TABLE OF CONTENTS At the beginning of Chapter 5, we remarked that it is expensive to achieve reliability by testing. The cost of test activities is primarily incurred during development. In contrast, achieving reliability by redundancy is not as costly during development but is usually very costly in production (every unit produced incurs the cost) and in operation and maintenance. The other significant difference in reliability improvement between testing and redundancy is that testing protects primarily against systematic failures (inherent in design or process), whereas redundancy protects primarily against random failures. In that respect, the two techniques are complementary.
Where random failures predominate, such as in most electronic equipment, redundancy is usually the only way in which low probability of failure can be achieved and demonstrated. A disadvantage of redundancy, in addition to the added cost, weight, and power, is the higher overall failure rate (increased maintenance expense) because more equipment is exposed to failure.
This chapter discusses material that is applicable to all or most implementations of redundancy. Broad classifications of redundancy are discussed, including detailed evaluation of several of the most frequently used structures, as well as alternatives to physical redundancy. A summary concludes the chapter.
In Section 2.3 we introduced block diagrams of components being operated in parallel for the purpose of maintaining service when one of the components failed. This section explores this concept in greater depth. For the time being we assume that there is a perfect voting or...
