You are on page 1of 11

1CHSME SYMPOSIUM SERIES NO.

144

THE USE OF RISK-BASED ASSESSMENT TECHNIQUES TO OPTIMISE


INSPECTION REGIMES
G R Bennett, M L Middleton, P Topalis
Det Norske Veritas. Stockport Technical Consultancy, Highbank House. Exchange Street,
Stockport, SK3 OET.

Regulators and insurance companies now recognise the acceptability


of a risk based approach to the optimisation of inspection and
maintenance intervals.
Qualitative approaches to Risk Based
Inspection (RBI) have been developed for general use and more
detailed quantitative methods exist for activities with major loss
potential or large preventive expenditure requirements.
DNV
pioneered the use of RBI in the chemical process industry in 1992 and
has produced a resource document for the API (API 581). DNV
subsequently further developed Quantified RBI software. A number
of case studies and client projects have been conducted and the RBI
techniques and software are becoming valuable tools for the oil and
chemical industry.
Keywords: Risk Based
Inspection, Inspection
Consequence, Likelihood, Corrosion

Planning,

WHY CONSIDER RISK IN INSPECTION PLANNING ?


"The first duty of business is to survive and the guiding principle of business economics is
not the maximisation of profit, it is the avoidance of loss"
(Peter Drucker)
All industrial organisations use people, equipment and property to add value to a
commodity and thus generate a return on their investments. If the returns are in excess of
the costs, the company is in profit, if they are less, a loss occurs and if the losses continue
to be sustained, the organisation will ultimately fail.
People and equipment are not infallible. People do make mistakes and equipment does
fail. All mistakes and failures have consequences, some small and others large. The
magnitude of those consequences influences the ability of the organisation to achieve a
sustained profit and thus survive. It is therefore good business practice to consider the
consequences of failures and how likely they are to occur, i.e. to systematically consider
the risks arising from failures.
All forward thinking organisations practice some form of loss avoidance policy.
Equipment maintenance, inspection and testing, and the training of personnel are all
undertaken to avoid failures and reduce or mitigate their associated losses

CKEMO S' MFOSrt'M SERIES N'

How many organisations however, know in detail how effective their investment in
loss prevention activity is ? Does the extent and frequency of inspection reflect the
magnitude of the consequences of an undesired failure ? Is the inspection activity likely to
identify the degradation mechanisms that exist ? Do the planned maintenance routines
actually influence the chance of the equipment breaking down ? Is money being spent on
inspecting and maintaining equipment which, if it fails, has very little effect on the
organisation ?
An organisation's ability to consider these questions has traditionally been driven by
legislative rather than business need. For example, inspection of pressure retaining
equipment or structures has been primarily calendar based, driven purely by legislative
requirements. This no longer needs to be the case. The new goal setting legislative
environment provides the opportunity to plan loss avoidance activities by linking any
increase in the level of activity, to the reduction in risk achieved by that increased level
of activity.
Both regulators and insurance companies now recognise the acceptability of a risk
based approach to the optimisation of maintenance.
RISK BASED INSPECTION METHODS
Risk Based Inspection (RBI) techniques have been developed along two complementary
routes. Qualitative approaches to RBI have been developed for general use and detailed
quantitative methods have been developed and are being refined for activities with major
loss potential or large preventive expenditure requirements. This paper will concentrate
on recent developments in the quantitative assessment approach.
QUALITATIVE RISK BASED INSPECTION
Qualitative RBI is based on answering a series of questions regarding likelihood of failure
and the consequences of failure and assigning notional levels ( High / Medium / Low ) to
the answers to place the item on a risk matrix, see Figure 1.
The closer an item is to the top right corner of the matrix, the more critical the item is,
and the greater the inspection activity warranted. It should be noted however, that changes
to inspection regimes can only effect the frequency of an event, and not its consequences.
Therefore if an item is in the high risk category primarily due to it's consequences, then no
amount of additional inspection will improve it, and design changes to the system may be
required instead. This statement is equally true for quantitative assessment techniques.
Software packages are now available which allow a qualitative estimate of inspection
frequency to be made. They vary in complexity, and in the degree of "engineering
judgement" to be applied, but they can be used as an effective screening tool in order to
determine which equipment should be subjected to a more detailed quantified assessment.
An example screen shot from one of the DNV software packages is shown in Figure 2.
The software is driven simply by the selection of various options from drop down
menus, which prompt the user to select the most appropriate category for equipment type.

504

SCHEME SYMPOSIUM SERIES NO. \44

location, fluid type and inventory, business interruption consequence, and specific details
relating to the material of construction of the equipment, and its susceptibility to various
failure mechanisms. Based upon these inputs, the software then uses a set of pre-defined
rule sets to calculate a risk ranking and a corresponding recommendation for the next
inspection interval.
THE QUANTIFIED RISK BASED INSPECTION METHOD
The Quantified Risk Based Inspection philosophy and method has been endorsed and is
being promoted by the American Petroleum Institute Committee on Refining Equipment.
The group is comprised of representatives from the following companies:
Amoco

Dow

Pennzoil

Aramco

DNO Heather

Petro Canada

Arco

DSM

Phillips

Ashland

Exxon

Shell

BP

Fina

Sun

Chevron

Koch

Texaco

Citgo

Marathon

Unocal

Conoco

Mobil

DNV pioneered the use of RBI in the chemical process industry in 1992. In 1993 the
API approached DNV with the request to jointly fund a larger development effort, aimed
at producing a resource document for how to establish risk based inspection in the
petroleum industry. DNV obliged, and in 1994 it produced the Base Resource Document
on Risk-Based Inspection. This document is now being promoted by API as a standard
referred to as API 581, it will be followed shortly by API 580 which will become an API
Recommended Practice.
In 1995, the API commissioned DNV to develop software for its members to allow
them to automate some of the processes required by API 581. This software based system
has now been used successfully on a number of studies conducted both by DNV, and the
sponsor group members. DNV decided to further develop the Quantified Risk Based
Inspection Software in 1996. The new software, called ORBIT was initially released in
March 1998 and is available to the industry as part of an integrated package of Risk Based
Inspection services. ORBIT provides a fast interpolation approach for consequence
analysis.
The quantified RBI approach provides a methodology for determining the optimum
combination of inspection methods and frequencies. Each available inspection method
can be analysed and its relative effectiveness in reducing failure frequency can be
estimated. Given this information and the cost of each procedure, an optimisation
program can be developed. The key to developing such a procedure is the ability to

- '

>SiUM SERIES NO. !44

quantify the risk associated with each item of equipment and then to determine the most
appropriate inspection techniques for that piece of equipment.
SYSTEMATIC APPROACH TO REDUCING RISKS
A fully integrated Risk Based Inspection system should contain the steps shown in Figure
3. The system includes inspection activities, inspection data collection and updating, and
continuous quality improvement of the system. Risk analysis is "state of knowledge"
specific and, since processes and systems are changing with time, any risk study can only
reflect the situation at the time the data were collected. Although any system, when first
established, may lack some needed data, the risk based inspection program can be
established based on the available information, using conservative assumptions for
unknowns. As knowledge is gained from inspection and testing programs and the
database improves, uncertainty in the analysis will be reduced. This results in reduced
uncertainty in the calculated risks.
The combination of elements required as inputs to a quantitative RBI analysis are
shown in Figure 4.
The two major elements of a quantitative RBI analysis, as with any risk based study are
an assessment of the probability (or likelihood) of an event occurring, and its
consequences should it occur.
LIKELIHOOD ANALYSIS
When considering the likelihood of a failure occurring, the RBI process utilises a series of
technical modules, in order to establish a damage rate for the equipment. The calculated
damage rate depends upon the item's material of construction, the process fluids it is
exposed to, its external environment and the process conditions (pressure, temperature
etc.). Details of the current technical modules are shown in Figure 5.
It is also necessary to consider the current inspection regime, and to identify when the
equipment was last inspected, how it was inspected, and what the results of those
inspections were.
For example, if an equipment is subject to damage by corrosion or erosion, and if no
inspections have been performed, then the likelihood of failure may be high. If however,
many inspections of sufficient quality have been performed ( and the equipment still meets
its design intent ) then the likelihood of failure will be quite low, even if there has been
significant corrosion, as the rate of corrosion will be well understood.
Inspection activity is not however guaranteed to provide precise details of actual
corrosion rates, each inspection has an error band associated with the particular technique.
These error bands can be established by trials and review of historical surveys. Inspection
activity does not change a corrosion rate, it reduces the error band and increases our
confidence that we know the actual corrosion rate.

506

ICHE.VIE SYMPOSIUM SERIES NO. 144

Statistical methods can be used to evaluate the likelihood that damage severe enough to
cause a failure could exist given the amount of appropriate inspection activity that has
been performed. As the damage rate is time based, future inspection techniques and
intervals can be planned based upon the amount of damage expected to be seen at some
point in the future. A proper balance must be established between advancing damage and
increased knowledge of the amount of damage, to ensure safe and economic operation.
An example of a screen shot of the equipment details from the equipment specification
module of the software is shown in Figure 6. A screen shot of the likelihood module is
shown in Figure 7.
CONSEQUENCE ANALYSIS
The consequence analysis conducted within the RBI software is based upon look-up tables
calculated using the DNV software package PHAST . Consequences are calculated in
terms of the area of equipment damage, and the areas within which personnel will be
adversely affected by flames, explosions, or the toxic effects of the product concerned.
Using the input data on process pressure and temperature, material properties, and
inventories, the system determines the release rate for a range of representative hole sizes,
and also determines the release type. After determining whether a release is continuous or
instantaneous (as in a vessel rupture), the software calculates the final phase in the
environment (liquid or gas) and then determines the toxic or flammable consequences. In
evaluating the consequences, the software also allows modeling of account mitigating
features such as isolation and shutdown systems.
An example of the Consequence data module can be seen in Figure 8.
RISK ASSESSMENT
Having evaluated both the likelihood of an event, and its consequences, the system then
combines this data to produce the overall risk evaluation for each piece of equipment.
This allows the assessment of the overall risk levels of the plant, and the identification of
where the high risk items on the plant are. so that inspection effort can be focused initially
on the high risk items. Various reports can be automatically generated to produce a wide
range of analyses. These reports include:

Action damage/mechanism summary reports.


Financial risks.
Inspection planning.
Risk ranking.

Graphs can also be produced for specific items of equipment showing the optimised
number of inspections (cost of risk per number of inspections versus years of inspection)
and the percentage of equipment versus the percentage of risk. An example plot is
illustrated in Figure 10. The curve demonstrates the general principle that 80-90% of the
risk is contributed by only 10-20% of a plants fixed equipment.

507

EGHEME SYMPC5TUM

3 O. 144

PLANT EXPERIENCE
It should be clear from the forgoing that regardless of whether a qualitative or quantitative
approach is followed, it cannot be implemented without the active involvement of
personnel familiar with the plant and its operation. Corrosion engineering experience is
required to determine the damage mechanisms possible. Inspection management personnel
are required to extract the knowledge of past inspection, and operations personnel are
required to assist with establishing the safety and production implications of failures. It
must be a team effort, in order to be effective.
CASE STUDIES
DNV has now completed a number of botii pilot and full scale studies, in order to validate
the data in the model, and to evaluate the usage of the technique. Results demonstrate that
the RBI techniques and software are becoming valuable tools for the oil and chemical
industry.
On one site, an RBI analysis of nearly 2,000 piping sections in an ethylene plant
showed that less than 10% fell into the high risk category. Failure of those items
constituted a business interruption and asset damage risk of SI 1.5 million per year. The
application of improved inspection techniques reduced this risk to $4.1 million per year, a
saving of S7.4 million per year. There was of course a cost associated with this risk
reduction, and the improved inspection technology utlised was estimated to cost $250,000
per year. In this case the benefits clearly outweighed the costs. On the other hand, a
review of the bottom 10% of risk items showed that the application of the same inspection
techniques could still result in a risk reduction from $12,000 per year, to $4,300 per year a
saving of $7,700 per year. However the costs of the improved inspection of those items
would still cost $250,000, and this was clearly not cost effective.
On another site. 10 vessels were removed from the annual inspection plan, at an annual
cost saving of $25,000 per vessel, and some pipework materials were upgraded at a 20%
increased cost of materials, but avoiding a possible $3 million loss in business
interruption.
SUMMARY
The RBI methodology and its supporting computer program has already gone a long way
toward an integrated risk management program.
The technology is still being developed by DNV, and the author wishes to take this
opportunity to thank the RBI sponsor group for their advice and support during the
development phase. The RBI philosophy and database has drawn upon the experience of
both DNV and the project sponsors in order to evaluate failure mechanisms, corrosion
rates, consequences etc.
The author would also like to acknowledge the contribution of his colleagues Mark
Middleton (DNV Stockport), Panos Topalis (DNV Software Products), Angus Lyon
(DNV Aberdeen) and Gert Koppen (DNV Rotterdam) to the production of this paper.

508

LCHEME SYMPOSIUM SERIFS N(

Figure 1 Example Risk Matrix

; MecJii um@l '%m

Increasing
Likelihood 3

.;..i.--'.V.

ws&.w

Mediun

1
::- -. .V... ,

1 2

Increasing Consequence
Figure 2 Example of Qualitative RBI Software

R~~
z qipmenH den#co(Jon

"3
CiM<wtnc- - . -

........

w5

13

000 psg

MateoiofCoriWucteK' | C x t x m S : i

| l o w Susceptible
1 MoOmald(i SuKeptitw
[Low Susceptible

La
'y

^ O . No rspoctiort in USM 10 X r .

L*j

In

|No!Su;cepts

CaraauenCs t h e o r y f c " ? ; R f c M & f e l

(I),:

|KwS..>cep'fele

1 No! Susceptible

[]

.JJ

M d High Ritk

509

C'lnepectiorvt 2>e*taoaf

-Irw&oCtcn RocomrandabcfT:
& r < f a ^ Frequency- J 4 y e a r *

ICHEME SYMPOSIUM SERIES NO. 1^

Figure 3

Risk Based Inspection Program for In-Service Equipment


PLANT DATABASE

RISK BASED PRIORITISATION

INSPECTION PLANNING

'

^^r

INSPECTION RESULTS

FITNESS FOR SERVICE

-.'

a ai-. -.

INSPECTION UPDATING
!

.-..,'...

SYSTEM AUDIT

Figure 4

Overview of Risk Based Inspection

PROCESS SAFETY
MANAGEMENT
IMPROVEMENTS

--

. - . . * .

125.vsd

510

SCHEME SYMPOSIUM SCRIES NO. 144

Figure 5 RBI Technical Modules

Technical Modules
Set-up

Corrosion
Under
Insullalion

Thinning
Module

Stress
Corrosion
Cracking

High
Temperature
Hydrogen
Attack

Caustic

CUI

HCI
HT Sulfidation

Amine

HT H.S/H.

Carbonate

HjSO,

Sulphide

HF
Sour Water

HIC/SOHIC - H3S
HSC/HF

Amine

HIC/SOHIC - HF

Brittle
Fracture

Fatigue

No
Mechanisms
Apply

Brittle
Fracture

Fatigue

HT Oxidation

Figure 6 General Equipment Module


'; Fquipment Detail Data
Likelihood

general

Equipment DimensionsWall Thickness: [5

- Tempetatuie and PressuteOp tempetature:


J293
Op pressuie:

Tank Shape:

102000

Design temperatuie: |263


|263

|323

Design pressure:

|103000

Internal Diameter lg.0

ZJ

j - Date
;
] Current Service Starting Data
11987-01 04 1

100-1000 Rev 1

; Insulated fr-es)
;

|3.rj

Material ot
I Carbon Steel
Construction;
PJID-

Internal Height

Pa

T empetatuie and Pressure Check


-Additional Data

|Verti

Internal Length: ("

Max.

Miri.

|99000

Results

Consequence

I~ PWHT [Vast

[""

Exterior Coating fTest 17 Notmalized [Yes):

: Vessel Lining (Yes):

l~" Impact Test (Yes):

T j
F \

Help-1

511

Notes

1CHEME SYMPOSIUM S5H;E5 NO

Figure 7 Likelihood Analysis Module


"8 Equipment Detail D a t a
eneial

Likelihood

Consequence

-Damage-Types

- Failure Freauency-

Jhinnkig
Leak Frequency
(peiyeaf|
SCC

Small

4.00E-05

Medium

1.00E-Q4

Large

100E-05

Rupture

200E-O5

Uiiltfe Fracture
No Mechanisms App(y

3r
Cfosa

Help

Figure 8 Consequence Analysis Module


I tqurpment Detail Data
Senetal

Chemical Name:

Lfcelihood

Consequence]_

Results

|KER0SENE

Lookup Table Name: |Keiosene


Inibal Fluid State:

jLiquid

Vessel Type: [Pressurized Liquid

Chemical. Vessel Type and Initial Fluid State


Inventory
<"

'

Item Inventory: J6000

r - ,:, -

-.-, I

Inventory Group:

Method: | Lookup Melhod


Elevation:

h 000

I Dike Area:
Surface Type: [ 5 ^ ^

f"

Calculate
Consequence Calculation

EJ

Scutate

Areas button is only available loi selection immediately loUbwing calculation.

Qose

512

Help

Notes

ECHEME SYMPOSIUM SERIES NO. 14^

Figure 9 Results Module

! Equipment Detail Data

:
Consequence Results

Ukdhood ResuKs
Ukelhood
3.81E-03 Events/yeaf

Categoiy

Area Based Results


Categoiy ;

i Consequence

Facta

368.001 s q .
2241
j Equipment Damage:

1.13E*Q2

|sqm

FataUy:

3.68E-02

I sqm

ToacAfea:

O.O0E+OQ | sqm" j

Inspection Planning Calculations


Inspection Planning

2241

CJoae

Help

Notes

Figure 10 Risk Results

Percentage of Equipment vs. Percentage of Risk

1
Is,

100
o~
80
70

m
50
40
30
20
10
0
i j j

P e r c e n t a g e of E q u i p m e n t

513

You might also like