Modeling the Effects of Future Tests and Redesign
In reliability based design optimization, we often design for a given uncertainty and compensate for that uncertainty by applying safety factors. However, uncertainty reduction measures, such as thermal and structural tests, are usually part of the design process and the effect of these measures is not taken into account in the design. Therefore, it should be beneficial to include the effect of a "future" test and consider its effect on the reliability during the initial design stage.
A test is performed to check the accuracy of a compuational model. Based on a test result, we can calibrate the computational model, and improve any estimate of the probability of failure. If the this improved model shows that the design is dangerous or overly conservative (unnecessarily large mass at too high a level of safety), the component can be redesigned.
We have developed a method of including a future test and redesign into reliability calculations during the initial design stage. We have shown that a test can reduce the probability of failure significantly when the objective is to redesign for the original safety factor. When the objective is to redesign for probability of failure, we have shown significant savings in weight.
The challenge of the designer is to determine the level of conservativeness of the design to balance the risk of redesign. We examine the trade-off between performance (e.g., mass) and development costs (measured by redesign probability) by seeking optimal design and re-design rules.