Reliability of electro-mechanical components subject to on/off cycling and continuous operation under multiple stresses

Power on/off and continuous operation are two important factors impacting the reliability of electronic and electromechanical components. In the reliability field, only limited work has been done to model component reliability involving transient phases and on/off cycling. This is due to difficulties in collecting data as well as establishing meaningful models which fit the data.

The purpose of this research was to study and model simple electro-mechanical components over a two-phase (on/off) life cycle under multiple stresses. In addition, this research was directed at determining a criterion for selecting between continuous and on/off operating policies. A reliability experiment was designed to study the on/off switching effect and continuous operation effect, under multiple load and voltage stresses, for simple electro-mechanical dc motors. A real-time C program was used for control and to monitor the lifetimes of the dc motors.

A failure mode analysis was used to study the physical causes by which the motors failed. Based on the results obtained, it was concluded that only one mode (a brush/commutator) failure mode was present. Continuous operation provided a longer life than on/off operation under the same environmental conditions, when off-time was removed. A general reliability model, based on the twoparameter Weibull proportional hazards life distribution, was developed. Due to the proportional hazards (common Weibull shape) nature of the model, a set of acceleration factors were developed to relate the stress levels. The calculated stress factor included multiple stresses.

Using this reliability model, different operating policies under different multiple stresses can be studied. The stress factor reliability model was extended to develop a simple policy decision rule. The rule compares the inverse of the stress factor with the demand duty cycle to determine the most effective operating policy. Hence, a quantitative analysis model, based on predicted reliability, has been developed to determine whether a continuous running policy or an on/off operation policy should be specified.