NDT RELIABILITY - A WAY TO GO

G.M. van Dijka and J. Boogaardb

aKEMA N.V., P.O.Box 9035, 6800 ET ARNHEM (NL)

bDSM Research b.v., P.O.Box 18, 6160 MD GELEEN (NL)

SUMMARY

Proper consideration of NDT reliability can only be achieved if reliability targets are specified in a relevant and achievable manner. A general approach in this respect is being presented taking into account reliability characteristics of NDT inspection systems and different inspection regimes. Limiting values for the applicable reliability characteristics are suggested considering a two class system.

1 INTRODUCTION

Reliability of products determines to a large extent their fitness for use. Inspection as well during manufacture as in service is an important tool to ensure this reliability. Proper inspection calls for a well-balanced inspection programme, i.e. choice of the inspection methods and the scope of inspection, with due attention to the cost factor. There is a growing need to assess these choices quantitativily. This is brought about on the one hand by general quality assurance principles requiring all relevant matters to be documented and traceable. On the other hand, this need is a natural consequence of the growing trend towards product certification, which provides easy-to-use benchmarks for technical product quality and reliability.

Up until now, the selection criteria for inspection methods have been limited to the aspects of sensitivity, procedure and qualification. The aspect of reliability of an inspection method has never been a criterion. In selecting an inspection system, it is well to aim at, and to quantify, the eventual overall effectiveness of the individual inspections. This requires quantification of the inspection reliability also. Although this may complicate the selection process, it offers important advantages to manufacturers, users, licensing authorities and inspection organisations. For licensing authorities, for instance, it provides a better means of appraisal. To manufacturers and users it provides the possibility of choice and optimisation, economic or otherwise. Furthermore, it enables inspection organisations to find economical alternatives by quantifying the value of the individual inspection procedures. An approach based on inspection effectiveness reveals what technical developments are most worthwile.

This presentation aims to illustrate in which way the aspect of reliability can be included in the selection of inspection programmes.

2 INSPECTION

2.1 Objective

Most usually, product properties show significant variations as a result of variations in the material properties, manufacturing process and in-service effects. Inspection aims to minimise the chances of negative excursions in the quality of a finished product through early detection of defects. A general distribution curve for the product quality is shown in Figure 1 (supposing that product quality is inversely proportional to the size of defects).

Figure 1 Probability of product quality value distribution and characteristic values.

In this figure the following characteristic values and their interdependences are important:

Figure 1 shows also the effect of inspection on the eventual product quality distribution. Apparently, in the example depicted here, it is only marginal. The quality is largely determined by the materials and the manufacturing processes employed. Accordingly, inspection will yield only a limited rejection rate. Inspection will gain importance when the product quality distribution changes. This may involve a shift in the average product quality, as shown in Figure 2, because of changes in the manufacturing parameters or deterioration of bare material properties.

Figure 2 Effect on probability distribution curve, due to a decrease in finished product quality.

Inspection will also gain importance with wider spreads in product quality (Figure 3) as may result from, for instance, wider manufacturing tolerances.

Figure 3 Effect on probability distribution curve due to an increase in manufacturing tolerances.

Inspection should be aimed at maintaining and assuring the quality of the product as delivered or retested during service. In actual practice, the defect detection score at the target quality value D often is less than 100% so that the inspection would seem to be inadequate and to fail to meet its objective.

We shall now discuss whether this conclusion is correct and whether it has any fuurther implications.

2.2 Inspection regimes

With inspections for assuring a given specified quality value S broadly two extreme scopes may be distinguished:

Symptom-oriented inspections

Here, inspection is mainly aimed at consistent detection of relevant defects to enable initiation of corrective action. A 100% defect detection score is not necessary. Examples are weld inspections of items that need not meet any specified rational quality values but where, rather, the quality is assessed for good workmanship. Inspections in mass production, usually focussing on process monitoring, fall into this category also. Another example are in-service inspections aimed at detecting not immediately critical degradation phenomena such as corrosion.

Characteristics of the inspections within this scope involve the following aspects:

- A constant detection characteristic. The detection score need not necessarily be 100%. It is essential, however, that the number of nonconformances observed in consecutive inspections truly reflect the actual number of nonconformances to enable appropriate corrective action.

- A low false rejection rate. Many inspection methods involve observation of what are believed to be nonconformances and of nonconformances that are immaterial to the product quality. A high proportion of good products being rejected can be costly and may lead to inappropriate corrective action being taken on false grounds.

- Coverage. The inspection must be well able to detect relevant nonconformances; in other words, specific nonconformances of any significance must not consistently be overlooked.

Product-oriented inspection

Here, the primary objective of inspection is to assure a defined minimum quality level, usually of single products. Examples are in-service inspections of loaded structures for fatigue cracks and inspections during manufacture of the wall thickness of parts impairing strength.

Characteristics of the inspections within this scope involve the following aspects:

- Very high coverage. In other words, the applied inspection system must not consistently overlook any of the relevant nonconformances (e.g. no radiography for crack detection).

- Ideally, the detection score is 100%, even if this entails some rejection of good products.

- A constant detection characteristic is essential for successive in-service inspections so that the observations accurately reveal the degradation trend.

3 RELIABILITY CHARACTERISTICS OF INSPECTIONS

3.1 Detection curve

A typical detection curve is shown in Figure 4. Here, the following parameters are of importance:

Figure 4 Detection curve.