Selection of the safety risk analysis technique most compatible with nature, requirements and resources of mining projects using an integrated Folchi-AHP method

There are numerous safety risk analysis techniques. Moreover, no ideal method exists for all companies; hence, the selection of the method most congruous with nature of the intended project, as well as the needs and resources of a mining company is of particular significance. To address the issue, a mathematical model has been developed with the aid of the Folchi-AHP method, whereby safety experts can opt for the best technique after multiplying an impacting factors matrix by a correlation matrix. The former is created by the safety team in the decision-making time, and includes 15 evaluation criteria, while the latter is comprised of the relative weight of each criterion to each technique. To find these weights, 22 methods were compared to each other in terms of 15 criteria by 10 safety experts using the analytic hierarchy process (AHP). To ease computations, an Excel program was developed and investigated in four mining projects.


Introduction
In an estimation by the International Labour Organization (ILO), 2.3 million out of 340 million occupational incidents occurred annually worldwide end in fatalities, i.e. there are 6000 deaths on a daily basis (ILO, 2021). Previous records indicate that various industries' employees are all exposed to work-related accidents in a different way (Dudarev et al., 2013). Owing to its unique working conditions, mining is counted among the most life-threatening professions in the world; therefore, adopting the appropriate precluding measures is imperative (Mijalkovski et al., 2020, Tripathy andAla, 2018). Safety has become a weighty issue within the mining industry, and so has its improvement for the authorities (Bagherpour et al., 2015). The Statistical Center of Iran reports 1775 mining accidents, 46 casualties and a fatality rate of 4.3 in 10000 workers in 2019 (SCI, 2021), while the fatality rate in a country such as the United States is 1.22, i.e. almost one third (MSHA, 2021), setting alarm bells off about safety in the mining sector.
In ISO 8402, safety is defined as a situation, in which the probability of damages to individuals or property has been contained up to an acceptable extent (ISO 8402, 1994). To put it differently, safety is a set of conditions to minimize the danger caused by an accident. The word accident signifies some difficulties in its prediction and prevention; hence, we deal with uncertainty or risk. The risk of a project is an incident posing negative or positive impacts on the project's main goals, including time, cost and quality if it takes place (Larson and Gray, 2015). The science of identification, assessment, control and response to risks in a project's lifetime is referred to as risk management, the centerpiece of which is risk analysis including hazard identification and risk assessment (Aven, 2015). Among dozens of miscellaneous risks in a project, work-related safety and health hazards are of great importance, deserving careful attention (ISO 31000, 2009).
Safety risk analysis determines systematically not only what hazards exist in a worksite, but it also estimates their occurrence and severity. Some techniques are adopted for hazard identification, and others for risk assessment. Of course, a number of methods demonstrate both capabilities. Furthermore, they are categorized into two general groups as quantitative and qualitative, and three subclasses of determinative, probabilistic and hybrid methods (Tixier et al., 2002). On this account, these techniques are varied in terms of approach and application scope, numbering above 100 (Ericson, 2015). For instance, Saat (2009) investigated 150 risk analysis methods for a construction site in Turkey. On the grounds of their high number, and various outputs, steps and functions, taking an accurate and effective approach for the selection of the best technique is necessary since the output of analysis changes with the meth-Rudarsko-geološko-naftni zbornik i autori (The Mining-Geology-Petroleum Engineering Bulletin and the authors) ©, 2022, pp. 43-53, DOI: 10.17794/rgn.2022.3.4 od taken Guneri, 2017, Guneri et al., 2015). Risk analysis should be predicated on the behavior of a system, and it must be compatible with the complexity of the system itself (Foussard and Denis-Remis, 2014, Rasmussen, 1997). The selection of the proper and efficient techniques in risk analysis and assessment is one of the most effective ways to diminish different accidents in mines (Siahuei et al., 2021). With all that being said, it begs the question as to how to choose the best technique with this vast diversity?
A plethora of studies have been conducted on safety analysis in different sectors and industries using only one particular technique, whereas articles considering the comparison of these techniques pale in number. AK (2020) drew a distinction among the analytic hierarchy process (AHP), Fine Kenny, failure mode and effects analysis (FMEA), process hazard analysis (PHA) and the Matrix Method in terms of five criteria in the construction sector, laying emphasis on the merging of multiple criteria decision making methods into conventional risk analysis techniques. Another comparison was made by Koçak (2019) in which four techniques including FMEA, Bowtie Analysis, Job Safety Analysis and the Matrix Method were evaluated by seven criteria in a coal mine using the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS), and the second technique was recognized as the most appropriate one. Shahba et al. (2016) prioritized environmental risk analysis techniques by adoption of the same approach, and FMEA turned out to be the most capable method with regard to six criteria. Guneri et al. (2015) compared three techniques with respect to four criteria by resorting to FAHP in a Turkish company, hallmarking FMEA as the most suitable method. Risk analysis techniques in the construction industry were reviewed and discussed by Sharma and Goyal (2015) suggesting the application of fuzzy logic for more precision in results. In another study, Chemweno et al. (2015) took an analytic network process (ANP) approach to propose a methodology for the selection of a technique among FMEA, fault tree analysis (FTA) and Bayesian Networks based on eight decision criteria in an asset maintenance decision-making domain. Mohamadfam et al. (2015) compared the two methods of management oversight and risk tree (MORT) and Tripod-Beta with respect to seven criteria using AHP in order to identify the root causes of fatal excavation accidents in the construction industry, and finally the latter was found to be superior. Foussard and Denis-Remis (2014) made a comparison among three widely-used risk assessment techniques (PHA, FMEA and Hazard and Operability Analysis (HAZOP)) within the energy sector in terms of symptoms and perspectives. They invited experts to combine these three methods in order to cover all perspectives in the triangulation of definition (functional, genetic and ontological) and avoid confusion before selecting a method in risk workshops.  (2008) first drew a differentiation between hazard analysis types and techniques, and then evaluated twenty two risk techniques from the perspective of eight attributes. Another study was devoted to mere review of the relative strengths and shortcomings of forty hazard identification techniques (Glossop et al., 2000).
However, the aforementioned studies are encountered with four chief drawbacks: 1) some of them have examined only a limited number of risk analysis techniques, 2) some have not considered an adequate number of evaluation criteria, 3) without introducing the best risk analysis technique, they have only classified the methods according to different criteria, 4) should they have come up with the best technique, their calculations have been predicated on the assumption of constant weights or importance for criteria, while the weight of a criterion is relative to the nature of the intended project as well as the needs and resources of the organization in charge. The high fatality rate in the mining industry on the one hand, and the shortcomings of previous studies on the other hand, emphasize the need for an accurate and effective framework for selecting the most appropriate risk analysis method according to a project's specifications, and the needs and resources of the organization implementing the project. The current study endeavors to fill the research gap by proposing a novel selection methodology explained in the next section.

Methodology
A schematic view of the general research procedure is depicted in Figure 1. First of all, a number of methods for the identification and assessment of hazards are selected from prevalent methods. Then, by studying previous research studies and relevant standards, an array of criteria for the evaluation of risk analysis techniques are considered. In the next step, a questionnaire will be designed and forwarded to safety experts in an attempt to gather input data for AHP so that a database is established in the output in which methods are classified according to each criterion. The weighted matrix obtained from AHP will provide the arrangements for the implementation of a method called Folchi. For ease of computation, a software will also be developed in the form of an Excel program. Finally, the performance of the proposed model in this research will be put to the test in four mining projects. The plentitude of Multi-Criteria Decision-Making (MCDM) Methods might beg the question as to why AHP and Folchi methods have been taken advantage of in this research, but it is worth mentioning that the conventional AHP method is not only straightforward but also well-known, rendering its application more convenient for both opinion surveyors and participants (experts). Regarding the Folchi method, it provides an opportunity to manipulate and justify the weights of criteria and alternatives according to the needs and resources of a safety team in the time of decision-making.

Risk Analysis Techniques and Evaluation Criteria
There is a myriad of methods for risk assessment and identifying hazards, of which 22 common methods were selected as follows: Preliminary  According to ISO 31010, in general, the appropriate risk analysis method should have the following characteristics: 1) it should be justifiable and appropriate to the situation or organization under consideration, 2) it should provide results in a form which enhances the understanding of the nature of the risk and how it can be treated, 3) it should be capable of use in a manner that is traceable, repeatable and verifiable (ISO 31010, 2010). Once the decision is made to perform a risk assessment, the following factors should be noted: the objectives of the study, the needs of the decision-makers, the potential magnitude of the consequences, the degree of expertise, human and other resources needed, and the availability of information and data. Other criteria can be sought and According to these research studies, the factors stated in the ISO standard, and after consultation with some safety experts, 15 criteria were selected to evaluate the risk analysis methods in this study. Each criterion has been defined and accompanied by the justification of its choice in Table 1.

AHP Method
Analytic hierarchy process is one of the most powerful multi-criteria decision-making techniques introduced in 1971 by Thomas L. Saaty, and was considerably welcomed by the scientific community (Saaty, 1980). In this method, a decision-making problem is divided into different levels of a goal, criteria, sub-criteria and alternatives. To build a decision model, at the top level is the goal, in the middle level or levels are the criteria, while the bottom level contains possible alternatives, and thus everything is placed to form a hierarchical structure. Experts are then asked through a questionnaire about the pairwise comparison of criteria with each other, and the comparison of each alternative in terms of each criterion to establish pairwise comparison matrices in which the relative weights of the elements are defined. The degree of consistency can also be calculated and judged as ac- ceptable or rejected. The allowable range of the inconsistency rate is less than 0.1 (Saaty, 2001). After receiving completed questionnaires from each of the experts, it is necessary to combine their individual answers to form a group decision-making matrix. Aczél and Saaty (1983) demonstrated that the geometric mean method is the best way to integrate judgments in the AHP. To expedite calculations, Expert Choice 11 software (developed by Expert Choice, Inc. in Arlington, Virginia, USA) was utilized.

Folchi Method
In this study, to select the most appropriate method of risk analysis, the concept of the Folchi method was utilized, which was first introduced by Roberto Folchi to express the environmental effects of an open pit mine in Italy (Folchi, 2003). This method consists of three main parts, including impacting factors (IFs), decision components (DCs) and a correlation matrix. The DCs can be ranked after the multiplication of the IFs by the correla- Applicability of the method to analyze a single task in the studied site Some methods are more suitable for a certain task ( To establish the correlation matrix, firstly, the importance or weight of these techniques in relation to the criteria previously determined by experts in the AHP is expressed in qualitative terms, and then will be converted into a quantitative form by performing some calculations as stated by Folchi. The matrix of evaluation criteria is comprised of magnitudes (numerical values) assigned by decision makers (safety experts) according to the needs and resources of the organization and the nature of the given project. From the product of the criteria matrix and the correlation matrix, risk analysis methods are prioritized, and therefore the best method(s) can be identified according to the intended criteria.
In order to prepare the correlation matrix, it is vital to turn the integrated opinions of experts on the importance of risk analysis methods in terms of each criterion into linguistic terms of Min, Med and Max. The elements of pairwise comparison matrices obtained from the AHP are in the form of decimal numbers in the range of 1 to 9. To convert the numbers to qualitative variables, a classification is implemented, so that the range of 1 to 9 is divided into three equal parts; hence, the ranges [1-3.7], [3.7-6.4] and [6.4-9] will be allocated to the terms of Min, Med and Max, respectively. According to Folchi's instructions, the elements of this matrix are quantified by defining the maximum effect, which is twice the medium effect, and the medium effect, which is twice the minimum effect. Then, the sum of these coefficients for each DC equals 10. To put it another way, the values of X, 2X, and 4X replace the Min, Med, and Max linguistic variables, respectively in the correlation matrix already filled with qualitative terms. The sum of these values in each column must be equal to 10. After solving this simple first-degree equation, the value of x is calculated for each column, and therefore the linguistic variables become quantitative elements. Next, the matrix of IFs is multiplied by the correlation matrix to compute the effect of IFs (evaluation criteria) on each DC (risk analysis method) (Equation 1). For clarification, suppose that the weight of the criterion 'Time' in the 'PHA' technique is 4.3 located in the second range. Therefore, the term 'Med' is entered in the correlation matrix, and then replaced by 2X. After solving an equation in the column of 'PHA', the value of X is calculated. This way, the element of 2X is obtained. The same goes for other elements. (1) Where: E -is a (1 × m) matrix in which each element represents the amount of overall impact on each DC, F -denotes a (1 × n) matrix in which elements represent magnitudes, C -is an (n × m) correlation matrix. The parameters n and m are the number of IFs and DCs, respectively.
The importance or weights of the alternatives in relation to the criteria can be determined using the AHP method, thereby ranking the risk techniques. However, the Folchi method is applied in this study owing to the fact that there is a remarkable distinction between the AHP and the Folchi method. In the former, one has to run the AHP process for every specific project. To put it differently, new questionnaires must be forwarded to experts every time a risk method selection is required, whereas in the latter, a decision maker can change the impact of the weights already designated by experts in the AHP method by assigning different magnitudes in the IFs matrix. In that manner, the correlation matrix remains unchanged, and only the impacting factors determined by decision makers according to the needs and resources of the intended organization are changed in every project, bringing comfort and speed. Hence, the final rankings by both methods will be the same, but it is a matter of convenience.

Questionnaires
The first step in the AHP is to establish a hierarchical structure in which an overview of the goal, criteria, and alternatives is graphically illustrated (see Figure 2). The goal of the AHP in this study is the classification of 22 risk analysis methods with respect to 15 criteria. The next step is the design of survey questionnaires. Here, instead of pair comparisons of criteria and alternatives, another approach will be taken into consideration in order to reduce the number of enquiries. That is to say, firstly, the grades of 1 to 9 are allocated by experts to the least and the most important criterion in that order. Then, other criteria are graded based on these two criteria. The same goes for the grading of the alternatives. Finally, one can calculate the preference of criterion i over criterion j by means of dividing the grade of i by the grade of j so as to achieve the pairwise comparison matrices required to run the AHP. For taking the influence of the personal characteristics of experts in the survey (the qualification level (B.Sc, M.Sc or PhD), discipline, and job experience) into account, experts are asked to compare the importance of these three factors at the beginning of the questionnaire. Then, the opinions of participants will be combined according to the weight obtained for each expert to be used in the formation of the final pairwise comparison matrices.
The structure of the questionnaire was designed so that the hierarchical structure was presented along with the tables related to the definition of criteria and alternatives at the beginning. Following this, experts were requested to complete their personal information before determining the importance of those three factors. The next step is the main part of the survey, in which the experts compare criteria and alternatives with each other. This questionnaire contains 350 enquiries and was sent to 10 experts in the field of health, safety and environment in Iran.

Results and discussion
This section consists of the risk techniques classification, the selection of the most appropriate technique, and the Excel program.

Classification of Risk Analysis Methods
It took experts nearly two weeks to fill out the questionnaires. The number of experts with bachelor's and master's degrees was 6 and 4, respectively. The discipline of industrial safety engineering had the highest frequency among participants. The average job experience was found to be 8.5 years. Experts considered the importance of job experience more than discipline, and the discipline more than the qualification level. Then, the influence of each expert was calculated and, their opinions were integrated using the method of geometric mean. In the next step, these combined grades were divided among each other to create elements of paired comparison matrices as input data for the Expert Choice software.
At the output of the software, risk analysis methods were classified according to various criteria, and a database was established to raise decision makers' awareness about the attributes of risk analysis techniques (see Table 2). It should be noted that the inconsistency rate in the software was turned out to be 0.035, indicating the accuracy and reliability of judgments. Sensitivity analyses made it apparent that these methods showed the highest fluctuations to the criterion of "Quantitative", and the least fluctuations to "Accuracy" and " Extent of evaluation". However, the main purpose of the current research is to create a framework for selecting the most appropriate risk analysis method according to the nature of the project, the needs and resources of the organization, which will be discussed in the next section.

Selection of the Most Appropriate Technique
In the previous section, 22 risk analysis methods were compared and classified in terms of 15 diverse criteria. However, in this study, the main purpose of implementing the AHP was to collect and integrate the opinions of experts to form a correlation matrix needed in the Folchi method. The numerical opinions of experts were replaced by linguistic variables, and then quantified by equaling the sum of each column to 10, as previously explained. The final correlation matrix is shown in Table  3. The criteria matrix is formed by a safety team in the studied project in the time of decision-making, with experts assigning a magnitude between 1 and 9 to each of the 15 evaluation criteria. The magnitudes of 1 and 9 convey the idea that a criterion is barely and highly important in the perspective of the safety experts, respectively. Then, a score is calculated for each risk analysis method by multiplying the criteria matrix by the correlation matrix using Equation 1. The method receiving the highest score is the most consistent method with the nature of a project, the needs and resources of the company conducting the project.

The Excel Program
In order to facilitate the computations of selecting the most appropriate risk analysis method by the Folchi method, a program was written in the Excel 2016 software (developed by Microsoft, Washington, USA), the overview of which is shown in Figure 3. The program file contains two worksheets. The first one contains tables explaining the criteria, risk analysis methods, and definitions of magnitudes ranging from 1 to 9. The sec- ond worksheet consists of three main parts: the correlation matrix, the criteria matrix and an answer section. After entering the magnitudes of the evaluation criteria in the relevant section, the final scores of the risk analysis methods are calculated automatically, and the top 5 methods are marked in green. At the same time, a bar chart of methods' rankings is drawn according to resultant scores. The program's user manual includes the following steps: 1) the perusing of tables related to the description of criteria, risk analysis methods and magnitudes in the first worksheet in the Excel file, 2) holding workshops by the safety team to contemplate and determine the magnitude of evaluation criteria according to the nature of the given project, needs and resources of the organization, 3) entering the magnitudes in the relevant section in the second worksheet, 4) observation of the risk analysis methods' rankings in a bar chart, 5) and review and making the final decision about the top 5 methods proposed by the program.

Case Studies
The proposed model was applied to four mining projects (two stone quarries, a metal mine and a coal mine) to evaluate its performance. To this end, a questionnaire was designed and sent to each mine's safety team, in which they were asked to first identify the topic of the risk analysis project they were working on. In the next step, they were requested to determine which of the 22 risk analysis methods considered in this study is more appropriate for their project based on their experience and knowledge. Finally, at the end of the questionnaire, they assigned a magnitude between 1 and 9 to each of the fifteen evaluation criteria as the input data to the Excel program. Having been run by the author of the present research, the program indicated the most appropriate risk analysis methods for each project in the output.
In Table 4, the proposed-top-five methods by the program have been compared with the proposed method of Only the method proposed by the safety team at the Bama Mine Processing Plant was the same as the method proposed by the Excel file on the grounds that the topic of interest in that mine was process-oriented, and since the HAZOP method is known for analyzing process systems, the safety team of this mine managed to  Total 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 select the technique properly, whereas in other mines, practitioners failed to opt for the best method fitting the nature of the project, the needs and resources of the mining company. It is noteworthy that the method proposed by the team in the Kulikosh Quarry is the fifth method suggested by the program. This point emphasizes the importance and necessity of applying the proposed model in mining projects. Since the approach adopted here incorporates a considerable number of criteria simulataneously, it enjoys more precision than conventional decision making methods. One cannot prescribe one specific technique to all projects. Dozens of contributory factors should be taken into account, and it will not be implemented unless a systematic approach is taken. Moreover, the Excel program needs only some input data to be run, and it doesn't seem to be a daunting task, but rather a tractable one. Thus, it stands to reason to be admitted that not only is the proposed model beneficial but also applicable.

Managerial implications
Having been encompassed with a plethora of miscellaneous risk assessment techniques, safety experts might be inflicted by bafflement upon the technique selection process, thereby failing to engage the appropriate method. Provision of the Excel program by managers will bring benefits to mining companies in terms of financial, accuracy and safety aspects. That is to say, one can take into account multiple factors related to the nature of the given project, and feasible resources of the implement-ing company, and subsequently choose a more suitable technique. For instance, assume that the time and budget of a company for risk assessment is limited. Then, the user of the computer program enters lower magnitudes for these criteria, and this way time-consuming and expensive methods are faded into insignificance; thus, a more congruous result will be obtained. Managerial implications can be summarized as follows: • the importance of drawing a distinction between various projects in terms of essence and available resources and needs; • the requirement of a systematic framework for the selection of the best risk assessment technique; • a greater level of safety can be realized using the novel approach proposed at a low budget. No particular financial investment is required, except for the trivial cost of the software.

Conclusion
Each project or organization has specific characteristics that are different from another project or organization; therefore, choosing the most compatible method with the nature of the project, needs and resources of the organization in the time of decision-making from hundreds of available methods is of paramount importance, as well as an arduous task. To address this bottleneck, a decision-making support model was introduced by combining the AHP and Folchi methods enabling the safety team in mining projects to select the most appropriate method based on the project's specifications and availa- Milling and flotation equipment hazards [9,5,1,9,8,7,5,7,2,3,7,7,6,5,7] HAZOP HAZOP LOPA OSHA SEIFT MORT

Tabas Coal Mine
Explosion in underground tunnels [5,9,4,7,5,6,3,4,3,4,5,7,5,3,8] FTA RCA CCA FMEA FTA HAZAN ble resources. A user-friendly program was developed in Excel software facilitating the computational process. The suggested model in this research was evaluated in four mining projects, presenting a successful performance. The most significant advantages of this model can be enumerated as reduction in costs, increased speed and accuracy in the process of methodology selection by safety practitioners in mining projects.