An innovative Prognostic Health Management system for turbojet

Turbine engine failures are the leading cause of class-A mechanical failures (loss of aircraft). The cost of these incidents is astounding: increasingly, commercial air maintainers are turning to Prognostic Health Management (PHM) systems to prevent these losses and to reduce maintenance costs. For this reason, the development of PHM systems is advancing rapidly. Taking into account the previous scenario, an innovative PHM system should develop and validate a turbojet Health Management System (HMS) able to process operational data acquired from sensor-equipped turbojet. The most common health monitoring parameter in the aerospace engine industry is vibration, making vibration monitoring a crucial component of any PHM system. The vibration measure can offer important information only if the data is obtained in correspondence of sensible points of the engine critical components: the choice of the sensor application points can be made only by means of a punctual dynamical model of the entire engine. Moreover, the validation and verification of detection/diagnostic/prognostic capabilities requires access to high quality data for both healthy and faulty systems. While generic data are often readily available, more specific values for healthy or faulty systems are much more difficult to obtain. In particular, seeded fault tests are time consuming, expensive, and are not always representative of operational condition. The development of the HMS should be carried out through the deployment of high-fidelity Vibration Validation Tool (VVT) able to supplement physical testing. Currently, physical testing is the only avenue to validate diagnostic/prognostic systems. However these tests have high costs and duration, and are therefore conducted on a very limited basis. As a result, the need for faulty data is great but their availability is limited. The VVT fills this need by generating realistic vibration data for faulty scenarios, data that can complement any available data from operation or test bed data. The VVT would reduce the costs and time required for the development of a HMS system providing simulated, yet realistic, vibration data for healthy and unhealthy critical components. Physics-based, data driven and hybrid vibration source models for the critical components can be merged to emulate the entire engine. VVT can then be used to generate realistic vibration signals for engine in different combinations of fault states and operating parameters. These results can be used to accomplish many difficult tasks, such as evaluating new technologies, assessing the most effective location for sensors, and determining the ability to detect a specific fault minimizing the false alarms and missed detections. The main aim of this article is therefore to illustrate the concept, objectives and relevance of this novel PHM system VVT based showing architecture and approaches. In addition, the development of a tool for prognostic/diagnostic based on vibrations can be extended to several domains.