The diffusion failure mechanism was designed out.
The handbook was not updated to account for new /other failure mechanisms.
It became obsolete quickly: field data could not be collected, analyzed, and logged rapidly enough.
Increasing design complexity led to increased costs and time required to perform the analysis.
Average constant failure rate data did not correlate to actual failures.
It did not address infant mortality, quality, or wearout issues, where up to 65% of failures occur (O'Connor 2012).
The handbook was declared unsuitable for new designs and phased out in 1994. A University of Maryland study documented its flaws and proposed it be replaced by a reliability physics approach. Avoid the use of MTBF or MTTF as reliability metrics. Manipulating these numbers is easy due to the adjustment of multiple quality factors present in the model. MTBF and MTTF are frequently misinterpreted and provide a better fit for logistics and procurement activities.
Reliability Growth Modeling
The old, traditional Western approach to reliability growth and modeling used a design‐build‐text‐fix (DBTF) cycle. This was essentially a trial‐and‐error approach where the product was designed, prototypes were tested, failures/defects were discovered, corrections were made in the design, more prototypes were made, etc. Traditional OEMs spent almost 75% of product‐development costs on this approach (Allen and Jarman 1999). Shortcomings of this approach include:
Design issues are often not well defined.
Early build methods do not match final processes.
Testing doesn't equal actual customer usage.
Improving fault detection catches more problems but causes more rework.
Problems are found too late for effective corrective action; quick fixes are often used.
Testing more parts and more/longer tests are “seen as the only way” to increase reliability.
OEMs cannot afford the time or money to test to high reliability.
In DBTF‐based product development and validation process, reliability growth continued well into production and the field and was a highly reactive process. Using this approach, design engineers worked independently, then transferred designs either “over the wall” to the next department or external to the company. Eventually, manufacturing had to assemble a product not designed for its processes. Since it was too late to make changes, manufacturing struggled to meet yield, quality, cost, and delivery targets. This required trial‐and‐error crisis management, followed by launch delays, and then quality and reliability issues. This approach has fallen out of favor and been replaced by a combination of concurrent engineering and reliability physics modeling approaches. The newer approaches have the goal of simultaneously optimizing the design across all the DfX disciplines.
Block Diagrams
Block diagrams are useful for performing system‐reliability models and calculations and may be simple, parallel, or redundant. A series reliability system model is used when one failure of one component results in the failure of the system. Let's calculate the reliability of a simple fuel system, Rfs, as shown in Figure 2.2.
Rfs(t) = Rfp(t) Rfl(t) Rfi(t) Recu(t) Rfw(t)
Rfs(t) = 0.980 0.998 0.985 0.975 0.964 = 0.905 90.5% reliability
F(t) = 1 Rfs(t) = 1 .905 = .0945 9.45% failure
For every 100,000 vehicles built, be prepared for 100,000
Parallel or redundant reliability system models are used where all the items in a parallel branch must fail in order for the system to fail. They can model backup systems used to maintain system availability of critical functions in case the primary system fails. A repair of the failed unit is still required, but critical functionality is maintained: for example, separate brake channels where the loss of one brake circuit degrades brake performance, but reduced braking capability remains. Let's calculate the reliability of a parallel brake system, as shown in Figure 2.3.
Figure 2.2 Block diagram for a simple fuel system.
Figure 2.3 Parallel brake system.
Rbt = 1 [1 Rbc1(t)] [1 Rbc2(t)]
= 1 [ 1 0.990] [ 1 0.990]
= 1 [ 0.010] [ 0.010]
= 1 0.0001 = 0.9999 or 99.99%,
then Fbc = 1 [1 0.990] = 0.02 and Fbt =.0001
For one failure out of two, the calculation is
Automated Design Analysis
Automated design analysis tools represent the latest frontier in reliability analysis and modeling. One example is the ANSYS‐DfR Sherlock automated design analysis software. It is a reliability physics‐based electronics design software that provides fast and accurate life predictions for electronic hardware at the component, board, and system levels in early design stages. Sherlock allows designers to simulate real‐world conditions and accurately model PCBs and assemblies to predict solder fatigue due to thermal, mechanical, and shock and vibration conditions. Approximately 73% of product development costs are spent on the test‐fail‐fix‐repeat cycle. Sherlock design software provides fast and accurate reliability predictions in the earliest design stages tailored to specific materials, components, dies, printed circuit
