Annals of Nuclear Energy 66 (2014) 13–19

Contents lists available at ScienceDirect

Annals of Nuclear Energy

journal homepage: www.elsevier.com/locate/anucene

New approach for risk based inspection of H2S based Process Plants
Gopika Vinod ⇑, Pavan K. Sharma, T.V. Santosh, M. Hari Prasad, K.K. Vaze
Reactor Safety Division, Bhabha Atomic Research Centre, Mumbai 400 085, India

a r t i c l e i n f o a b s t r a c t

Article history: Recent trend in risk informed and risk based approaches in life management issues have certainly put the
Received 29 January 2013 focus on developing estimation methods for real risk. Idea of employing risk as an optimising measure for
Received in revised form 30 July 2013 in-service inspection, termed as risk based inspection, was accepted in principle from late 80s. While
Accepted 27 August 2013
applying risk based inspection, consequence of failure from each component needs to be assessed. Con-
Available online 21 December 2013
sequence evaluation in a Process Plant is a crucial task. It may be noted that, in general, the number of
components to be considered for life management is very large and hence the consequence evaluation
Keywords:
resulting from their failures (individually) is a laborious task. Screening of critical components is usually
Hydrogen sulphide
Risk-based inspection
carried out using simplified qualitative approach, which primarily uses influence factors for categorisa-
Risk matrix tion. This necessitates logical formulation of influence factors and their ranges with a suitable technical
Consequence analysis basis for acceptance from regulators. This paper describes application of risk based inspection for H2S
based Process Plant along with the approach devised for handling the influence factor related to the
quantity of H2S released.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction as a basis for prioritizing and managing the efforts in an inspection

program. Risk based inspection focuses the utilization of risk quan-
1.1. Background tification in formulating an In-Service Inspection (ISI) plan thereby
emphasizing the importance of surveillance and maintenance
Deterioration of structures, systems and components in Process activities on plant risk. RBI would be able to establish an effective
Plants has raised the concerns over their ability to with stand oper- structural integrity management programme, which reduces plant
ational, environmental and accidental conditions. To protect the down time, industry and regulatory burdens, and continue to
public, financial investment and environment from accidents, it maintain plant safety.
is essential to have an ageing management programme for achiev-
ing safety in operation. In formulating the life management pro- 1.2. Objective
gramme priorities have to be assigned based on operating
experience on ageing and premature failures. An effective life man- Existing in-service inspection programme of H2S based Process
agement programme depends on the ability to detect degradation Plant is based on ASME Section XI guidelines. H2S is highly toxic
and initiate mitigating measures for maintaining functional capa- and inflammable gas. Accidental release of even a small amount
bility without compromising safety margins as per design. Existing of H2S has enormous consequences on plant personnel and public
information on deterministic analysis/risk analysis and engineer- in the plant vicinity apart from production loss, repair and mainte-
ing judgment etc. can aid in prioritization process in life manage- nance costs. With this in view, extensive ISI is carried out on all
ment programme according to safety significance. toxic element-carrying components. This often results in carrying
Last decade saw a trend where life management programmes out excessive inspection or ineffective inspection being carried
are globally moving from prescriptive/time-based towards risk out on some components carrying toxic gas. Risk prioritization of
based decision making. Risk analysis finds use/application in deci- these components enables to determine the optimum level of
sion making, for operation, maintenance and regulatory activities. inspection or inspection effectiveness required to maintain the risk
This methodology has been applied in planning maintenance activ- at its present level as the component ages. Various standards have
ities such as testing time, repair time, inspection interval etc. When emerged for providing guidelines on applying risk based inspection
this is applied to inspection planning, it is termed as Risk based such as API 581 (1998), ASME (2003), and CWA (2008).
inspection. Risk Based Inspection (RBI) is a method for using risk Traditionally API 581 is considered as guideline for applying H2S
based Process Plants. API 581 describes three approaches for risk
⇑ Corresponding author. Tel.: +91 22 25593362; fax: +91 22 5505151. based inspection: (1) qualitative (for screening), (2) semi-
E-mail address: [email protected] (G. Vinod). qualitative and (3) quantitative analysis (detailed analysis). Since

0306-4549/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.anucene.2013.08.042
14 G. Vinod et al. / Annals of Nuclear Energy 66 (2014) 13–19

RBI looks into each and every component in a plant for categorisa- data. The general form of Baye’s theorem is written as (Siu and
tion, qualitative approach is an important step for screening out Kelly, 1998):
non-critical components. While applying risk based inspection
using qualitative approach of API 581 for H2S based Process Plant, LðE=ki Þf ðki Þ
certain limitations were observed in consequence estimation. This
f ðki j EÞ ¼ X ð1Þ
LðE=ki Þf ðki Þ
paper describes the experience on applying risk based inspection i
on H2S based Process Plant and approach adopted to circumvent
the shortcoming. where f(ki|E) is the probability of ki, given evidence E, (posterior dis-
tribution); f(ki) the probability of ki, prior to having evidence E,
2. Risk based inspection in H2S based Process Plant (prior distribution); L(E|ki) is the probability of the evidence E, given
ki, (likelihood function).
Risk is defined as ‘‘the likelihood of a specified undesirable Likelihood function denotes the plant operating experience and
event occurring within a specified period or in specified prior denotes the generic data base. Typically, Log normal distribu-
circumstances.’’ tion is considered for generic data and Poisson distribution is con-
sidered for likelihood function. Using these techniques, PoF of
X
Risk ¼ likelihood of undesirable event its consequence equipments were estimated.
accident scenario
(ii) Remaining life model for pipelines.

For applying the framework of risk based inspection, it is re- Corrosion being a predominant degradation mechanism in H2S
quired to estimate likelihood of failure of components in Process based Process Plant, ANSI/ASME B31G model (Caleyo et al., 2000;
Plant and their consequence, in terms of damage to the equipment Santosh et al., 2006), has been used to estimate the remaining
and impact of toxic release to public. strength of pipeline containing corrosion defects. All the failure
RBI uses the risk to plan, justify and aid in the assessments of pressure models are concerned with the estimation of remaining
results from inspection, testing and monitoring. RBI is the process strength but not the failure probability of pipelines containing cor-
of identifying and quantifying the consequences and the probabil- rosion defects. For this purpose, reliability analyses were required
ity of failures. The method applies both qualitative and quantita- to assess the remaining life of corroded pipelines with further cor-
tive approaches to prioritizing first analysis efforts and then rosion growth. First Order Reliability Method (FORM) has been
inspection activities. The primary difference between the qualita- used for reliability analysis of pipelines. The main assumption of
tive and quantitative approach is the level of resolution. The qual- the reliability analyses for the corroded pipelines is the random-
itative procedure requires less detailed information about the ness of the load and resistance parameters determining the limit
facility and as a result, its ability to discriminate is much more lim- state function (LSF). The LSF or performance function is defined
ited. The qualitative technique would normally be used to screen for this mode of failure as the difference between the pipeline fail-
components for detailed quantitative RBI studies. API 581 provides ure pressure Pfp and the pipeline operating pressure Pop, i.e.
influence factors in quantifying likelihood and consequence in qual-
itative approach. Typically, expert judgment is used in selecting LSFðPfp ; Pop Þ ¼ Pfp  Pop ð2Þ
the values of influence factors for various ranges.
In this study, semi-qualitative approach is followed, wherein In this study, the modified B31G model is employed to estimate
likelihood is ranked using the quantitative (assessed using quanti- the pipeline failure pressure.
fied from operating experience data and structural reliability tech-
!
niques) and consequence is ranked using the qualitative (influence ðd þR ðTT ÞÞ
2ðY s þ 68:95Þt 1  0 dt 0
factor approach). LSFðPfp ; Pop Þ ¼  Pop ð3Þ
D 1  ðd0 þRðtMÞ
d ðTT 0 ÞÞ

2.1. Estimation of likelihood of failure of components The folias factor, M is dependent on defect length, L, pipe diam-
eter, D and pipe wall thickness, t. Pop is the operating pressure. Rd is
In this study, estimation of likelihood of failure of components the radial corrosion rate T0 is the time of last inspection. Ts is the
was carried out using two approaches: pipeline elapsed time Ys is the yield strength of the pipe material.
Various parameters for failure pressure model were also fixed
(i) Statistical modelling for equipments such as towers, heat based on research studies and analysis. These factors were dis-
exchangers, etc. cussed in detail before finalisation. Details of this model can be
found from (Caleyo et al., 2000; Santosh et al., 2006).
Estimation of failure probability of equipment (PoF) is an Categorisation of failure probabilities with respect to level of
important step. Eventhough use of generic information is sug- severity is required for applying RBI. Typical categorisation framed
gested at the design stage, due to the absence of instances of failure for all type of components is shown in Table 1, based on consensus
in operating period, it is used for estimating failure probability of of expert from regulator, utility and analysts’ end.
piping equipment in operating plants. Data bases such as OREDA
(2002) are the result of various collaborative efforts taken towards
methodical collection of operating experience information, which
can be termed as generic data base. Service Data Analysis based Table 1
Probability of failure categories.
on operating experience is one of the popularly employed methods
used for this purpose. Probability of failure (PoF) value Category
When sufficient operating experience is available, these generic 1e4 to 1.0 5 Very high
estimates can be updated with the plant experience to arrive at 1e5 to 1e4 4 High
plant specific estimates for equipment failure probability. To up- 1e6 to 1e5 3 Medium
1e8 to 1e6 2 Low
date the generic information with plant experience, Baye’s theo-
<1e8 1 Very low
rem is used, which facilitates integrating different sources of
 G. Vinod et al. / Annals of Nuclear Energy 66 (2014) 13–19 15

2.2. Estimation of consequence of components After determining the PoF and CoF category, it needs to be ap-
plied to Risk matrix to establish the inspection category.
For estimation of consequences from a Process Plant, qualitative
approach based influence factors from API 581 serve as the most 2.3. Limitation in applying the current approach
suitable option. Influence factors considered account for the prop-
erties related to chemicals (toxicity, flammability, etc.), detection The categorisation for quantity factor specified in API 581 (for
systems installed in the plant site, isolating systems provided to both Damage Consequence Factor and Health Consequence Factor)
mitigate accident situation, damage potential and health effects. is a general one to suit the needs of all type of industries. Since the
In a H2S based Process Plant, due to the highly toxic and inflamma- total inventory of H2S handled in the Process Plant falls in the low-
ble nature of H2S, consequence in terms of damage and toxic needs est category as per API 581, the criterion was not suitable for
to be assessed. applying to the plant under consideration. Moreover, API 581 does
Table 2 and 3 present the various factors used except quantity not have any influence factor accounting for the height of release,
factor, the values chosen and the technical basis for the choice which plays a crucial role in dispersion. A need was felt to give a
for damage as well as toxic consequence, as suggested in (API different treatment to ground level and stack level releases. This
581, 1998). directs us to redefine the quantity factor to suit the scenario exist-
ing in the plant under consideration.
Damage Consequence Factor ¼ Chemical þ Quantity þ State
þ Escalation þ Credit 3. New approach for scaling the quantity factor

A great deal of research continues into developing methods for

Health Consequence Factor ¼ Toxic Quantity þ Dispersibility estimating or calculating the severity of the consequences of haz-
ardous incidents with increasing precision. In the process indus-
þ Credit þ Population tries, it is possible to perform calculations of the impact at a
Table 4 presents the categories for ranking of damage and distance, for any incident, since the consequences are governed
health consequence factors (API 581, 1998). These categories will to some extent by physical and chemical laws. This possibility is
be used in Consequence matrix, as shown in Fig. 1, for establishing harnessed to overcome the limitation in quantity factor and for
the combined consequence (CoF) category for all components. developing a new categorisation scheme.

3.1. System description

Table 2
Factors for damage consequence category.
The H2S based Process Plant has exchange towers, which handle
Sl. Factor Definition large quantities of H2S. In addition there are storage tanks, and pipe-
no: lines in ground level. Also, if there is any accident situation, H2S can
1 Chemical Measure of a chemical‘s inherent tendency to ignite be released through stack, which is 122 m in height. This makes it
2 Quantity Largest amount of material which could be released essential to consider the height of release for a realistic consequence
3 State Indication of the fluid’s tendency to vaporize and disperse
categorisation. There also exists a gradation in concentration of H2S
when released into the environment
4 Escalation Penalty applied to fluid that is processed at a temperature from 75 kg to 2 MT depending on point of release. Hence the cate-
above its autoignition temperature gorisation scheme should consider the quantity of release along
5 Credit Engineered systems in place which can reduce the with height of release. Keeping this in view, following steps are in-
damage from an event
volved in formulating the categories as per the new approach:

1. Postulate accidental release of H2S either from stack level or

Table 3 ground level.
Factors for health consequence category. 2. Use CFD heavy gas dispersion model to estimate concentration
Sl. Factor Definition in ppm of H2S, exposure time values at the required distance for
No: prevailing weather conditions.
1 Toxic Measure of both the quantity of the chemical and its 3. Find the probit numbers from the probit relationship.
quantity toxicity 4. Associate the probit numbers with quantity factors in API 581.
2 Dispersibility Measure of the ability of the material to disperse,
given typical process conditions
3 Credit Accounts for safety features that reduce the 3.2. Release scenarios of H2S
consequences of a toxic release by detection, isolation
and mitigation It is required to estimate critical quantities corresponding to
4 Population Measure of the potential number of people that can be
various severity levels for release of H2S from stack level and
affected by the toxic event.
ground level. Release quantity from 75 kg to 2 MT at a ground level
and from 0.12 MT to 20 MT at stack level release scenarios were
analysed. Release type can be of instantaneous or continuous. For
Table 4 this analysis, instantaneous type of release is considered for atmo-
Categories of damage and health consequence. spheric dispersion. The concentration of H2S at 1.6 km was kept as
Damage consequence Health consequence Category Definition the boundary condition for categorising the severity level of quan-
factor factor tity factor used in API 581.
0–19 <10 A Very low
20–34 10–19 B Low 3.3. CFD model for atmospheric dispersion
35–49 20–29 C Medium
50–79 30–39 D High
H2S being a heavy gas, Gaussian model (which is applicable for
>80 >40 E Very high
neutral gas dispersion) may not be applicable, and Heavy Gas
16 G. Vinod et al. / Annals of Nuclear Energy 66 (2014) 13–19

Damage A B C D E

Health

C E Very high

D High

D C Medium

B Low

E A Very low

Fig. 2. Schematic of source.

Fig. 1. Consequence matrix.

Dispersion models need to be used for analysis. Data generated

from well-recognized/validated computer software (i.e. Fire
Dynamics Simulator) is considered suitable for heavy gas disper-
sion. Fire Dynamics Simulator (FDS) is a field model based software
developed by NIST (USA) and has built-in dedicated computational
fluid dynamics (CFD) model to describe and analyse fire in multiple
compartment geometries. The code solves numerically the set of
governing equations of mass, momentum (Navier–Stoke) and en-
ergy conservation appropriate for low-speed, thermally-driven
flows of multispecies gas mixture to describe the smoke and heat
transport arising from fires. The details of formulation of the equa-
tions and the choice of numerical algorithm available are contained
in a companion document, called Fire Dynamics Simulator – User’s
Guide (Kevin et al., 2010). Fire Dynamics Simulator has two mod-
ules, the first, called simply FDS, is a Fortran 90 computer program Fig. 3. Dense gas mass fraction at start of release (xy plane).
that solves the governing equations described in elsewhere (Kevin
et al., 2010) and second, called SMOKEVIEW which is an OpenGL
graphics program that allows one to visualize the results. All of
the input parameters required by FDS to describe the particular
scenario of interest are inferred via one or two input text files cre-
ated by the user, in the format described in the input manual. A de-
tailed systematic validation and verification exercise for the FDS
hydrodynamic model has been carried out by (Kevin et al., 2007;
Sharma et al., 2004). The FDS predictions have demonstrated very
good agreement with the experimental data (Sharma et al., 2004).
The computational domain was 4400 m (length)  2000 m
(width)  480 m (height). The source is released from stack and
ground depending on the assumed accident scenario with a range
of quantities of heavy gas release. The left side boundary was mod-
elled as inlet where the available wind velocity (constant wind)
was applied. The scalar was released from the stack at a height
of 122 m. H2S release was considered for 30 min duration at a con-
stant rate. Constant Velocity (3 m/s) normal to inlet plane entering Fig. 4. Dense gas mass fraction during release (xy plane).
into the domain was applied at inlet boundary (refer Fig. 2). Other
three sides are modelled as open to atmosphere. A large set of sim-
ulation have been performed assuming a perfect flat terrain with 3.4. Find the probit numbers from the probit relationship
no obstructive components and structures and by varying the
source strength (see Figs. 3–5). If a population is exposed to a hazardous event, such as the de-
The Fig. 6 depicts the mass fraction just above the ground level fined level of heat radiation for a particular duration, or a concen-
in xy plane at various phases of dense gas release for a particular tration of toxic gas for a particular duration, the effects on all the
source strength. people would not be identical. Some would be more susceptible,
CFD analysis estimates the concentration and the exposure time and some much less. This is accounted in probit mathematics.
at the boundary for various releases scenarios postulated. A typical The effects of a toxic gas depend on both the concentration and
instantaneous release (at ground/stack level) attains a peak con- the duration of exposure. Probit mathematics can be used to pre-
centration in a finite time and the concentration decreases to insig- dict the probability of fatality for a defined exposure (concentra-
nificant levels, as shown in Figs. 7a and 7b as time passes. tion, duration). The general form of probit relation is
 G. Vinod et al. / Annals of Nuclear Energy 66 (2014) 13–19 17

Fig. 5. Dense gas mass fraction after release (xy plane).

Fig. 7b. Concentration profile at 1.6 KM at Ground level receptors for stack level
release.

Fig. 6. Dense gas mass fraction clearing phase (xy plane).

Fig. 8. Typical probit curve.

3.5. Associate the probit numbers with quantity factors in API 581

From CFD analysis, it was found that concentration at a distance

is controlled by the point of release. Different cases from stack and
ground release were able to capture the probit relation graph.
Some of the cases performed and the subsequent probit relation
is shown in Table 5.
This exercise has helped to design quantity factor as per the
inventory handled in the plant. Factors are retained as per the
API 581 guidelines, but the ranges are modified according to the
inventory handled in the Process Plant. Constant velocity of 3 m/
s is considered during the entire transient period. The quantity fac-
tors are compared with 10 point probit values (10%, 20%, 30%, 40%,
50%, 60%, 70%, 80%, 90% and 100%). Relation between quantity re-
leased and % fatality was derived based on the results presented in
Table 5. Typical relation for ground level release is given in Eq. (2).
Fig. 7a. Concentration profile at 1.6 KM at Ground level receptors with constant
velocity and different source strength of GL75.
Quantity released ¼ 605 þ 22:78ð%fatalityÞ  0:4086
2 3
 ð%fatalityÞ þ 0:00331ð%fatalityÞ ð5Þ
Y ¼ k1 þ k2 log eðC n tÞ ð4Þ
Using this relation appropriate quantity for 10 point probit val-
where k1, k2, and n are constants depending on the gas, C is the con- ues is deduced. Similar exercises have been carried out for estimat-
centration in parts per million, and t is the exposure time in ing the ranges for quantity in stack level release scenario. Modified
minutes. quantity factor for damage consequence category for ground level
The typical probit graph is given in Fig. 8. and stack level release are shown in Table 6.
Probit number is used to represent the severity of conse- Similar exercise is carried for modifying the quantities in toxic
quences, which gives the percentage of fatalities for a particular consequence categories. The quantity factors so obtained are used
probit value. For H2S, k1, k2, and n are 36.2, 2.366 and 2.5 respec- in place for the quantity factors given in Table 2 and Table 3 for
tively (Tweeddale, 2003). damage and toxic consequence category.
18 G. Vinod et al. / Annals of Nuclear Energy 66 (2014) 13–19

Table 5
Evaluation of probit parameters from results obtained from CFD calculations.

Case Conc in ppm (C) Time in min (T) CnT Probit value Fatality (%)
Ground release
75 kg 2.40E+01 1.47E+00 4.13E+03 1.65E+01 9.54E60
150 kg 1.86E+02 8.67E01 4.09E+05 5.63E+00 1.46E14
250 kg 2.98E+02 9.00E01 1.38E+06 2.75E+00 1.24E07
750 kg 7.76E+02 9.67E01 1.62E+07 3.08E+00 6.98E+00
1000 kg 9.46E+02 1.02E+00 2.80E+07 4.37E+00 3.14E+01
1500 kg 1.24E+03 1.08E+00 5.84E+07 6.11E+00 8.04E+01
1700 kg 1.32E+03 1.12E+00 7.13E+07 6.58E+00 8.88E+01
Stack release
0.12 MT 6.60E02 4.50E01 5.03E04 5.42E+01 0.00E+00
5 MT 4.38E+00 1.82E+01 7.29E+02 2.06E+01 1.17E84
10 MT 1.69E+02 2.83E+01 1.05E+07 2.05E+00 1.16E+00
15 MT 2.59E+02 3.33E+01 3.60E+07 4.97E+00 4.90E+01
20 MT 4.51E+02 2.83E+01 1.22E+08 7.86E+00 9.86E+01

Table 6
Modified the quantity factor of damage consequence category.

Material released (ground level) Material released (stack level) Quantity factor Probit (10 point scale)
<795 kg <5 MT 15 10
796–924 kg 5–7 MT 20 20
925–1010 kg 7–9 MT 25 30
1011–1070 kg 9–11 MT 28 40
1071–1140 kg 11–13 MT 31 50
1141–1220 kg 13–15 MT 34 60
1221–1330 kg 15–17 MT 39 70
1331–1510 kg 17–18.5 MT 41 80
1511–1760 kg 18.5–20 MT 45 90
>1760 kg >20 MT 50 100

Risk based inspection (RBI) in Process Plant

Probability of failure (PoF) Consequence of failure (Failure)

1. Service data analysis 1. Damage
Quantity factor
2. Remaining life models 2. Toxic from new
approach

Integrate the effect of consequences

(Use Consequence matrix)

Risk Matrix

Fig. 9. Overall approach for RBI in process plant.

4. Risk categorisation 2 and 1 respectively. Depending on the category the component

falls into, the frequency, scope and method of inspection adopted
After determining the PoF and CoF category, it needs to be ap- on that component is determined. The overall approach adopted
plied to Risk matrix to establish the inspection category. Risk Ma- for risk based inspection in a Process Plant is summarized, along
trix can be defined as the decision matrix, which has two elements: with risk matrix defined is shown in Fig. 9.
likelihood/probability and consequence of failure. Likelihood and
consequence of failure have been categorised individually to 5 cat-
egories, which makes Risk matrix dimension as 5  5. In the Risk 5. Experience on risk categorisation in h2s based Process Plants
matrix, 4 risk categories are defined, which has its own inspection
type, scope and frequency of inspection. Four categories are High, This study covers approximately 245 m gas lines of large diam-
Medium, Low and Very low, which are denoted by numbers 4, 3, eter (750–900) & 465 m H2S saturated liquid lines, two Exchange
 G. Vinod et al. / Annals of Nuclear Energy 66 (2014) 13–19 19

Table 7
Results of the pilot study on RBI of Process Plant.

Items Number of items Consequence categorya Failure probability category Risk Current ISI
15 yr 25 yr 15 yr 25 yr
Process HX 4 Nos M VH VH M M H
Coolers, chillers steam heater 5 Nos M VH VH M M H
Gas lines 6 Nos H M H M H 4H 2M
Liquid lines 30 Nos 2H 4H 27 VH 28M 1H 10H
28M 24M 1H 2L 29M 20M
2L 2M
a
VH – Very High; H – High; M – Medium; L – Low.

Towers, nine Heat Exchangers, 1 Booster (Compressor) apart from method of inspection. Risk based inspection has provided a quan-
many other components such as valves, flange and instruments. titative basis for framing the inspection programme for compo-
Pipe Line inventory in the system is about 1100 kg gas and nents as well as plants. The approach proposed in the paper
55,000 kg liquid. RBI was applied based on the approach described tailors the qualitative approach prescribed in API 581 to be used
in Fig. 9. in any type of Process Plant. The paper introduces a basis for mod-
Each component is assigned an inspection category, which els used in simplified approaches, which is required to improve the
specifies the severity of inspection to be employed in it. Risk cate- acceptability of risk based inspection, by plant personnel as well as
gory has been estimated for 15 and 25 years and the results of this regulators.
analysis are presented in Table 7. In Table 7, the term (for example)
4H indicates that four components are falling in the High category. References
Risk impact analysis has been brought out by comparing with the
existing inspection plans (Kevin et al., 2010). API 581, 1998. API Base Resource Documentation—Risk-Based Inspection, American
Petroleum Institute.
Current inspection plan puts almost 50% of its equipments con- ASME CRTD – Vol. 41, 2003. Risk-based Methods for Equipment Life Management:
sidered in pilot studies in high category. But RBI puts on 4% of An Application Handbook, ISBN 0791835073, ASME International, New York.
equipments in high category at the end of 15 years and 20% of CWA 15740, 2008. Risk Based Inspection and Maintenance Procedures for European
Industry, CEN Workshop Agreement, April 2008.
equipments in high category at the end of 25 years. It has been OREDA, 2002. Offshore Reliability Data base.
found that large quantum of 48 inspection is reduced by placing Siu, Nathan O., Kelly, Dana L., 1998. Bayesian parameter estimation in probabilistic
the equipments in inspection category based on RBI methodology. risk assessment. Reliability Engineering and System Safety 62, 89–116.
Caleyo, F., Gonzalez, J.L., Hallen, J.M., 2000. A study on the reliability assessment
methodology for pipelines with active corrosion defects. International Journal
6. Conclusion of Pressure Vessels and Piping 79, 77–86.
Santosh, Vinod, Gopika, Shrivastava, O.P., Saraf, R.K., Ghosh, A.K., Kushwaha, H.S.,
2006. Reliability analysis of pipelines carrying H2S for risk based inspection of
The objective of this study is to optimise the ISI by classifying heavy water plants. Reliability Engineering and System Safety 91, 163–170.
various components in different inspection categories based on Kevin, McGrattan, Randall, McDermott, Simo, Hostikka, Jason, Floyd, 2010. Fire
consequence and failure probability. There are lot of uncertainties dynamics simulator (Version 5), user’s guide NIST special publication 1019-5.
In: Cooperation with: VTT Technical Research Centre of Finland National
in estimation of piping failure frequency as well as in consequence Institute of Standards and Technology, Gaithersburg, Maryland.
estimation. Since risk based inspection mainly looks into the rank- Kevin, McGrattan, Simo, Hostikka, Jason, Floyd, Howard, Baum, Ronald, Rehm, 2007.
ing for inspection planning, certain amount of inaccuracies in these Fire dynamics simulator (Version 5) volume-3, verification, technical reference
guide NIST special publication 1018-5. In: cooperation with: VTT Technical
estimates can be tolerated. Research Centre of Finland National Institute of Standards and Technology,
Risk based inspection methodology has demonstrated substan- Gaithersburg, Maryland.
tial reduction in inspection. It should be noted that risk to the plant Sharma, Pavan K., Markandeya, S.G., Ghosh, A.K., Kushwaha, H.S., 2004. Approaches
for Modelling of Dispersion of Pollutants in Atmosphere with Emphasis on
has not increased by reducing these inspection, since the high Computational Fluid Dynamics. NEHU, Shillong.
safety significant components are still following the higher fre- Tweeddale, Mark, 2003. Managing Risk and Reliability of Process Plants. Gulf
quency of inspection. Risk based inspection has organized the cat- Professional Publishing, ISBN:0-7506-7734-1.
egory of inspection for each component thereby changing its