|
|
Baseline Reliability Prediction at the Early Product Design Stage
Standards based reliability prediction is an important tool for predicting reliability for systems and devices. It uses globally recognized military or commercial standards at the early design stage to evaluate whether the design meets certain reliability goals. As part of the Synthesis Platform, Lambda Predict can share the prediction reliability models with other applications such as BlockSim for further analysis. In this article we will use an example to show how the analysis result in Lambda Predict can be used by BlockSim.
Example
A light control device maker designed a new light controller using the PWM modulation technique. You are a reliability engineer and have been asked to evaluate the reliability of the new design. No testing or any other failure information is available for this device at this early design stage. Therefore, using failure data to evaluate its reliability is not possible.
To estimate the baseline reliability of this device, you decide to use a standards based reliability prediction method and use the Telcordia SR-332 Issue 3 standard to do the prediction for this controller.
Based on the design layout of the controller, you build the system hierarchy for the controller in Lambda Predict as shown next.
The light control system consists of three subassemblies: the signal detection subassembly, the light control subassembly and the PWM generation subassembly. All the subsystems and the components under them are in a series configuration.
In almost all reliability prediction standards, all the components are assumed to have constant failure rates. In other words, the failure time of each component is assumed to follow an exponential distribution.
Since each component in a given subsystem has a constant failure rate and all the components are in a series configuration, the failure rate of a subsystem is the sum of the failure rates of all the components under it. For example, the Signal Detection Block subsystem in the above system structure has its failure rate calculated as:
The overall system failure rate is calculated as:
Therefore, the failure rate for the system is 261.4611 FITs (failures per billion hours).
Next you decide to transfer this reliability prediction model to BlockSim for further analysis in order to answer the following questions:
- What is the reliability of the device for a 15 year period?
- If the 15 year reliability requirement for the controller is 98.5%, how many redundant controllers are needed?
- Since the PWM generation block has the highest failure rate, instead of using redundancy for the entire controller, how many redundant PWM blocks are needed in a single controller?
To answer these questions, you first need to publish the failure rate model for the entire system. To do this, select the Light Control System block in Lambda Predict then choose Prediction Tools > Share > Publish Branch.
You then open the same project in the BlockSim software and choose Insert > Build from Synthesis > Build RBDs and FTs from Synthesis. In the Build RBDs and FTs from Synthesis window, you choose to build an analytical diagram from Lambda Predict, as shown next.
You then select the Light Control Prediction check box, as shown next.
BlockSim will create one diagram for the overall system that is linked to one subdiagram that contains the subsystems, as shown in the following pictures.
Diagram 1: Light Control System
Diagram 2: Light Control Subsystems
You then analyze the overall light control system (Diagram 1) and open the Quick Calculation Pad to calculate the reliability at 15 years. As shown next, it is 96.4%.
Since the reliability requirement of 98.5% at 15 years is not met, you modify Diagram 1 to add one redundant controller, as shown next.
You then reanalyze the RBD. As shown next, the calculated reliability is 99.87%.
This result shows that when you add one redundant controller, the system reliability meets the requirement.
To answer the third question, you modify the subsystem diagram (Diagram 2) to include a redundant PWM generation block, as shown next.
You then analyze the RBD. As shown next, the calculated reliability of 98.83% with the two redundant PWM generation blocks meets the 98.5% requirement.
Conclusion
In this article, we used an example to illustrate how to share the failure rate data obtained in a reliability prediction created in Lambda Predict with BlockSim for advanced analysis. Using the same method of publishing models, the failure rate data could also be used for other advanced reliability analysis in Xfmea, RCM++ or RBI.
