Statistics Seminar on Wednesday 6th October 2004

This seminar consists of two short talks

• Speaker: Dr Mordechai Jaegar, University of Haifa, Israel
• Time: 4:00p.m. Wednesday 6th October 2004
• Venue: Red Centre Building Room RC-3084, near Barker Street Gate 14

Title 1: IDENTIFYING SENSITIVITY CHANGES BETWEEN EXPLOSIVE BATCHES

A simple test procedure for the identification of variation in shock
sensitivity between batches is suggested. Thia can be adapted to both
acceptance and aging tests. A normal distribution of the shock sensitivity
threshold is assumed, whose parameters are known from past experience.
Sensitivity test results are of the 'go/no go' type; therefoe plannig a
testinvolves determining the test shock levels and defining a success
criterion. The test procedure is required to be statistically powerful and
yet easy to apply.

Title 2: DESIGN OF EXPERIMENTS FOR SYSTEMS RELIABILITY DEMONSTRATIN AND GROWTH

A systematic reliability test program is a part of the design program of a new
product. Existing models deal with an abstract reliability growth function or
with "demonstration" tests. Our approach is to correlate betwee the overall
product reliability and the occurrence probability of major failure modes. We
present a model for planning an "optimal" development testing program. It is
a Bayesian reliability-cost model, which refers to uncertainties on failure
modes probability evaluations and the cost of the tests.