Availability Improvement of Construction Vehicles based on Global Machine Data Analysis
In both construction and extraction processes, mobile machines are part of a logistical transport chain in which the failure of one machine results in the stoppage of other machines, resulting in high downtime costs for the operator, even in the event of minor damage.
Available condition monitoring systems are cost-intensive and have mainly been established for stationary applications as of now. Development of a new type of intelligent condition monitoring system that provides a basis for decision-making makes it possible to reduce cost-intensive failures and to minimize and help planing necessary maintenance work.
Reduction of cost-intensive failures
|Detection of component damage in test rig|
|Early planning of maintenance work||Acquisition of global signal patterns of the complete vehicle under operating conditions|
|Knowledge of the remaining service life of individual machines||System modelling to simulate the lifespan of the machine
and to identify component damage
|Lifespan modelling of the drive train|
Challenges in the development of a condition monitoring system
Throughout its use, a machine is subject to a certain amount of wear and tear. If the wear exceeds a certain limit the machine will fail. At this point the wear reserve decreased to a point at which the function fulfilments under defined conditions are no longer achieved. The figure shows the reduction of the wear reserve due to use and the restoration of the initial state by repairing or improvement. Further reasons for machine failure are weak points of the machine. In addition to the natural wear and tear of the machine, spontaneous overloads can also lead to component failure.These spontaneously occurring conditions are critical, but difficult to identify in the context of predictive condition monitoring.
Available condition monitoring systems are cost-intensive and havemainly been established for stationary applications. Today, mobile machines are equipped with a large number of sensors and control units whose signals have a high information content. Nevertheless, because of high costs and lack of evaluation methods, the sensor signals are usually solely monitored for limit values as a protective function against critical operating states. The recorded data are not analyzed or used to predict the machine condition. Typical signs of wear and tear of a construction machinery drive train are shown below and have to be identified within the context of the research project.
Objective of the project ProMachineData
The research project "Verfügbarkeitssteigerung mobiler Arbeitsmaschinen durch Prognose des Maschinenzustandes auf Basis globaler Daten" used a new, indirect approach. All evaluable sensor and operating data available on the machine today were to be used to detect correlations between component damage and overall signal patterns and to determine the remaining lifespan of the individual machine based on the actual operating loads. This new approach to machine condition determination is based on two main types of information. On the one hand, the remaining component lifespan as a result of previously experienced stress and, on the other hand, the global signal behavior when component damage occurs. This enables a direct statement to be made on the probability of damage and the probability of failure of each component and assembly at any time during operation. The knowledge of the wear progress enables the prediction of the machine condition for the near future. An existing maintenance strategy can be optimized on the basis of real-time information and can be integrated into a higher-level planning system due to the condition prognosis. This marks the beginning of a new, fourth development stage in maintenance, in which knowledge-based sensor data utilization and real-time-based liefespan calculation enable intelligent onboard diagnostics that provide robust condition forecasts. By using this condition information for higher-level planning systems, greater planning reliability and thus an increase in process reliability can be achieved. The manufacturers of mobile machinery gain a clear technological advantage for machines equipped with such a CM-System, especially compared to the cheap competition from the far east. On the highly competitive international market for mobile machinery, this can compensate higher acquisition costs for a German mobile machinery in international comparison. In addition, reliable condition monitoring allows better utilization of the wear and tear of components, thus saving resources.In the proposed research project, the theoretical and technical requirements for the realization of the new intelligent CM system were researched and developed. Based on this, the participating companies will have the opportunity to develop the system to market maturity in the following years after the end of the project.
As part of the project, a study of component damage in underground loaders was carried out at the beginning. This made it possible to classify the components on the basis of their failure probabilities. A stationary test rig consisting of the components with the highest failure rates was set up and test cycles were performed. The cycles were repeated after defective components were installed. During all tests, all measurement signals from the sensors were recorded. These included sensors already present on delivery of the machine as well as newly installed sensors. In a second test phase, an underground loader was equipped with the same sensors and tests were run.
The evaluation of the signals using machine learning methods from the test bench phase showed a high degree of accuracy in the detection of damage. The knowledge gained could thus be transferred to the measurements from the underground tests. Here, it was also possible to detect damage to the cardan shaft by means of amplitude increases in vibration signals.
The study ProMachineData- Examination of the hydraulic components was funded by the European Union and Nordrhein-Westfalen. The ifas would like to thank all project partners, the European Fund for Regional Development and the State of NRW.
Wöll, Lothar, et al. "Reliability Evaluation of Drivetrains: Challenges for Off‐Highway Machines." System Reliability. InTech, 2017.
Duensing, Yannick, Rodas Rivas, Alejandro; Schmitz, Katharina, “Machine Learning for failure mode detection in mobile machinery”, KIT Scientific Publishing (2020)