Application of the FMEA in insurance of high-risk industries: a case study of Iran's gas refineries Shamsi Ghasemi, Rahim Mahmoudvand & Kazem Yavari
Stochastic Environmental Research and Risk Assessment ISSN 1436-3240 Stoch Environ Res Risk Assess DOI 10.1007/s00477-015-1104-7
1 23
Your article is protected by copyright and all rights are held exclusively by SpringerVerlag Berlin Heidelberg. This e-offprint is for personal use only and shall not be selfarchived in electronic repositories. If you wish to self-archive your article, please use the accepted manuscript version for posting on your own website. You may further deposit the accepted manuscript version in any repository, provided it is only made publicly available 12 months after official publication or later and provided acknowledgement is given to the original source of publication and a link is inserted to the published article on Springer's website. The link must be accompanied by the following text: "The final publication is available at link.springer.com”.
1 23
Author's personal copy Stoch Environ Res Risk Assess DOI 10.1007/s00477-015-1104-7
ORIGINAL PAPER
Application of the FMEA in insurance of high-risk industries: a case study of Iran’s gas refineries Shamsi Ghasemi
1 •
Rahim Mahmoudvand2 • Kazem Yavari3
Springer-Verlag Berlin Heidelberg 2015
Abstract The purpose of this paper is to categorize and analyze various risk factors in Irans gas refineries for insurance purposes. Using the failure modes and effects analysis method as a subset of probability risk assessment technique and gas refineries data for the period March 2011 till March 2012, risk priorities numbers are calculated from the perspectives of both the insured party (gas industries) and the insurer (insurance companies). Our empirical results indicate that various property damage risk factors embodied in gas refineries including fire, explosion, error and omission, and machinery breakdown are insurable risks. Risks of pressurized vessels defects are in safe category and can be tolerated by the industry owner. The policy implication of this paper for Iranian policy makers in the energy sector is that, gas refineries are insurable in the market with reasonable risk premium. Insuring gas refineries will definitely reduce capital losses which can otherwise be enormous for the economy in general and for oil and gas industries in particular. Keywords
Risk Refinery Insurance
& Rahim Mahmoudvand
[email protected];
[email protected] 1
Centeral Insurance of IR Iran, Tehran, Iran
2
Department of Statistics, Bu-Ali Sina University, Hamedan, Iran
3
Faculty of Management and Economics, Tarbiat Modares University, Tehran, Iran
1 Introduction Energy is closely related to environmental risk. A rising fuel price in the 1970s had hurt consumers and caused disturbance to the natural environment (Youn et al. 2011). Gas is currently one of the major sources of worlds fuel consumption and its share in the world energy is expected to rise in future. According to the British Petroleum (BP) official statistics, more than 24 % of the world energy comes from gas (BP Statistical Review of World Energy 2012). Gas refining is naturally a risk-embodied business due to the existence of the risk factors such as fire, explosion, and toxic propagation. With the increase in natural gas production in recent years, some sources, including the US Environmental Protection Agency, have suggested that upstream methane emissions are increasing (Shahriar et al. 2014). In addition, high temperature and pressure increase hazards and uncertainties in this energy sector. Occurrence of these incidents in gas refineries will inevitably lead to massive losses in production, manpower, chemical materials, and equipments. Therefore, to maintain safety and health standards, identification and assessment of risk factors are very important in various development stages of gas industries. Watching for safety standards and protection of oil and gas industries are vital now and in future for Irans economy. Irans oil and gas refineries have had many losses and catastrophic accidents such as explosion in Isfahans gas refinery, oil and gas blowout from well No. 24 of NaftShahr, blowout and fire from well No. 104 of Maron, explosion of the boiler No. 13 of Oil Terminals Company (OTC), so far. Moreover, existence of 8 oil refineries and 13 gas refineries in Iran pose a significant danger to the communities in which they operate, and the above examples just show how much risks they pose. Historical data on the occurred losses in Irans oil and gas refineries clearly
123
Author's personal copy Stoch Environ Res Risk Assess
show that the insurer companies must apply more effective risk assessment methods in this area. Whether risks in gas refineries can be transferred by owners to insurance companies and insurance companies are willing to take such risks, is an important issue from perspectives of both the owners and insurance companies. The possibility of insuring gas refineries is also an important issue for energy policy makers. The main purpose of this study is to reveal this possibility from the insured and insurer perspectives. Since dealing with risks generally requires probability analysis, we have used a suitable probability risk assessment (PRA) technique for risk assessment of Irans gas refineries. This technique is the so-called failure modes and effects analysis (FMEA) method which has been recommended by the International Organization for Standardization (ISO) and the Occupational Health and Safety Assessment Systems (OHSAS). Although some researchers such as Khan and Abbasi (1998), Bernatik and Libisova (2004), Guo et al. (2009), Casamirra et al. (2009) and Cheng et al. (2012) have applied the FMEA method in their studies, research on the application of the FMEA in oil, gas, and petrochemical industries particularly from an insurance perspective is very sparse. The closest work to this paper is the study that Ghasemi and Mahmoudvand (2013) have done by applying the results of FMEA for categorizing each risks in safe, insurable and uninsurable risks. The FMEA method is an efficient method in a sense that both the insured and the insurer can use this method for risk assessment from their own perspectives and concerns. However, their assessments of the embodied risks may be not the same. This is the problem that Ghasemi and Mahmoudvand (2013) and other studies in this area have not considered. In addition, FMEA involves human subjectivity, which introduces vagueness type uncertainty and necessitates the use of decision-making under uncertainty. Note that there is this problem for other methods in the area of risk assessment such as AHP (see for example, Tesfamariam and Sadiq 2006). Here we deal with these problems and provide solutions for them. The rest of the paper consists of the following sections. Section 2 introduces the research method. An application of the FMEA for insurance purposes is presented in Sect. 3. In Sect. 4, we focus empirically on the case of Irans gas refineries by categorizing property damage risk factors and calculating risk probability numbers (RPNs) for insurance purposes. Finally, Sect. 5 brings the summary and conclusions.
2 Research method: the failure mode and effects analysis technique (FMEA) The American army began using FMEA in the 1949. The first guideline was Military Procedure MIL-P-1629 ‘‘Procedures for performing a failure mode, effects and
123
criticality analysis’’ dated November 9, 1949, and in 1974 it produced the army standardMIL-STD-1629 ‘‘procedures for performing a failure mode effects and criticality analysis. In 1980, there also was a second print of MIL-STD1629A. In 1990; the international organization for standardization (ISO) recommended the use of FMEA for design review in the ISO9000 series (Chang 2009; Cice and Celik 2013). This method has proven to be a useful and powerful tool in assessing potential failures and preventing them from occurring (Sankar and Prabhu 2001).The FMEA is an analytical technique for defining, identifying and eliminating the known and/or potential failures, problems, and errors and so on from system, design, process and/or service before they reach the customer (Stamatis 1995). Since introduction of FMEA as a support tool for designers, it has extensively been used in a wide range of industries, including aerospace, automotive, nuclear, electronics, chemical (especially in oil and gas industries), mechanical and medical technologies industries (Chang and Cheng 2010, 2011; Chin et al. 2009b; Sharma et al. 2005). The purpose of FMEA is to prioritize the failure modes of the product or system in order to assign the limited resources to the most serious risk items. Risk assessment in the FMEA is traditionally carried out by developing a RPN. To develop this, the first step in the FMEA is to identify all possible potential failure modes of the product or system by a session of systematic brainstorming. In the next step, an analysis is performed on these failure modes taking into account the risk factors such as severity of failure (S), occurrence of failure (O), and detection of failure (D). Following that, the RPN is defined and constructed by the multiplication of the S, O and D of a failure. That is, RPN ¼ S O D:
ð1Þ
For obtaining the RPN of a potential failure mode, three risk factors are evaluated using the 10-point scale described in Table 1. Details for this table are provided in Appendix Tables 5, 6 and 7. The higher the RPN of a failure mode, the greater the risk is for product/ system reliability. With respect to the scores of RPNs, the failure modes can be ranked and then proper actions will be preferentially taken on the high-risk failure modes. RPNs should be recalculated after the corrections to see whether the risks have gone down, and to check the efficiency of the corrective action for each failure mode (Liu et al. 2013). To calculate RPNs, a team of experts in the area of risk assessment will review all processes of the industry and provide an instruction of hazard identification and risk assessment according to the ISO (2002) and OHSAS (2007). The risk assessment team determines factors S, O and D according to Table 1 and the RPN is subsequently calculated using Eq. (1).
Author's personal copy Stoch Environ Res Risk Assess Table 1 Commonly scales for risk factors in FMEA Severity
Occurrence
Detection
Scale
No
Almost never
Almost Certain
1
Very slight
Remote
Very high
2
Slight
Very slight
High
3
Minor
Slight
Moderately high
4
Moderate
Low
Medium
5
Significant
Medium
Low
6
Major
Moderately high
Slight
7
Extreme
High
Very slight
8
Serious
Very high
Remote
9
Hazardous
Almost certain
Impossible
10 X 10 X 10 X
fsod 500g
10
Since each factor (S, O, and D) ranges between 1 and 10, the RPN varies between 1 and 1000. The RPNs will have a multimodal distribution. Assuming independency of S, O and D, the probability of each RPN can be obtained as: PðRPN ¼ kÞ ¼
contract (Mcguinness 1969). Here, RPN is a representative of loss and therefore we use the probability distribution of this variable to obtain the insurable bound. It is also worth noting that the probability pertinent to the PML involves only one tail—the upper end—of the relative frequency distribution of losses. Applying this rule on the probability distribution of the RPNs, we get a PML around 500 in confidence level 94%, when the components of RPN follow uniform distribution. That it: 10 X 10 X 10 X 1 940 PðRPN 500Þ ¼ ¼ 0:94 ¼ 1000 1000 s¼1 o¼1 d¼1
PðS ¼ sÞPðO ¼ oÞPðD ¼ dÞ
s¼1 o¼1 d¼1 fsod¼kg
ð2Þ where k = 1, 2,…1000. Note that the independence between severity and occurrence of losses is an usual assumption in risk analysis, although in some situations such as catastrophe modeling it might be plausible to consider the dependency between these components. Under the uniform distribution for all S, O and D, we have for example: 10 X 10 X 10 X 1 PðRPN ¼ kÞ ¼ 1000 s¼1 o¼1 d¼1 fsod¼kg
Note that RPNs do not attain all values in set 1, 2,…1000. For example, we wont have any RPN between 900 and 1000. A few computation show that only 120 numbers could be generated for RPNs. Of course we have 1000 values for RPNs but many of them are repeated several times. Figure 1 shows the frequency distribution of RPN, when the components of RPN are all uniformly distributed on set 1, 2,…10. According to this figure, values 60, 72 and 120 have the highest frequency to be identified as RPN. In addition, this figure indicates that small RPNs have higher probability than larger ones. Moreover, vertical axis of this figure shows that the distribution of RPN is multimodal. Now, we can use the probable maximum loss (PML) to find the upper bound of insurability by the insurer. Recall that, the probable maximum loss under a given insurance contract is that proportion of the limit of liability which will equal or exceed the amount of any loss covered by the
ð3Þ Note that, this bound is not only attainable by uniform assumption for the components of RPN, but it can be obtained by other distributions. See the empirical results in the subsequent sections. This means that 500 is the upper insurable bound of the insurer. Let us to justify this bound from a heuristic perspective. Loss acceptable ratios, i.e., the acceptable ratios of occurrence probability are very important and challengeable in insurance industry. Insurance is an economic business whose main goal is to maximize profit as much as possible given risks. It is clear that the risks, whose probability occurrences occupy the rank between 6 and 10, are not insurable by the insurance industry unless the insurer covers such risks with higher premium rates under different conditions. Since, there are no certain data for the severity and detection of the risks with the failure rate more than 0.0125 (1 in 80) (see Appendix Table 5), therefore, in order to obtain the maximum RPN acceptable by the insurer, the rank 10 is chosen for the severity (O) and detection (D) of such failure modes. As a result, the risks with RPN less than 500 are just insurable. Also, the more the RPN is the more proper safety and correction actions will be performed by health, safety and environment (HSE) sector of the industry. According to another similar idea by the experts in gas refinery, we can consider most probable scales as the occurrence to be less than or equal to 6, the severity effect to be less than or equal 8 and detection scale to be considered less than or equal to 6. By these assumptions, we can consider binomial distribution with probabilities 0.8, 0.6 and 0.5 as well as parameter size 9 for random variables S, O and D, respectively; that is: 9! 0:8s1 0:210s ; ðs 1Þ!ð10 sÞ! 9! 0:6o1 0:410o ; PðO ¼ oÞ ¼ ðo 1Þ!ð10 oÞ! 9! 0:5d1 0:510d ; PðD ¼ dÞ ¼ ðd 1Þ!ð10 dÞ! PðS ¼ sÞ ¼
123
Author's personal copy Stoch Environ Res Risk Assess
15 0
5
10
Frequency
20
25
Fig. 1 Frequency distribution of RPN, when its components come from uniform distribution
1 48 108 175 243
315 384
480
560 630 700
800
900
1000
RPN
where s, o and d attain values in set 1, 2,…10. Applying Eq. (2) and a computational code, it is easy to see that PðRPN 500Þ ¼ 0:947 which is closed to the value given by uniform distribution in Eq. (3). A graph of probability plot for RPN is given in Fig. 2. Besides, according to the regulation of some industries (such as gas refineries in Iran), risks with RPN less than 100 do not require safety and correction actions. These risks are called safe risks in the industry. To summarize, the risks in the selected industry are categorized as follows: – –
0.04
–
Uninsurable risks: risks with RPNs greater than 500, Insurable risks: risks with RPNs between 100 and 500 that need corrective actions by the industry, and Safe and tolerable risks: risks with RPNs less than 100.
The FMEA method is an efficient method in a sense that both the insured and the insurer can use this method for risk assessment from their own perspectives and concerns. However, their assessments of the embodied risks may be not the same. If risks are estimated only on the basis of ideas of insured experts, then the insurer company does not play any role in implementing the FMEA method. In this case, the insured party may underestimate risks, aiming at paying lower premium. This problem will be addressed in this section by elaborating on the methodology of our research and in particular the way of calculation of RPNs. Based on the risk classification carried out in the industry, RPN of each risk group is calculated to determine the type of each group; uninsurable, insurable or safe risks. Since each risk group itself includes some risks, the risk factors S, O and D are averaged over risks within each group according to Eq. (4): n
0.02
0.03
j 1X Sj ¼ Sij ; nj i¼1
0.00
0.01
Prob
3 Application of the FMEA for insurance purposes
0
200
400
600
800
1000
RPN Fig. 2 Probability distribution of RPN in the cases that RPN’s component come from binomial distribution
123
n
j X j ¼ 1 O Oij ; nj i¼1
n
j X j ¼ 1 D Dij nj i¼1
ð4Þ
where nj is the number of risks clustered in group j; Sij, Oij and Dij are referred to as the severity, occurrence and detection of risk i in group j, respectively. The calculated O and D) for each group are acceptable average values (S; representative of the rate of severity, occurrence and detection of that risk group. Then, RPN of each risk group is calculated using Eq.1 based on the calculated j and D j ) from the viewpoints of average values (Sj ; O both the insured and the insurer. Therefore, we encounter two RPNs for each risk group; one calculated by experts
Author's personal copy Stoch Environ Res Risk Assess
of the insured party, RPNdj , and the other obtained by the insurer experts shown as RPNrj . The calculated values of RPNdj and RPNrj are most likely different and there is no mechanism to prefer one over the other. In addition, agent-principal problems such as moral hazard and adverse selection might arise during risk assessment from both sides of the insured party and the insurer. To overcome such problems and to capture the effects of both perspectives in the final calculation of RPN, a new formula is proposed. This is as follows: RPNdj ; RPNFj ¼ RPNr 1 lj RPNj
j ¼ 1; 2; 3; 4; 5:
ð5Þ
where RPNFj is the final RPN for risk group j. Furthermore, RPN* represents the maximum acceptable RPN by the insurer based on our discussions in the Sect. 2. RPN* equals 500 in this study. Also, lj is called the effective coefficient which captures the effect of insurers RPN on the insureds one, and, is defined as follow: lj ¼
RPNrj RPNdj RPNrj þRPNjd pffiffi 2
ð6Þ
Note that, lj is the coefficient of variations of values RPNrj and RPNdj . In other words, lj shows the deviation of the viewpoint of the insurer from the viewpoint of the insured party. It is easy to see that if RPNrj ¼ RPNdj ; then lj ¼ 0 and RPNFj ¼ RPNdj ¼ RPNrj : Also, RPNrj RPNdj in all situations since the insurer likes more than the insured party to attribute a higher RPN to a risk group. So, the Eq. (5) is constrained by lj 0: Note also that, lj is bounded from above, too; because for every two nonnegative values a and b, where a [ b we can write: ða þ bÞ2 a2 þ b2 ða þ bÞ2 : 2 Using this inequality, we get: 0
ab 1: aþb
Applying this conclusion with a ¼ RPNrj and b ¼ RPNdj we get: pffiffiffi 0 l j 2: The above statement indicate that the proposed RPNFj can be considered as a compromise among the insurer and insured perspectives.
4 A case study of Irans gas refinery Iran is a country with more than 151.2b barrels of oil proved reservations and 33.1 trillion cubic meters of natural gas at the end of 2011 (BP Statistical Review of World Energy 2012). According to these numbers, Iran ranks the third and the second, worldwide, from the perspective of oil proved reservations and natural gas sources, respectively. Despite growth of non- oil production and exports over the past decades, oil is still the main source of Irans external income. About 80 % of external income and more than 40 % of governments public budget come from oil exports (Central Bank of Islamic Republic of Iran 2011). In this research, a gas refinery in Iran was chosen as the case study in order to implement the FMEA method described earlier. The case study is investigated from two different points of views. In the first viewpoint, the RPNs of the categorized risks were obtained by the experts of the gas refinery as the insured organization. Since the insurer might disagree with the results represented by the insured party, a group of experts in the insurer company elicited the RPN of risks separately and without interference of the insured organization. Both results were then compared and finally, a unique group of RPNs is calculated using the proposed formula of RPN in Sect. 3. selected gas refineries data are used for the period March 2011 till March 2012. 4.1 Categorization of Irans gas refinery risks One of the main contributions of this research paper is to categorize more than 300 risks into five risk groups. To cluster the risks, exact information on the process of the gas refinery is required. In a gas refinery, different operational units are designed and constructed for producing different products. Generally, these units are divided into three major ones including process, utility and lateral units. The main source of risks in gas refinery is due to hazards of the hydrocarbon processes. The Probability of occurrence of such processes decreases by strengthening the actions of hazard safety, prevention and incidents control, however, the safety will not reach 100 %. Often, risks in gas refinery have low frequency and high severity. Therefore industrial installation must be somehow designed, and constructed that people, environment and future generation feel secure from the undesirable occurrences. This aim is achievable by applying the exact risk assessment continuously. Generally, two types of loss occurring in gas refineries are: Property damage and bodily injury which are defined as follow: –
Property damage This damage leads to two kinds of loss: direct loss and indirect loss. Direct loss is the financial loss that occurs in different sectors and
123
Author's personal copy Stoch Environ Res Risk Assess
majority of risks pertain to fire and machinery breakdown groups in Irans gas refinery.
Explosion Errors and omission
4.2 Empirical results Fire Machinery breakdown
The empirical results of this research are presented from perspectives of both the insured party and the insurer. The RPNs of risk groups calculated by both the insured and insurer are illustrated in Tables 2 and 3.
Pressurized vessel defects
Fig. 3 Proportion of different risk group in the gas refinery
–
indirect loss includes business interruption which also results in loss of profit. Bodily injury Experts and experience manpower may be injured and die because of accident in production process (Habibi 2009).
Most prevalent risks in gas refinery are: – – – – – – – – – – –
Fire and explosion Leakage Hazardous chemicals The existence of limited spaces Cleaning storages CO2 hazards Human errors and omission Machinery breakdown Pressurized vessels defects Natural hazard including such earthquakes, flood, storm, and lightning. External hazards comprising war and terrorism, radioactive contamination, electromagnetic weapons, cyber-attacks and sanctions.
Iranian insurance companies only cover the main and consequential risks such fire, explosion, machinery breakdown, error and omission, pressurized vessels defects and natural hazards such as earthquakes, flood, storm, and lightning. External hazards are excluded in insurance policy and will not be covered by insurer. This research paper categorizes, from the insurer viewpoint, all risks identified by the FMEA into five classes including fire, explosion, error and omission, machinery breakdown and the pressurized vessels defects that occurs in operational unit of gas refinery. Natural and indirect risks as well as human injuries were excluded in this study. According to the above categorization, the number of risks in each category was found in Irans gas refinery during the period of study. We have found that there are 156 (55 %), 21(7 %), 18(6 %), 84(30 %), and 5(2 %) types of risk in risk categories of fire, explosion, error and omission, machinery breakdown and pressurized vessels defects respectively. This risk breakdown is shown in Fig. 3 which shows that the
123
Note that the total of each factor calculated by weighted average. In addition, we have applied coefficient of variation to capture the variability of estimators that we used here. Recall that the coefficient of variation, cv, is defined as: cv ¼
Standard deviation of the data : Average of the data
Meanwhile, Mahmoudvand and Hassani (2009) have shown that the coefficient of variation is bounded from above when the data are non-negative. They have shown that: pffiffiffi 0 cv n; where n is the number of data. Because of this property, we have reported pcvffiffin in Tables 2 and 3 which is comparable with 1. So, the cv close to zero and far from 1 indicate the small variation. As it can be seen in these Tables, all estimators have a relatively small variations, as well as the total risk indicate a small variations in both insured and insurer perspectives. The empirical findings are the following: (a)
(b)
c)
All calculated RPNs are less than 500, indicating that there is no uninsurable risk group in the selected gas refinery. Suitable safety policies, good fire extinguisher infrastructure and approximately high technology in most units of gas refinery prove why all groups of risks are attributed with low RPNs. Therefore, all of the identified risks are insurable from the perspective of the insurer company. Although, most property risks pertain to fire group (see Fig. 3), the explosion group has the maximum RPN and therefore, taking sufficient preventive actions through process units of gas refinery, specifically, in Gas Treatment Unit (GTU), Steam and Supply Water, Utility Analog and Flare unit seems to be highly required. Also as per the idea of experts, the capability of the detecting of the failure modes related to explosion group should be enhanced by strengthening the control system of equipment, specially, in power generation and compressor units. The RPNs less than 100 for the two risk groups of the pressurized vessels defects and error and omission
Author's personal copy Stoch Environ Res Risk Assess Table 2 RPN obtained by the insured party in the gas refinery
RPN d
cv pffiffiffij nj
Type of risk group
1
Fire
152
6.12
4.76
4.85
115.21
0.04
Insurable
2
Explosion
20
8.85
3.45
5.15
157.24
0.11
Insurable
3
Error and Omission
18
6.94
3.56
3.39
83.68
0.18
Safe
4
Machinary breakdown
83
6.16
4,25
3.89
101.90
0.06
Insurable
5
Pressurized vessels defects
5
8.60
2.20
3.40
64.33
0.14
Safe
278
6.43
4.39
4.47
126.20
0.04
Insurable
No
Group of risk
nj
Sj
j O
j D
RPN r
cv pffiffiffij nj
Type of risk group
1 2
Fire Explosion
152 20
6.79 9.44
4.85 4.40
6.21 7.36
204.5 305.70
0.04 0.07
Insurable Insurable
3
Error and omission
18
7.35
4.43
5.63
183.32
0.11
Insurable
4
Machinary breakdown
83
7.74
4.87
5.13
193.37
0.05
Insurable
5
Pressurized vessels defects
5
8.85
3.95
5.45
190.52
0.09
Insurable
278
7.34
4.78
5.93
207.87
0.03
Insurable
reveal the fact that there is an adequate safety in units involved with these risks. Despite of high effects of severity in group 5, which occurs particularly in GTU, Steam and Supply Water and Overhaul units, high detection chance and remote probability of failure make such risks safe and tolerable for the refinery, but, based on the internal regulations of the gas refinery, the safety of units with rank of severity about 10 must be improved by appropriate corrective actions. Also, risk group 5 is considered as supplementary hazards in insurance policy and the insured party must pay a high premium. While this group of risks is known as safe risks, the insured is recommended to avoid insuring this risk group. On the Error and Omission group, there are many factors contributing to being safe for a gas refinery such as good training of manpower, observing safety cautions, and using adequate alarm systems through the refinery. Machinery breakdown risk group takes the second place in terms of RPN. According to Appendix Tables 5 and 6, the failure rate of this risk group is between 1 in 400 and 1 in 2000 and its likelihood of detection by the design control is almost moderately
Table 4 Calculated final RPNs
j D
nj
Total
(d)
j O
Group of risk
Total
Table 3 RPN obtained by the insurer of the the gas refinery
Sj
No
high. To improve the safety of this group, performing the more regular services and maintenance and fundamental overhauls as well as using safety valves with high control capability can be effective. As we discussed previously, to get a concrete conclusion about insurance, we need to get a unique RPN for each risk category. This is done by applying the formula (4) proposed in Sect. 3. The calculated unique RPNs of risk groups using Eq. 4, are reported in Table 4. Table 4 reports final RPNs for various values of effective coefficient which captures the relative difference between RPNd and RPNF. It shows that the first four risk groups including fire, explosion, error and omission, and machinery breakdown, are insurable and the risk of last group (pressurized vessels defects)is tolerable by the gas industry. It is worth mentioning that the goal of calculating the final RPN is to determine the rate of premium based. In other words, the insurance policy is somehow designed to characterize the premium rate in terms of final RPN of the gas refinery. From this viewpoint, the difference between RPNd and RPNF is justified, because, this difference may alter the rate of premium. Hence, both sides including the
No
Group of risk
nj
RPN d
RPN r
lj
RPN F
Type of risk group
1
Fire
152
141.05
204.50
0.26
157.81
Insurable
2
Explosion
20
157.24
305.70
0.45
217.57
Insurable
3
Error and Omission
18
83.68
183.32
0.53
103.76
Insurable
4
Machinery breakdown
83
101.90
193.37
0.44
122.69
Insurable
5
Pressurized vessels defects
5
64.33
190.52
0.70
87.74
278
126.20
207.87
0.49
147.39
Total
Safe Insurable
123
Author's personal copy Stoch Environ Res Risk Assess
insured party and the insurer can easily agree on the insurance policy and premium level based on the above methodology of RPN calculation.
5 Summary and conclusions Various risk factors affect performance and output of gas refineries. Human and physical capital losses resulting from risk factors such as fire, explosion, human errors and omission, machinery breakdown, pressurized vessels defects and many natural and external hazards can be enormous. Hence, gas refineries must be among top priorities of energy policy makers. Although owners and shareholders of gas refineries are willing to avoid such high risks, they are not transferable unless insurance companies show their desire to buy them with reasonable risk premium. To reach a practical way for insuring such high- risk businesses, we need to properly identify, categorize and evaluate embodied risk factors by the subset of probability risk assessment technique. This research paper applied the FMEA as the most relevant PRA technique in Irans gas refineries for insurance purposes. Using information and data for period March 2011–March 2012, it identified and categorized various property damage risk factors. Based on risk categorization of Irans gas
refineries, the risk probability numbers (RPNs) were calculated from perspectives of both the insured party and the insurer and then manipulated to obtain unique and final RPNs. The empirical results clearly indicate that various property damage risk factors embodied in gas refineries including fire, explosion, error and omission, and machinery breakdown are insurable risks with reasonable premium. Risks of pressurized vessels defects are in safe category and can be tolerated by the industry owner. The results imply that gas refiners, despite of high risks, are generally insurable in the market and Iranian policy makers in the energy sector should pay particular attention to this vital issue. Insuring gas refineries with reasonable risk premium will definitely reduce capital losses which can otherwise be enormous for the economy in general and for oil and gas industries in particular. Acknowledgments The author would like to acknowledge the Research and Technology Directorate and Supervision and Coordination of Gas Production Directorate in National Iranian Gas Company (NIGC) for providing research opportunity and the use of required resources for this study.
Appendix See Tables 5, 6 and 7
Table 5 Ratings for severity of a failure mode Effect
Criteria
No
No effect
Scale 1
Very slight
Very minor effect on product or system performance
2
Slight
Minor effect on product or system performance
3
Minor
Small effect on product performance. The product does not require repair
4
Moderate
Moderate effect on product performance. The product requires repair
5
Significant
Product performance is degraded. Comfort or convince functions may not operate
6
Major
Product performance is severely affected but functions. The system may not operate
7
Extreme
Product is inoperable with loss of primary function. The system is inoperable
8
Serious
Failure involves hazardous outcomes and/or noncompliance with government regulations or standards
Hazardous
Failure is hazardous, and occurs without warning
Table 6 Ratings for the occurance of a failure mode
123
Probability of failure
9 10
Failure rates
Scale
Almost never
1 in 1,500,000
1
Remote
1 in 150,000
2
Very slight
1 in 15,000
3
Slight
1 in 2000
4
Low
1 in 400
5
Medium
1 in 80
6
Moderately high
1 in 20
7
High Very high
1 in 8 1 in 3
8 9
Almost certain
1 in 2
10
Author's personal copy Stoch Environ Res Risk Assess Table 7 Ratings for the detection of a failure mode
Detection
Criteria
Scale
Almost certain
Design control will almost certainly detect a potential cause of failure
Very high
Very high chance the design control will detect a potential cause of failure
2
High
High chance the design control will detect a potential cause of failure
3
Moderately high
Moderately high chance the design control will detect a potential cause of failure
4
Medium
Moderate chance the design control will detect a potential cause of failure
5
Low
Low chance the design control will detect a potential cause of failure
6
Slight
Very low chance the design control will detect a potential cause of failure
7
Very slight
Remote chance the design control will detect a potential cause of failure
8
Remote
Very remote chance the design control will detect a potential cause of failure
Impossible
Design control does not detect a potential cause of failure
References
Bernatik A, Libisova M (2004) Loss prevention in heavy industry: risk assessment of large gas holders. J Loss Prev Process Ind 17:271–278 BP Statistical Review of World Energy (2012), 6 and 20, http://www. bP.com/statisticalreview/ Casamirra M, Castiglia F, Giardina M, Lombardo C (2009) Safety studies of a hydrogen refueling station: determination of the occurrence frequency of the accidental scenarios. Int J Hydrog Energy 34:5846–5854 Central Bank of Islamic republic of Iran (2011). http://www.cbi.ir/ Chang KH, Cheng CH (2010) A risk assessment methodology using intuitionistic fuzzy set in FMEA. Int J Syst Sci 41:1457–1471 Chang KH, Cheng CH (2011) Evaluating the risk of failure using the fuzzy OWA and DEMATEL method. J Intell Manuf 22:113–129 Chang KH (2009) Evaluate the orderings of risk for failure problems using a more general RPN methodology. Microelectr Reliab 49:1586–1596 Cheng Y, Hu G, Wu GW (2012) Development of a risk-based maintenance strategy using FMEA for a continuous catalytic reforming plant. J Loss Prev Process Ind 25:958–965 Chin KS, Wang YM, Poon GKK, Yang JB (2009b) Failure mode and effects analysis using a group-based evidential reasoning approach. Comput Oper Res 36:1768–1779 Cicek K, Celik M (2013) Application of failure modes and effects analysis to main engine crankcase explosion failure on-board ship. Saf Sci 51:6–10 Ghasemi S, Mahmoudvand R (2013) A New Insight into the FMEA and its Applications in Insurance: a case study in gas refinery. Int J Energy Stat 1(4):257–268 Guo L, Gao J, Yang J, Kang J (2009) Criticality evaluation of petrochemical equipment based on fuzzy comprehensive evaluation and a BP neural network. J Loss Prev Process Ind 22(4):469–476
1
9 10
Habibi H (2009) Variety of hazard and insurable risks in oil, gas and petrochemical industries. In: 2th seminar on insurance and risk management in oil, gas and petrochemical industries, pp 7–8. (in Persian) ISO (2002) Petroleum and natural Gas Industries Offshore Production Installations Guidelines on Tools and Techniques for Hazard Identification and Risk assessment. BE EN ISO 17776 http://www.shana.ir Khan FI, Abbasi SA (1998) Techniques and methodologies for risk analysis in chemical process industries. J Loss Prev Process Ind 11:261–277 Liu H, Lui L, Liu N (2013) Risk evaluation approaches in failure mode and effects analysis: a literature review. Expert Syst Appl 40:828–838 Mahmoudvand R, Hassani H (2009) Two new confidence interval for the coefficient of variation in a normal distribution. J Appl Stat 36(4):429–442 Mcguinness J (1969) Is probable maximum loss (PML) a useful concept. Proc Casual Actuar Soc lv1(105 & 106): 31–40 OHSAS 18001:2007, BSI, British Stand Sankar NR, Prabhu BS (2001) Modified approach for prioritization of failures in a system failure mode and effects analysis. Int J Qual Reliab Manag 18:324–336 Shahriar A, Sadiq R, Tesfamariam S (2014) Life cycle greenhouse gas footprint of shale gas: a probabilistic approach. Stoch Environ Res Risk Assess 28(8):2185–2204 Sharma RK, Kumar D, Kumar P (2005) Systematic failure mode effect analysis (FMEA) using fuzzy linguistic modelling. Int J Qual Reliab Manag 22:986–1004 Stamatis DH (1995) Failure mode and effect analysis: FMEA from theory to execution. ASQC Press, New York Tesfamariam S, Sadiq R (2006) Risk-Based Environmental decision making using fuzzy analytic hierarchy process (F-AHP). Stoch Environ Res Risk Assess 21(3):35–60 Youn H, Lim BI, Lee EY, Choi Y (2011) The impact of the risk environment and energy prices to the budget of Korean households. Stoch Environ Res Risk Assess 25(4):323–330
123
