Abstract
This paper introduces a small scale experiment, which simplifies validating the effectiveness of reliability approaches combining performance and reliability measures based on diagnosis, reliability estimation, and reliability prognosis. The experiment consists of a classical slot car setup. The vehicles motor power is mapped on an overloaded light-emitting diode (LED), which is operating in parallel to the direct current (DC) motor. The rule of the race is: If the LED fails during a race, the driver lost the game.
First of all LED reliability properties depending on constant power load are tested. For every power level an a posteriori mean time to failure (MTTF) value is calculated. The linear damage hypothesis approach is applied here estimating semiconductor reliability, namely an a priori estimated value of the time to failure. Based on this measure, it is easy to conduct a reliability prognosis. Prognosis can be conducted during the race. Comparing MTTF, estimated time to failure, and prognosis values, results are pretty interesting. However, individual vehicle prognoses are based on individual diagnosis data and mean value-based reliability statements. A driver should keep that in mind if he/she controls his/her vehicle based on prognosis values calculated during a race.
