Safety Analysis of different industries using Fuzzy AHP

In recent days, we march towards a new occupational health and safety era in which work cultures are directed towards positive safety values. It is predicted that the safety analysis techniques now in place are quite difficult to address the potential risks which weakens the era. A novel approach of analyzing different crucial criteria in different industrial sectors is analyzed carefully in this paper. In this unique approach, fuzzy AHP(Analytic Hierarchy Process) technique is applied to determine the respective weights of three main criteria and seventeen sub-criteria as a way of enriching the decision making process while in a dilemma. A survey was initiated in different industrial sectors to obtain reliable data for the research. The results shows that the main criteria „human safety‟ acquired a weight of 72.5% while the respective weights of main criteria machine safety and work environment safety falls to 8.9% and 18.4%. The weight of the main criteria, human safety indicates that the sub-criteria such as eye protection, manual lifting, material handling practices, fire fighting drills, training and safety officers are implemented to a greater extent in most of the surveyed industries.


INTRODUCTION
Are we safe at the working places? Many a time"s industrial activities are performed in challenging and critical environments. Workers who are exposed to potential hazards at their working places are at the risks of occupational injuries and illness1, 2). Over the centuries we have been a long way from industrial revolution. Undoubtedly, these growths of industries in and around India confirm that we are heading towards our economic excellence. But the occurrence of occupational accidents appears to be an alarming issue in the industries today.
As per the statistical year book, India 2016, in the year 2012-13, there are 222,120 factories in India with a total of 10,051,626 workers3). In the year 2013, 955 factory/machine accidents are reported4). As per the report of ILO (International Labor Organization), the occupational health and safety have been improved in industries over the past 20 to 30 years. But these statistics are comparatively imprecise in developing countries because of the gaps in accident identification, reporting and records. These accidents pave way for moral, legal and financial disputes in an industry. Hence the need for improvement in industrial safety is realized at this point.
Britain Standards Institute defined risk as a combination of occurrence and results of a hazardous event). In order to demonstrate the need for this paper, some of the post major occupational accidents in India are described which includes, Bhopal Gas Tragedy took place at Bhopal in 1984 due to the leak of the toxic Methyl Iso Cyanate (MIC) resulting in 558,125 injuries including 38,478 temporary partial injuries and approximately 3,900 severely and permanently disabling injuries. A huge fire broke out at leather factory at Kolkata on November 22, 2006 which resulted in 10 fatalities and 18 injuries. A Boiler Explosion took place at a tyre-melting unit in Coimbatore on February 12, 2016. As a result of it, six workers suffered severe burns all over the body. High severity level of occupational accidents is observed in workplaces where more than 50 employees are engaged6). The lessons learned from these kinds of accidents are not considered for improving safety performance. This is also one of the reasons for the recurrence of occupational accidents7).
The following studies on occupational accidents should be mentioned: Efthimia K. Mihailidou.et.al. 2012 in his article recorded 319 major industrial accidents all over the world8). Valeria Casson Moreno.et.al. 2016 developed a database containing information about 167 accidents in bio-gas plants. The authors concluded that there is a need for up gradation and implementation of safety standards, safety culture and to promote awareness on risk reduction9). Romina Process (ANP) and Fuzzy Linguistic approach) for risk analysis to evaluate the safety performance of hot environment in foundry industry33). Among these techniques Fuzzy AHP is one of the simple method and easy to use. Thomas Saaty first applied the fuzzy AHP method for solving problems containing multi decision criterion34) Application of fuzzy AHP in different fields: AHP is used for addressing multi criterion vague problems that may be either qualitative or quantitative. Metin Dağdeviren.et.al. 2008 proposed fuzzy AHP for a real manufacturing company for determining the Faulty Behaviour Risk (FBR) in work system. He weighted faulty behaviour with triangular fuzzy numbers through pair wise comparisons and evaluated the factors using fuzzy linguistic variables. He concluded that this is the paramount way through which faulty behaviour is prevented and safety of work system is improved37). Guozhong Zheng.et.al. 2012 applied trapezoidal Fuzzy AHP for hot and humid environments with the criteria work, environment and workers for identifying the workers performance38). Zeyang Song.et.al. 2014 employed trapezoidal and the triangular extent fuzzy AHP methods for identifying the early warning system for self-ignition risks in coal piles. In a comparison of trapezoidal fuzzy AHP with triangular extent fuzzy AHP, the authors concluded that the triangular fuzzy AHP is more reliable for evaluation of self-ignition risks of coal39 It can be realized from the above discussions that fuzzy AHP has wider scope of influence in safety analysis. Hence it can be recommended as an alternative for existing safety analysis techniques. However it is important to note that fuzzy AHP cannot be used as a substitute for risk analysis techniques. It is true from the above discussions that AHP could be applied for analysis of safety in industries. Hardly few researches focus on application of fuzzy AHP in safety analysis. Unfortunately, of those few, no gathered literatures have hands on analysis of safety through evaluation of crucial criteria in different industries. Hence in this paper, three main criteria and seventeen sub-criteria are proposed for analysis of different industries through fuzzy AHP. Fuzzy AHP is used for estimating and ranking the respective weights of proposed criteria. The reliable data for the work is obtained through a questionnaire survey. The final judgments are arrived based on the rank of the individual criteria. Most exclusively, this paper explains the present status of defined criteria in various industries through the leverage of data from survey and knowledge of industrial experts.

MOTIVATION FOR RESEARCH
Owing to large manpower, resources and good economic conditions India seems to be a best place for investing and starting industries. Due to these reasons, business people from all over the world are attracted towards India. And also, the industrial revolution has contributed to large number of industries all over India. However these industries play a vital role in contribution to India"s economy, it has been observed that there are large number of occupational accidents in these industries. Some of them are listed: Bhopal Gas Tragedy which took place at Bhopal on 1984 due to the leak of the toxic Methyl Iso Cyanate (MIC) resulting in 1984 558,125 injuries, including 38,478 temporary partial injuries and approximately 3,900 severely and permanently disabling injuries. A huge fire broke out at leather factory at Kolkata on November 22, 2006 which resulted in 10 fatalities and 18 injuries. The fire broke when the hydrocarbon and wielding gas came into contact and soon after triggered an explosion at IPCL plant on 06 June, 2008. Four people were killed and 46 others were injured. A chimney collapse occurred on 23 September, 2009 in a construction under contract for the Bharat Aluminium Co Ltd (BALCO) killing 45 people. Two explosions broke out at Ankleshwar-based chemical dye manufacturing unit on Tuesday 06 January, 2009 which killed three workers and severely injured two other. The entire unit has been damaged in the explosion. A fire broke out at Indian Oil Corporation on 29 October 2009 in an oil depot tank. The depot fire raged for 11 days, 12 people were killed and over 200 were injured resulted in losses worth INR 2.80 billion and during the period half a million people were evacuated from the area. A huge fire broke out at a pharmaceutical company in Andhra on December 19, 2011 and spread to neighboring factories. More than six workers had been injured in the fire. A fire accident broke out at a private thermal power plant in Tuticorin on 15 August 2011. Four employees were killed and six were severely injured. On 5 September 2012 an explosion broke out at a fireworks factory in Sivakasi. 40 people were killed and more than 70 were injured. People killed included factory workers and local villagers who walked in after the initial fire. An industrial incident at Ambuja Cement"s plant due to a fly-ash hopper situated on the fifth floor was allegedly overloaded during a maintenance operation and collapsed and crashed four floors below. There was a huge blast in the reactor at a pharma unit on September 28, 2015 and the factory was engulfed in smoke. Two workers were killed and five wounded. Of the five wounded, the condition of one worker was said to be critical. A demand of compensation of INR 30 lakh a piece was given to the families of the dead workers. A Boiler Explosion took place in a tyre-melting unit at Coimbatore on February 12, 2016. As a result of it, six workers suffered severe burns all over the body. One of the most important phases in health and safety is the assessment of risks in an industry. It can be observed from the above case studies from the period 1984-2016, though lot of safety management systems and risk assessment techniques are in practice, the occurrence of occupational accidents goes on continuing. It can be realized that these systems are insufficient in addressing the risks entirely.

FUZZY AHP
Which one to select? Or which one is best among a set of alternatives? Solutions for these questions are obtained through fuzzy AHP. Despite going in-depth, literal points related to fuzzy AHP mechanisms are briefed. Fuzzy AHP is one of the most widely used technique in numerous research papers as it provides solid advice for solving Multi Criteria Decision Problems (MCDM -linguistic variables that are vague and uncertain)46). Prof. Thomas L. Saaty first introduced AHP (1970) for solving MCDM problems and it seems to be effortless. Saaty defined AHP as a method of "measurement through pair wise comparisons and relies on the judgments of experts to derive priority scales" 47,48,49,50).
Through the application of simple maths, AHP yields both quantitative and qualitative results. AHP involves dividing problems into hierarchy of criteria followed by calculation of their respective weights. Then based on the weights obtained, the criteria are compared and ranked through pair-wise comparison. Finally, decisions are obtained based on the rank of the criteria. AHP Algorithm (Rosaria de F. S. M. Russo, Roberto Camanho, 2015) the sequence of steps involved in AHP is stated as51): Step 1: Definition of Problem: (i) Initially the problem to be analyzed is chosen.
(ii) The corresponding criteria and sub-criteria relevant to the problem are identified through any data collection methods.
Step 2: Organizing the Decision Hierarchy: (i) The Hierarchical structure consists of three stages: The sub-criteria are defined in relation to the main criteria as shown in the figure 1.
Step 3: Building Comparison Matrix: Comparison is built as follows: (i) The criteria in stage II is compared with the sub-criteria in the stage III respectively i.e. each criterion is compared with all the sub-criteria irrespective of the criteria it is defined with. (ii) A matrix is developed for the each and every criterion in stage II. (iii) A rating scale is defined with qualitative and quantitative data as shown in table 1.
(iv) In case of criteria in the column is preferred to the criteria in the row, then the inverse of the rating is defined i.e. if row is preferred than to column, row is rated at the exact rating as defined in the scale or else if column is preferred than to rows, the inverse of the rating value is considered. The lower triangular matrix is filled by using the reciprocal of the upper diagonal. Let aij is the element if row "i" and column "j", if so, the lower diagonal is defined as:

SAFETY EVALUATION OF THE SYSTEM
The essential evidence for the research is obtained through a questionnaire survey conducted in five different industries (Heavy engineering, automobile, manufacturing, and foundry and textile industry). In the initial step, the fine points about industries in southern part of Tamilnadu are collected and scrutinized. This study yields the result that the above mentioned industries are the major industrial sectors that covers most of the industries in the local area. This motivated the authors to consider these industries. The population involved in the survey includes personnel"s such as casual labours, contractors, technicians, maintenance supervisors, shift supervisors, production managers and safety manager. Table 2 shows the particulars of participants involved in the survey.
A sample filled in questionnaire is shown in Appendix 1. The results of the survey are analyzed by using a team of five experts who have vast industrial experience in the areas of production, maintenance, engineering, quality and safety. The final pair wise comparison matrix is developed based on the decisions of the expert"s team.

EVALUATION OF CRITERIA
Safety atmosphere differs from industries to industries. The selected industries are those which may give the most and least priority to safety in their business. In order to narrow down the areas to be focused in these industries, several criteria were defined initially based on suggestions from a team of experts. It includes three main criteria and seventeen sub-criteria. These criteria are the elements to look at in these industries which are grouped under a label called main criteria namely Human Safety, Machine Safety and Work environment Safety attributes. The grouping is done as follows: 1. Human safety attributes takes interest in sub-criteria such as eye protection, manual lifting, material handling practices, Fire Fighting drills, Training and Safety officer, 2. Machine safety attributes includes sub-criteria such as fencing, revolving parts protection, safe work speed, pressure plant protection, power cut-off devices, 3. Work environment safety attributes includes sub-criteria such as manhole protection, explosion safety, lightening protection, flammable dust prevention, pits, sumps protection and portable light usage. The decision model projecting main criteria and their respective sub-criteria is shown in figure 2.

CASE STUDY
In this paper, application of the proposed model includes evaluation of a real time problem faced by industries today. Heavy engineering industry included in the survey involve in production of construction & mining machineries including compact dump trucks, excavators, backhoe loaders, motor graders, bulldozers and skid steer loaders and industrial machineries employing more than 1600 people. Whereas automobile industry manufactures auto components such as clutch plates, chains and sprockets, fly wheel housing, gear housing, lube oil cooler cover assembly, filter head, air connectors, clutch housing, filtration module casting, turbo charger, compressor cover assembly, fuel pump housing, crank case, cylinder head etc. with a total of more than 1,100 employees. Manufacturing industry referred to in this survey own a business of hand tools, metal forgings, metal stampings etc. with 790 employees working round the clock. Foundry involves in casting of components for textile, automobile, machine tools etc. with manpower of 550 people. Textile industry involves in the business of production of yarn from cotton fibres and poly ester with strength of around 663 workers. The goal of this paper is to rank the criteria and sub-criteria based on their respective weights and to decide on the criteria that still needs improvements. The linguistic terms used for construction of pair wise comparison matrix is shown in the table 3.

AHP WORK OUT
AHP enhances the interpretation of decision making problems. The proposed AHP involves the following steps: Initially the evaluated criteria are disintegrated into a decision hierarchy as shown in Fig 1 which includes the objectives, main criteria and sub-criteria defined under them. Then the pair wise comparison matrix is formed for each main criteria and sub-criteria for determining their respective weights. Through pair wise comparison each main criterion is compared with other main criteria and in a similar way each sub-criterion is compared with the other relevant sub-criteria. Table 4 shows the pair wise comparison matrix of main criteria and sub-criteria. In the next step, the resultant values drawn from pair wise matrices are normalized. The final step is the calculation of the consistency index and consistency ratio (CR < 0.1) to evaluate consistency of the constructed pair wise matrix. Table 5 shows the respective weights and consistency values of main and sub-criteria respectively. Table 12 shows the respective weights of main criteria, sub-criteria and industries. The following data are extracted from the table 12. The respective weights of the three main criteria are Human safety attributes (0.72), Machine safety attributes (0.089), and work environment safety attributes (0.184).
It can be observed from the second column of table 12 that the human safety attribute tops the weights with a weight of 72.5%. It indicates that the sub-criteria of human safety eye protection (0.189), manual lifting (0.442), material handling practices (0.053), fire fighting drills (0.086), training (0.040), and safety officer (0.191) are found to be mostly followed by all industries. In addition to it, it is practical that most of the sub-criteria material handling practices, fire fighting drills, training and safety officer under the main criteria human safety falls under administrative controls and it requires an experienced or competent persons to train the workers which costs low and he may be the safety manager. The respective weights of the machine safety and work environment safety are 8.9% and 18.4%. The weight of machine safety attribute seems to be very low which depicts that the sub-criteria fencing (0.105), revolving parts protection (0.171), safe It can be observed that the management"s shows no or less interest in automation of machines. This may be due to involvement of huge investments in modification or replacement of machines. It shall be appreciable that if employers concentrate on engineering controls and proper work methods for reducing workplace hazards and risks. It is also concrete that sub-criteria under work environment safety manhole protection (0.064), explosion safety (0.350), lightening protection (0.172), flammable dust prevention (0.259), pits, sumps protection (0.050), and portable light usage (0.102) are being done during the erection and commissioning phases of an industry. The low weight of the work environment safety attributes (18.4%) may be due to improper maintenance work done in preserving the conditions of the sub-criteria.
It can be observed from the fourth column of table 12 with respect to human safety attributes that the manual lifting sub-criteria leads with a weight of 44.2%. This may be due to the fact that most of the materials in the surveyed industries are lifted manually. It can be observed that proper trainings are being provided to the workers during their induction period and refreshed periodically for manual lifting activities. The weights of the remaining sub-criteria eye protection, material handling practices, firefighting drills, training, and safety officers are found to be low which may be due to improper control methods for identified hazards and non-availability of competent persons for training the workers.
With respect to machine safety attribute that the sub-criteria pressure plant protection has a maximum weight of 40.2%. This is because of the point that the manufacturer of pressure plants ensures essential safety devices are in-built into the pressure plants before delivering to its customers. However the responsibilities of monitoring and maintenance of the devices falls on the responsibility of management of the individual industries. As already stated the sub-criteria fencing, revolving parts protection, power cut-off devices involves phases of purchasing or fabrication, installation, execution and monitoring costs for all the machineries in a plant which involves huge costs. So, the management shows less importance in implementation of the above sub-criteria.
The sub-criteria explosion safety under work environment safety attribute has an imperative function with a weight of 35%. This could be possible because it can be observed during the survey that the flammables and explosives are stored in an isolated area with essential safety precautions. The respective weights of the remaining sub-criteria under main criteria work environment safety attribute are manhole protection, lightening protection, flammable dust prevention, pits, sumps protection and portable light usage are 6.4%, 17.2%, 25.9%, 5% and 10% respectively. The low weights of sub-criteria manhole protection, lightening protection, pits, sumps protection may be due to inadequate maintenance resources and persons for preserving their conditions. When viewed heavy engineering industry as a separate chapter, it is liable that these industries shows an upper hand while comparing with other industries. This could be possible due to the fact that most of the heavy engineering industries are multinational corporations where they have distinct management systems and guidelines for managing industrial safety. It can be fathomable that the implementation and monitoring of these systems has a positive impact on safety which could be realized through the comparison with different industries. In this survey, it is also likely that the textile industry stands last in most of the sub-criteria. From the analysis and feedbacks from relevant industry experts and workers it can be grasped that the production is seen as a more vital factor than safety in these industries. In addition to the above theme, most of the textile industries falls under small scale segment and hence the attitude is narrowed to profit making i.e. return on investment. Hence there is a need for immediate drift for improving the sub-criteria to cope up with other industries performance. The ultimate implication could be that the heavy engineering and the textile industries has their trails on two opposite extremes in implementation of safety.
Whereas from the analysis it is clear that the foundry positions themselves a step ahead of textile industries. But as similar to textile industries, foundry has to travel a long way for achieving safety excellence. It can be witnessed that the manufacturing has a lower hand in comparison with heavy engineering but shows an upper hand in comparison with automobile. It signifies that the manufacturing industry should concentrate on all the sub-criteria except fencing for continual improvement in safety. In a similar comparison, automobile industry lags behind manufacturing but ranks ahead of foundry. Hence the automobile industry should caution on all the sub-criteria except safe work speed for improving the performance of safety.

CONCLUSION:
Through this novel approach, safety performance of different industries is analyzed using fuzzy AHP to determine the uncertainty in decision making process. The respective weights of main criteria and sub-criteria were calculated based on the data obtained through the survey. These weights are then analyzed to arrive at the rankings of individual criteria and sub-criteria. An innovative approach for analysis of safety criteria in industries is visualized though this paper. This approach will be in place as a guide for the researchers and industry professionals for exact analysis and ranking of safety parameters based on individual priorities.    Very Strong Important (5,7,9) Absolute Important (9,9,10)