The first step in the reliability engineering process is to specify the required reliability that the equipment/system must be designed to achieve. The essential elements of a reliability specification are:
- a quantitative statement of the reliability requirement.
- a full description of the environment in which the equipment/system will be stored, transported, operated and maintained.
- the time measure or mission profile.
- a clear definition of what constitutes failure.
- a description of the test procedure with accept/reject criteria that will be used to demonstrate the specified reliability.
Quantitative Requirement
To be meaningful, a reliability requirement must be specified quantitatively. There are four basic ways in which a reliability requirement may be defined:
- As a “mean life” or mean time between failure, MTBF. This definition is useful for long life systems in which the form of the reliability distribution is not too critical or where the planned mission lengths are always short relative to the specified mean life. Although this definition is adequate for specifying life, it gives no positive assurance of a specified level of reliability in early life, except as the assumption of an exponential distribution can be proven to be valid.
- As a probability of survival for a specified period of time, t. This definition is useful for defining reliability when a high reliability is required during the mission period but mean time to failure beyond the mission period is of little tactical consequence except as it influences availability.
- As a probability of success, independent of time. This definition is useful for specifying the reliability of one-shot devices such as the flight reliability of missiles, the detonation reliability of warheads, etc. It is also specified for these items which are cyclic such as the launch reliability.
- As a “failure rate” over a specified period of time. This definition is useful for specifying the reliability of parts, units, and assemblies whose mean lives are too long to be meaningful or whose reliability for the time period of interest approaches unity.
The reliability requirement may be specified in either of two ways:
- As a nominal or design value with which the customer would be satisfied, on the average; or
- As a minimum acceptable value below which the customer would find the system totally unacceptable and could not be tolerated in the operational environment a value based upon the operational requirements.
Whichever value is chosen as the specified requirement, there are two rules that should be applied;
- when a nominal value is specified as a requirement, always specify a minimum acceptable value which the system must exceed,
- when a minimum value alone is used to specify the requirement, always insure that it is clearly defined as minimum. In MIL-STD-781, the nominal value is termed the “upper test MTBF” and the minimum acceptable value is the “lower test MTBF.” Of the two methods, the first is by far the best, since it automatically establishes the design goal at or above a known minimum.
Example:
A complex radar has both search and track functions. It is also possible to operate the search function in both a low and high power mode. The reliability requirement for this system could be expressed as:
The reliability of the system shall be at least:
- Case I High power search 28 hours MTBF
- Case II Low power search 40 hours MTBF
- Case III Track 0.98 probability of satisfactory performance for 1/2 hour
The definition of satisfactory performance must include limits for each case. These are necessary since if the radar falls below the specified limits for each case, it is considered to have failed the reliability requirement. An important consideration in developing the reliability requirement is that it be realistic in terms of real need, yet consistent with current design state-of-the-art. Otherwise, the requirement may be unattainable or attainable only at a significant expenditure of time and money.
Environment
The reliability specification must cover all aspects of the use environment to which the item will be exposed and which can influence the probability of failure. The specification should establish in standard terminology the “use” conditions under which the item must provide the required performances. “Use” conditions refer to all known use conditions under which the specified reliability is to be obtained. Examples include:
- Temperature
- Penetration/Abrasion
- Humidity
- Ambient Light
- Shock
- Mounting Position
- Vibration
- Weather (wind, rain, snow)
- Pressure
- Operator Skills
The “Use” conditions are presented in two ways:
- Narrative. Brief description of the anticipated operational conditions under which the system will be used. The electronics pod must be capable of withstanding exposed airborne environments encountered while suspended from the aircraft wing for periods up to three hours. This includes possible ice loading conditions, in subzero weather, etc.
- Specific. Itemized list of known or anticipated ranges of environments and conditions. When changes of environment are expected throughout an operating period, as in an aircraft flight, an environmental profile should be included.
Time Measure
Time is vital to the quantitative description of reliability. It is the independent variable in the reliability function. The system usage from a time standpoint in large measure determines the form of the reliability expression of which time is an integral part. For those cases where a system is not designed for continuous operation, total anticipated time profile or time sequences of operation should be defined either in terms of duty cycles or profile charts.
Example:
The mission reliability for the airborne radar system shall be at least 0.9 for a six hour mission having the typical operational sequence in the appendix defining mission profile.
Definition of Failure
A clear, unequivocal definition of “failure” must be established for the equipment or system in relation to its important performance parameters. Successful system (or equipment) performance must be defined. It must also be expressed in terms which will be measurable during the demonstration test.
Parameter measurements will usually include both go/nogo performance attributes and variable performance characteristics. Failure of go/nogo performance attributes such as channel switching, target acquisition, motor ignition, etc., are relatively easy to define and measure to provide a yes/no decision boundary. Failure of a variable performance characteristic, on the other hand, is more difficult to define in relation to the specific limits outside of which system performance is considered unsatisfactory. The limits of acceptable performance are those beyond which a mission may be degraded to an unacceptable level. They must be defined in clear, unequivocal terms. This will minimize the chance for subjective interpretation of failure definition, and post test rationalization (other than legitimate diagnosis) of observed failures.
Reliability Demonstration
It is not enough to merely specify the reliability requirement. One must also delineate the test(s) that will be performed to verify whether the specified requirement has been met. In essence, the element of reliability specification should answer the following questions:
- How the equipment/system will be tested (the specified test conditions, e.g., environmental conditions, test measures, length of test, equipment operating conditions, accept/reject criteria, test reporting requirements, etc.)
- Who will perform the tests (contractor, Government, independent organization)
- When the tests will be performed (development, production, field operation)
- Where the tests will be performed (contractor’s plant, Government organization)
References:
1. MIL-HDBK-338, Electronic Reliability Design Handbook, 15 Oct 84
2. Bazovsky, Igor, Reliability Theory and Practice
3. O’Connor, Patrick, D. T., Practical Reliability Engineering
4. Birolini, Alessandro, Reliability Engineering: Theory and Practice