Risk-Based Inspection Pressure Relief Devices

Moiss Sebastio Jones

ARTICLE INFO

ABSTRACT


Keywords:
Pressure Relief Devices
RBI API
Weibull parameters
Probability of failure
RBI software

Risk-based inspection (RBI) methodology constitutes a fundamental point in the
inspection planning in order to control and mitigate risks that involves process
industry. Risk is used as priority criterion in the forecast of optimal inspections
intervals.
Several methods of RBI have been developed to use appropriate methodologies
for applying to various industry components. American Petroleum Institute (API),
as an institution has developed a standard document, as a guide to RBI,
applicable to a wide range of equipments inside processes industry. This
document, however, does not present a systematic understanding or simple and
immediate application, that the reason why it is always welcome to industry, any
initiative indicating simplified risk assessment methods for give to operators
some tools, to better quantify, assess state of components and help them in the
vast resources of inspection available, for reducing risk.
In this work a survey of current state of art of RBI methodology applied to
Pressure Relief Devices (PRDs) in processes industry was made. It was also
developed the identification of critical parameters of analysis as well as its
determination.
It was also made a synthesis of API RP 581, API RP 580 and an approach on
the Weibull analysis and the determination of probability of failure by this
method.
Finally, the methodology was applied in practical cases, through application of
software, which was developed based entirely on API RP 581 standard, very
easy to use, quantify and determine probability of failure, and parameters that
enabled determination of such probability.

1. Introduction
Since we were born all of us make risk decisions. Risk is associated to any adverse event
(consequences) what can happen to everyone, so as its possibilities of occurrence (probability of
failure) can be great or just little. PRD, Pressure Relief Devices, are the equipment in high-risk
industrial platforms, so, optimal inspection interval is the key to minimize the risk. A calculation of risk
with the RBI (risk-based inspection) API (American Petroleum Institute) involves determining the
probability of failure combined with its consequence.
Failure of a PRD is, not fulfilling the function what was supposed to do, which must occur when the
equipment does not perform in any way its function, or performs it badly, not comply with the
objectives for which it was designed.
Defining risk as the combination of the probability of a failure during a time interval with the
consequences, as follow:
= (1.1)
The probability is known as dimensionless. The consequence of failure, however, is measured in
monetary values per unit of time (annual monetary losses). So the risk is almost always measured in
Moiss Jones / Risk-Based Inspection Pressure Relief Devices

2

/year, can also be measured in area/year, the area affected by the damage for one year. But in this
work the unit of measurement used is /year.
1.1 Project of the PRDS
PRDS is the devices designed to help external pressure or vacuum in a container does not exceed a
predetermined value, doing that by transfer of fluid to a closed system or atmosphere.
There are main types of PRDS which are divided into three groups, such as a) Reclosing type b)
Vacuum type and c) Non-reclosing type.
1.1.1 Inspection programs and risk management for the PRDS
Inspection for PRDS must be done to provide protection in case of emergencies. This inspection
should include necessary factors which may affect performance of valves.
They are considered following factors:
Temperature variation, environmental and system factors
Vibration
Residue in the inner parts of valve
Elements of valve subjected to tension
Turbulence in fluid
Sizing and configuration of discharge pipe
Sizing and configuration of intake pipe
Design
2. Analysis of Weibull model
Weibull model is an empirical expression developed by Ernest Hjalmar Wallodi Weibull, Swedish
physicist, who in 1939 presented the model in planning of statistical fatigue of materials. Its
usefulness, to allow:
Represent typical initial failures (infant mortality), random failures and failures due to wear.
Get important parameters for configuration of failures.
The probability density function of Weibull distribution, known as the Weibull function with three
parameters, has the following form:

() =

(2.1)
Where (e =2.7183), , and t
0
are parameters with following meanings:
t
0
- Parameter of location, corresponds to the lowest value given by t (for example, in the case of
failure modes that had a cause a wear or fatigue, failure may occur only after some time of operation).
Shape parameter, it reflects a mechanism of degradation, indicates the shape of the curve of
probability and the characteristic of failure.
Moiss Jones / Risk-Based Inspection Pressure Relief Devices

3

- Parameter of scale (or characteristic life), corresponds to the life characteristic value, the time
interval between t
0
and t "which occur 63.2% of failures, leaving therefore 36.8% of items without
failure.
When we consider data modes failure of a system, we often consider that the lower limit of life t
0
=0,
and it is fairly acceptable. Accordingly, the above expression simplifies, moving to contain only two
parameters, it is considered that the equipment has no operating time, back to the pristine
state (t
0
=0) for the inspection in question.

() = 1

(2.2)

Fig. 1: Cumulative probability function of Weibull with two parameters
This function is called the cumulative probability function of failure, sometimes referred as unreliability.
The assumption used to determine the parameters of Weibull pattern is that the PRDS on similar
services have the same probability of failure on requesting, P
fod
, and the probability of leakage, P
l,
is
also similar. Therefore, rate of failure data of industry may be used as a basis to establish initial value
(or default) probability of failure for a specific device.
2.1 Bayesian Updating
Update for the values of Weibull parameter, , is made statistically based on the Bayes theorem, by
calculating Bayesian updates. The Bayes theorem is a corollary of total probability theorem which
allows us to calculate the following probability:
(|) =
(|) ()
()
(2.3)
Where:
Pr(A|B) are the priori probabilities of A conditioned to B.
The Bayes rule shows how to change the priori probabilities based on new evidence in order to obtain
a posteriori probability.
Moiss Jones / Risk-Based Inspection Pressure Relief Devices

4

3. Software for applying to RBI for PRDs
In accordance with regulatory requirements for testing and inspections this software aims to provide a
basis for calculation of specific Weibull parameters for PRDS, it is not, however, a software for a
complete API RBI evaluation covering all the API RP 581 standard for this equipment.
The algorithm of software, is represented in the diagram of Fig. 2

Fig. 2: Di agram of calcul ati on i n the software
The software has a main menu (Figure 3) where user makes a choice to what failure mode analysis
will proceed, whether it is failed to open (FAIL) analysis or Leak (LEAKAGE) analysis.

Fig. 3: software main menu
This menu gives the user the possibility to choose which of the failure modes will make study, it can
select between LEAKAGE or FAIL mode.
To analyze the parameters and probability of failure for the failure mode chosen, it will appear one of
the following window:
Moiss Jones / Risk-Based Inspection Pressure Relief Devices

5


a)

b)
Fig. 4: a) Wi ndow anal ysis Fails to open, b) Window anal ysis Leakage
In the first panel of these two windows recognizes of entity to which this application was developed is
made, and describes its usefulness and purpose. In the vertical left, user enters values previously
established by simple analysis of tables and conditions what can be seen in the API 581, and values
arising from the experience of inspector. These values are, therefore, the inspection interval, the
parameters of Weibull default (or initial, can be found in the API 581), adjustment factor, maximum
allowed working pressure (MAWP), excess pressure, adjustment factor environmental factors
confidence factor (CF), test duration or time of operation of device without inspection and the number
of inspections or tests. In this panel we can see a pop-up menu that allows choosing which standard
of review that device has experienced, this is done using the table of effective inspection, table 7.7,
API 581. Finally, this panel there is a value that is given through internal calculations of Life program
and the last feature adjustment factor of pressure.
Then we see the right three Pusshbuttom, estimate, close and save the results as. In the chart area,
gives the probability of failure analysis, and device in question, the probability of failure that will be
used to determine the risk and so in order to make the analysis of RBI, obeying standards API RP
580 and API RP 581.
To analyze the parameters and probability of failure for the mode LEAKAGE you'll see a window with
the configuration window shown in Figure 3a.
4. Application to cases of safety valves studied
It is necessary to ratify and consolidate the study of risk-based inspection of PRDS made so far by an
application in practical cases, far below, will be made the determination of Weibull parameters for
evaluating the probability of failure caused by any failure to open or how or by leakage.
4.1 Data from a processing unit in a refinery
Initially a survey was done only with data from the TR 69 valves (49 and safety relief and relief of 20)
was examined in stopping a refining unit in 1998 after a successful operation during 3 years. These
data are presented in Table 1.
Moiss Jones / Risk-Based Inspection Pressure Relief Devices

6

Analysis of these data shows what can be summarized in the following table:
Situation of PRDs
Safety and relief
valve
Relief
valve
Total
Qty. % Qty. % Qty. %
Opened in the tolerance range and kept the
tightness
19 39 4 20 23 33
Opened in the tolerance range and didnt kept
the tightness
15 31 3 15 18 26
Opened 20% above set pressure 4 8 2 10 6 9
Opened 20% to 40% above set pressure 0 0 8 40 8 12
Opened 40% above set pressure 3 6 2 10 5 7
Opened 95% down set pressure 2 4 1 5 3 4
Had leakage before open 6 12 0 0 6 9
Table 1: Results of i nspecti on and TR after stoppi ng at a refi nery in 1998
At the time were not identified because it is not considered, the causes that led to poor results in the
TR of these valves.
Then, in 2000 and 2004, were made during and after they stop working for a period of 2 and 4 years
respectively, a more detailed survey of the results of the TR safety and relief valves, in order to
determine the availability valves and the optimal inspection intervals, as had the aim of adopting a
range of 5 years for such unit.
After this analysis it was decided, as a general rule, establish inspection intervals for 2 years for the
valves that failed. For the other valves was recommended to monitor the period of inspection of the
equipment protected.
4.3 Documentation required at prior
Data collection is essential at prior to proceeding API RBI analysis, as well as standard provides the
inspector must be able to make the compilation of this information means that through the API RP
580 or API RP 581 can meet all the conditions that facilitate the fast and efficient access to them and
make the evaluation process more feasible.
4.4 Results applying software
Calculations in that software primarily, aimed determining the parameters of Weibull and ultimately
the failure probability or FAIL LEAKAGE, are based on the procedure for calculating of API RP 581
and give us the probability of failure for a range specified inspection.
Consider one of the valves in the refinery case cited above. As we exposed the same PRDS
underwent three TRs at different times, the first in 1998, then, in 2000 and 2004, it is clear that many
valves were replaced and some added ways to improve safety, however, at least most of them still
continued their normal operation.
We will identify these valves by the letters X and Y, in fact, the valves are considered, a relief valve
and other safety and relief valve, all of them conventional.
4.4.1 Evaluation for X valve (safety and relief)
Moiss Jones / Risk-Based Inspection Pressure Relief Devices

7

The following input data were the result of field research by a surveyor by the codes tabulated in API
RP 581.
Confidence factor
(CF
i
)
Duration
(Years)
CF
leak
CF
noleak
T
dur,i
0,50 2
0,50 4
0,70 5
Table 2: History val ve study by consulti ng the API RP 581
So here we are in condition to make the calculations using the software for this valve just seeing API
RP 581.
Results:

[Years]

[Years]

[Years]
5 1,80 23,90 1,25 0,50 14,94
Table 3: Val ues i ntroduced to the program and changed life cal cul ated by him for LEAKAGE

[Years]
12,03 11,02 5,45
[Years] 2 6 11

0,04 0,28 0,97

Table 4: Cal culated by program to determi ne the probabili ty of LEAKAGE
These results only confirm advantages of this methodology for analysis of risk-based inspection
through probability density function of Weibull. It was expected that the characteristic life to degrade
as the years pass and the valve was failing, in the case, leaks. Note that for the first two inspections
there was no leakage, which justifies the fact that the life of the component does not decrease
significantly. However in the third inspection, as shown in Table 2, there have been an increase in
operating time without inspection, which justifies the fact that the life of the PRD significantly decrease
from 11 to 5 years. See the figure below the trend curve of probability of leakage.

Fig. 5: Evol uti on of probabil ity of l eakage
The graph in Figure 5, although few data, gives us a sense of how would be the evolution of
probability of failure, for this valve as time goes on, making it necessary shorten on the inspection
intervals.
0
0,5
1
1,5
0 5 10 15
P
r
o
b
a
b
i
l
i
t
y
Time [Years]
Moiss Jones / Risk-Based Inspection Pressure Relief Devices

8

4.4.2 Evaluation for the Y valve (relief)
For this case it has more data, it will improve perceptions of evolution for probability of failure over
time and will see how it is easier to interpret data and analyze them if they are available in detail. The
following input data were the result of field research by an inspector through codes tabulated in API
RP 581.
Confidence factor
(CF
i
)
CF
pass
0,50 0,50 0,50 0,50 0,50 0,50 0,50 0,50
CF
fail
0,70 0,70
Duration (Years) T
dur,i
3 3 3 3 3 2 2 2 2
Table 5: Historical data of studi ed through API RBI 581
Results:

[Years]

[Years]

[kPa]

[kPa]

[Years]
3 1,80 23,90 1,25 1500 6000 0,80 0,50 7,17
Table 6: Val ues i ntroduced to the program and changed life cal cul ated by him for FAIL

[Years]
6,80 6,40 6,20 5,90 5,80 2,30 4,40 4,20 4,00 2,10
[Years] 3 6 9 12 15 17 19 21 23 25

0,58 0,85 0,97 0,99 1 1 1 1 1 1

Table 7: Val ues calcul ated by the program to determine the probabi lit y of FAIL

Fig. 6: Evol uti on of probabil ity of fail ure to open
This latter case is easier for understand because there is more data to study than the first case. Note,
in this case, the probability that the equipment would fail grow fastest than the first case, this can be
explained by the fact that there are many factors that contribute PRD does not open when prompted,
and that this failure mode includes other failures to open as late opening of the device, valve stuck
open or partially are failures that occur frequently.
5. Conclusions
In a RBI API assessment, it is recommended to review the attached tables, API RP 581. They are
simple perception as to the quantitative evaluation of the software presented in this paper is
0
0,5
1
1,5
0 10 20 30
P
r
o
b
a
b
i
l
i
t
y
Time [Years]
Moiss Jones / Risk-Based Inspection Pressure Relief Devices

9

concerned to solve it in order to facilitate the process of calculation and which conducts the
evaluation. We considered details of the safety valves and relief systems pressures, made the
approach practical recommendations and analysis, introduction to risk-based inspection. The software
is simple to apply, once collected the information required by him, can be used on any operating
system, as it was made an executable to run it on any computer. However it should not be construed
as a software evaluation RBI API, but as a useful aid.
Studied cases were, here generalize the application of software and leads to the perception of the
beginning of an RBI API evaluation. More cases could be studied, but the lack of data and its dificult
to obtain it conditioned the consolidation of this work. But it should be noted that here was satisfied
the requirements for the originally intended, namely the determination of Weibull parameters for later
analysis of API RBI.
The main difficulty in this study was to evaluate available data, which was insufficient and sometimes
inadequate for determining the probability of failure and to proceed RBI API evaluation. It is hoped
that with this work a collection of appropriate information in accordance with API RP 580 and API RP
581 can be make to proceed a RBI API.
Evaluation of API RBI becomes complete by determining the risk that each component where PRD
are subjected. That is why in future we can continue quantitative determination of consequences of
failure and the risk associated with these consequences. Can be developed further, a software more
complete analysis of the API RBI for PRDS.
In future it can be make too an adequacy, simplification and determination, with the help of some
software, the consequences of failure of PRDS for the two failure modes considered.
References
damo Santana, Bayesian Inference, 2006.
American Petroleum Institute, API Recommended Practice 580 - Risk-Based Inspection, First edition
May 2002, Revision 01 2007.
American Petroleum Institute, API Recommended Practice 581 - Risk-Based Inspection Technology,
SECOND EDITION, SEPTEMBER 2008.
American Petroleum Institute, API RP 520 PT II - Sizing, Selection, and Installation of Pressure-
Relieving Devices in Refineries Part II - Installation, USA, 1 AUGUST 2003.
ASME PTC 25-2008 (Revision of ASME PTC 25-2001) Pressure Relief Devices Performance Test
Codes, J anuary 13, 2009.
Chi-Hui Chien, Chun-HungChen, Yuh J . Chao, Reliability Engineering and System Safety - The
strategy for the risk-based inspection of pressure safety valves, September 2008.
Lynne Kaley, RBI: Risk-Based Inspection Reassessment, 2009.
Mohammad A. Malek, Ph.D., Pressure Relief Devices ASME and API Code Simplified - PE 2006.
Panos Topal, F. Ghassab Alajmo, Yong Seng Teo, Ijaz Ul Karim Rao, IMPLEMENTATION OF AN
INTEGRATED RISK BASED INSPECTION (RBI) SYSTEM INSTALLATION IN AN ONSHORE, 2006
Moiss Jones / Risk-Based Inspection Pressure Relief Devices

10

PC Gregory, Bayesian Logical Data Analysis for the Physical Sciences A Comparative Approach
with Mathematica
TM
Support, Department of Physics and Astronomy, University of British Columbia,
Cambridge University Press 2005.
ReliaSoft Corporation, http://www.weibull.com, last access: 02/09/2009.
Rui Assis, Decision Support in Maintenance Management, Reliability and Maintainability, Lidel 2004.