At a time of rising energy and material costs, manufacturing process efficiency is becoming increasingly important. This is one reason why cutting and especially sawing processes, usually the first step in most manufacturing chains in discrete part production, have to be investigated more intensively. Due to problems with runout and poor surface finishes, raw material is conventionally cut to oversize by either circular sawing or band sawing. This oversize has to be removed by following processes, costing extra energy and wasting material. Since the problem of runout increases with increasing tool wear due to higher deflections of the thin and compliant tools, an even larger oversize is required. This paper describes an approach to reduce the need of oversize even with increasing tool wear in band sawing by tilting the saw band in order to compensate for tool deflections during cutting. To achieve this, it is necessary to measure and understand the saw band runout. The next step is to present a system design and controller for tilting the saw band. Finally, tests are carried out to analyse the effectiveness of the system. In addition, the approach allows the use of increased process parameters to the end of the tool life without losing more material. The tool can therefore be used productively for a longer time period.

Häufig ist der Trennprozess von Halbzeugen mittels Bandsägen der erste Schritt in der Teilefertigung. Aufgrund bisher fehlender simulativer Untersuchungen wurden die Bearbeitungsparameter für diesen Prozess meist aufwendig experimentell ermittelt. Mit Analogieversuchen, die den zugrunde liegenden Prozess weitestgehend abdecken und mit viel geringerem Aufwand durchgeführt werden können, kann dies effizienter gestaltet werden. Der Beitrag untersucht, wie diese Versuche durch künstliche Intelligenz (KI) unterstützt und sich die Anzahl notwendiger Versuche reduzieren lassen. The cutting process of semi-finished products by means of band saws is often the first step in producing parts. So far, the lack of simulative investigations usually meant that machining parameters for this process had to be determined experimentally at great expense. This can be done more efficiently by means of analogy tests. They cover most of the underlying process and are more easily executed. This article examines the extent to which artificial intelligence (AI) can be used to support these experiments and reduce the number of experiments required.

In band sawing, monitoring signals are commonly averaged over the whole band and therefore cannot localise damaged teeth. This study presents an in-process optical pipeline that images moving teeth, segments the visible wear region with a compact grayscale U-Net, condenses each mask into three lateral cutting-edge zones and assigns detections to a weld-seam-referenced band coordinate. Four findings generalise beyond the present setup. The choice of input modality matters more than network size: an otherwise identical colour model collapses under cutting fluid and reflections, because the colour channels encode the appearance of the process rather than the wear; classical thresholding is not a viable alternative either. Wear extent and its lateral distribution behave differently and should be kept as separate descriptors, since the summed wear land width grew on every tooth while concentration and left–right imbalance stayed tooth-specific. The uncertainty of the position assignment is measurable rather than assumed, and because it spans several teeth, the tooth group is the smallest unit such a map can reliably resolve. Finally, values from single frames scatter too strongly for an isolated verdict, which makes consolidation over repeated revolutions part of the method rather than an optional refinement. Segmentation reached a mean Dice similarity coefficient of 0.817 on moving-band images at about 3 ms per image, giving spatially resolved, uncertainty-aware wear information for targeted inspection.

In metal band sawing, higher cutting speeds increase frictional heat at sliding guide blocks. Recirculating water-miscible metalworking fluids (MWFs) often lack fine filtration and accumulate debris that can enter the guide–band interface. A 1 L coolant sample collected after 22.5 m² of cutting contained a particle load of 0.438 g/L; optical sizing yielded a number-median maximum Feret diameter of 345 µm, with particles up to 1.5 mm. Compared with typical guide clearances (~0.1 mm), these sizes imply frequent ingress/bridging and three-body interactions. The coolant viscosity follows an Andrade relation and decreases by ~2% K⁻¹ around 40 °C. HFRR tribometry indicates low steady-state friction (µ ≈ 0.12), comparable to cutting oil. Together, these results provide quantitative design inputs for next-generation guide clearances and targeted filtration/coolant-delivery concepts in high-speed band sawing.