Table of Contents ECNDT '98
Economic assessment of inspection - the inspection value methodM. Wall, F.A. Wedgwood
AEA Technology plc, National NDT Centre, Abingdon OX14 3DB, United Kingdom
Corresponding Author Contact:
AEA Sonomatic BV, Krombraak 15, 4906 CR, Oosterhout (Netherlands),
Phone: +31 (0)162 42 55 88. Fax: +31 (0) 162 42 43 43
Email: Wilbert.Martens@aea-technology.nl , URL: http://www.aeat.co.uk/ndt
|TABLE OF CONTENTS|
A performance based economic model of inspection based on cash value is described. This is called the Inspection Value Method. Three examples of plant inspection are shown to illustrate the method.
Inspection plays a vital role in manufacturing quality control and plant maintenance but managers are increasingly aware that economy both in manufacture and plant management are vital to competitive success. At the same time changes in government regulation have given these managers more flexibility and responsibility in assuring the safety of products and operating plant. This means that in planning inspection a method of economic assessment will become increasingly valuable so that a method of quantifying the benefits and costs of NDT is worthwhile.
Since Inspection by its nature is an investment; pay now, benefit later, it would be ideal if decisions on inspection were carried out, however simply, on standard investment accounting methods. For this reason ideas about trying to quantify costs and benefits of inspection have been developed by the UK National NDT Centre (1). We call this the Inspection Value Method (IVM) and its main emphasis is to quantify all the important Costs and Benefits in terms of cash. We then define Inspection Value as:
In this paper the outline for carrying out IVM is set out. The applications are very varied, ranging from setting threshold levels in a single inspection procedure to strategic decisions on plant management, so that understanding the method is best done using a series of examples. In the following sections the formalism is outlined and examples on oil tanks, pipelines and pressure vessels are shown. These examples were all carried out using spreadsheet programs.
Definitions of Cost and Benefit
We define the following:
|C:||The total Cost of inspection including direct costs and indirect costs such as plant shut down caused by the inspection.|
|B:||The Benefit of the inspection which includes the benefit of compliance with legislation and reduction of risk of failure.|
|V:||The added Value given by inspection; defined as V = B - C|
|P:||The probability of failure due to defects in manufactured items or operating plant.|
|P:||The reduction of P due to inspection|
|CF:||The Consequence of the failure, preferably given in cash terms. Sometimes called the severity of failure. This should be expressed in cash terms if possible.|
|R:||The Risk of failure defined as the product P×CF|
Since the Benefit of inspection frequently accrues at a later date than the cost the usual financial planning formulae can be used. In particular the added value can be given as Net Present Value:
where t is the elapsed time in years and the sum is continued for a period of T years; Bt
and Ct are the annual benefits and costs and r is the discount rate %. T is the total time for which the inspection can be considered effective.
Frequently it is difficult to make cash estimates of the severity CF so we make use of the relative value of inspection VR where:
We define four factors which determine Inspection Performance (1). These are:
In risk based inspection the benefit of inspection is given as some fraction of the risk of failure. Risk is usually categorised into 5 different levels (based on a 3×3 rank matrix) but no attempt is made to quantify these levels in cash terms. We consider that it is worth attempting to make some cash estimate of this risk since it can then be compared directly with the inspection cost.. Given a cash value for a risk the benefit of inspection can be calculated as the reduction of that risk by inspection. For a single mode of failure of a single component, a suitable inspection would reduce this risk by the factors POD, the probability of detection, and F, the coverage. Then the Benefit, is given by:
Magnetic (MFE) floor-scanning has become accepted as the standard method for inspection of oil storage tanks for corrosion pitting which could lead to leakage. The tank bottom is screened using the floor-scanner with ultrasonic inspection then used to characterise any defects found. This two-stage method has produced substantial savings in the cost and time of inspecting tanks.
The critical parameter for the operator is the setting of the threshold for reporting (indicating) a suspect area during the magnetic survey process. A low threshold will give a high probability of detecting the real defect but a large number of "false positive" indications which have to be followed up unnecessarily by the UT inspection; too high a threshold will greatly reduce these but increase the chance of missing the real defect. The threshold is set relative to a standard reference defect in a test plate: for example a half wall penetration 120 ° cone.
The inspection performance, populations and costs have been collected from records of calibration measurements and utilisation of some of the 30 or more MFE Floorscanners built by AEAT and operated over the past 8 years. As an example, estimates have been made of the various costs for inspection of the floor of a typical 40m diameter storage tank. The calculation is carried out using the following cost data: Cost of inspection of tank; £2800, Cost of UT reinspection per indication; £1, cost of repair per confirmed defect £1000. The consequence of a subsequent leak due to a missed corrosion pit is estimated to be £50,000.
The signal probability distribution calculated from the experimental measurements on real and reference defects was used to estimate the POD dependent on the threshold chosen. Similarly the false-call rate (NFI) was calculated from the measured noise distributions. The IVM method has been used to calculate the Value and the optimum setting of the threshold. No account was taken of the time delay between inspection and possible occurrence of a leak, e.g. NPV was not estimated.
The results for three cases are shown in Figure 1. The cases correspond to progressively increasing levels of noise relative to the signal from a true critical pitting defect. The first case uses actual noise data from a relatively clean floor-plate; the second corresponds to a doubling of this noise (typical of many as-found plates) and the third to a trebling of noise level as has been observed on plates with seriously corroded and scaled top surfaces or on floors too thick to allow the magnet fitted to the Floor-scanner to magnetically saturate the plate. The figure shows that a higher optimum value for threshold should be used the noisier the plate.
|Fig 1: Value of Inspection for low, medium and high noise conditions|
In the first case the optimum setting of the threshold (40% of the mean critical standard signal) gives a probability of 96.8% of detecting the one threatening defect and a net value of £44,350. Under these conditions the predicted number of false positive indications requiring checking is of the order of 300 (corresponding to about 3 per plate). Note in Figure 12 that lowering the threshold rapidly leads to poor added-value as this number of false positives escalates without a significant improvement in the probability of real defect detection.
In the second case there has been a falling off of the probability of detection of the critical defect to near 90%, but the net value of carrying out the inspection is still high. In the third case, the occurrence of false positives from the noise has driven the optimum threshold setting so high that there is a 60% probability of missing the critical defect. Interestingly, the net value is still positive, although greatly reduced.
Due to the age and condition of many iron water mains they are being replaced or lined with plastic. This is
a major investment with a very long payback period so it is well worth making sure the worst pipes are
replaced first. Inspection is one option for choosing which sections of pipe to replace. Firstly we need to define
a measure of "Pipe Condition". This can be defined as the number of corrosion pits greater than 50 % wall thickness
per km. In most cases the corrosion is on the outside of the pipe and is highly variable due to ground condition
variables. Suppose there are on average 1000 pits per km which are potential leaks (nc). A way of sorting out sections of pipe into "Good" and Bad" is to choose those with more than the average number of pits as Bad and replace or reline those sections only (i.e n >nc). The problem now is how accurately can we sort out the sections into good and bad?
The accuracy of counting defects depends on the square root of the number detected, and thus on the square root of
EOD. Suppose the actual number of pits in a section of pipe is na and we measure this using an inspection method with EOD given by
Figure 2 shows the probabilities of correctly retaining a good kilometre of pipe and correctly rejecting a bad one based on Gaussian statistics. The area below the curves gives the proportion of cases where the correct decision is made which we call the benefit factor. In figure 3 we have assumed that relining pipe costs about £100, 000 per km and choosing the right sections to reline is about 20% of this. Full inspection (EOD = 100%) could cost between £10,000 and £50,000 depending on method . Here it is set at £40,000 and it can be seen that the optimum setting of EOD is about 5% and the added value is £14,000 per km. Note that to achieve a EOD of 2% does not put high demands on POD: for instance we could use an inspection with POD 50% and coverage 10%.
This example illustrates the use of the IVM method to evaluate different methods of inspecting a typical pressure vessel (Figure 4) used in an offshore gas platform. It is an inlet separator which removes sand and liquid from the gas. This was one of a set of vessels where contractors had carried out a full risk analysis and had been classified as "medium" Probability of Failure (P) and "medium" Consequence of Failure (CF). Medium consequence is interpreted as "Requiring shutdown of system for up to 2 days" meaning a loss of production of, typically, £1,000,000. In this spreadsheet we calculate the NPV for a period of 20 years making the following assumptions.
Finally the spreadsheet calculates two total costs; that for no inspection and that for inspection every 3 years. The difference between these is the added value which is multiplied by the discount factor and plotted to give annual discounted value . This is summed for 20 years to give the NPV. Assuming the input parameters are reasonable we then ran the program for the three inspection methods for various inspection intervals ranging from 1 to 10 years. These are summarised below.
|CORROSION HISTORY : 20 Year NPV(£1000)
||MANUAL UT EXTERNAL
This can be used as a basis for choosing which methods to use and for planning . for instance the present method uses a combination of internal visual and external UT inspection at 3 year intervals. This has a combined NPV of over £200k which is significantly larger than the external UT C-Scan method. this argues for no change. However it should be noted that the ultrasonic methods do not require access to the pressure vessel; in the present calculation it was assumed that there is no extra cost for access. If the UT C-Scan were carried out without shutdown there could be a considerable advantage over the present invasive visual inspection.
It is important to check the calculation to see how sensitive it is to input data which is not well known. It can be seen below that the separator vessel calculation is very sensitive to corrosion pi growth rate (g) but insensitive to the number of pits (n). This means that effort is needed to get a reasonably accurate value of g from observations on this or similar vessels.
|Sensitivity Check on pitting NPV|
|UT C - Scan, 5 year inspection,||Vary g and n|
|g mm/yr||NPV £k||defects/sqm||NPV|
The Inspection Value Method of assessing inspection has been outlined and three examples shown from plant inspection. These range from simple applications to more complex ones where considerable data input on inspection performance and costs are required. The single bottom line is the Value or Net Present Value which gives an unambiguous decision on inspection choice. Since many of the input data are uncertain it is usually necessary to check the sensitivity of the result to variations in that data. As long as this is done we have found many applications where the IVM method is valuable in both detailed and strategic planning of inspection.
The authors would like to thank the members of the Harwell Offshore Inspection R&D Service who have part-sponsored the development of the economic assessment method (IVM) mentioned in this paper. The current HOIS members are Amerada Hess, British Gas, BP, DNV, Norsk Hydro, Phillips Norway, Phillips UK, RTD, Saga, Saudi Aramco, Shell, Statoil, Texaco Britain & HSE.