Using probabilistic methods for solving the problem of ensuring leak tightness of heat exchanger tubes of nuclear power plant steam generators - Probabilistic Safety Assessment for Optimum Nuclear Power Plant Life Management (PLiM) - page 290

measurement error. Insufficient detectability of the defect is characterised by
the probability of defect detection (PDD) which is a function of defect size.
Currently, the PDD can be determined, according to the authors, only
by experiments on special samples with artificial defects whose dimensions
are known beforehand. Statistical analysis of nondestructive testing of this
model produces a PDD estimate.
In Ref. 131 it is porposed to describe the PDD function by the functions
shown graphically in Fig. 9.17

exp( b +b ln a )

Pa () =

0 1

;

[9.5]

d.d

1 − exp( b +b ln a )

0 1

− 1

 

 π ln a −µ 

 

Pa ( ) =+− 1 exp .

[9.6]

 

 

 

d.d

s

3  

   

In [9.5] and [9.6], values β 0 , β 1 , μ and σ are constant and can be
determined by statistical analysis. Interdependence of the constants has
the following form:

µ = −b b

/;

[9.7]

0

( )

[9.8] s=π b / / 3.

1

It is easy to estimate the physical dimensions of the cracks using the ISI
data if the detectability of inspection is known. In single inspection it is
also easy to estimate the number of actual cracks using the curve of defect
detection probability. Nevertheless, some problems may be encountered
in this case. For example, cracks can be detected only after they reach a
certain minimum size. However, some cracks are detected immediately after
reaching this size, while others are detected only after a few inspections. To
count the number of actual cracks, it is necessary to know the history of
inspection of each crack. This requires a special technique. It should be

￿ ￿ ￿ ￿ ￿

P d.d.

Defect size

9.17 Generalised form of the PDD curve.