This is a procedure that I use when I have to demonstrate a
specified reliability confidence statement, and at the same time to modify the design to
prevent reoccurrence of failure types when I find them. I believe it is applicable to any
type of device, assembly, or component. The procedure is supported by software to develop sampling plans for reliability.
I use a sequential
sampling plan to guide the debugging and acceptance testing
of software that I write. It is based on the operating
characteristic curve that expresses my required confidence statement. The OC-Curve's
metric is either MTBF or reliability. Each "bug" that I discover is fixed right
- When the sequential sampling plan is still in the continue zone, I just keep testing
more options and example data.
- When the sequential sampling plan rejects, I start it over at n=0 and continue testing
all options with new example data.
- When the sequential sampling plan accepts, the required reliability confidence statement
is confirmed and I stop testing.
This process will continue, with corrections being made,
until the data substantiates the required reliablity confidence statement.
The rationale for using operating characteristic curves
is described further in this applications area of this
A more thorough written report on this application is
available at no charge from H & H Servicco Corp.
Request the software reliability example.