification methods. These are used to determine and compare the influencing variables that are relevant for the structural durability. Examples include the average values and frequency of amplitudes of different magnitudes of any load-time functions. The signals are consolidated through different mathematical methods and the results displayed graphically, whereby resulting information losses are taken into account. Today, one would no longer resort to a term such as “classification method”. Instead, it is now referred to as “big-data analysis” – but the procedures and results are the same.

One challenge in experimental structural durability is the handling of what are sometimes very small sample sizes. From a mathematical point of view, it is impossible to derive robust characteristics and specify accurate scatter bands. In the 1960s, Erwin Haibach presented the standardized Wöhler line based on preexisting experimental results and experiences. Today, this method would also be described as a result of big-data analyses.

Early digital twins

Numerous research projects have shown that one prerequisite for accurate lifespan assessment of cyclically stressed components and structures is the consideration of the manufacturing process chain or its effects on the cyclic material behavior and thus on the component behavior. To emphasize the importance of the interaction between material, design, production, load and their influence on the structural durability, a department called “Component-related material behavior” was founded in 1984 at Fraunhofer LBF; this name is still an integral part of the institute’s organizational chart.

With the emergence of computers, the analytical methods were expanded significantly, and towards the end of the 1980s Fraunhofer LBF, using numerical manufacturing simulation, was able to numerically examine the influence of different production parameters during rolling. The experimentally verified and validated models therefore made it possible to simulate, over the lifespan, the influence of manufacturing parameters, which are often inaccessible experimentally. This set the basics for a digital twin, i.e. the numerical mapping of a component with all its essential properties. The commercial marketing of a digital twin lay, however, in a different area.

“Wheel strength”

Since the mid-1970s, durability of wheels has been an integral part of the research portfolio at Fraunhofer LBF. In addition to the development of special wheel test benches, their further development or adaptation to current and future operating requirements, the development of a suitable sensor for detecting the wheel loads and load data analysis to derive the operational stresses for strength verification, this formative experiment has also been digitized and has been available to the user since 2000 as the “Wheel strength” software. “Wheel strength”, the structural durability of a digital twin wheel, replicates the experimental test in a virtual environment, reducing the number of time-consuming iteration loops with prototypes. Although this method for evaluating the structural durability of wheels can be described as an industry standard, it should be noted that, particularly for safety-relevant components, an experimental durability approval is mandatory.

Manufacturing influences on component-specific material behavior

The integration of manufacturing influences in the wheel evaluation requires interfaces with which the local component-specific material behavior can be taken into account. Whether the required information or data comes from measurements, preferably non-destructive, or from manufacturing simulations is secondary, as long as it helps qualify the actual material condition. In the field of cold forming, it has been repeatedly demonstrated that the integration of a forming simulation to determine the local component properties can be a suitable method. On the contrary, a casting simulation is only a first step since numerical prediction of imperfections such as voids, pores, and throttles is currently possible only to a limited extent. For safety-relevant and large cast products, the use of non-destructive testing technology is state of the art. Dr. Christoph Bleicher has been working very successfully on correlations between the findings from the non-destructive testing and structural durability, which in turn allows for an accurate numerical proof of strength or manufacturing control. This in turn helps determine if the decision on whether a component with imperfections must be declared as a reject can be made reproducibly, based on appropriate quantifiable properties. Intensive work is thus being done to correlate the results of non-destructive testing with the resulting fatigue strength and hence a digital twin in order to further refine the mapping of process-related properties, especially the lifetime, in a virtual environment.

An improved understanding of the individual production steps and their influence on the component-related material behavior and thus on the component behavior is advantageous for the durability approval.

Transient material behavior

In terms of structural durability, Industry 4.0 not only means gaining access to the quantifiable properties for optimal production, but also process-related deviations in order to estimate the actual, component-specific material behavior during operation. Here, it is important to be able to describe the influences and their effects that are relevant for the component-specific cyclic material behavior. Besides simulations to describe the influence of manufacturing, more and more non-destructive measurement methods are becoming increasingly interesting, since these methods and procedures not only detect the actual state of the material after production but can also determine changes in its properties under cyclic loading. The importance of methods of detecting and accounting for the transient material behavior increases as long as special events and their effects on the component-specific material behavior must be considered when approving a component. This is because safer, reliable lightweight structures can be realized only by jointly considering the load and load capacity or stress and stress capacity.

Virtual test environment

In 2005, LBF took a further look at the term ‘digital twin’ and decided to move from the digital component to the digital system. LBF set out to realize a holistic virtual test environment using the example of a 12-channel axle test bench, which was later implemented by Marc Wallmichrath. This required linking of simulation techniques from widely different disciplines. In addition to the numerical analysis of stress and structural durability, modeling examples from the fields of hydraulics, control engineering, software as well as structural and system dynamic, which interact via suitable interfaces, should also be mentioned here. One focal point of the research activities was therefore the realization of a common platform for the integration of the individual simulation programs, the control of data transfer, and the targeted evaluation and visualization. The developed methodology for constructing digital twins of test systems was subsequently implemented in a large number of industrial projects for the digital mapping of the test benches.

Big data in fatigue life curves

Evaluating the service life for applications with very high cycle numbers remains of particular importance as long as the trend towards an extension of the service life persists. This trend was not disregarded in LBF’s research planning, especially following the recommendation by Gaßner and Pries in 1941, stating that cyclic tests should be performed with at least N_G=1·10⁸ cycles since the fatigue strength can still deteriorate after N=1·10⁷ cycles. Subsequently, in 2005, Sonsino proposed to reduce the fatigue strength by 5% or 10% per decade after the break point if there are no experimentally verified test results for this range. To be able to carry out a lifetime assessment based on a continuous Wöhler line from low cycle fatigue to high cycle fatigue, Wagener presented the Fatigue Life Curve in 2017, which perfectly meets these requirements. This description method for a lifespan curve is based on big data analysis.

Industry 4.0 and structural durability

It therefore remains to be noted that, since its foundation as a laboratory for structural durability, the fundamental idea behind Industry 4.0 at Fraunhofer LBF has been inextricably linked to the developments in durability. Over the 80 years of intensive fatigue research, the possibilities of considering the component-specific material behavior in lifetime assessment have increased significantly, but the basic idea remains the same.