Automatisierte KI-basierte Leckage-Erkennung/Automated AI-based leak detection

Obwohl Druckluftleckagen jährlich hohe Kosten verursachen, ist deren automatisierte und aufwandsarme Erkennung immer noch nicht möglich. In diesem Beitrag wird das Konzept einer automatisierten KI-basierten Leckage-Erkennung vorgestellt und auf einen Druckluft-Labordemonstrator angewendet. Erste Validierungen der Vorgehensweise zeigen, welche Potenziale und Herausforderungen sich für das Leckage-Management an Druckluftmaschinen ergeben.   Though incurring substantial costs every year, compressed air leakages still cannot be detected automatically and without much effort. This paper presents a method for an automated AI-based leak detection system, which is applied to a compressed air laboratory demonstrator. First validations of the approach show what potential and challenges there are in the leakage management of compressed air machines.

Modeling Complex Contact Conditions and Their Effect on Blade Dynamics

Dry friction devices such as underplatform dampers are commonly included in turbine bladed disks designs to mitigate structural vibrations and avoid high cycle fatigue failures. The design of frictionally damped bladed disks requires adequate models to represent the friction contact. A widely used approach connects contact node pairs with normal and tangential springs and a Coulomb friction law. This simple model architecture is effective in capturing the softening behavior typically observed on frictionally damped structures subjected to increasing forcing levels. An unexpected hardening behavior was observed on the frequency response functions (FRFs) of a two-blades-plus-damper system tested by the authors in a controlled laboratory environment. The reason behind this unexpected behavior will be carefully analyzed and linked to the damper kinematics and to the dependence of contact elasticity on the contact pressure. The inadequacy of contact models with constant spring values will be discussed and alternatives will be proposed. The importance of being able to represent complex contact conditions in order to effectively predict the system dynamics is shown here using a laboratory demonstrator; however, its implications are relevant to any other case where large contact pressure variations are to be expected. The nonlinear steady-state simulations of the blades-plus-damper system will be carried out using an in-house code exploiting the multiharmonic balance method in combination with the alternating frequency time method.

Modeling Complex Contact Conditions and Their Effect on Blade Dynamics

Dry friction devices such as underplatform dampers are commonly included in turbine bladed disks designs to mitigate structural vibrations and avoid high cycle fatigue failures. The design of frictionally damped bladed disks requires adequate models to represent the friction contact. A widely used approach connects contact node pairs with normal and tangential springs and a Coulomb friction law. This simple model architecture is effective in capturing the softening behavior typically observed on frictionally damped structures subjected to increasing forcing levels. An unexpected hardening behavior was observed on the frequency response functions of two-blades-plus-damper system tested by the authors in a controlled laboratory environment. The reason behind this unexpected behavior will be carefully analyzed and linked to the damper kinematics and to the dependence of contact elasticity on the contact pressure. The inadequacy of contact models with constant spring values will be discussed and alternatives will be proposed. The importance of being able to represent complex contact conditions in order to effectively predict the system dynamics is shown here using a laboratory demonstrator, however its implications are relevant to any other case where large contact pressure variations are to be expected. The nonlinear steady state simulations of the blades-plus-damper system will be carried out using an in-house code exploiting the Multi-Harmonic Balance Method (MHBM) in combination with the Alternating Frequency Time (AFT) Method.

Internet of Energy Training through Remote Laboratory Demonstrator

In this paper, a new learning tool is proposed to train professional figures, such as entrepreneurs, engineers, and technicians, who need to improve their skills in the field of Internet of Energy. The proposed tool aims to cover the lack of experimental knowledge on new energy systems and to layer proper skills, which are useful to deal with challenges required by smart energy management in the new complex distributed configuration of the electric power systems, characterized by demand response services. This tool is based on a small-scale laboratory demonstrator, representative of a smart rural house, equipped with a measurement and control system. This demonstrator can be remotely accessed, through web server applications based on a low cost single-board computer. Trainers can have direct experience on the main concepts related to smart grids, renewable energy sources, electrochemical storage systems, and electric vehicles, through the use of the proposed tool managed by the web software interface.

Impurity detection in polymer parts for the semiconductor manufacturing industry

The very strict purity requirements of polymer parts that are used in the semiconductor manufacturing industry lead to an excessively low yield rate. To decrease the costs in the production chain, detailed knowledge about possible impurities during each processing step is essential. This work provides an overview of non-destructive testing techniques that are capable of either detecting or even identifying these impurities. The most promising techniques were then combined in a semi-automatized laboratory demonstrator that can be used as an in-line diagnostics tool to prevent processing of faulty parts. Results from these measurements are presented and discussed.