In the Oil & Gas Industry, large fleets of centrifugal pumps are used for different services working in diverse process conditions. More specifically in oil refineries, they have many characteristics in common since they are centrifugal machines that handle liquids using the same operating principle and because their design is highly standardized by API 610/ISO 13709 std for centrifugal pumps and API 682 std for sealing systems. As well, operating units and refining processes are well known and do not differ much regardless of where they are installed. Due to their criticality in the refining process, the reliability of these assets is of the utmost importance, being the MTBF (Mean Time Between Failures) one of the most used KPI (Key Performance Indicator) for evaluating it. Considering the characteristics indicated above, the possibility of predicting the MTBF of centrifugal pumps based on historical failure data, design features and expected operating conditions with Cox Proportional Hazards Model (PHM) is suggested.

Nowadays production companies are in a difficult situation since batch sizes are decreasing, the number of product variants is growing, and the demand is difficult to forecast. New technologies enable to design more complex production systems capable of handling these challenges, but these “Industry 4.0 solutions” are often very expensive and hard to implement. In Bosch Power Tool Ltd. (Hungary), a flexible manufacturing concept was implemented, which enables to produce efficiently even with batch size one, requires much less space from the shop floor, prevents disruptions in production, and last but not least realized in from a relatively small budget. The concept introduced in this paper can be used as a best practice for manufacturing companies facing similar challenges.

Wearable technology in healthcare refers to medical devices for real-time monitoring of a wide variety of biomedical parameters of the human body. The Health 4.0 scenario has led to the rapid development of these systems in all fields of medicine. However, a wearable non-invasive device for skin health monitoring for skin hydration control or cancer prevention is still an open challenge. In this regard, this paper addresses the design and the implementation of a low-cost portable system for skin diagnostics purposes which is based-on microwave reflectometry technique. For this purpose, a specific sensing element (SE), connected to a miniaturized vector network analyzer (m-VNA) has been designed and assessed for the optimal detection of the variation of the dielectric properties of the skin. The proposed system has been validated, first, through full wave simulations, then, directly on human skin, demonstrating a good potential for achieving a fully wearable system for skin monitoring and diagnostics.

The no-clean flux-residues pose a high reliability risk, thus controlled strictly in the modern electronics assembly industry. A series of customized Surface Insulation Resistance (SIR) Experiments and AC impedance analysis under various test patterns demonstrate the dramatic impact of the partial activation of the fluxes unevaporated solvents, and non-decomposed activators on the reliability of the final assembly. Although the mainstream no-clean pastes and liquid fluxes are qualified under all the standard SIR and electrochemical migration (ECM) reliability tests, the solder mask filled SIR patterns are more realistic and the tests results more accurate. The spaces are filled with solder mask thus the ionic migration analysis is more complex. From this perspective, we demonstrate a thorough examination of these more realistic structures with classic DC SIR method and AC impedance measurements. This study examined the impact of no-clean flux residues on capacitance and resistance in case of solder mask filled comb patterns.

The induction motors are a part of the bulk critical actuators in metallurgy, engineering and other industries. Generally, industries technological processes limit a redundancy of the critical actuators. Therefore, condition monitoring of the critical actuators parts is the basis of efficiency and reliability of the processes. In turn, the special operate conditions of the actuators, like unsteady speed and load, affects the motor condition monitoring success significantly. Induction motor rotor defects such as squirrel cage damage are minor but leads to unpredicted shutdown and greater maintenance costs. The present study presents a reliable method for detecting defects in squirrel cage rotor bars of the actuator induction motor which runs at various speed and load. The method is based on higher-order space harmonics processing in the motor current signal. The method combines normalization, filtering by variational mode decomposition, wavelet transform and train a convolutional neural network. The method generates a diagnostic model which allows to diagnosis motor rotor bar fault at various speed and load. At the same time, the model requires only one frequency and load for training. The experimental results show the model, which is trained at 35 Hz rotary speed, detects a rotor bar fault at 15 to 50 Hz rotary speed and up to 36% load of the nominal torque with 97% average accuracy. The proposed method is effective for the real equipment operating conditions which have limits of completeness datasets acquisition for training.