Introduction
The Reliability Case is becoming an ever more popular document for the UK MOD to get industry to produce as part
of a contract. The principle aim of the Reliability Case is for the supplier to guarantee that their product will
meet an agreed set of in-service reliability requirements. The onus of responsibility is on the supplier to build
the case by gathering evidence, that the equipment will meet the reliability requirements. The reliability case is
held by the supplier and is not submitted to the customer. The supplier then develops a Reliability Case Report, which
contains a summary of the Reliability Case with supporting evidence.
It is this report that the supplier will have to get the customer to agree to, and the better the evidence contained
within the report, the more likely it is that the supplier will accept the argument that the equipment will satisfy the
reliability requirements. Different techniques should be employed, dependent on the type of equipment that is being
designed, and if similar equipment has been designed previously.
The different techniques that can be used to build your argument contained within
the Reliability Case are detailed below. Some of these techniques (e.g. In-Service Data) are more likely to provide a
stronger case for reliability than others (i.e. calculation). The choice of technique is primarily dependent upon the quality
and quantity of information available. Invariably, a combination of these techniques will yield a much stronger Reliability
Case than any single technique.
Index
Click on the links below to view the various techniques mentioned above.
In-Service Data
Calculation
Testing
Simulation
Analysis
Expert Opinion
Best Practice
In-Service Data
Data that has been collected through the study of similar equipment being used by the customer. Historical data
related to failures can be collected by both the customer as well as the supplier of the equipment and can provide a
very strong case for reliability. In-Service Data can sometimes form the strongest argument for reliability however,
this is very much dependent upon such factors as:
- Similarity / applicability of the In-Service data to the new design.
- Quality of the data in terms of accuracy and traceability.
- The quantity of data available.
When analysing such data, it is important to filter data which is relevant to the design. Although often important
when assessing re-design, failures induced by mishandling or operation outside the intended operating environment can
be discounted, when calculating in-service reliability.
Calculation
Reliability data can be calculated by the analysis of each component within a system against a pre-defined model.
The two primary methods of performing these calculations are:
- Parts Count – summates all similar components and multiplies this by the number of components. This is a
simplistic method and used more often at the earlier stages of a project.
- Parts Stress Prediction – used once the design has been firmed up. Each component is taken in turn and
analysed using the stress levels and environmental factors to predict the reliability.
There are a number of different standards that these predictions can be based on. For the purposes of electronic
components, most UK military predictions are based on Mil-Standard 217F Notice 2, though Belcore, Telcordia and RDF are
all valid, if they suit the prediction that you are developing. Analysis of mechanical components will often involve
the use of NPRD 95.
Testing
Testing is often a costly and time consuming exercise, and appropriate test routines are often difficult to define.
The results of a reliability prediction can be used to focus testing on specific areas of the design and also assist in
defining appropriate test conditions. Testing of these areas can then verify whether or not the reliability prediction
has identified potential flaws in the design or production processes. Any resultant design changes then need to be
reflected in the reliability prediction. The whole process is iterative and should hopefully result in significant
improvements in reliability at minimal cost.
There are many different testing methods and profiles however; most of these will involve some sort of temperature and
vibration testing which simulate possible worst case field conditions. This process is intended to highlight any
weaknesses in the design before it goes into service. Another approach involves overstressing the design to accelerate
the life of the equipment in order to induce potential age related failures.
Simulation
System simulations can be performed when other methods are incapable of modelling strong dependencies between failures.
Simulation can also be used with components that are repairable that exhibit non-constant failure rates.
Some Reliability software tools have incorporated a Simulation facility. The Monte Carlo method utilises a random
number generator to replicate the way the system may fail whilst in-service.
Analysis
Any design can be improved by focusing on weak areas and through analysing the total design focal points can be
established for improvement. One method of analysing the design is by performing Fault Tree Analysis (FTA) which can
identify critical paths within the design.
Expert Opinion
Past performance can be used to build a case of reliability. In-service data of similar systems can be used as a
bench mark providing an indication of future reliability.
Best Practice
Techniques should be used through the design process to increase reliability, by eliminating flawed design practices.
Best practice can be utilised in component selection, the maximum stress levels for components and the location of
sensitive components. There are many of these best practices that should be followed and by doing so many rudimentary
errors could be avoided.
