FIDES & Confidence Level

6 posts / 0 new
Last post
FIDES & Confidence Level

Hi,How can we interpret the FIDES calculation results and a confidence level requested 50%, 75%,90% ?For me the confidence level is only related to a reliability test results. But can we take as assumption that FIDES has a confidence level of 50%?Or Is there any other correlations? Thanks in advance AnthoTR


You are right, confidence level is a statistical concept not applicable to a reliability prediction like FIDES. Confidence level applies to intervals or their bounds. There is no rationales to say the FIDES prediction is the upper (or lower?) bound of a confidence interval. 
The only way proposed to build confidence levels would be to do Monte-Carlo simulations on the prediction after defining key parameters as distributions instead of fixed values. For example the temperature in the life profile could be proposed as gaussian. Unfortunatly no reliability tools propose this key innovation. If our new FIDES tool was a true excel tool (as I requested...) it would have been possible to use tools like SIMCAB or Crystal ball... but it is not the case.

Confidence level

Lambda's answer is more in terms of coverage of the life profiles than confidence, but it may be the same. If I use the 90th percentile of each "key parameter" for my failure rates, I should be certain that the result should cover 90% of life situations for any type of parameter. It will give a "very worst case". MonteCarlo approach is obviously statistical. I think that if not too much components are studied, Expertool may compute with one shot several results taking into account normal distributions of parameters. Using either differents "types" of components or differents phases in the mission. And then one computes average and deviation of the results...Regards

A standard output of models

A typical output of statistical models is the standard error of each of the variables.  Then one could do 'prediction intervals'.

Shiv Prasad
Regarding Phase 1 Application Field

Hi Lambda,

Could you please help me know whether I can use FIDES for part count method.

In part count method we did apply the application voltage or rated power, so in case of FIDES should keep zero value in field of application phase 1.

Hope for early reply



Dear Shiv Prasad,The FIDES

Dear Shiv Prasad,

The FIDES methodology has a "Part count" methodology (you can find it in page 227 to page 240) and there is also a specific sheet in ExperTool named "PART COUNT" in order to performed this kind of calculation.

I am really not sure to understand your question :

"In part count method we did apply the application voltage or rated power, so in case of FIDES should keep zero value in field of application phase 1."

but i hope this little text will help you.


In MIL-HDBK-217F notice 2 there is also a part count methodology (defined in page 128). In this methodology, you must count how many components you have in every generic part multiplied by a quality factor and for a specific environement. The result is obtained by reliability feedbacks.

In fact for FIDES, there are some similarities in the way to performed it. You must defined the application you want to make of the system by defining a mission profile (do not forget to set the Pi_application for each phase of your mission profile), how many components you have in every generic part proposed by FIDES, defined the Pi_PartManufacturing (equivalent to the quality factor) for every generic part and a generic Pi_placement. Then the results will be obtained by a physic calculation, but all the parameters required (including application voltage or rated power) are defined in pages 231-240.

So there is no need to know the stress applied to your components to perform this calculation. The values are defined in the FIDES Part Count methodology section.


Best regards.