Production at the Leading Edge of Technology

The Tailored Forming process developed in the presented research enables the production of axial bearing washers with AISI 1022M (C22.8/1.0460) base material and AISI 52100 (100Cr6/1.3505) cladding on the rolling contact surface. By limiting the use of the bearing steel to the highly loaded surface, significant amounts of alloyed steel can be saved. The cladding is applied through plasma transferred arc (PTA) welding and subsequently formed to improve its properties. The challenge in developing a hot upsetting process lies in the high difference in flow stress of the two materials, since the harder bearing steel is merely pressed into the softer base without sufficient deformation. In order to equalise the flow stress of both materials, an adapted temperature gradient is induced over the washer height before upsetting. Due to this, a higher cladding temperature is set while the base material remains significantly cooler. This is realised by means of local inductive heating of the cladding and different transfer times to the upsetting process. The process variants are applied in an automated forging cell and subsequently evaluated in metallographic analysis of cross sections after welding and after forming. The results show the most favourable material properties after forming when local inductive heating of the cladding is simultaneously combined with cooling of the base material and the transfer time between the heating stage and forming is minimized.

Many metallic powder materials can be processed fast and cost-effectively using the conventional powder metallurgy method. For this purpose, the metal powder is pressed into a compact and then sintered in a furnace to produce a finished component. Gamma titanium aluminides are an exception to this. Due to their brittleness, they cannot be compacted in classical die pressing. A promising approach is the addition of elemental significantly more ductile alloy powder. The aim of this work is to investigate the influence of the admixture of elemental powder on the compressibility and the properties after the sintering process. Within the scope of the work, commercially available pre-alloyed TiAl48-2-2 (GE48) powder, which is applied e.g. for turbine blades in aircraft, is used. The powder alloy is mixed with elemental titanium, aluminium, chromium and niobium powder according to its composition and then pressed to a compact. Selected samples are sintered and metallographically characterised. By varying the pressing load and the proportion of elemental powder, as well as the proportion of elemental powder mixtures, the influence on the compaction behaviour and the mechanical properties is investigated. It is possible to produce compacts with sufficient mechanical properties by adding specific proportions of different elemental powders depending on the element and the compaction parameters. The results show a significant dependence of the relative density and tensile splitting strength on the proportion and type of elemental powder added.

Process-induced residual stress is a major challenge in today's additive manufacturing (AM) processes, such as powder bed fusion by laser beam melting of metal. After the AM process, the exact stress state is usually unknown, and parts often require heat treatment to relieve residual stress. In-process measurement of residual stress is currently not possible. This paper presents a concept to derive the measurement of the residual stress by analyzing the structure-borne sound induced during the AM process. The first step of the concept is to integrate a device into a build plate to set a defined mechanical load during the manufacturing process. Then, samples can be fabricated on this build plate in several steps. By applying mechanical load with the device, the stress state in the samples can be changed between the fabrication steps. During this stepwise fabrication process, the structure-borne sound signal is recorded. Subsequently, the correlation between the stress states and the acoustic process emissions is analyzed using FFT, STFT and cross-spectral analyses. The overall goal is to establish a model to determine residual stress in AM components by evaluating the acoustic process emissions.

The combination of aluminium (AlSi1MgMn) and titanium (Ti6Al6-4V) allows producing components with high lightweight potential and at the same time high strength and chemical resistance. Upon joining of dissimilar materials, intermetallic phases (IMP) can form. These are comparatively hard and brittle and represent a weak point in the hybrid component. Along the process chain for manufacturing a hybrid bearing bushing made of AlSi1MgMn and Ti6Al-4V by co-extrusion, die forging and heat treatment, the joining zone is exposed to high thermal loads. As a result, the individual process steps can lead to the growth of IMP reducing the compound’s quality. In order to investigate the formation and the growth of IMP at process-relevant temperatures and contact times in detail, experimental analogy tests were carried out. Subsequently, the specimens were examined by scanning electron microscopy. Due to the constant temperature and the respective contact time, the diffusion coefficient was calculated from the determined phase thickness using the Einstein-Smoluchowski equation. This allowed describing the diffusion coefficients as a function of temperature and implementing them into a finite element model via a subroutine. To validate the subroutine, further tests were carried out and the calculated phase thickness was validated with experimentally determined phase thickness, which exhibited good correlation.

Planning of machining operations for fibre reinforced plastics components today entails expensive trials in order to meet quality, productivity and cost requirements. The use of existing data for modeling and simulation has so far been severely limited due to the lack of universal process models that capture fundamental mechanisms in a process-independent approach and thus allow data to be merged across different cutting processes. Recently, a universal model describing the engagement conditions in oblique cutting of unidirectional FRPs has been developed. The model closes the gap described and builds the basis for cross-technology data merging from different cutting operations, which has been common practice for homogenous materials for a long time. In case of mostly thin FRP components and often poor clamping conditions the generated forces in cutting operations are crucial as they may lead to dynamic process instabilities and to unfavorable part deflections impeding part precision. Furthermore, the quality of the machined surface depends on the engagement conditions, which usually change significantly during machining. Force and quality data from different sources and across various cutting processes and FRP materials were merged using the universal engagement model to reveal generally applicable relationships. These will enable faster, more reliable and more cost-efficient planning of cutting operations for FRP components in the future.

Prestressed, fibre-reinforced mineral cast has the potential to replace steel in the field of structural components for machine tools. High damping properties, a low density compared to steel and adjustable creep properties coupled with a low CO2 equivalent and a low primary energy requirement make the hybrid material a suitable material for the future. However, the pre-stressing of the reinforcing fibres requires high tensile forces corresponding to around 5% of the compressive strength of the mineral cast. The installation situation, which also schedules integrating the mechanisms for a subsequent readjustment of the prestressing into the machine tool, requires a minimum installation space for the prestressers with a maximum base area of 25 × 25 mm2 and a minimum height. The problematic clamping conditions of the carbon fibers, which should only be loaded along the fibre direction due to their low transverse strength, require a novel tension mechanism. In this paper, specially developed pretensioning mechanisms for the carbon fibers used as reinforcement are investigated and the different mechanisms are compared to each other.

The manufacturing of high-quality cold forged components with respect to environmental and economic requirements requires a high process reliability and long tool life. For this purpose, digitization offers new opportunities to increase process understanding by implementing inline data acquisition. Measurement of process parameters close to high loaded forming zones poses a serious challenge. Furthermore, tool wear cannot be detected inline. Therefore, integrated and wear-resistant thin-film sensors represent an innovative approach to realize in situ data acquisition. Hence, a sensor design will be introduced enabling both temperature measurement with spatial resolution and real-time wear detection directly in the forging zone. First, a thermal simulation is performed to determine the temperature range expected in the forming zone to conceive sensor structures and layers. Subsequently, various standard tool coatings are evaluated regarding their tribological and mechanical characteristics. Finally, the embedded system was conceptualized to read out the sensors and send the preprocessed data via USB-interface or BLE.

About 95% of all steel products are rolled at least once during their production. Thus, any further improvement of the already highly optimized rolling process, for example reduction of energy consumption, has a significant impact. Currently, most rolling processes are designed by experts based on their knowledge and heuristics using fast analytical rolling models (FRM). However, due to the complex interactions between the processing constraints e.g. machine limits, the process parameters as well as the product properties, these manual process designs often focus on a single optimization objective. Here, novel methods such as reinforcement learning (RL) can detect complex correlations between chosen parameters and achieved objectives by interacting with an environment i.e. FRM. Therefore, this contribution demonstrates the potential of coupling RL and a FRM for the design and multiple objective optimization of rolling processes. Using FRM data e.g. the microstructure evolution, the coupled approach learns to map the current state, such as the height, to process parameters in order to maximize a numerical value and thereby optimize the process. For this, an objective function is presented that satisfies all (technical) constraints, leads to desired material properties including microstructural aspects and reduces the energy consumption. Here, two RL algorithms, DQN and DDPG, are used to design and optimize pass schedules for two use cases (different starting and final heights). The resulting pass schedules achieve the desired goals, for example, the desired grain size is achieved within 4 µm on average. These meaningful solutions can prospectively enable further improvements.

Hot forging dies are subjected to high cyclic thermo-mechanical loads. In critical areas, the occurring stresses can exceed the material’s yield limit. Additionally, loading at high temperatures leads to thermal softening of the used martensitic materials. These effects can result in an early crack initiation and unexpected failure of the dies, usually described as thermo-mechanical fatigue (TMF). In previous works, a temperature-dependent cyclic plasticity model for the martensitic hot forging tool steel 1.2367 (X38CrMoV5-3) was developed and implemented in the finite element (FE)-software Abaqus. However, in the forging industry, application-specific software is usually used to ensure cost-efficient numerical process design. Therefore, a new implementation for the FE-software Simufact Forming 16.0 is presented in this work. The results are compared and validated with the original implementation by means of a numerical compression test and a cyclic simulation is calculated with Simufact Forming.

Surface layers of forging dies are subject to thermal, mechanical and tribological influences during forging. These loads occur combined and result in a variety of tool damages, which shorten tool life. The predominant cause for tool failure is wear. While abrasive wear and crack formation directly cause tool failure, adhesive wear can be equally disruptive as it results in a geometrical deviation of the tool and the formed work piece. However, adhesive wear can also be beneficial in acting as a regenerating, protective layer to the surface of forging dies. This paper deals with the influence of process parameters and billet material on the formation of adhesive wear on forging dies. As adhesive wear is facilitated at elevated temperatures, high thermally loaded dies with a mandrel geometry are investigated in forging tests. During forging, thermal, mechanical and tribological loads on the tools are varied by changing cooling parameters, steel billet material and lubrication strategies. The study presents adhesion-promoting process parameters and tool areas of increased adhesive wear. The results show, that the formation of adhesive wear occurs predominantly at high tool temperatures and in areas with increased material flow, while lubrication and the billet material show little to no impact.

Collars with internal threads are used in a wide range of products. For the production of threads in sheet metal, collar drawing with subsequent thread cutting or forming is used in progressive tools. For a reduction of the manufacturing costs, collar drawing and thread forming can be combined into a single process step. Since the punch for collar drawing must be retracted after thread forming, an undersized collar must be drawn. The thread forming must be adapted to the collar as well as the process kinematics. Thus, FE simulation is used to analyse the process interactions. Parameters such as feed rate, rotational speed, pre-hole diameter, die diameter, sheet thickness are varied, and their influence on the process are investigated. Based on the findings the best combination of the process parameters is determined. The results will be used to manufacture a tooling system for collar drawing with integrated thread forming in the future.

The quality of deep-drawn parts is subject to uncontrollable fluctuations, triggered by material property variations and process deviations, which occur despite extensive quality controls along the entire process chain. Monitoring and controlling the draw-in of the sheet material—which is an indicator of a faultless deep drawing process—would allow for a significant increase in process robustness. However, this requires sensor systems suitable for the industrial environment, which so far do not exist. This paper presents a newly developed inductive sensor in thin-film technology for measuring the flange draw-in. The sensor was designed with the aid of finite-element-analysis and then manufactured using thin-film processes. After integration into a deep-drawing tool, the system was tested and validated. Afterwards, the detection of typical deep-drawing defects was investigated. It was demonstrated that the sensor system can reliably detect both cracks and wrinkles as well as the time at which they occur.

A newly developed in-situ-hybridization single-step process for the manufacturing of formed fiber-metal-laminates (FML) was introduced in previous works. During the deep drawing process, the fabric layer is infiltrated with a low-viscous thermoplastic matrix in a resin transfer molding process. The matrix polymerizes after the forming is completed. First parts could be manufactured successfully, but the influence of many process parameters continues to be unknown. The interaction of fiber and metal layer (DC04) on the formability of the FML is experimentally investigated by the deep drawing of FML parts without matrix injection. Parameters tested were the blank holding force, tool lubrication as well as different surface treatments of the metal sheet. Fiber breakage was observed after deep drawing of the dry FML. The deep drawn metal sheets were analyzed by surface strain measurements. The formability was then assessed by comparing the measured surface strains to a forming limit curve obtained by Nakajima-tests of the metal-fiber-metal stack. The results of the parameter investigation during dry deep drawing are analyzed to understand the influence of the process parameters on the in-situ hybridization process containing matrix injection.

The socio-political demand for more sustainability is putting a lot of pressure on the packaging industry. In addition to logistics and marketing, packaging today serves preparation, resealability and more purposes. Though more sustainable, paper packaging produced by folding, for example, falls far behind plastic packaging in terms of geometric variety, among other things. These deficits are to be remedied by sustainable formed packaging. The determination of process parameters and material settings in the forming process for paper is complex due to the inhomogeneous natural material and limited formability. Consequently, process layout is mainly based on empirical knowledge. Moisture affects the forming behavior of paperboard significantly. Therefore, process simulations at different moisture levels and distributions within a sample allow more targeted selection of process parameters. This paper paves the way for simulations of cardboard behavior at different moisture levels and reveals the influence of moisture distributions on properties of deep-drawn products.

Deep drawing is a common forming method, where a sheet metal blank is drawn into a forming die by a punch. In previous research, conventional deep drawing was extended by the introduction of an additional force in the bottom of the cup. The force transmission initiates a pressure superposition in critical areas resulting in a delayed crack initiation. For numerical investigation of the considered process, an accurate modelling of the material failure is essential. Therefore, the parameters of the modified Mohr-Coulomb criterion were identified for the two high-strength steels HX340LAD and HCT600X by means of tensile tests with butterfly specimens. In this research, the fracture modelling is applied in the simulation of deep drawing with and without additional force transmission to enhance the failure prediction. The fracture criterion is validated by experimental deep drawing tests. Finally, the influence of the additional force on the prevailing stress state is evaluated.

Increasing weight and energy efficiency requirements drive the use of novel composite materials. For this, metal-polymer sandwich plates are particularly promising. However, their widespread application is hindered by limited formability and a lack of efficient joining technology due to the combination of materials with vastly different mechanical properties. This paper presents an innovative joining element, addressing the special characteristics of sandwich composites by locally compensating the influence of the polymer core. The presented joining elements act as “lost punches” inserted into the sandwich material, opening up the possibility of manifold connection possibilities. In a two-stage hole-flanging process, which can be realized with conventional forming machines, a form and force fit between the punch and the sandwich composite is established. This challenging forming process is discussed, extensively numerically analyzed and the joint geometry is optimized with respect to the resulting joint strength. Furthermore, the achievable limit loads are discussed. Concluding, first prototypes offer an outlook on the industrial application.

Deep rolling is an efficient process to increase the service life of highly stressed components such as crankshafts or roller bearings by inducing compressive residual stresses. The residual stresses correspond to the deep rolling force applied. Monitoring the deep rolling force enables the processing result to be assessed. The rolling force is a two-dimensional vector. However, current approaches only allow the measurement of one dimension. Thus, this article presents a force-sensitive deep rolling tool that can measure the applied deep rolling force in two axes. This article, describes the principle of the sensory deep rolling tool and its calibration process. Finally, the sensory properties are evaluated.

In the freeform bending process, the obtained geometry is determined by the kinematics of the bending head. In order to derive the motion profiles leading to a part within tolerances, the correlations between bending results and machine settings are investigated. This inverse problem is solved by generating calibration curves, whose aim is to correlate the bending radii with the head movement from experimental tests results. Nevertheless, slight adjustments in the shape of the calibration curves show a significant impact on the bending results. In this paper, the robustness of the calibration process is investigated. First, the effect of different interpolating methods is considered. In addition, the influence of the experimental points is examined by comparing the performance of global and local data for the interpolation. Finally, calibration curves obtained with different ratios between the translation and the rotation degrees of freedom are compared. In this way, the interactions between the given parameters are investigated and a more efficient process for calibrating the freeform bending machine can be determined. This allows the reduction of the experimental effort for determining the relation between the machine parameters and the bent result as well as optimizing the process with respect to the geometrical deviations and dimensional stability.

Punching is one of the most commonly used production processes in sheet metal working industry. Here, major criterion for the quality of cutting surfaces is a high clean cut proportion. However, the disadvantage of conventional punching processes is that they can only produce clean cut proportions up to 20–50% of the sheet thickness. Until today, more complex processes such as fine blanking are therefore required for a higher cutting surface quality. The content of this paper is a numerical and experimental investigation for a new tool design called “concave punch nose design”. The idea of the concave punch nose design is to optimize the cutting edge geometry of conventional punches in order to enlarge clean-cut proportion along the cutting surface despite a process sequence similar to conventional shear cutting. The numerical and experimental investigations presented in this contribution show, that the concave punch nose design increases compressive stresses in the shear-affected zone and therefore significantly raises the cutting surface quality. Compared to conventional punching, concave punch nose design increases clean cut proportions by more than 100%.

The economic efficiency of cutting processes is generally determined by tool wear. Empirical investigations on tool wear in bevel gear cutting, however, are limited to face milling plunging. Hence, the objective of this paper is to examine the influence of the process design on tool wear for an industrial application of face hobbing plunging. For this purpose, first the influence of cutting speed and feed on process characteristics is analyzed with the help of the manufacturing simulation BevelCut. Subsequently, cutting trials using the same process parameter variations are performed and their influence on tool wear and machining time is assessed. Finally, the simulation results are compared to the results of the cutting trials in order to identify correlations between the process characteristics and the tool wear.

The in-mold assembly process can be used for the production of lightweight hybrid components made of metals and plastics. The connection between the different materials is often realized by a form fit joint. Conventional through-injection points enable the load transfer between the materials. However, through-injection points have disadvantages in the transmission of multiaxial loads. Furthermore, notch effects often occur under load, which can lead to premature failure in the material interface. As a result, the dimensions of the hybrid component or the amount of through-injection points are oversized. In order to increase the bond strength, the use of a friction-drilled bushing was investigated. First, friction drilling tests for varied parameters were performed and analyzed. Second, lap shear tests on hybrid components for appropriate bushings were carried out. The findings obtained have been transferred to the design of a demonstrator. Here, the connection quality between metal and plastic was determined by means of quasi-static and impact load tests. The joint using a friction-drilled bushing thereby confirms the advantages of the enlarged effective area for load transfer compared to conventional through-injection points.

In drilling EDM different flushing methods and electrode geometries are applied to improve the process stability and avoid form deviations as a consequence of discharges in the lateral working gap. These methods are usually reviewed indirectly by determining target parameters of the process, the tool electrode or the bore hole. A localization of discharges along the bore hole however provides a quantitative measure for the ratio of discharges in the frontal and lateral working gap. This paper presents a setup for localizing discharges by use of a workpiece electrode consisting of electrically insulated segments and performing signal analyses. Tool electrodes with interior and exterior flushing channels are analyzed with this experimental setup to compare the effectiveness of the resulting flushing conditions. It was found that the use of a helix tool electrode leads to an increased material removal rate by at least 11% and a considerable reduction of lateral discharges.

Due to the attractive material properties such as high specific strength and very good corrosion resistance, titanium and its alloys are frequently used in the aerospace, automotive and chemical industries. However, the low thermal conductivity leads to high thermomechanical tool loads during machining, making an optimal cooling lubricant supply to the cutting edge essential. For an optimized design of the tools a realistic simulation is necessary. Since the viscosity of deep hole drilling oils is strongly temperature dependent, it plays a significant role in the comprehensive simulation of the process. In this study the thermomechanical tool loads on TiAlN-coated helical deep hole drills during machining of the titanium alloy Ti-6Al-4V (Grade 5) are investigated and will serve as input for a three-dimensional finite element method (FEM) chip formation simulation focusing on the temperature distribution. The experimental investigations are carried out with successively varying process parameters of cutting speed, feed rate and cooling lubricant pressure. The knowledge gained in this study is of fundamental importance, as it serves as the basis for the future development of a fluid-structure interaction (FSI) simulation in order to be able to take the temperature influence on the cooling lubricant flow into account.

Milling is one of the most frequently used manufacturing processes. End milling cutters, which are available in a wide variety of designs and sizes, are used very frequently due to their universal applicability and low manufacturing costs. During NC programming and the associated path planning, the tool to be used for machining a specific area is determined. The choice of tools plays a crucial role here. In particular, their diameter has a significant influence on the efficiency of processing. In order to choose the best tool, it is necessary to know which tool can be used for machining a specific part of the differential volume. The geometric accessibility sets an upper limit for the tool diameter. This paper presents an algorithm for calculating the largest possible tool diameter for end mills for any point within concave polygons. The heuristic determines the largest possible tool diameter and the associated position of the tool guide point of the end mill within an adjustable error corridor. The algorithm allows for the efficient calculation of the distribution of the largest possible tool diameters for any arbitrarily shaped 3-axis milling feature. Possible areas of application are optimization of tool usage, automated tool selection, feature recognition, automated NC programming and automated process planning.

Climate change, scarcity of resources and sustainability are increasingly becoming the focus of social and political attention. In this context, the importance of timber construction in particular is increasing. The use of wood as a building material offers additional storage capacity for the greenhouse gas CO2, which is bound during tree growth. In this way, another extremely effective CO2 reservoir can be created alongside the natural forest reservoir. In order to promote the establishment and further development of timber construction, there is a particular need for action in the machining of construction elements. To this end, it is necessary to investigate and develop solution approaches in terms of machines, processes and tools in order to optimize the manufacturing processes in timber construction and increase productivity. In the industrial environment, wood materials are usually machined dry. Unfavorable process parameters can lead to thermal problems that can have a negative effect on the machining qualities. In contrast to dry machining, no scientific findings are yet available for wood machining when using minimum quantity lubrication (MQL). This paper discusses the results on the influence of the use of minimum quantity lubrication when grooving spruce beams. Within the scope of experimental tests, different process parameters were varied and the effects on process forces and surface characteristics of the workpieces were analyzed.

This paper contains the first steps for an analysis of the fluid dynamics in a circular sawing process with an internal cooling as part of an ongoing project. The analysis takes place with a test rig for orthogonal cutting of the IfW of the University of Stuttgart, which is modified to be able to create an artificial narrow closed cutting gap. It is built with sapphire glasses on the sides to make the fluid dynamics and the chip formation in the gap visible and analyzable. The investigation of the fluid dynamics carried out with the PIVLab software, which is a module in Matlab.

The use of fiber-reinforced plastics (FRP) is essential to meet future lightweight construction requirements in the automotive industry. Therefore, resource- and cost-saving as well as innovative manufacturing processes are needed, which enable large-scale series production. One example is the manufacturing of such components from glass mat-reinforced thermoplastics (GMT) and unidirectional fiber tapes (UD tapes) by combined compression molding and thermoforming. It may be necessary to insert multiple GMT pieces into the mold to improve mold filling and extend the process limits. In the forming process, the different material fronts of the GMT blanks collide and are joined by fusing and consolidating the matrix material of the composite. This area is called the weld line. In this paper, the influence of the weld line on the tensile strength of the workpieces was investigated. First, two GMT blanks were placed next to each other and formed in a heated plate mold. On the basis of tensile specimens that were cut across the weld line, it was found that the weld line causes a marked decrease in tensile strength. In further investigations, the UD tapes were added into the process. Due to the reinforcement of the UD tapes, the decrease in tensile strength due to the weld line was significantly reduced.

Clinching is a well-established joining technology, e.g. in automotive production, because of its cost-efficiency and the ability to join different materials at low cycle times. Nowadays, a detailed quality inspection of clinch points is usually carried out ex-situ, e.g. via macroscopic examination after joining. However, only 2D-snapshots of the complex three-dimensional and time-dependent forming and damaging phenomena can be made. The closing of cracks and the resetting of elastic deformations due to unloading and specimen preparation are also disadvantageous. In contrast, the use of non-destructive in-situ testing methods enables a deeper insight into the joint deformation and failure phenomena under specific load conditions. In this paper, progressive damage is observed during the single-lap shear testing of a clinch point using in-situ computed tomography (CT) and transient dynamic analysis (TDA). The TDA can continuously monitor the characteristic dynamic response of the joint, which is sensitive to damage and process deviations. In-situ CT creates 3D images of the inner structure of the clinch point at specific process steps. In this work, the sensitivity of both testing methods to detect damage in joints with EN AW 6014 and glass fibre reinforced polypropylene (GF-PP) is evaluated. As a reference, joints with both joining partners made of aluminium alloy (EN AW 6014) are analyzed. It is shown, that TDA and in-situ CT has the potential to identify joint quality as well as critical processing times.

The accuracy of the machined workpiece depends on the true conditions of the hydrodynamic linear guides. Linear guides based on hydrodynamic lubrication are still in use due to their high damping coefficient and high load carrying capacity. Measuring the true conditions of the hydrodynamic linear guide is important to achieve high accuracy. Until today, pressure measurement has not been established for linear guides. Oil film pressure is one of the important factors explaining the operating conditions of the hydrodynamic linear guides. The main goal of this study is to develop a method in which pressure sensors were installed in the lubrication gap to measure the oil film pressure under realistic condition of a hydrodynamic linear guide. A hydrodynamic linear guide testing rig with varying load capability was used as main test device. Multiple miniature pressure sensors were installed in a stationary rail in different manners to get the realistic oil film pressure distribution along the length of the slide. Additionally, the sensors were also calibrated hydrostatically and hydrodynamically with variable frequency and amplitude. Due to the problem of the enormous influence of air inside the lubrication gap of the testing rig, the measured pressure by the sensors showed that the numerical results have to be adapted to the experimental oil film pressure, which is lower. A new sensor’s integration method has shown great improvements in estimating the oil film pressure of hydrodynamic guides experimentally.

Various sensors as well as the numeric control serve as sources for the acquisition of process data in operating machine tools. Since the manufacturing industry acknowledges the value immanent to the data, numerous approaches to analyze and exploit large amounts of data have been developed. Generating comparable data sets represents a general challenge when collecting data in real production environments. Multiple external interferences, such as interventions by the operator, alter the manufacturing process and the data set. In order to ensure the transferability of results, a standardized preprocessing and the comparability of data sets, such interferences need to be eliminated by a synchronization algorithm. In this paper, a novel approach is presented, which allows for a synchronization of numeric control process data sets considering multivariate raw data. The approach is able to align data sets of different lengths considering local process modifications. Reliably synchronizing data sets, the presented algorithm aims to support data preprocessing in manufacturing environments and, thus, facilitates the application of data-driven solutions for production optimization.

Industrial product development today is faced with the challenge of achieving shorter creation cycles to keep up with international competition. This causes constantly changing requirements for the geometry of the components and thus for the used forming tools. These tools must be designed much faster so that customer requirements are met quickly, which is feasible through a parametric CAD design. As part of a cooperative research project involving the GFaI and the IFUM, a fully parametric CAD model for a sheet-bulk metal forming process was developed. With this tool it is possible to produce cylindrical components with internal and external gearing by combined sheet and bulk forming operations. For this purpose, the CAD model of the tool system is divided into different assemblies. Each assembly consists of various components which relate to each other. Furthermore, the dependencies between the assemblies were built up parametrically via global constrains. An initial structure of the CAD model including constraints is described in this paper. In addition, various process limits are determined by means of experimental tests and calculations. In the first stage of the forming process, blanks are deep-drawn into cups. Due to the geometry of the gears, round cup forming tests were conducted to examine the drawing ratio for different materials (DC04, DP600 and HC260LA). The characteristic values are converted into parameter limits for the new CAD model. Thus, the forming tool can be designed depending on the material used and the required gear size, which can reduce the development time in the future.

At present, optical components are costly and complex to manufacture. The costs often are a decisive factor in developing and manufacturing of optical components and sensors. The goal of the cluster of excellence PhoenixD, a major cross-disciplinary initiative, is the time- and cost-efficient production of optical systems. One promising approach is the accurate molding of micro- and nanostructures in a precisely controllable embossing process. Embossing as a manufacturing process for structured functional surfaces enables high output rates at low costs per component. However, embossing of micro- and nanostructures in particular requires high demands concerning the precision of the used machines and tools as well as on the precision of the positioning accuracy of actuated active parts. Machine- and tool-related disturbances are often unavoidable—these include guide inaccuracies, bearing clearances or temperature-related expansions in the powertrain. All these effects can be counteracted by means of an active process control. For this reason an embossing device is being developed which enables the die to be positioned precisely so that micro- and nanostructures can be transferred reproducibly with a high quality. In addition to the high positioning accuracy, this embossing device should also provide high embossing forces. This leads to an expansion of the material spectrum in microembossing and enables a variety of new applications. In this paper various concepts are presented and analyzed concerning their suitability for the precise embossing of fine structures by means of multi-body-simulation with regard to their deformation under load. In addition, a test bench of an electromagnet-spring system is introduced.

State of the art drive controllers, based on numerical and programmable logic controllers (NC and PLC), have not yet established standardized and easily accessible endpoints to capture status and process variables, such as motor current, torque or position values. Direct access to those data sources is limited to proprietary tools or licenses and requires modern control hardware. Besides, available data sources are limited to sample periods down to the NC or PLC cycle time, that varies between 1 and 10 ms. In this paper, we introduce a low-cost, low-tech and low-effort solution for monitoring feed axes based on contactless current sensing. We deploy split-core current transformers onto motor power cables of a variable frequency drive achieving sample rates of 50 kHz. This provides a retrofit solution for feed axes monitoring. Also, we outline the required signal processing to show the solution’s potential for further applications like anomaly detection. As a result, we enable a low-cost monitoring solution for machine tools using a physic-based model.

Conventional industrial robots (IR) represent a cost-effective machining alternative for large components. However, due to the serial kinematics and the resulting high tool deflections, they usually lack precision. Model-based simulation and control methods are used to increase the accuracy of IR regarding both planning and the process itself. The majority of the applied models include the compliances of the gears and bearings but neglect the deformations of the manipulator bodies. This paper introduces an approach to directly measure and evaluate the deformation of robot bodies in the presence of process forces. The measurement setup contains multiple Integral Deformation Sensors (IDS), which provide the change of length due to deformations of the respective body. Subsequently, the measurements are fed to a beam model (BM), which calculates the body’s 3D Cartesian deflections. The presented approach is validated by static tensile tests on a conventional six-degree-of-freedom (DOF) robot manipulator.

During machining, process forces cause form deviations on the workpiece depending on the interacting stiffnesses of all components involved. In order to avoid tolerance violations, it must be ensured that the resulting deflection of tool and workpiece are within the tolerance limit. At the same time, process parameters must be selected in such a way that a cost-efficient and productive production is possible. Stiffness can be determined experimentally by applying a specific force and measuring the resulting deformation. Due to the large variety of tools and tool holders as well as their combinations, it is generally too expensive to determine the stiffness in this way. During quality control, the workpiece geometry is measured, for example, with coordinate measuring Machines (CMM), so that resulting dimensional and form deviations can be determined. Furthermore, there exist approaches to predicting process forces based on machine internal data such as motor currents. In this paper, an approach is presented that enables a determination of the resulting system stiffness at the TCP based on machine-internal and quality data. During the milling process, machine internal data is recorded and a dexel-based material removal simulation (MRS) is performed. Therefore, the estimated force vector is calculated for each dexel. After the simulation, the virtual part is compared with the real part measured by a CMM to determine the deviation vector. By solving a linear equation system, the resulting stiffness is calculated. To reduce the influence of other effects, they are modeled in the MRS or usually reduced so that they are negligible.

Energy-bound forming machines such as forging hammers tend to vibrate due to abruptly applied process forces, which is particularly noticeable in form of intense vibrations of the machine environment. This paper presents a new concept of shock absorbers for forming machines, using dampers filled with magnetorheological fluids. Magnetorheological fluids are suspensions of magnetizable particles in a non-magnetizable carrier fluid. By applying a magnetic field, the internal structures and thus the rheological properties of the fluid can be varied. Using an evolutionary based control strategy, the damping can be adjusted depending on the excitation. The dependencies as well as challenges in the design process of magnetorheological dampers for forming machines are described. In addition, simulation results of foregoing studies concerning damper design and the evolutionary control strategy are presented.

In the automotive industry, due to fast changing markets and push for implementing novel materials, competitiveness relies increasingly on economical and short planning cycles of machining process. Digitalization has created opportunities for automotive industries to reduce time to market with the help of computer simulations of manufacturing process. In the past decade, particle methods like Smooth Particle Hydrodynamics (SPH), and Smooth Particle Galerkin (SPG), among many others, have been used to simulate machining process. The particle methods have an advantage over classical Finite Element Methods (FEM) as particle methods do not require remeshing or continuous mesh adaptation. Thus, particle methods eliminate the mesh entangling problems in simulation of large plastic deformations, such as machining. This paper describes the application of original SPG and the new Momentum-Consistent-SPG (MC-SPG) method in the orthogonal machining simulation of 1.4837D casted steel material. Furthermore, a study was conducted to understand how different SPG parameters affect simulation results, particularly force components, chip form and temperature. In the end, a Design of Experiment (DoE) was created to study the effects of cutting velocity, feed, and rake angle on force components. The simulation results were experimentally validated, and a good agreement (for cutting forces mean deviation 0–11% and feed forces 4–15%) was found between experimental and simulation results.

Process failure and part distortion are some of the main challenges in the additive manufacturing process laser powder bed fusion (PBF-LB). This is leading to increased part costs due to the need of post processing or a redesign and restart of a build job. Process simulation can enable engineers to predict possible build failures before the build job has been started. One of the primary problems with existing commercially available simulation approaches is the need for experimental data to calibrate the process model. To eliminate the need for calibration specimen, a new simulation technique was developed and is presented in this paper. Using a multiscale simulation, the calculation time can be significantly decreased compared to a single scale approach. On the micro scale, a high fidelity thermomechanical process model is developed to predict the associated inherent strains for the chosen process parameters and geometrical conditions. In addition, the material model is adapted to match the different phases present in the process. On the macro scale, a pure mechanical approach is used to predict part distortion and possible build failure due to recoater contact. In contrast to commercially available solutions, the scanning path is explicitly considered on both scales of the model to examine the influence of different scan strategies on the final parts properties. The simulation model were tested and validated against defined test specimen, which, as known from previous examinations, cause a recoater failure. All examinations were conducted with the commercially available aluminum alloy AlSi10Mg.

Manufacturing processes in sheet metal forming industry are subject to process-related variations, which can adversely influence the manufacturing costs and the quality of products. For example, during sheet metal forming of car body components, variations in material characteristics of the semi-finished product and in process parameters can lead to variations in the springback behavior of the sheet metal parts and therefore restrict the tolerances that can be realized with sufficient process reliability. In the assembly process, the springback variations of the individual sheet metal parts can also affect the dimensional accuracy of the joined sheet metal assembly and therefore the quality of the car body. In the course of digitizing the process chains in car body manufacturing, one of the objectives is to visualize such springback variations occurring after forming and joining processes of the individual sheet metal parts as well as of the sheet metal assembly at an early stage of development. On the one hand, this allows sheet metal parts and parameters with highest influence on the assembly to be identified and robustly designed, resulting in time and cost savings in hardware phase. On the other hand, tolerances for “less-important” parts of the assembly could be opened up, which may lead to additional cost and time reduction during die manufacturing. Against this background, the present paper provides an approach for modelling the manufacturing process chain of a forming and joining process considering variations in process parameters and material characteristics using finite-element-method and method of metamodeling. Here, metamodeling is used to predict the process behavior and thus reduce the required simulation effort. Based on the metamodels, Monte-Carlo simulation is carried out in order to perform variation and tolerance analysis.

In modern manufacturing processes, the environmental impact becomes an increasingly important aspect. The aim of developing new coolant strategies is therefore an approach to increase the efficiency of the machining process while reducing coolant consumption. The priority is to optimize the supply of coolant to the tool-workpiece interface. In case of cryogenic machining, low-temperature liquefied gases are used to cool the tool’s cutting edges and to decrease the overall process temperatures. The high cooling rates of this technology can reduce the thermomechanical loads for tools and workpieces especially in machining operations. Since cryogenic medium have no lubricating effect, additional lubrication strategies, e.g. Minimum Quantity Lubrication (MQL), are necessary to enhance the application limits of the cryogenic cooling technology. Nevertheless, in deep hole drilling, using small diameter twist drills, it is impossible to supply internal cryogenic coolant and MQL simultaneously. Therefore, this paper deals with a novel combination of cryogenic Minimum Quantity Lubrication (cMQL) by determining the lubricant’s efficiency according to its solubility in liquid CO2. The strategy leads to a significant increase in performance during deep hole drilling of difficult-to-cut materials and shifts the process limits in terms of tool life and feasible cutting parameters using environmentally friendly MQL techniques.

Due to their low rigidity compared to machine tools, industrial robots (IRs) are less suitable for dynamic processes such as machining. In order to benefit from the flexibility and large workspace of IRs for e.g. machining of large workpieces, a model-based feedforward control can be used to compensate the deviations of the tool center point. This control algorithm does not require additional components but a dynamics model of the IR, incorporating effects such as mass inertia and friction. This paper focuses on friction modeling of robotic drive trains comparing analytical, e.g. the LuGre model, and data-driven models, e.g. Long Short-Term Memory networks. The models are parametrized and trained using data from the first axis of a 6-degree-of-freedom IR and evaluated regarding their ability to model dynamic nonlinear friction effects like stick-slip. This serves as a basis for structured learning of robot dynamics by combining analytical and data-driven models.

In metal forming technology, faster development and production cycles are complicated by the production of deep drawing tools. Those are already mandatory even for small series. Compared to conventional concrete materials, ultra-high performance fiber reinforced concrete (UHPFRC) is characterized by its strength properties, which are close to the requirements for use in metal forming technology. The present work investigates the potential of UHPFRC in terms of strength properties and tool manufacturing to be used for prototyping as well as for small and very small series production in metal forming technology. For this purpose, the flexural strength and the compressive strength of UHPFRC materials is increased by adding carbon fibers of different lengths and different volume ratios. The material is evaluated by means of three-point bending tests and compression tests. Additionally, a novel indirect rapid tooling approach, a combination of fused deposition modeling (FDM) and room-temperature-vulcanizing (RTV) silicone molding, is introduced. This approach enables the casting of near-net-shape deep drawing tools with high dimensional accuracy.

In recent years, Machine Learning (ML) applications for manufacturing have reached a high degree of maturity and deal as a suitable tool for improving production performance. In addition, ML applications can be used in many other areas of production to enhance sustainability within the manufacturing process. One specific area is the storage and transportation of bulk materials with Intermediate Bulk Containers (IBC). These IBCs are currently used solely for their primary purpose of storage and transportation for raw and finished goods. But for a major part of their handling cycle time these IBCs are a black box, and therefore do not add additional value to manufacturers. By equipping those containers with sensor technology, new data can be generated along the entire supply chain, taking the sustainability of production to a new level. Within the research project smart.CONSERVE we use this additional data to prevent waste of resources through storage of production goods in defective IBCs through predictive maintenance. In this publication, we describe how the use of such smart IBCs in the food industry increases supply chain visibility and reduces food waste by presenting a number of use cases that are possible due to the new data availabilities. Additionally, we provide insights into the transferability of these use cases to other industries and the many opportunities for manufacturers to develop new smart services and ML applications based on the collected data to increase sustainability.

The availability of effective and eco-friendly powertrain systems for electrification of passenger and commercial traffic is a crucial requirement for achieving current climate targets. With increasingly limited energy resources, fuel cell technology is gaining interest as an alternative to conventional electrical drives. Especially for heavy-duty and long-distance vehicles, where the required payload and range would require enormously heavy batteries, fuel-cell-technology offers a promising solution. Critical components of such modern fuel cells are metallic bipolar plates (MBPP) manufactured by high-precision embossing of thin metallic foils. The critical point is that even slightest fluctuations within the manufacturing process can lead to forming defects and result in unacceptable springback of metallic bipolar plates. Combined with the dimensional accuracy required for MBPP, extensive quality assurance and thus relatively low cycle times are inevitable in today´s production of these components. In this context, this paper deals with an approach to actively control the manufacturing process of MBPP based on numerical data sets. For this purpose material characterization of 0.1 mm stainless-steel foil (1.4404) was performed, allowing for comprehensive modelling of the embossing process and the springback behavior. In order to maintain a robust forming process aimed at increasing productivity, a numerical analysis was then conducted under variation of different geometric parameters using AutoForm R10. It was found that variation of selected geometric parameters such as channel width, channel height, draft angle and tool radii can remarkably reduce thinning and springback in MBPP production in compliance with tight tolerance specifications. Furthermore, the investigations show that active control of the lubrication conditions offers an additional possibility for subtle adjustments of the dimensional accuracy of produced components.

In the automotive sector, an important strategy for reducing CO2-emissions is lightweight construction. In this regard, body-in-white parts offer a high potential for weight reductions by substituting conventional steel parts with tubular, aluminum-based components. Due to their high strength-to-weight-ratio and high crashworthiness, tube profiles are often used as safety-relevant car body components. Therefore, an exact determination of the material properties is necessary, in order to achieve a high prediction accuracy of FE-simulations. In contrast to the testing of flat semi-finished parts, there are only few standardized methods for the material characterization of tubular components. Furthermore, a profound knowledge regarding the influences of geometry and manufacturing process of the tube profiles on the material properties has to be gained. Thus, AA7020 tubes are analyzed in this research work. Tensile specimens are cut out of a tube by a laser-cutting machine. As a result, the cross-section of the samples is curved as well. To ensure a proper material testing, the clamping jaws for the tensile test are adjusted to the curvature of the samples. For the characterization of the tube properties, caused by the manufacturing process, optical measurements are performed. The mechanical properties of the different grain structures are determined by the thermo-mechanical simulator Gleeble 3500 in T6 condition. Therefore, tensile specimens with a reduced wall thickness are prepared. Afterwards the curved tensile samples are tested in T6 and W-temper condition. To validate the comparability of the determined material properties, a comparison is conducted with flat specimens prepared from the semi-finished tube profiles.

Conventional die casting tools are often oversized with regard to their mechanical stability, which results in increased energy requirements for kinematic processes and tool temperature control systems. In addition, these die casting tools often lack flexibility and modularity. One way to counter these disadvantages are lightweight structures, which can be built up in a modular manner depending on the application. Due to their larger cross-section compared to conventional lattice structures, plate lattice structures (PLS) offer higher mechanical load-bearing capacity. The aim of this work is to investigate such PLS with regard to their manufacturability in an additive manufacturing process. The specimens made from 1.4545 are first examined for dimensional accuracy and defects using computed tomography. Subsequently, an elastic model of these structures is numerically generated and calculated to obtain the structural stiffnesses. With this knowledge, a simplified numerical feasibility study was carried out.

The application of machine learning models in production systems is continuously growing. Hence, ensuring a reliable estimation of the model performance is crucial, as all following decisions regarding the deployment of the machine learning models are based on this aspect. Especially when modelling with datasets of small sample sizes, commonly used train-test split variation techniques and model evaluation strategies encompass a high variance on the model’s performance. This difficulty arises, as the available amount of meaningful data is severely limited in production engineering and can lead to the model's actual performance being greatly over- or underestimated. This work provides an experimental overview on different train-test splitting techniques and model evaluation strategies. Sophisticated statistical sampling methods are compared to simple random sampling, and their impact on performance evaluation in production datasets is analysed. The aim is to ensure a high robustness of the model performance evaluation, even when working with small datasets. Hence, the decision process for the deployment of machine learning models in production systems will be improved.

This paper presents a blockchain-based approach to machine data capturing and storage using Hyperledger Fabric with a low-cost, esp32 microcontroller. Our research focuses on the immutability of captured data to ensure that data on the blockchain is valid for use by customers and manufacturers. This requires an embedded implementation to sign collected data before it reaches modern IT environments like IoT gateways or programmable logic controllers (PLC). We discuss the challenges and advantages of using Hyperledger Fabric compared to an already validated implementation based on Ethereum technology. The proposed method demonstrates a solution to overcome the challenges that result from the Hyperledger Fabric communication protocol in an embedded environment. We thus illustrate both the technical design of the implemented logic on the microcontroller and the implementation of the distributed messaging protocol within Hyperledger Fabric. In our study, we validate the implemented method and perform a quantitative comparison with existing solutions from information technology. At last, we discuss the limitations of the proposed method and give an outlook on approaches that can potentially solve them.

The demand for individualized products with high quality and low costs is a challenge for manufacturers. Classic production engineering approaches can no longer meet these requirements and often require long and complex start-up cycles, while producing scrap and limiting product changeovers. This challenge becomes more acute when factoring in recycled products and materials. One solution to this is the use of Changeable Production Systems, which can be planned, controlled and monitored on the basis of data and data-based algorithms. This paper presents a strategy for constructing and utilizing virtual quality gates (VQG) in the context of Changeable Production Systems. Their logic is based on artificial intelligence and they are placed at the edge of the network to facilitate fast decision making.

As a result of the steadily growing importance of data-driven methods such as digital twins, approaches for automated parameter identification in production equipment are becoming increasingly important. Previous work has shown that AI-based approaches for classification are increasingly reaching their limits. As a result of new developments, CNC reference runs with a high information content that can be specifically identified via an ID can be generated. In this context, it was possible to achieve an oscillation state on a test machine that is particularly well suited for identification. In this paper, an analytical approach is presented which, in addition to classification, can assign the signal to the respective source and therefore establish interdependencies between signals. Here, on the test machine tool, with successfully excited oscillations, all signals could be classified and assigned via the ID with high accuracy. If the oscillation state cannot be reached, classification accuracies of over 90% could be achieved, depending on the motion generation.

Over the last years, artificial intelligence (AI) and machine learning (ML) became key enablers to leverage data in production. Still, when it comes to the utilization and implementation of data-driven solutions for production, engineers are confronted with a variety of challenges: What are the most promising application areas, scenarios, use cases, and methods for their implementation? What openly available data sets for the training of ML and AI solutions do exist? In this paper, we motivate the challenges of applying AI and ML in production and introduce an extended taxonomy of application areas and use cases, resulting from a comprehensive literature review. In addition, we propose both a process model and a concept for an ML-Toolbox that are tailored to cope with the specific challenges of production. As a result, from an extensive study, we present and launch a comprehensive collection of currently more than 130 datasets that we make openly available online to serve as a continuously expandable reference for production data. We conclude by outlining three key research directions that are decisive for a widespread adoption of real-world ML. The contributions of this paper establish a foundational development framework that allows to identify suitable use cases, gain experience without having suitable in-house data at hand, improve existing data-driven solutions and promote applied research in this challenging field of ML in production.

The complexity of products is increasing and key functions can often only be realized by using micro components. The requirements of high-precision components often reach technological manufacturing limits. This is of particular importance for micro components with complex geometries, such as micro gears, where manufacturing deviations are relatively large compared to the component size and therefore have a large influence on the functional characteristics of the assembled product. In this paper, an approach is presented to predict and optimize the functional characteristics of assembled micro gear pairs in terms of Noise, Vibration and Harshness (NVH), based on optical in-line measurements of the entire topography of the gears. The overall quality is optimized by individually selecting the gears to be assembled with regard to minimising predefined NVH parameters. For implementation, a large number of possible combinations must be predicted. It is proposed to develop a meta-model with machine learning (ML) methods, which enables the near-real-time prediction of the NVH parameters of micro gear pairs, based on the optical in-line measurements.

In many conventional areas in mechanical engineering, such as mechanical design, there are process models for engineers like VDI 2221 that guide through the process with methodological support, provide criteria for evaluating the results and thus ensure quality. Generalized process models such as CRISP-DM, KDD and SEMMA already exist for Data Mining, as well as DMME, DAPLOM or ISO 17359 specifically for production engineering. However, these only focus on the sequence of the necessary tasks in several phases without naming adapted methods or without considering aspects of data analysis. Furthermore, the transferability to new use cases or the reuse of the developed solutions has not yet been addressed. In this paper, based on the stages of the DMME, adapted methodical guidelines for enabling machines to acquire data that is suitable for Data Mining are provided. The methods focus on the identification and prioritization of analysis goals and the design of measurement chains and experiments for the acquisition of training data based on the process and the machine structure. In terms of reusability, approaches to transfer the results into templates will be discussed. The methods are applied in a condition monitoring project for a concrete mixing machine.

The increasing demand for customized and complex products with small batch sizes confronts the manufacturing industry with new challenges, which can only be handled by flexible and dynamic manufacturing processes. As a major part of the process chain, autonomous machine tools can contribute to satisfying these requirements. Although there are many contributions on autonomous machine tools in research, the development of a self-learning, autonomous AI-integrated machine tool has been implemented neither in the industry nor in research. This paper proposes an industry-oriented concept of a self-learning machine tool. The system architecture consists of the existing CAD, CAM and CNC process chain and extends it with a knowledge base and an intelligent CAPP system for domain knowledge representation and decision making. Process knowledge is represented by using a continual learning process simulation approach for small-batch production.

This paper examines and benchmarks different approaches in their ability to detect and predict faults in manufacturing processes, based on real-world use cases and with respect to their differing dataset properties. Knowing about the occurrence of faults becomes more and more important in manufacturing due to increasing quality demands and legal guidelines. In addition, the complexity of manufacturing processes is constantly increasing. This stems from a higher product variance resulting from individual and customized products as well as additional external influences such as human errors, environmental factors and tool wear. As a result, today’s process data is often no longer normal distributed. Furthermore, data volume steadily increases, thereby opening new opportunities for data-driven analytics approaches. Frequently applied control charts for statistical process control (SPC) often lack the ability to deal with multiple variables and non-normal distributed data at the same time, since multivariate and nonparametric control charts are underrepresented in past research. Consequently, there is a need for new process control methods in manufacturing that are suitable for large amounts of data and cover diverse and dynamic distribution models. Therefore, machine learning models have been recognized as feasible approaches to meet these requirements. For comparison a Hotelling’s T2 control chart, a K-Chart, an Isolation Forest, an ARIMAX model and a Neural Network have been implemented. We evaluate each method by missed detection rate (MDR), false alarm rate (FAR) and whether signals occurred before or after the faults. Real-world data sets of a commercial vehicle manufacturer serve as benchmarking basis.

Due to the ongoing trend of digitization and the strong increase in the number of Industrie 4.0 use cases, the use of machine tool process data in data-driven applications, namely process or condition monitoring, is on the rise. However, the provision of data such as motor currents or the positions of a machine’s axes—which are essential to these applications—can in many cases only be achieved under high individual effort. This is largely due to two aspects. Firstly, due to the heterogeneous nature of communication infrastructures and information models in the machine environment, there is no one-size-fits-all solution for the acquisition of data. Secondly, in cases where the denomination of the sought-after parameters is not known in advance, an implementation of data-driven applications becomes challenging. Thus, the objective of the research work presented is the development of a system capable of extracting time series data from a variety of machines and related data sources and determining their identity (e.g. current, position, etc.) based on their characteristics using an intelligent approach. A prototypical software application with a modular structure is presented, which envelops functionalities from the discovery and acquisition of data to its intelligent identification.

Highly competitive markets urge manufacturing companies to enhance the performance of their business and production processes. As part of business process improvements, companies examine as-is-process models for weaknesses (process analysis) and then derive measures for an improved to-be-process (process design). The process design phase is typically conducted manually in workshops, which makes it liable to a set of challenges. First, it is highly dependent on the participants’ knowledge about measure types and their suitability for addressing process weakness types. This knowledge is often heterogeneous and deficient. Second, participants with proprietary positions expose the workshop-based approach to subjective influences. Third, open brainstorming on potential measures is time-consuming. Fourth, the impact of measures on the business process is barely quantifiable at the time of process design without using process data. Existing approaches for event log-based weakness detection use this data to semi-automate the process analysis phase. Their outputs are process weakness types detected in event logs on business processes. For the phase of process design, such semi-automated approaches are not available. This paper presents a concept for data-based measure derivation for business process design. Therefore measure types for process design, that impact the process flow and are context-independent, are identified in literature. In a second step, they are allocated to process weakness types that they can address. In the third step, the measure types’ impact on business processes is quantified using event log information. Thereby, the concept enables semi-automated measure identification and process performance related decision support for business process design.

With the continuous rise of industry 4.0 applications, artificial intelligence and data driven monitoring systems for machine tools proved themselves as highly capable alternatives to classical analytical approaches. However, their precision is limited to a number of crucial aspects. One of the main aspects revolves around the lack of meaningful data, which leads to imprecise and false model predictions. This issue is closely linked to production processes and machine tools in production engineering, as the available amount of meaningful real data is strongly limited. The usage of simulation models to acquire additional synthetic data is able to fill this lack. This work looks into improving the prediction accuracy of a deep learning model for temperature forecasting of a 3-axis precision machine by combing and comparing real process data with domain randomized simulation data. The used thermal simulation model is based on finite element models of the machine assemblies. Model order reduction techniques were applied to the FE models to reduce the computational effort, increasing the simulation-to-reality gap. The approach is evaluated on unseen real data, demonstrating the underlying potential of the inclusion of synthetic data from simulation models of machine behavior.

The digitization of manufacturing and the economic potential from the utilization of data sets result in the monetarization and trading of process data. In data marketplaces, a suitable pricing mechanism ensures the definition of a fair value of data for the data consumer and the data owner. Game theoretic concepts ensure this fairness by determining the value functions of the participants in a transaction, which is based on the quality and quantity of data. In this paper, cooperative games are used to model data transactions. First, empirical methods are used to realistically determine the value of data sets. Using the example of a data set of a grinding machine, a function is set up depending on the quality dimensions, with the help of which declarations about the usability of the data set for regression models are possible. This is followed by the pricing of a data set based on the Kalai-Smorodinsky solution. It offers the advantage that the quality of a data set does not have to be assumed as a variable and that pricing is possible individually for already existing data sets whose quality is not to be adjusted further. Finally, the chosen approach is compared with the common method for price discovery, the Stackelberg game, with respect to the concession of the actors, based on the negotiation result to the maximum possible benefit. In the outlook, potential further developments of the approach based on Kalai-Smorodinsky are discussed.

The aim of the Industry 4.0 initiative is to secure Germany’s future as an industrial location and to strengthen its competitiveness compared to other countries. In contrast to large companies, it is more difficult for medium-sized ones to implement the migration from Industry 3.0 to 4.0, as they do not have the financial and human resources to fully replace all systems currently in operation. Therefore, the migration needs to be executed evolutionarily by retrofitting existing production facilities so that new acquisitions can be avoided. In this paper, such a retrofit will be analyzed based on a machine for forming copper rings, which is part of a process for manufacturing valve seat inserts for combustion engines. Since production is carried out under a high cycle time and the rings meet tight tolerances, condition monitoring is to be implemented to detect failures at an early stage. For this purpose, the approaches Design of Experiments (DoE) as well as Machine Learning (ML) are considered. Both options are evaluated based on a complexity analysis using the environment concept of an intelligent agent by RUSSEL & NORVIG. Finally, suitable supervised ML algorithms are selected.

The disassembly of complex capital goods is characterized by strong uncertainty regarding the product condition and possible damage patterns to be expected during a regeneration job. Due to the high value of complex capital goods, the disassembly process must be as gentle as possible and being adaptable to the varying und uncertain product's state. While methods based on data mining have already been successfully used to forecast capacity and material requirements, the determination of the product’s or component's condition has become apparent in the recent past. Despite the rapid increase in sensor technology on capital goods such as aircraft engines and their use for condition monitoring due to countless interfering effects, it is only possible to react spontaneously to the product’s condition. So far, we have concentrated on product condition-based prioritization of disassembly operations in a logistics-oriented sequencing strategy. In this article, we present an approach to predict disassembly process-planning parameters based on operational usage data using machine learning. With the prediction of disassembly forces and times, processes, tools and capacities can be efficiently planned. Thus, we can establish a component-friendly disassembly process adaptable to varying product conditions. In this article, we show the successful validation on a replacement model of an aircraft engine.

Uncoordinated decisions that have a long-term impact on the production network lead to inefficient structures and limit the ability to change. However, the ability to change is a basic prerequisite for future decisions. At the same time, the world is becoming more volatile, uncertain, complex, and ambivalent. To counteract this, external and internal influencing factors must be considered in the early stages of planning global production networks (GPN). The design of GPN is on the one hand associated with a large number of degrees of freedom and on the other hand with a large number of influencing factors. Influencing factors can thereby be known and predictable, but also unknown and unpredictable. To make production networks capable to change in the long term, influencing factors and their effect on the network design must be considered. The combination of influencing factors with consideration of uncertainty still needs further research in the context of network design. Thus, this article aims to develop a method for network design that does not only take external and internal influences into account at an early stage but also leads to a network configuration that considers these influences and increases resilience. To achieve this, the influencing factors should first be represented in scenarios using the receptor theory. Subsequently, the scenarios can be incorporated into the optimization of the network configuration by choosing a solution from a predefined solution space. The process of solution selection and testing can be supported by a digital twin. The result is an initial concept that merges these different steps into a continuous process that can be used to design adaptable GPN in the future.

The democratization of desktop 3D printing has opened the domain of manufacturing to the masses. Today individuals can design and manufacture a variety of products in their living rooms. However, scaling a product from prototype to production and setting up a small-scale manufacturing business is often hindered by the expensive machinery and high upfront capital investment required. This paper presents the findings of a unique experiment that was carried out to understand the process of prototyping a relatively complex product (in this case, a 3D printer) in a home setting and then scaling it up to a small-scale production (10 units). In order to partially automate the manufacturing processes, two open source machine tools (OSMT), whose blueprints are freely available in the internet, were built, namely a CNC laser cutter and a CNC milling machine. The experiment reveals the particularities of starting a small-scale production in a home setting and the potential of OSMT to affordably scale up production, while also highlighting the challenges of OSMT adoption.

The production of highly individualized engineer-to-order products has special characteristics that lead to a significant increase in the complexity of production planning and control. Therefore, aggregate resource planning is a dynamic and complex process that must always deliver reliable results. But without appropriate tools, these predictions can only be achieved with significant manual effort. Therefore, this paper presents a holistic method that predicts and schedules the required manufacturing resources for new customer orders based on a type representative by means of product modularization and data preparation of approximately identical historical manufacturing orders. This allows the actual processing status of the current customer project to be derived from the preplanning by means of a concurrent calculation in order to be able to initiate countermeasures at an early stage in the event of project delays and also to reduce the lead time of the customer order by preallocating the required production resources.

The example of the automatization ramp-up of the hydrogen electrolyzer production illustrates the complexity of plant engineering from scratch to production. Collaboration takes place in an interdisciplinary environment of several business and engineering units simultaneously. A complex chain of a multitude of process steps is created, as well as the exchange of data from various programs and simulations. With virtual commissioning (VC) and digital models of the plant, new possibilities arise to continuously validate the results of the planning process. As a result, there is a time advantage compared to traditional physical commissioning. There are always sudden market ramp-ups of products, for example, solar cells, car batteries, and fuel cells. Particularly due to the strong increase in demand for hydrogen electrolyzers, it is important to design the product and the production plant simultaneously. A plant manufacturer can respond to the increasing demand by pre-designing a configurable and scalable pool of production modules based on black-box approaches to the product in its various sizes, weights, etc. as is being investigated in the research project FertiRob. The aim of this paper is to present a concept of designing configurable production modules for a rough solution space. Templates created for a production module are shown, which can be accessed by a future plant configurator to customize the configurable modules and retrieve certain master data. Future research activities are aimed at securing the templates via a neutral data exchange platform (e.g., AutomationML, PLCopen XML, COLLADA) so that access is guaranteed via a configurator and all other engineering software.

The plant engineering process is characterized by high complexity, diverse interfaces and multidisciplinary processes. Today, there is still no standardized reference process architecture. A proven procedure is to use and adapt existing planning documents with certain similarities to a new project. This involves lots of effort not only in the design phase but much more in the semantic adaptation, resulting from redesign and reprogramming. Due to these challenges and the increasing importance of data as gold of digital age, a continuous flow of data and information is becoming even more important in plant engineering. Creating a solid base of engineering and operation data free of waste supports effectiveness and efficiency during the entire product life cycle. A new approach is to design general and configurable production modules. Just as virtual commissioning brings considerable time and quality benefits, the design of defined process steps in advance of a customer order is intended to bring forward some of the tasks, standardize multidisciplinary work and unify the data base. The aim of this paper is to present the lean-based concept of configurable production modules. Thereby, a focus is especially on lean information management to achieve an effective as well as an efficient plant engineering process and to create the requirements for a lean production process. The concept of configurable production modules is applied to the example of the plant design process of automated production plants for hydrogen electrolyzers.

The production environment is in a constant state of change. This results in a continuous change of production processes. A key factor in mastering change is to increase flexibility. To achieve this, the targeted training of employees is essential. Within the framework of the research project “reQenrol”, research is being conducted to sustainably design personnel development based on the competence and tasks of the employees. Manufacturing companies face the challenge to efficiently training their personnel for an increasing and dynamic range of tasks. Training measures must be adapted to personal skill level of employees as well as to requirements of individual tasks in production. As a basis for a competence-based workforce deployment and the realization of targeted training measures, a survey was conducted on the current training situation and the relevance of competences in production. The results are placed into the context of the concept for an assistance system that enables manufacturing companies to perform a dynamic, competence-based workforce scheduling and realize targeted employee training.

To reduce energy costs in manufacturing, load forecasting plays a decisive role, for example, as a central instrument for load management or in the proactive marketing of demand-side flexibility. To allow for accurate forecasts and the ability to incorporate the latest available information, very short-term load forecasting is applied. While a 15 min time resolution standard is predominant in the energy sector, much higher input resolutions are available due to modern sensors, declining data storage prices, and increasing computing power. Nevertheless, an increase in the resolution does not have to improve the forecasting accuracy since higher-resolution data series are subject to stronger stochastic fluctuations. It is unclear up to what level of resolution refinement forecast improvements can be expected. This paper systematically examines the effects of varying the input data resolution with respect to the prediction performance of the resulting load forecasting models. We propose a method for identifying the optimal resolution for very short-term load forecasting applications and validate it on electrical load data of three companies from the industrial sector, each sampled at different resolutions.

In this article, a methodology for estimating both the product and the production-side lightweight design potential is presented, which can be used at an early stage of the product development process due to the limited amount of data required. This can help companies to increase the performance of their production facilities through the proper use of their potential and, on the other hand, to identify the lightweight construction potential in their products. This allows for faster integration of lightweight construction in sectors not typically associated with lightweight construction due to the reveal of hidden possibilities in production and ultimately leads to resource savings in industry. For this purpose, possible influencing factors and existing potential analyses are examined first, the requirements for a methodology in the early phase of product development are analyzed and the use cases of calculation for a given component and calculation without a determined component are identified. From the information obtained, a linkage and relevance analysis is used to derive key factors influencing the lightweight design potential of the product and production. The methodology is developed on the basis of these key factors, with a division into potentials of geometric and material lightweight design. Parameters from both areas and their effects on product design and production were taken into account. The lightweight design potential of the production equipment and products is then given as a percentage of the optimal degree of fulfillment.

Caused by the trend of shorter product lifecycles, higher numbers of product variants and volatile markets, production systems face increasingly short periods with unchanged requirements. Therefore, the capability of manufacturing systems to reconfigure fast and cost-efficiently to changed requirements becomes a crucial factor for companies to maintain their competitiveness. Currently, reconfigurations of manufacturing systems are, on the one hand, limited due to technical constraints of the used hardware and software. On the other hand, reconfigurations require a lot of time due to manual engineering processes, planning procedures and inefficient deployment of changed production system configurations. Well-known response mechanisms for reducing reconfiguration efforts are the concepts of flexibility and changeability. This paper shows how the challenges of applying these concepts, such as managing complex modular systems or handling high reconfiguration frequencies, can be addressed with introduction of a new approach. With the paradigm shift towards software-defined manufacturing, the full potential of flexibility and changeability can be accessed. Software-defined manufacturing allows to largely decouple the production task from the operating production hardware and to manage the configuration of the production system via a continuous and highly digitized adaption process. By exploiting technologies like data mining and digital twins, the digital planning process determines new configurations of the production that fulfill changed requirements. Subsequently, the new configuration can be validated and procedures for the deployment to the production system can be determined.

Manufacturing companies in high-wage countries face a variety of challenges. International competition exposes them to constantly increasing pressure to offer new products that meet customer requirements in increasingly shorter intervals and at competitive prices. Particularly in high-wage countries, where personnel resources are the second highest cost factor it is important to utilize these resources efficiently in order to be competitive in global markets. An important lever to increase the efficiency of personnel resources is the specific allocation of employees according to their competencies and not according to their function in the company. Employees with a set of competencies that matches well with the characteristics of an activity or a process are able to achieve the same results in less time compared to employees with a less matching set of competencies. Therefore, the goal of this paper is to develop a methodology for the improvement of personnel resources efficiency by assessing the time needed for specific activities based on the competencies of personnel resources. This enables a better resource management since activities and processes can be assigned to the employees who are able to finish them fastest.

Machining sequence planning for milling, also called operation sequence planning, can be considered one of the most important tasks of manufacturing process planning. Computer-Aided Process Planning (CAPP) is one of the application areas of machining sequence planning and is also an important interface between computer-aided design and computer-aided manufacturing. The planning tasks are multidimensional, but they are often handled in a linear way, which is one of the problems of conventional CAPP systems. This problem leads to limited solution space. The solution can be far away from the optimum or even not represented in reality due to resections caused by technical reasons. Multiple planning tasks cannot be combined in every way. They are restricted by the technological properties of the machining process, which makes the solution even more complicated. In contrast to the conventional approach, this paper generates valid sequences of operations based on a graph (Hamiltonian path) using a Simulated Annealing algorithm. Simulated Annealing is meta-heuristic, which finds global extrema within a graph by approximation. The algorithm’s goal is to minimize the number of setups and tool changes. To evaluate the validity of the method, the Simulated Annealing algorithm was tested on parts with known experimental machining sequence optimum.

Various production parameters such as inhomogeneous material properties or varying lubrication lead to deviations in manufacturing. Quality management must ensure that the required geometric dimensions and tolerances are maintained. In many cases, the inspection is carried out randomly, manually and at the end of the production chain, which prevents early detection of rejects and intervention. The problem complexifies due to the increasing demand for 100% testing of workpieces, e.g. for safety-relevant components in the automotive industry. This leads to additional effort regarding time, personnel and logistics. One solution to these problems is the inline measurement of the workpiece geometry. Due to rough environmental conditions in forming machines, the implementation presents particular challenges. In this publication, the disturbance variables occurring in presses are described and requirements are derived which result for the applied sensor technology. Based on this, a methodology for measuring small rotationally symmetrical workpieces is presented.

Scholars have highlighted the importance of decarbonizing manufacturing industries for several years already. Industry accounts for about 20% of the EU’s greenhouse gas emissions. In order to meet the targets set in the Paris Agreement, industry must reduce emissions to almost zero by 2050. A wide range of measures can be taken to achieve climate neutrality consisting of three categories: reducing greenhouse gases by adapting business models, substituting products or offsetting the emitted greenhouse gases. Companies have to determine the optimal set of measures taking into account their individual situation as well as available resources. From this, a complex optimization problem arises and the proposed decision model offers significant sup-port for the selection of decarbonization measures. By using the decision model, companies can achieve the greatest possible emissions reduction with a minimal set of resources according to their target system, thus taking into account net present value, benefits, and risks. This paper introduces a novel modeling of measures that incorporates relevant evaluation criteria. The arising decision model is solved by using Mixed-Integer Programming. The presented approach was validated in a case study with an industrial corporation.

In the context of sustainable management, organizational resilience is gaining importance. Manufacturing companies are increasingly exposed to external disturbances. Crisis-resistant product development is of particular importance, as innovative products offer a promising opportunity to create competitive advantages and thus secure the company’s existence or even enable a company to increase its market share in the event of a crisis. At the same time, corporate functions today are usually geared towards efficient execution. In this context, the paper presents a concept for the alignment of product development in the conflict between efficient goal achievement and the prevention of the impacts of disturbances. For this purpose, the design elements, goals and relevant disturbances of product development are taken into account. Based on the interdependencies of these elements, a methodological approach for a company-specific determination of the target characteristics of the design elements is presented, in order to enable an alignment in the conflict between efficient goal achievement and resilience. The concept is designed to the alignment of product development, but can be transferred to other corporate functions and corporate divisions.

In view of the increasing scarcity of resources and global efforts to reduce CO2 emissions, production management approaches are focusing on enabling a circular economy. The remanufacturing of used products to the quality standards of a new product is one key enabler. Remanufacturing offers economic and ecologic advantages by reducing the amount of resources used in production. Thus, associated manufacturing costs are reduced and the dependence on imports of critical raw materials decreases. To do so, remanufacturing requirements must be considered in the early development phases of products and production processes. In practice, companies focus on the economic perspective in the development phase, methodically supported by highly iterative product and production process development (HIP3D) approaches. However manufacturing companies neglect the inclusion of the ecological perspective in the development phases, partly due to missing methodical support. This paper presents a framework for the industrialization of remanufacturing in the HIP3D. Since the feasibility of remanufacturing is defined at the early stages of product and production process development, this paper aims at integrating remanufacturing requirements in the development phase. First, the requirements arising from remanufacturing are identified through a systematic literature review. Subsequently, it is examined to what extent HIP3D already covers these requirements. For non-fulfilled remanufacturing requirements, adaptions and extensions to the HIP3D approach are derived and described in design guidelines. This results in a framework for the industrialization of remanufacturing in the HIP3D, enabling manufacturing companies to exploit their economic and ecological potential.

In the energy transition context, the manufacturing industry moves into the spotlight, as it is responsible for significant proportions of global greenhouse gas emissions. The consequent pressure to decarbonize leads to suppliers needing to report and continuously reduce the energy consumption incurred in manufacturing supplied goods. To track the energy footprint of their products, manufacturing companies need to integrate energy data with process and planning data, enabling the tracing of the product-specific energy consumption on the shop floor level. Since manufacturing processes are prone to disturbances such as maintenance, the energy footprint of each product differs. Meanwhile, the demand for energy-efficiently produced products is increasing, supporting the development of a sustainability-focused procurement by OEMs. This paper addresses this development and outlines the technical requirements as well as how companies can identify product-specific energy consumption. Furthermore, a case study is conducted detailing how to determine the product-specific energy footprint.

Sustainability is important aspect of management. Public awareness for climate change and other aspects of sustainability (resource efficiency, energy efficiency, and social responsibility) has forced companies to integrate sustainability considerations into their day-to-day management activities and overall enterprise strategy. Historically, economic aspects have been the focus of all management activities and thus controlling mechanisms and management systems have evolved to provide all kinds of data to facilitate management activities and decision-making. Sustainability data, however, is currently not available for management with comparable ease. Therefore, this paper describes a service-oriented hub (EcoHub) to enable a sustainability management system and to facilitate management decision-making based on sustainability data in manufacturing companies. The focus is on data and service requirements of real use cases and the respective requirements for a data system based on the Asset Administration Shell (AAS).

Production and sustainability represent a challenge that still exists today. The demand for more efficient use of resources and operating materials is clear, and possible through the pragmatic integration of digital technologies and the approach of the circular economy along the entire process chain. However, when leaving individual processes, the complexity of data increases since causal effects between the process steps and their impact on the resulting KPIs must be considered simultaneously. This is where data-driven analysis unfolds its full potential. For this purpose, in addition to the manufacturing process itself, it is imperative to consider the often-neglected peripheral systems, including the provision of raw materials, consumables and supplies. In addition to the necessary consistent and cross-process-step data, manufacturing companies and especially small and medium-sized companies lack usable digital services for the demand-oriented control of process and periphery and for event-based instead of time-controlled recommendations for action for staff, maintenance, and management to achieve an increase in resource efficiency. This work provides an approach that addresses prediction, classification, and anomaly detection using modular machine learning models trained on heterogeneous data in the electroplating industry and gives a conceptual outlook for transforming these models into robust edge services for control systems in manufacturing.

The reduction of CO2-emissions is an essential need in the automotive industry. Progressive die stamping offers a large potential to reduce CO2 emissions due to potential savings in material scrap after the punching operations. This work presents a methodology to calculate the material loss in progressive die stamping and illustrates strategies to rearrange the stamped parts to determine the potential of saving material. The share of scrap in the conventional process ranges from 16 to 60%. The average savings potential with a redesigned component layout in sheet metal band lies between 28 and 41%, revealing a high potential to reduce material loss and CO2-emissions.

A digital twin enables the accessibility of data, information, models, and simulations for a physical object. Therefore, digital twins become increasingly relevant for different areas of production. In particular, the production of battery cells with its high complexity could benefit from digital representations such as digital twins. Still, there is no coherent definition for digital twins in the battery cell production yet. In this paper we introduce the first concept of digital twins in battery cell production. For this we combine existing ideas for the digital twin with the characteristics of the battery cell production. The concept consists of digital twins for buildings, products, and machines or assets, which we validate based on different use cases. By this, we demonstrate the benefits that arise from the implementation of digital twins in battery cell production—from increased productivity, faster ramp-ups, to increased sustainability.

Increasing concerns for a more sustainable future have led to a fast-growing demand for high quality lithium-ion batteries. In order to expand available manufacturing capacities to the desired magnitudes within a reasonable timeframe, the concept of Digital Twins is seen as a possible solution. With the purpose of better understanding the abilities of this concept and showcasing how it can be used to accelerate the ramp-up process of manufacturing technology, this paper presents an analysis of existing approaches to Digital Twins within battery cell manufacturing. Available case-studies and scientific publications along with novel concepts will be used to identify and cluster the potentials and benefits of the technology. The resulting framework provides an overview for possible Digital Twin implementations as well as an opportunity to identify future areas of research. Based on the performed analysis, two conceptual use cases will be presented. It is shown, that Digital Twins can help transform both the mixing of electrode slurry as well as the production sequence of separating and stacking battery electrodes from traditional discrete processes to continuous production flows.