JCHR Journal of Community Health Research 2017; 6(2): 85-92.

Use of Failure Mode and Effects Analysis

in Improving Safety: A Case Study
in An Oil Company
Downloaded from jhr.ssu.ac.ir at 17:19 IRST on Tuesday January 22nd 2019

1 1 1
Malihe Kolahdouzi , Gholam Hossein Halvani , Ebrahim Nazaripour Abdehgah ,
2 3
Maryam Rostami Aghdam shendi , Mohsen Yazdani aval

1. Department of Occupational Health Engineering, School of Public Health, Yazd Shahid Sadoughi University of
Medical Sciences,Yazd, Iran.
2. Department of Occupational Health Engineering, School of Public Health, Tehran University of Medical Sciences,
Tehran, Iran.
3. Department of Occupational Health Engineering, School of Public Health and Research Center for Health
Sciences, Hamadan University of Medical Sciences, Hamadan, Iran.

ARTICLE INFO ABSTRACT

Original Introduction: The Failure Mode and Effects Analysis (FMEA), first of all,
identifies dangerous modes in a process and, then, provides proper control
Received: 1 Jan 2017
measures that are required to reduce the level of risk. This study aims to
Accepted: 23 May 2017
investigate the risk level before and after the implementation of control measures
in an edible oil factory.
Methods: This cross-sectional study was conducted in an edible oil factory in
Tehran, Iran. The FMEA technique was used to identify failure modes related to
equipment that existed in five units of this factory. After that, in all units, Risk
Priority Numbers of all the processes were assessed on the basis of the FMEA
method. With regard to the Risk Priority Number (RPN), some control measures
Corresponding Author: were taken to reduce the risk of events. After nine months, risk assessment was
Mohsen Yazdani Aval repeated, and primary and secondary RPNs were compared with each other to
[email protected] investigate the effects of interventions.
Results: The results showed that the highest probability of danger was related to
the installation unit. The probability of danger in this unit before and after
interventional actions was 4.5 and 3.25 respectively and the highest degree of
severity was related to tool production. The marginal homogeneity test showed
the positive effect of control measures on the risk level in the factory. There was
a significant inference between before- and after-data in RPN in the unit tool
production, neutralization unit, and installation unit. The Tests of Repeated
Measure showed that there was a statistically significant inference between
before- and after-data in the total average of RPN criterion (pvalue < 0.001).
Conclusion: It can be concluded that the high risk level in the factory units was
reduced after the application of interventional corrective measures. These actions
have been useful. It can be mentioned that the FMEA has been successfully able
to identify and control the level of risk in this factory.

Keywords: Risk Assessment, FMEA, Edible oil factory, RPN

How to cite this paper:

Kolahdouzi M, Halvani GH, Nazaripour Abdehgah E, Rostami Aghdam shendi M, Yazdani aval M. Use of Failure
Mode and Effects Analysis in improving safety: A case study in an oil company. J Community Health Research. 2017;
6(2): 85-92.

Copyright: ©2017 The Author(s); Published by Shahid Sadoughi University of Medical Sciences. This is an
open-access article distributed under the terms of the Creative Commons Attribution License
(https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any
medium, provided the original work is properly cited.
 Use of Failure Mode …

Introduction
The identification of dangers to evaluate and properly carried out. Moreover, proposing the
manage risks is a crucial step in any working proper control actions to reduce the risk level in
process. It helps managers define corrective the industry and taking actions to prevent
actions and implement them to mitigate the risks. probable events in future is necessary. In view of
Based on this step, proper safety plans can be the above statements, this study aimed to identify
Downloaded from jhr.ssu.ac.ir at 17:19 IRST on Tuesday January 22nd 2019

formulated as well. To achieve better safety the dangers, proposing corrective measures, and
performance in the process, more precise implementing as also evaluating the efficiency of
identification of dangers is necessary (1). There is the measures at an edible oil factory in Tehran,
a lot of methods to implement risk assessment, using FMEA.
but it should be considered whether a method
Methods
would be useful for this as a particular goal, and This cross-sectional and interventional study
whether it is appropriate for the activities and was conducted in an edible oil factory in Tehran
processes in the system (2). The Failure Mode and in 2014. The FMEA technique was used to
Effects Analysis (FMEA) is a widely used identify failure modes related to the equipment in
technique, with the application of which five units and to evaluate them in the factory. To
manufacturers are able to improve the quality, achieve this goal, necessary coordination with the
reliability, and safety of their products (3). The management team of the factory was carried out
technique is used to predict failures in working to achieve the requisite justifications to enter the
processes in several industries (3–6). The FMEA units. It should be noted that the FMEA was used
initially identifies dangerous modes in the process in the present study because the units in the
and then provides proper control measures factory that were studied comprised mechanical
required to reduce the risk level. On the other compartments instead of chemical processes.
hand, the activity-related failure modes and their After visiting and studying different parts of
potential effects are identified with the help of the factory, five units were selected for the
this technique, and some control measures are implementation of FMEA, based on whether there
proposed to eliminate or significantly reduce the was dangerous equipment in the units. Also, the
likelihood of failure (3, 7). Many studies have prioritized duties related to equipment were
shown that the technique is a powerful method to considered, using the accident frequency rate
prevent the occurrence of accidents; it is used (AFR). Using FMEA Worksheets (3), the risks
frequently in high-risk industries, such as involved in the units were examined in detail, and
aerospace and nuclear power plants (8). It should the likelihood, severity, and detectability of
be noted that conducting the FMEA is a time- dangers—the three parameters required to
consuming process and the effective calculate the RPN score—were quantitatively
implementation of the technique to evaluate a studied.
process requires that some information be To identify dangers in the selected units,
precisely gathered (9). including the units for the production and filling
The edible oil factories are considered among of cans, tools production, neutralization, and
the important industries in Iran. However, enough installation, we evaluated workers’ duties and the
studies have not been conducted to identify and to ways in which these were done. We checked the
assess dangers—and, more importantly, to control current instructions and regulations in the factory,
existing dangers in such factories. Therefore, safe evaluated the documented data related to the
work procedures in such industries are very accidents that had occurred in the past, inspected
essential. To achieve this goal, the identification the units, and also held interviews with the
of dangers and the risk evaluation of the dangers workers and their supervisors. The likelihood of
are among the major actions that should be danger was determined on the basis of documents

86
 Kolahdouzi M, et al Journal of Community Health Research 2017; 6(2): 85-92.

that were available and the repeatability of required. To verify the accuracy of the results of
failures related to the dangers was considered. the repeated measures model, Levene’s test of
The severity and detectability of dangers were homogeneity of variances, Mauchly’s sphericity
also determined after considering the documented test, and the Q-Q plot for normality of residuals
information. After achieving the scores for all were used. The significance level was considered
three factors, the RPN score was calculated, using 0.1 for the nonparametric Wilcoxon test used for
Downloaded from jhr.ssu.ac.ir at 17:19 IRST on Tuesday January 22nd 2019

Equation 1 (RPN = Occurrence × Severity × units with small sample sizes, and because of the
Detection). Finally, the risks of identified dangers weakness of the test to clarify probable
were prioritized using the RPN number and a differences between before- and after-data. The
team consisting of researchers, health, safety and significance level was considered at 0.05 for
environment (HSE) experts, a production parametric tests.
engineer, and one of the most experienced
Results
workers argued about control measures. The team In the interventional study, conducted in an
proposed some corrective measures, using edible oil factory in Tehran, the dangers (73
brainstorming, which could mitigate the RPN for dangers) in five units of the factory were
any danger. identified and the risk level was also evaluated
Some of the most important corrective actions using RPN equation before and after
included the use of equipment and tools designed implementing corrective actions using FMEA.
in a safer way, the elimination or limitation of
dangers, the deployment of guards, the use of The comparison of danger parameters
audio-visual alarm signs, the use of specific before and after interventions in the factory:
education programs and instructions, and, finally, The Tests of Repeated Measure were used to
the application of personal protective equipment clarify differences between before- and after-data
(PPE). After nine months, we reevaluated the in probability and severity and the detectability of
safety status of the units and RPN2 was again danger and RPN number in all units. The results
calculated for all the dangers and primary and showed that there was a statistically significant
secondary RPNs were compared with each other inference between before- and after-data in
to investigate the effects of the interventions. probability in all the units (p value < 0.001).
There was also a significant inference between
Statistical analysis: the average of probability in different units
For the descriptive and analytical analysis, (p value = 0.08).
SPSS 23 was used. We compared primary and The paired sample T-test was used to clarify
secondary RPNs to evaluate changes in the risk how the probability of danger was different from
levels. the others and in which unit. The results,
we used the median as a central tendency to according to Fig. 1, showed that the probability of
describe the parameters of danger, when the dangers had a statistical significant inference in
nonparametric Wilcoxon test was implemented the installation unit with the can production unit
for statistical analysis. The Marginal (p value = 0.031).
Homogeneity Test was used to compare primary There was a statistically significant inference
and secondary RPNs before and after the between before- and after-data in severity in all the
implementation of corrective measures. To units (p value < 0.001). There was no significant
compare parameters of danger before and after inference between t he average of severity in
interventions in all the units, the Tests of different units (p value = 0.08). This means that the
Repeated Measure were used and the variable unit average of dangers in all units was similar.
was considered a disturbance variable. Tukey’s The results showed that there was no
test was also used for the further analysis that was significant inference before and after between

87
 Use of Failure Mode …

detectability of the average score in different units respectively 4.5 and 3.25. The lowest rate of this
(p value < 0.124). There was also no significant parameter was related to the can production unit
inference between the average of detectability of and the numbers before and after interventional
danger in different units (p value = 0.133). actions were respectively 3.6 and 2.8. The results
There was a statistically significant inference are provided in Table 1.
between before- and after-data in a total average The results showed that in the tools production
Downloaded from jhr.ssu.ac.ir at 17:19 IRST on Tuesday January 22nd 2019

of RPN criteria (p value < 0.001). But there was unit, there is no statistically significant inference
no statistically significant inference between between before- and after-data in the probability
before- and after-data in the average of RPN criterion (p value = 0.1). In the two units—
criteria in different units (p value = 0.077). This neutralization and installation—in contrast, there
means that the average of dangers in all units was was a statistically significant inference between
similar. The results are presented in Fig.1. the before- and after-data in the probability
criterion (p value = 0.059). In two units—can
The comparison of danger parameters
production and can filling—there was also a
before and after interventions in different
statistically significant inference between
units:
before- and after-data in probability criterion
The results show that the highest degree of
(p value < 0.001).
severity was related to the tool production unit
There was a statistically significant inference
and the average of the severity of danger in this
between before- and after-data in RPN in the unit
unit before and after interventions was 4.75 and 3
tool production (p value = 0.066), neutralization
respectively. The lowest amount of severity of
unit (p value = 0.042), and installation unit
danger was related to the can production unit,
(p value = 0.068). A significant inference was
such that the amounts before and after
also seen between before- and after-data in RPN
intervention were respectively 4.07 and 2.42. The
in the units of can production and can filling
highest probability of danger was related to the
(p value < 0.001). The results are provided in
installation unit. The probability of danger in this
Fig. 1.
unit before and after interventional actions was

Figure 1. The average of probability and severity of danger and RPN score before and after corrective actions in the units.

88
 Kolahdouzi M, et al Journal of Community Health Research 2017; 6(2): 85-92.

All the dangers identified in all the units with risk levels. Before implementing the control
the minimum and maximum RPN scores are measures, low unacceptable risk level was
shown in Table 2. It is clear that the highest and 15.07%, and relatively acceptable and
lowest RPNs were respectively 80 and 4. The acceptable risk levels were 45.21% and 39.73%
marginal homogeneity test was used to clarify respectively. After the implementation of the
the effects of interventions on the factory’s control measures, the amounts of low
Downloaded from jhr.ssu.ac.ir at 17:19 IRST on Tuesday January 22nd 2019

safety status. The test simultaneously tested the unacceptable risk level and relatively
equality of marginal proportion for before- acceptable risk level were reduced to 2.77% and
and after-, and the results show that there is a 5.56% respectively while the amount of
significant inference between marginal acceptable risk level increased to 91.67%. The
proportion (distribution) for before- and after- results showed the positive effect of control
(p value < 0.001). measures on the risk level in the factory. The
Considering these results, risk levels were results are provided in Fig. 3. A control table
categorized at three levels, including was designed at the end of this study on the
acceptable, relatively acceptable (with basis of the information obtained in the oil
modifications), and unacceptable risk levels. factory. The risk levels and RPN numbers were
The latter was categorized into three categories, calculated and classified. The results are
including low, average, and high unacceptable provided in Table 3.

Figure 2. Frequency of risk levels before and after the interventions.

89
 Use of Failure Mode …

Table 1. Comparison of scores of occurrence and severity average before and after interventions in different units.
Average±STD of Average±STD of
severity/occurrence and severity/occurrence and
Unit p value
detection score before detection score after
intervention intervention
Tools production
Severity 4.75±0.5 3±0 0.06
Downloaded from jhr.ssu.ac.ir at 17:19 IRST on Tuesday January 22nd 2019

Can production 4.07±0.79 2.42±0.57 0.001

Neutralization 4.4±0.89 2.8±0.44 0.038
Can filling 4.18±0.84 2.75±0.61 0.001
Installation 4.25±0.95 2.75±0.5
0.063

Tools production 3.75±0.5 3.25±0.95 0.157

Can production 3.57±0.8 2.8±0.69 0.001
Neutralization 3.8±0.83 3±0 0.05
Occurrence
Can filling 3.84±0.61 3.2±0.59 0.001
Installation 4.5±0.57 3.25±0.5
0.05
Tools production 3.25±0.5 3.25±0.5 1
Can production 3.12±0.65 2.85±0.67 0.01
Detection Neutralization 3.2±0.44 3±0.7 1
Can filling 3.33±0.59 3.33±0.59 1
Installation 2.75±0.5 2.75±0.5 1

Table 2. Frequency of risk levels in the factory units

The number of Minimum risk number Maximum risk number

Unit
dangers Before After Before After
Tools production 4(5.47%) 48 24 60 36
Can production 26(35.6%) 18 4 80 48
Neutralization 5(6.84%) 36 18 80 36
Can filling 34(46.57%) 27 6 80 64
Installation 4(5.47%) 30 12 75 36

Table 3. Control levels

Risk level Range
Unacceptable 60 > RPN > 45
)ALARP( As Low As Reasonably Practicable RPN < 45
Acceptable RPN ≤ 60

Discussion evaluated by comparing primary and secondary

In the present study, the FMEA was used to RPN scores before and after applying the
identify and evaluate the risk of dangers in an measures.
edible oil factory in Tehran. After considering the The results show that the highest risk level was
probable occurrence of events based on the related to the tools production unit. The average of
RPN scores, some corrective measures were RPN scores in this unit was 57 before
implemented to reduce the risk level in the factory. interventions, and the score was reduced to 30.57
The effects of the proposed measures were after the application of corrective measures. The

90
 Kolahdouzi M, et al Journal of Community Health Research 2017; 6(2): 85-92.

lowest average of RPN score was related to can primary score) and the scores for the two other
production (46.39). After the implementation of abovementioned risk levels were reduced to 72%
the corrective actions, the score was reduced to and 53% respectively, less than their primary
19.65. The results of the other studies— scores. The results of the present study are in line
Khoshakhlagh et al. (10) and Pareek et al. (11)—have with the two abovementioned studies (9). It can be
showed that the highest risk level was related to concluded that the corrective measures that were
Downloaded from jhr.ssu.ac.ir at 17:19 IRST on Tuesday January 22nd 2019

the installation unit. Similar results were observed implemented in this study had a useful effect on
in the present study, showing that there are the risk level in the factory.
instantaneously several risk factors and dangerous It should be noted that the involvement of the
conditions (material and equipment for lubrication, managers in the risk assessment process and the
welding, electric tools and splurge caused by repair consideration of their viewpoints in offering
of equipment), which together caused the results corrective measures facilitated the implementation
achieved in this unit. of the offers. The results of the present study
The results of the present study showed that showed that if the corrective measures proposed by
there is a statistically significant inference between researchers have taken into consideration the
RPN1 and RPN2. The results are in line with industry’s economic and physical conditions,
the other studies (12, 13). These show that a proper convincing the manager to accept the offers
risk assessment and implementation of the becomes easier. The implementation of such
corrective actions could lead to a reduction in corrective measures also minimized the potential
the RPN score. risks by reducing the probability and probable
Several corrective measure were taken in the consequences of dangers in an industry.
present study. Some of the interventions included
Conclusion
guarding, replacing tools and equipment, using Finally, based on the findings of the present
personal protective equipment (PPE), limiting study, it can be concluded that the high risk level
dangers by changing the distances between in the factory units that were studied was reduced
personnel and equipment, installing alarms and after the application of interventional corrective
symbols such as auditory or visual warning alarms, measures. These actions have been useful. It can
and providing special instructions and educations also be mentioned that the FMEA has been
programs. In a study, Bonfant et al. (14) showed that successfully able to identify and control risk levels
a significant reduction in the RPN score occurred in the factory.
after the implementation of some corrective It must be considered that the implementation of
measures at a hospital in Italy. It should be noted a proactive program to reduce the risk level of
that some of the corrective measures—the use of dangers in the processes is indispensable. One of
specific operational instructions and procedures the most important elements to prevent the
and educational programs for beginner workers— occurrence of accidents in the industries is the use
which were taken up in the present study—were of appropriate safety programs. A good safety
similar to those taken in Bonfant et al.’s study (14). program, apart from describing the duties of all
In a study, Mohammadian and Hashemi Nejad personnel in detail, can improve the level of
(9)
showed that the RPN score was reduced to near responsibility of the people in an organization,
zero after corrective measures were applied. including top managers, headmen, supervisors, and
Hodaasadat Hosseini et al., in a study, showed that contractors. To achieve this goal, it has been
the RPN scores for three levels of risk—including necessary to use novel methods of assessment to
low, average, and high risk levels—were 62.7, monitor and evaluate all existing jobs, equipment,
31.6, and 5.7 respectively. After applying machinery, and behavior of personnel in the
corrective measures in their study, the RPN scores working project.
for the low risk increased (30.6% more than the

91
 Use of Failure Mode …

Acknowledgment participants and the entire staff of the oil factory

This article is extracted from Master’s Thesis. for their gracious cooperation.
The authors would like to acknowledge the
Conflict of Interest
financial support of Yazd Shahid Sadoughi Authors declare that there is not any competing
University of Medical Sciences. As well, the interest.
authors would like to appreciate all honorable
Downloaded from jhr.ssu.ac.ir at 17:19 IRST on Tuesday January 22nd 2019

References
1. Ebrahimzadeh M, Halvani G, Mortazavi M, Soltani R. Assessment of Potential Hazards by Failure Modes and Effect
Analysis (FMEA) Method in Shiraz Oil Refinery. Occupational Medicine Quarterly Journal. 2011;3(2):16-23.
2. Halvani GH ZM. Safety system engineering and risk management. Tehran: Asare Sobhan Publication; 2008.
3. Ebrahemzadih M, Halvani G, Shahmoradi B, et al. Assessment and Risk Management of Potential Hazards by Failure
Modes and Effect Analysis (FMEA) Method in Yazd Steel Complex. Open Journal of Safety Science and
Technology. 2014;4(3):127-135.
4. Paparella S. Failure mode and effects analysis: a useful tool for risk identification and injury prevention. Journal of
Emergency Nursing. 2007;33(4):367-71.
5. MU SJ, JIANG Cm, WU Cg. The interrelationships between SDG and process hazard analysis. Acta Simulata
Systematica Sinica. 2003;15(10):1381-84.
6. Chiozza ML, Ponzetti C. FMEA: a model for reducing medical errors. Clinica Chimica Acta. 2009;404(1):75-8.
7. Alimohammadi I, Adl J. The comparison of safety level in kilns in two gypsum production factories by Failure modes
and effects Analysis (FMEA). Iran Occupational Health. 2008;5(1):77-83.[Persian]
8. Ashley L, Armitage G, Neary M, et al. A practical guide to failure mode and effects analysis in health care: Making
the most of the team and its meetings. The Joint Commission Journal on Quality and Patient Safety. 2010;36(8):
351-8.
9. Mohammadian M, Hashemi Nejad N. Hazard Analysis in Poly Van House Factory Using FMEA Method. Journal of
Jiroft University of Medical Sciences. 2014;1(1):49-58. [Persian]
10. KHoshakhlagh A, Halvani Gh, Mehrparvar A, et al. Effectiveness of control measures to reduce the level of risk
the dangers of working in the ceramics industry. tebe kar yazd. 2013;5(2):51-9. [Persian]
11. Pareek PK, Nandikolmath TV, Gowda P. FMEA Implementation in A Foundry In Bangalore To Improve Quality
And Reliability.Mechanical Engineering and Robotics Search.2012;1(2):82-87.
12. Liu HT, Tsai Yl. A fuzzy risk assessment approach for occupational hazards in the construction industry. Safety
science. 2012;50(4):1067-78.
13. Liao CJ, Ho CC. Risk management for outsourcing biomedical waste disposal–Using the failure mode and effects
analysis. Waste management. 2014;34(7):1324-9.
14. Bonfant G, Belfanti P, Paternoster G, et al. Clinical risk analysis with failure mode and effect analysis (FMEA)
model in a dialysis unit. JN journal of nephrology. 2010;23(1):111-118.

92