Sources for Electronics Design Guidelines and Rationale for Component Derating

I've been asked to help review and update the design guidelines that my company uses for electronics design to produce reliable products. The list is fairly complete (I believe that it was based on a set of military or automotive guidelines) but one thing that we'd like to do is to add links to external sources to provide reasons for the guidelines.For instance we have a lot of guidelines of the form:
<component type>; electrical parameters derated by xx% for non-critical
applications, and yy% for critical applications.What I'd really like is a rationale for how much to derate components.
So, I'm looking for any links you might have to lists of design guidelines and/or explanations of why these guidelines are a good idea.

This is a big topic, I know because I am literally just writing a course on the FPGA / Digital aspects of it, but I have done plenty other areas too.The first rule is to identify what standards you are going to need to be compliant with they all result in many of the same steps but there are suitable differencesStandard areas are 1) part stress ensure your parts are de rated this includes voltage, temperature, power current etc. ESA, NASA and the US DoD provide some good reference de rating standards 2) Common Causes of failure make sure you remove them all from your design - This may or may not introduce redundancy 3) Consider the impacts of the environment, you can see some very strange effects - from things like vibration 4) Select the right parts quality - this can get expensive but sometimes it is necessary 5) Analysis to perform  - Preliminary Hazard Analysis, Fault Tree (top Down) FMECA (Bottom up), Part Stress Analysis, Functional Hazard Assessment, Independent Protection Layers, Layer of Protection Analysis 6)Worse case analysis - Analysis of the design to demonstrate functional performance over worst case conditions. Addresses design variability 7) Design Rules and Guidance - This is key to set the design team in the right direction 
You might be interested in the presentation Reliability in an hour I gave at FPGA Kongress last year 

ElizabethWhen I wrote my book on Excel, I developed a model for calculating MTBF based on a Bellcore Technical Reference . That very dated booklet was numbered TR-NWT-000332, Issue 3, September 1990. Most components were listed with their all the reliability parameters including stress details. Noticeably absent were electrolytic capacitors, I am told, because their reliability was insufficient for military applications.  Of course this may have changed. If you google "Bellcoe/Telcordia" (Telcom reliability) to should get to publications that will give you your answers. There is also MIL-HDBK-217 (it appears F is the latest revision) (Military reliability)It seems this article is quite good.I hope this is what you are looking for.If you like I can let you have the chapter which covers how MTBF is calculated using the factors in Excel ( but I don't remember a thing about it).

For smaller-value electrolytic capacitors, use hermetically-sealed tantalum capacitors. Make sure that they never experience reverse polarity. Momentary reverse polarity can start whiskers growing, which ruins long-term reliability.I once had to design a power supply for an avionics computer, without using electrolytic capacitors. They were willing to provide three-phase 120-volt 400-cycle AC power, so I used a three-phase bridge rectifier driving a Vicor DC-DC converter.

The standards you seek are generally best found from some of the major manufacturers of parts in that particular product area.  For instance, I find info on capacitors invaluable from companies like AVX, Kemet and Cornell Dubilier.  Risistors, Vishay, IC's companies like Texas Instruments and Linear Tech/Analog devices. For Inductors I look to Taiyo Yuden, Bourns, Pulse electronics (there has been a lot of considation in this area of components as well).
Simple derating factors are OK for when criticality doesn't require a more detailed analysis to get a good answer fast.   Some parameters you may have no choice but to use a simplified derating factor as its affect on performance may not be documented (like altitude on some capacitors).  For products that will need to last a very long time, it makes sense to dig deeper and understand ALL the specs that may be relevant to a design and then the choice becomes more obvious and not based on a simplified rule of thumb (for instance ESR affect on heating not important when currents are very low).When in double or when lacking other guidance you need to derate at least 2X the tolerance to insure parameters important to you stay in safe range.

I use www.kalcmate.com. Does Electrical Load and Fault Calculations

I have been using www.KaclMate.com for all my load and fault calculations.  Great tool and huge time saver. They offer a free trial.