A hybrid risk assessment method for mining sector based on QFD,fuzzy inference system,and AHP

Abstract

Conventional risk assessment methods are widely used for industrial safety applications. However, there are serious obstacles to their usage as; (i) all of the potential hazards are considered as an independent event, (ii) various risks are identified based on these hazards, (iii) risk magnitudes of these risks are obtained without considering interdependencies among the hazards, and then (iv) the protective measures against the defined risks are taken based on these risk magnitudes. Therefore, conventional methods do not provide any assessment for overall risks in the working environment. Furthermore, although an accident may cause different severity such as loss of working days, loss of limbs, occupational disease, and death, the conventional methods do not consider all potential consequences of any accident, simultaneously. The main objective of this paper is to propose an effective risk assessment approach by using the fuzzy set theory, Analytical Hierarchy Process (AHP), Fuzzy Inference System (FIS), and Quality Function Deployment (QFD) methods to quantify the risk of any hazard considering interdependencies among all potential hazards and consequences in working environment. Within the scope of this research, an application in the mining sector has been presented to illustrate the validation and the effectiveness of the proposed approach ** .

Keywords

Risk assessment mining sector QFD fuzzy logic AHP

1 Introduction

The magnitude of any risk is quantified by the combination of the probability of loss arising from potential dangers and the severity caused by these losses on the employ and the system in terms of occupational health and safety. It is possible to create a safe working environment by eliminating or minimizing potential risk factors. Therefore, occupational health and safety specialists aim to provide a safe operation through various methods that eliminate the adversities. Various analytical methods have been identified to assess the qualitative and quantitative data obtained through measurement and observation [1].

The fundamental goal of occupational health and safety practices is to prevent workplace accidents and occupational diseases. Any operations may cause workplace accidents producing both significant material and immaterial losses including injury, loss of limbs, death, and as well as occupational diseases. These losses are mostly experienced in the mining and construction sectors in Turkey and in the rest of the world. According to the data provided by the International Labor Organization (ILO), mining workers are 3–6 times more at risk of accidents than the workers in other fields [2]. In Turkey, the mining sector is also one of the worst sectors in terms of the fatal and permanently incapacitated accidents. According to the Social Security Institution (SSI) statistics in Turkey, around 9% of all work-place accidents have occurred in the mining sector. Furthermore, 18% of these accidents are permanently incapacitating accidents and 28% of them are fatal work accidents. Therefore, the mining sector is one of the three priority sectors identified by the Ministry of Family, Labor, and Social Services in terms of fighting workplace accidents. In addition, according to the Hazard Class Directive, which is prepared by the ministry, the mining sector is in the “Very Hazardous” category [3]. Therefore, effective proactive approaches to prevent occupational accidents in the mining sector must be developed and implemented immediately.

Risk management including risk assessment, risk analysis, and risk evaluation is the most important and proactive approach to eliminate workplace accidents. Current approaches used for risk management take the source of the hazard as a point of reference, identify the various risks associated with that source and specify the control measures to be taken to counter the risks involved. Furthermore, many different analytical methods are used for risk analysis and these methods have conducted just only activity-based analysis or risk-based analysis [5, 6]. When the risk analysis methods used in the literature are examined, it can be concluded that “Fine-Kinney” and “Failure Mode and Effect Analysis (FMEA)” are the most used methods in the mining sector. Fine Kinney method calculates risk magnitude by considering probability, severity, and frequency while FMEA provides risk magnitude by considering probability, severity, and detection parameters. Both methods quantify risk as a categorical structure. However, the current method is far from being able to present an assessment of the main risk across the entire operation [4]. Therefore, a novel approach based on Fuzzy Analytical Hierarchy Process (FAHP), Fuzzy Inference System (FIS), and Quality Function Deployment (QFD) methods have been proposed to literature in order to cope with the difficulties given above.

QFD Method which is proposed by Akao [8] in 1966 for the design of ships at Mitsubishi’s Kobe shipyards in the early 1970s [9] is widely used to improve the quality of both the system and products in the industry. The method is actively used to increase customer satisfaction by translating customer expectations or requirements into product and design parameters [7]. The QFD method covers the processes carried out on a schematic called the “House of Quality” including rows for customer requirements, columns for the technical characteristics used to meet those requirements and a matrix in which the degree of relationship between them is identified. In the relationship matrix, the technical characteristics and customer requirements are associated with one another by assigning numerical values to the expressions as: weak, medium and strong. In the correlation matrix, which forms the roof for “The House of Quality”, the degree to which the technical characteristics will affect one another is identified as positive, negative or neutral. This association makes it possible to see how other characteristics and the performance of entire system will be affecting by improving any technical characteristic. Once the matrices are worked out the priorities, those areas that need to be tackled are identified by listing the technical characteristics in order of importance [10].

When the studies in the literature are examined, the QFD approach is used in many different areas such as customer-oriented product design, improving the existing system, and competition analysis. When a study in which the QFD method was used in conjunction with the existing methods for risk analysis and assessment. It was used only for prioritizing purposes by associating risks with precautions. However, the method does not offer an assessment for the overall system [11].

Based on the literature, the conventional methods, used for risk analysis, consider hazards as a reference, identify various risks depending on these hazards, and specify the protective measures to be taken against the determined risks. However, these methods don’t provide any assessment for overall risks in the working environment. Furthermore, these methods do not consider simultaneously all of the possible consequence of an accident although it may cause different severity such as loss of working days, loss of limbs, occupational disease, and death [12]. In the scope of this study The House of Safety approach, inspired by the “The House of Quality (HoS)”, has been proposed in order to analyze potential risks with a possible consequence at different levels of the working environment and evaluate the combined effects of those risks simultaneously. The House of Safety has been developed by embedding the FAHP and FIS into the QFD method. FIS is used to evaluate the importance degrees of the related activity while both FAHP and FIS are utilized together to determine the potential risk of possible consequences. The importance degrees of related activities are obtained based on the workload of the activity, duration of the activity in the process, and interdependencies among activities by using FIS.

The rest of the paper is organized as follows: a brief literature review on the applications of risk analysis methods are given in Section 2. In Section 3, a novel risk assessment approach named “House of Safety” is presented. Section 4 provides an application of occupational health and safety assessment in mining sector. Finally, concluding remarks and suggestions are presented in Section 5.

A part of this study is presented at INFUS 2019 and is given an extended version for the special issue [12].

2 Literature review

Risk assessment techniques are classified into three categories as quantitative (PRAT Technique, DMRA Technique, Risk Measures of Social Risks, QADS, ORA Technique, CREA Method, PEA Method, WRA ... etc.), qualitative (Check Lists, What-If Analysis, Safety Audits, HAZOP, Task Analysis, STEP Technique ... etc.) and hybrid (HEAT/HFEA, FTA, ETA, RBM ... etc.) [13].

The risk assessment matrix, which is one of the most commonly used approaches, has been developed in order to meet the safety requirements of the US military standards. Matrix diagrams are evaluation tools which are used to analyze the relationship between two or more variables. One of the simplest matrix methods is L type (5×5) matrix which is multiplied by ascribing them values using the figures identified for probability and intensity. The assessments are made in line with the resulting scores. X type matrix is more inclusive (probabilities, statistical data, costs of preventions ... etc.) matrix method and it can be done by experts who have at least 5 years of accident investigation experience [14, 15].

One of the most commonly used risk assessment methods is “Fine Kinney Method” which is suitable for all areas of the industry and easily applicable. This method was revealed by Kinney in 1976. Fine-Kinney Method includes scales for probability, intensity, and frequency. There are point intervals identified for every scale and a score is obtained as a result of multiplication by similarly ascribing value, and the assessment is made according to the score intervals [16].

FMEA (Failure Modes and Effects Analysis) is the other method which is the most widely used methods in industrial risk analysis. The basis of the method; total or part of any system is considered; the parts, tools, and components that may occur in these components are examined and the results that can be affected are analyzed. The types of FMEA are System, Design, Process, and Service. The FMEA method uses coefficients for occurrence, severity, and detection [17].

ETA (Event Tree Analysis) and FTA (Fault Tree Analysis) are quantitative weighted hybrid techniques and widely used in risk management. ETA is selected to see where an accident will proceed with operator errors and system disturbances. Logic calculation system is used. It is the main technique used in the result analysis since it shows the conditions before and after the accident. FTA examines error by subcomponent, identifies mechanical, physical, chemical or man-made errors and mechanisms of these errors, schematizes probable sub-events with a logical diagram and is used in conjunction with reliability and probability theorems [18].

HAZOP (Hazard and Operability) and ICCT (International Chemical Control Toolkit) are commonly used risk assessment methodologies of the chemical industry. Both methods are qualitative. HAZOP examines the parameters in chemical processes and performs a comprehensive analysis. ICCT focuses on chemical risk factors and prioritizes by risk levels [13 –19].

Multi-Criteria Decision Making [AHP (Analytic Hierarchy Process), ANP (Analytic Network Process), TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) etc.], Fuzzy Logic and its combinations also have used for risk assessment of safety management in the literature. Multi-Criteria Decision-Making Methods allow the consideration of both objective and subjective factors in selecting the best alternative. Fuzzy logic is a branch of artificial intelligence and product of simulation works on the human brain. It is a mathematical system which is based on the fuzzy set theory. Control system used fuzzy logic is based on the logical expressions and the connections between the mentioned expressions. Fuzzy logic doesn’t need any mathematical model of the system and it can be controlled by using the logical expressions [20–22 , and 31].

3 Proposed method

In this paper, a new risk assessment approach, “House of Safety (HoS)”, has been proposed to determine priorities of the total risks on the working environment and to present importance degrees of the precautions which must be taken. In this section, the calculation process of the proposed approach has been explained step by step in detail. The structure of the proposed method is given in Fig. 1.

Fig. 1

Structure of the proposed approach.

Phase 1. Establishment of risk assessment team:

In this step, the members of the risk assessment team are determined and the risk assessment process is planned. In the planning stage, the types of data to be collected, data sources used in the study, and team members’ assignments and responsibilities are identified. In this step, importance degree of each member in the team is equal to each other since it is assumed that they have the same expertise.

Phase 2. The identification of the activities and their importance degrees:

At the work environment, there are various activities involving different tasks and responsibilities that affect project success. Among these, workload of any activity, duration rate in the total project time, and interdependencies among the activities are key factors on success. Therefore, in order to determine importance degrees of the activities conducted in the work area, the workload of the related activity, the duration of each activity within the predicted total project time, and interdependencies of the activities have been considered, simultaneously. In this phase, Mamdani fuzzy inference mechanism [23, 24] is used to calculate importance degrees. The model characterizes a rule-based system and the general structure of the rule-base used in the model is given in the following equation. The rules for the inference mechanism are presented in Table 1.

Table 1

Fuzzy rule base which is used to determine risk magnitude

$\begin{matrix} if x_{1} = Z_{i 1} and x_{2} = Z_{i 2} and x_{3} = Z_{i 3} and \dots \\ x_{n} = Z_{in} THEN y = P_{i} . i = 1, 2, 3, \dots, k \end{matrix}$ (1)

In the equation, x_n (n = 1, 2, 3, … m) expresses the input data set, Z_iandP_i are the linguistic expression defined by the membership functions, the y presents the output value, and k symbolize the number of the rules utilized in the rule-base. If multiple discrete rules in the system are activated at the same time, a MAX-MIN operator is used to obtain the result. The MAX-MIN operator is given by the following equation [25].

$\begin{matrix} μ_{Pk} (y) = maks [\min [μ_{Z 1 k} (x_{1}), μ_{Z 2 k} (x_{2})]], \\ k = 1, 2, 3, . . n \end{matrix}$ (2)

The μ_Pk, μ_Z1k, μ_Z2k given in the equation are the degrees of membership for the output y, and the inputs x₁ x₂, and x₃, respectively.

The fuzzy output value obtained from the model needs to be defuzzified. The centroid or area or center of gravity (COG) method, which is widely used in practice, will be used in this study for the defuzzification procedure.

The equation for this method is as follows [26]: $Z_{COZ}^{*} = \frac{\int_{z} μ_{X} (x) xdx}{\int_{z} μ_{Z} (x) dx}$ (3) $Z_{COG}^{*} = \frac{\sum_{i}^{q} μ_{Z} (x_{i}) x_{i}}{\sum_{i}^{q} μ_{A} (x_{i})} i = 1, 2, 3, \dots, q$ (4)

The $Z_{COG}^{*}$ given in the equation shows the exact value obtained from the system.

Detailed information for the rules of the inference mechanism used in our study is given in Table 1.

Step 2.1. Identify Vulnerability level (v): Occupational health and safety specialists first identify the vulnerability levels for the corresponding activities in the work site. The vulnerability level for an activity is the value that expresses the safety threat level for the activity. In this step, threat levels can be constructed using the pairwise comparison method. These levels can also be directly presented by the specialists as in a linguistic form. When there are more than one specialist in the risk assessment process, the specialists’ preferences are pulled one by one and arithmetic mean method is used to obtain co-decision matrix on the preferences.

Step 2.2. Determine the activity rate (ar): The next step is the calculation of the activity rate. The activity rate is defined as the proportion of the activity’s time in relation to the entire project’s time. It is predicted from the work plan at the beginning of the project. The following equation is used to obtain the activity rate $(\tilde{r_{i}})$ : $\tilde{r_{i}} = \frac{{\tilde{t}}_{i}}{{\tilde{T}}_{p}} = \frac{{\tilde{t}}_{i}}{\sum_{i; = 1}^{n} {\tilde{t}}_{ai}} i = 1, 2, 3, \dots, m$ (5) where, ${\tilde{T}}_{p}$ and ${\tilde{t}}_{i}$ expresses the predicted total project duration and total duration of the i^th activity within the project and $\tilde{r_{i}}$ for its part gives the i activity’s rate within the overall project duration. The reason why the numbers here are fuzzy is due to the possibility of the activities and thus the entire project being delayed due to workplace accidents, supply delays and adverse weather. Generally speaking, the durations planned in many projects overrun to various degrees. Therefore, in order to consider the uncertainties in the activity durations, the values calculated in our studies will be expressed as fuzzy values.

Step 2.3. Calculate the degree of membership relating to vulnerability, activity rate, and interdependencies: If the data obtained for threat and activity durations is expressed in fuzzy numbers, they will not be able to be used directly in the Mamdani inference system. The rule base given in Table 1. is made up of linguistic expressions and the degree of membership of these linguistic expressions for the obtained data must be calculated. Therefore, the chart given in Fig. 2 is used to calculate degrees of membership. The intersection point for the membership functions given in Fig. 2 with fuzzy figures gives us degrees of membership.

Table 2

Potential severities caused by risk

Abbr.	Description
LWT	No-loss of working time
LWD	Loss of working days
LWW	Loss of working weeks
LL	Loss of limb /Occupational disease
D	Death/Deaths

Fig. 2

Triangular membership functions for linguistic scale.

Step 2.4. Calculate the activity-based risk magnitude: In this step, the degree of activity-based risk formation is calculated by inputting the degree of threat for every activity plus that activity’s duration into the fuzzy inference mechanism. If the general expression given in Equation 1 is written taking into account the parameters given in Table 1 any rule can be expressed in the following way:

$\begin{matrix} if v_{i} = μ_{vi} and {ar}_{i} = μ_{ari} then w_{i}^{a} = μ_{a_{i}} . \\ i = 1, 2, 3, \dots, k \end{matrix}$ (6)

The expression μ_vi, μ_ari, and μ_{a
_i} as given in Equation 6 is the values for threat, activity rate, and degree of membership for the activity’s degree of importance, respectively. Multiple rules can become active depending on the input values in the fuzzy inference mechanism. In this situation, the fuzzy inference mechanism generates results by using the minimum operator as given in Equation 7.

$\begin{matrix} μ_{Ri} (x, y) = μ_{vi} (x_{1}), μ_{ari} (x_{y}), μ_{a_{i}} (y) . \\ i = 1, 2, 3, \dots, k \end{matrix}$ (7)

One single result is obtained depending on the results obtained as a result of the rules that were validated depending on the input values, and this result is used in Equation 8. $μ_{R} (x, y) = ⋁_{i = 1}^{k} μ_{Ri} (x, y) .$ (8)

Since the values obtained are degrees of membership they need to be clarified so that the data can be used in the QFD table. Equation 9 is used for the clarification operation [25]. $W = \frac{\sum_{i = 1}^{q} Y_{i} μ_{R} (y_{i})}{\sum_{i = 1}^{q} μ_{R} (y_{i})}$ (9)

Phase 3. Grading the intensity of the adverse outcome that will be caused by the risk (Hazard Meter):

When an accident occurs, it can result in various negative outcomes such as permanent incapacitation, loss of limb and even death. In other words, accidents can result in different outcomes each having different probabilities. The probable outcomes as a result of any risk are given in Table 2.

Conventional methods only base the intensity parameter on the most likely outcome. In a departure from the literature in our study, we work out the intensity value by taking the possibilities of happening for all the adverse outcomes into account. The steps to be followed when calculating intensity are:

Step 3.1. Calculate the intensity values: The intensity values for the possible outcomes that could take place in the event that a risk identified with a place of work emerges are calculated using the pairwise comparison method. The Buckley AHP method is used for this calculation. The steps for this method are as follows [27 –29]:

Let the matrix showing the pairwise comparisons be ${\tilde{C}}_{i}$ ${\tilde{C}}_{i} = [\begin{matrix} \begin{matrix} 1 & {\tilde{c}}_{12} \\ {\tilde{c}}_{21} & 1 \end{matrix} \begin{matrix} \dots & {\tilde{c}}_{1 n} \\ \dots & {\tilde{c}}_{2 n} \end{matrix} \\ \begin{matrix} ⋮ & ⋮ \\ {\tilde{c}}_{n 1} & {\tilde{c}}_{n 2} \end{matrix} \begin{matrix} ⋮ & ⋮ \\ \dots & 1 \end{matrix} \end{matrix}]$ (10) $i = 1, 2, 3, \dots, l ve n = 1, 2, 3, \dots, p$

Here ${\tilde{C}}_{i}$ is the pairwise comparison matrix for the nth expert relating to the possible outcomes that will result if the relevant risk occurs. In the matrix, the value shown as ${\tilde{c}}_{ij}$ is the linguistic expression that shows the probability of i outcome happening when compared to j outcome. The l and p given in the equation respectively show the number of experts and the possible outcomes if the risk occurs. The linguistic expressions used in the review were adapted from the study by Hsieh et al. [29] and the scale is given in Table 3. Just as reviews can be made separately by every single expert, so a matrix can also be created with the experts’ joint decision.

Table 3

Linguistic scale for comparison matrix [29]

Linguistic scales	Abbreviation	Fuzzy numbers
Equally important	(Eq)	(1,1,3)
Weakly important	(Wk)	(1,3,5)
Essentially important	(Es)	(3,5,7)
Very strongly important	(Vs)	(5,7,9)
Absolutely important	(Ab)	(7,9,9)

The fuzzy weights are calculated with the aid of the equations given below. ${\tilde{r}}_{i} = ({\tilde{c}}_{i 1} \otimes {\tilde{c}}_{i 2} \otimes \dots \otimes {\tilde{c}}_{in})^{1 / n}$ (11) ${\tilde{w}}_{i} = {\tilde{r}}_{i} \otimes ({\tilde{r}}_{1} + {\tilde{r}}_{2} + \dots + {\tilde{r}}_{n})^{- 1}$ (12)

In Equation 11, ${\tilde{r}}_{i}$ shows the calculated geometric average result of the i outcome that could happen as a result of risk compared to other outcomes and ${\tilde{w}}_{i}$ for its part shows the occurrence weight of the i class outcome. Equations (11) and (12) are repeated for the outcome of each identified risk and the values for the risk intensities are calculated. If there are multiple experts in the review process and if their expert judgments are not gathered by forming a consensus, pairwise comparison matrices for each expert are generated. Afterward, a joint decision matrix is obtained by calculating a geometric average. Finally, the occurrence weight values for the outcomes that will happen because of the relevant risks will be determined by applying Equations (11) and (12). Since the value for every form of risk outcome is fuzzy, Equation 13 is used to clarify the fuzzy values. $w_{i} = \frac{{\tilde{w}}_{i}}{\sum_{j = 1}^{n} {\tilde{w}}_{j}} = \frac{(w_{il} + w_{im} + w_{iu})}{\sum_{j = 1}^{n} {\tilde{w}}_{j}}$ (13)

Step 3.2. Calculate the resultant risk intensity for the risk outcome: In this step, the degree of intensity for the relevant risk is calculated taking the intensity values for every possible outcome for each risk into account. The scale is given in Fig. 3 and the equation given below is used for the resultant risk intensity [25]. $RS = \frac{(\sum_{i = 1}^{q} w_{i} \times μ (y_{i}))}{\sum_{i = 1}^{q} μ (y_{i})}$ (14)

Fig. 3

Linguistic scale for activity weight (w).

The RS value given in Equation 14 shows the risk intensity, while the μ (y_i) value shows the central point of the linguistic expression given in Fig. 3.

Phase 4. The House of Safety

In this stage, a schema is created for the “House of Safety” approach risk assessment in which the weighted activities are associated with the risks. The activities whose weights are known are evaluated together with the likely risks using the relationship matrix. At this phase, the total weight of the risks is also determined. “The House of Safety” is given in Fig. 4 schematically.

Fig. 4

Calculates of The House of Safety (HoS).

The numerical comparisons of the degrees of relationship in the relationship matrices are categorized as strong relationship (9), medium relationship (3) and weak relationship (1). In calculating the degrees of importance, the absolute degrees of importance are worked out by multiplying every relationship degree by every weight and adding them together.

The relationships in the correlation matrix that makes up the roof of the Safety House are associated as positive (+) or negative (-). The presence of a negative relationship states that the measures to be taken in order to prevent risks cause to cancel each other out. For this, a special effort should be taken to provide the relevant precautions.

4 Application

In this section, the proposed method is applied to evaluate the risks of a mining project. An open pit mine operation was selected for the application of the proposed method. Then, the activity and risks to be weighted were determined based on the literature survey and investigation of industrial applications. The activities for project risk assessment are given in Table 4 by specifying the possible risks and the measures to be taken. To minimize the subjective differences, a risk assessment team consisting of two B-Class and one A-Class work safety expert was created. The risk assessment team evaluated every single parameter respectively, and the results were interpreted according to these data.

Table 4
Activities and Possible Risks

ACTIVITIES RISKS

A1 Excavation R1 Hurt by falling materials.

A2 Drilling R2 Hurt by falling from high up.

A3 Transportation R3 Hurt by falling from low height.

A4 Detonating R4 Hurt by being pierced by pointed object.

A5 Loading R5 Hurt by getting a limb trapped.

A6 Wall Work R6 Hurt by flying debris.

A7 Drainage R7 Hurt by collision.

A8 Mineral Processing R8 Hurt by electrocution.

A9 Fine Work R9 Hurt by explosion.

A10 Mechanical Work R10 Hurt by fire.

A11 Electrical Work R11 Hurt by collision with a vehicle.

A12 Ground Work R12 Hurt by exposure to loud noise.

A13 Support Services R13 Hurt by exposure to vibration.

R14 Hurt by exposure to air flow.

R15 Hurt by exposure to dust.

R16 Hurt by exposure to chemicals.

R17 Hurt by exposure to radiation.

R18 Hurt by infectious diseases.

R19 Hurt by lightning strike.

R20 Hurt due to ergonomic problems.

ACTIVITIES	RISKS
A1	Excavation	R1	Hurt by falling materials.
A2	Drilling	R2	Hurt by falling from high up.
A3	Transportation	R3	Hurt by falling from low height.
A4	Detonating	R4	Hurt by being pierced by pointed object.
A5	Loading	R5	Hurt by getting a limb trapped.
A6	Wall Work	R6	Hurt by flying debris.
A7	Drainage	R7	Hurt by collision.
A8	Mineral Processing	R8	Hurt by electrocution.
A9	Fine Work	R9	Hurt by explosion.
A10	Mechanical Work	R10	Hurt by fire.
A11	Electrical Work	R11	Hurt by collision with a vehicle.
A12	Ground Work	R12	Hurt by exposure to loud noise.
A13	Support Services	R13	Hurt by exposure to vibration.
		R14	Hurt by exposure to air flow.
		R15	Hurt by exposure to dust.
		R16	Hurt by exposure to chemicals.
		R17	Hurt by exposure to radiation.
		R18	Hurt by infectious diseases.
		R19	Hurt by lightning strike.
		R20	Hurt due to ergonomic problems.

Phase 1. At this stage, the threat level for each activity is defined by the risk assessment. The threat levels defined for the activities are described in Table 5. First, the linguistic expressions are turned into fuzzy numbers by using Fig. 3 and then a common decision matrix is obtained by taking the arithmetic averages (Table 5).

Table 5

Occupational Health and Safety Specialists’preferences

Activities	S1	S2	S3	S1	S2	S3	Aggregated value
A1	M	H	H	(0.25,0.5,0.75)	(0.5,0.75,1)	(0.5,0.75,1)	(0.42,0.67,0.92)
A2	VH	H	VH	(0.75,1,1)	(0.5,0.75,1)	(0.75,1,1)	(0.67,0.92,1)
A3	VH	H	H	(0.75,1,1)	(0.5,0.75,1)	(0.5,0.75,1)	(0.58,0.83,1)
A4	H	H	H	(0.5,0.75,1)	(0.5,0.75,1)	(0.5,0.75,1)	(0.5,0.75,1)
A5	H	H	H	(0.5,0.75,1)	(0.5,0.75,1)	(0.5,0.75,1)	(0.5,0.75,1)
A6	VH	VH	VH	(0.75,1,1)	(0.75,1,1)	(0.75,1,1)	(0.75,1,1)
A7	VH	VH	VH	(0.75,1,1)	(0.75,1,1)	(0.75,1,1)	(0.75,1,1)
A8	VH	VH	VH	(0.75,1,1)	(0.75,1,1)	(0.75,1,1)	(0.75,1,1)
A9	H	H	VH	(0.5,0.75,1)	(0.5,0.75,1)	(0.75,1,1)	(0.58,0.83,1)
A10	H	VH	VH	(0.5,0.75,1)	(0.75,1,1)	(0.75,1,1)	(0.67,0.92,1)
A11	H	VH	VH	(0.5,0.75,1)	(0.75,1,1)	(0.75,1,1)	(0.67,0.92,1)
A12	M	M	M	(0.25,0.5,0.75)	(0.25,0.5,0.75)	(0.25,0.5,0.75)	(0.25,0.5,0.75)
A13	L	VL	VL	(0,0,0.25)	(0,0,0.25)	(0,0,0.25)	(0,0,0.25)

The estimated project duration and completion periods for the activities are given in Table 6. The estimated values for the activities are obtained using Equation 5. Since the durations are estimates, the fuzzy durations in which the uncertainties in the activity periods can be considered are given in Table 6.

Table 6

Fuzzfied activity durations

Activities	${\tilde{t}}_{i}$	$\tilde{r_{i}}$	TFNs
A1	600	0.55	(0.44,0.55,0.68)
A2	350	0.32	(0.25,0.32,0.4)
A3	350	0.32	(0.25,0.32,0.4)
A4	200	0.18	(0.15,0.18,0.23)
A5	150	0.14	(0.11,0.14,0.17)
A6	300	0.27	(0.22,0.27,0.34)
A7	150	0.14	(0.11,0.14,0.17)
A8	500	0.45	(0.36,0.45,0.57)
A9	750	0.68	(0.55,0.68,0.85)
A10	900	0.82	(0.65,0.82,1.02)
A11	1000	0.91	(0.73,0.91,1.14)
A12	500	0.45	(0.36,0.45,0.57)
A13	1100	1	(0.8,1,1.25)
${\tilde{T}}_{p}$	1100

The values in Tables 5 and 6 must be converted into degrees of membership by using Fig. 3 in order to place them in the fuzzy inference mechanism. The degrees of membership for the linguistic classes calculated to assess both duration and threat are given in Table 7. When the degrees of membership obtained for the threats and durations were entered into the system the rules expressed in Equation 6 and given in Table 1 become active and the degree of importance for the activities is calculated together with the activated rules. The degrees of membership corresponding to the linguistic scale given in Fig. 3 and used to represent the degree of importance are calculated. Afterward, the activities’ degrees of importance in forming the foundation for risk are obtained using Equation 3. The results of the calculations are given in Table 7.

Table 7

Threats, Activity Duration, Importance Degree of Membership and Activity Weights

	v					ar					w				W
	VL	L	M	H	VH	VL	L	M	H	VH	VL	L	M	H
A1	0.00	0.17	0.67	0.84	0.33	0.00	0.17	0.93	0.47	0.00	0.17	0.67	0.84	0.33	5.99
A2	0.00	0.00	0.17	0.67	0.75	0.00	0.78	0.45	0.00	0.00	0.17	0.75	0.45	0.00	4.61
A3	0.00	0.00	0.34	0.83	0.60	0.00	0.78	0.45	0.00	0.00	0.34	0.78	0.45	0.00	4.21
A4	0.00	0.00	0.50	1.00	0.50	0.37	0.76	0.00	0.00	0.00	0.50	0.76	0.00	0.00	2.81
A5	0.00	0.00	0.50	1.00	0.50	0.50	0.60	0.00	0.00	0.00	0.50	0.60	0.00	0.00	2.64
A6	0.00	0.00	0.00	0.50	1.00	0.11	0.94	0.26	0.00	0.00	0.11	0.94	0.26	0.00	4.34
A7	0.00	0.00	0.00	0.50	1.00	0.50	0.60	0.00	0.00	0.00	0.50	0.60	0.00	0.00	2.64
A8	0.00	0.00	0.00	0.50	1.00	0.00	0.40	0.86	0.18	0.00	0.00	0.40	0.86	0.18	6.54
A9	0.00	0.00	0.34	0.83	0.60	0.00	0.00	0.54	0.84	0.25	0.00	0.34	0.83	0.60	7.44
A10	0.00	0.00	0.17	0.67	0.75	0.00	0.00	0.24	0.84	0.60	0.00	0.17	0.67	0.75	8.09
A11	0.00	0.00	0.17	0.67	0.75	0.00	0.00	0.06	0.63	0.81	0.00	0.06	0.63	0.75	8.44
A12	0.00	0.50	1.00	0.50	0.00	0.00	0.40	0.87	0.19	0.00	0.40	0.87	0.50	0.00	4.17
A13	1.00	0.50	0.00	0.00	0.00	0.00	0.00	0.00	0.40	1.00	1.00	0.50	0.00	0.00	2.00

Phase 2. In this step, the potential outcomes that could happen in the event that any risk occurs are assessed by the risk assessment team using the pairwise comparison method. The possible outcomes as the result of any risk were given earlier in Table 2. By way of example, the pairwise comparison values for the possible outcomes of the risk of being hurt by falling objects are given in Table 8. When equations 11 –13 are applied to the data in Table 8 the values 0.45, 0.32, 0.08, 0.03, and 0.12 are obtained for LWT, LWD, LWW, LL, and D respectively. Thus, in the case of falling materials the weight for a loss-free working day is 45%, for short-term working day loss 32%, for long-term working day loss 3%, for loss of limb 3% and for death 12%. Equation 14 is used to obtain a single intensity value that combines all these outcomes.; Using equation 14, RS is obtained as 3.103. $\begin{matrix} RS = \frac{(\sum_{i = 1}^{q} w_{i} \times μ (y_{i}))}{\sum_{i = 1}^{q} μ (y_{i})} = \frac{0.45 \times 1 + 0.32 \times 2.5 + 0.08 \times 5 + 0.03 \times 7.5 + 0.12 \times 10}{0.45 + 0.32 + 0.08 + 0.03 + 0.12} = 3.103 \end{matrix}$

Table 8

Damage Due to Material Fall Parameters

	LWT	LWD	LWW	LL	D
LWT	(1,1,1)	(1,1,3)	(5,7,9)	(7,9,9)	(3,5,7)
LWD	(0.33,1,1)	(1,1,1)	(3,5,7)	(7,9,9)	(1,3,5)
LWW	(0.11,0.14,0.2)	(0.14,0.2,0.33)	(1,1,1)	(3,5,7)	(1,1,0.33)
LL	(0.11,0.11,0.14)	(0.11,0.11,0.14)	(0.14,0.2,0.33)	(1,1,1)	(0.2,0.14,0.11)
D	(0.14,0.2,0.33)	(0.2,0.33,1)	(3,1,1)	(9,7,5)	(1,1,1)

The fuzzy weights, net weight and intensity values obtained for all the assessments pertaining to all risks are given in Table 9.

Table 9

Fuzzy Weights of Risks, Net Weight and Severity Values

Risk	w _i					RS
	LWT	LWD	LWW	LL	D
R1	0.45	0.32	0.08	0.03	0.12	3.10
R2	0.04	0.05	0.27	0.10	0.55	7.74
R3	0.44	0.36	0.13	0.04	0.03	2.60
R4	0.39	0.33	0.18	0.07	0.03	2.93
R5	0.33	0.28	0.20	0.15	0.04	3.52
R6	0.35	0.30	0.22	0.09	0.04	3.27
R7	0.40	0.30	0.22	0.03	0.06	3.02
R8	0.09	0.03	0.28	0.16	0.45	7.21
R9	0.04	0.05	0.15	0.29	0.48	7.84
R10	0.07	0.08	0.30	0.10	0.45	7.04
R11	0.08	0.09	0.31	0.13	0.40	6.80
R12	0.60	0.05	0.04	0.32	1.00	3.29
R13	0.60	0.29	0.12			1.90
R14	0.60	0.29	0.12			1.90
R15	0.48	0.26	0.13	0.05	0.09	3.02
R16	0.34	0.29	0.20	0.06	0.12	3.71
R17	0.51	0.19	0.10	0.09	0.11	3.31
R18	0.05	0.34	0.40	0.06	0.14	4.79
R19	0.09	0.11	0.80			4.37
R20	0.60	0.29	0.12			1.90

Phase 3. “The House of Safety” in which the weighted risks and the activities are associated is given in Fig. 5.

Fig. 5

Calculates of The House of Safety (HoS).

As a result of the activities and the risks whose weighted degrees of importance as associated by the “The House of Safety” method were identified, the risks were ranked as: “ R2 >R8 >R10 >R1 >R11 >R6 >R5 >R7 >R9 >R4 >R12 >R15 >R3 >R16 >R20 >R18 >R14 >R19 >R17 >R13” (Fig. 5). Prioritization for this listing was carried out by taking the relative importance scores as a baseline. Accordingly, the most important risk element equating to R2 was identified as “Being hurt by falling from high up” while the other important risks were identified as R8: “Being hurt by electrocution”, R10: “Being hurt by fire” and R1: “Being hurt by falling objects” respectively. R13, R17, and R19, corresponding to “Hurt by exposure to vibration”, “Hurt by exposure to radiation” and “Hurt by a lightning strike”, were identified as low probability risks. When the statistics are examined this prioritization fully matches the mining sector in terms of likely risks.

5 Conclusion

In the literature, there are lots of methods obtaining risk magnitude based on severity and probability parameters. In addition to these parameters frequency and detectability parameters are used in Fine-Kinney and FMEA methods, respectively. Furthermore, Fine-Kinney Method and FMEA are also the most widely used methods for risk analysis and risk assessment at the mining sector. Although these methods well work to provide risk magnitude for each risk factor, individually, these methods are far from providing overall risk at mining site including the interdependencies among activities. Furthermore, conventional risk analysis methods consider just potential severity of any risk factor. The potential consequence of any risk factor are varies including different possibilities. For this, in this paper, a novel approach including QFD, fuzzy AHP, and fuzzy inference systems in order to cope with the deficiencies defined above. In this perspective, “House of Safety” has been structured for the analysis. The main advantages and the originalities of the proposed approach has been summarized as follows;

The proposed approach considers the interdependencies among the activities at working environment while calculating risk magnitude. However, the conventional risk analysis methods cannot consider interdependencies.

The proposed approach calculates risk magnitudes based on all potential consequences of any risk. However, the conventional risk analysis methods consider only the most possible consequence.

The proposed approach includes fuzzy inference system to provide better reflect human inference system in the decision process. Hence, this provides sensitivity on the obtained result. However, the conventional risk assessment methods work based on categorical data set. Therefore, these methods are far from the sensitivity.

When the conventional methods are applied, since there are too many analysts’ preferences, meaning subjective factors, it is possible to obtain very different results when the same assessment is made by different people. However, since the subjective factors are reduced to a minimum by making a detailed association in the “The House of Safety” method that is being proposed, it is possible to calculate real risk weights.

In this paper, the proposed approach has been applied to mining sector in order to illustrate the applicability of the method. Since the proposed method is not unique to any specific sector, it can easily be applied for any sector in order to analyze risks at working environment. For the further study, the cost analysis of the control measurements can be added to proposed approach. In addition, the proposed approach can be extended by using extensions of ordinary fuzzy sets such as intuitionistic fuzzy, hesitant fuzzy, type-2 fuzzy, Pythagorean fuzzy set, or neutrosophic sets.

References

Ozkılıc

, Occupational Health and Safety and Risk Assessment Methods, TISK, Istanbul, 2005.

ILO (International Labour Office), http://laborsta.ilo.org. [23.07.2019]

Ceylan

, Analysis of Occupational Accidents According to The Sectors in Turkey, Gazi University Journal of Science 25(4) (2012), 909–918.

Covello

V.T.

and Merkhofer

M.W.

, Risk Assessment Methods: Approaches for Assessing Health and Environmental Risks; LLC, ABD, 2013.

Gurcanlı

G.E.

, Bilir

and Sevim

, Activity based risk assessment and safety cost estimation for residential building construction projects, Safety Science 80 (2015), 1–12.

Souza

, Almeida

N.M.

and Dias

L.A.

, Risk-based management of occupational safety and health in the construction industry – Part 2: Quantitative model, Safety Science 74 (2015), 184–194.

Erginel

M.N.

, A model of Failure Mode and Effect Analysis, Journal of Industrial Engineering 15(3) (2002), 17–26.

Costa

A.I.A.

, Dekker

and Jongen

W.M.F.

, Quality Function Deployment in the Food Industry: A Reviev, Trends in Food Science & Technology (November), (2001), 306–314.

Besterfıeld

D.H.

, Total QualityManagement. Prentice-Hall Inc., 2nd Edition., Upper River-New Jersey USA, 1999.

10.

Bazaati

, Bayrmi

and Oral

, Quality Function Deployment and an Application of Construction Sector, Cukurova University Journal of Engineering 29(1) (2014), 53–61.

11.

Toptanci

and Erginel

, Risk Assessment for a Construction Company with Failure Mode And Effects Analysis and Quality Function Deployment, Journal of Engineering Sciences and Design 5 (2017), 189–199.

12.

Çınar

and Çebi

, A Hybrid Risk Assessment Method for Mining Sector Based on QFD, Fuzzy Inference System, and AHP, INFUS 2019, Advances in Intelligent Systems and Computing (AISC 1029), Springer, (2020), pp. 1198–1207.

13.

Marhavilas

P.K.

, Koulouriotis

and Gemeni

, Risk analysis and assessment methodologies in the work sites: On a review, classification and comparative study of the scientific literature of the period 2000-2009, Journal of Loss Prevention in the Process Industries 24 (2011), 477–523.

14.

Akpinar

and Cakmakkaya

B.Y.

, Employers’ Risk Assessment Obligation in Terms of Occupational Health and Safety, Journal of Labor and Society (1) 2014.

15.

Seber

, How to do Risk Analysis in Occupational Health and Safety?, Journal of Electrical Engineering (445) 2012.

16.

Kinney

G.F.

, Practical Risk Analysis for Safety Management, Calhoun: The NPS Institutional Archive, China Lake, CA: Naval Weapons Center, 1976.

17.

Xiao

, Huang

H.Z.

, Lİ

, He

and Jin

, Multiple failure modes analysis and weighted risk priority number evaluation in FMEA, Engineering Failure Analysis 18(4) (2011), 1162–1170.

18.

Ferdous

, Khan

, Sadiq

, Amyotte

and Veitch

, Fault and Event Tree Analyses for Process Systems Risk Analysis: Uncertainty Handling Formulations, Risk Analysis 31(1) (2011), 86–107.

19.

ILO (International Labour Office), “International Chemical Control Toolkit”, Programme on Safety and Health at Work and the Envıronment, Safework,

20.

Silvestri

, De Felice

and Petrillo

, Multi-Criteria Risk Analysis to Improve Safety in Manufacturing Systems, International Journal of Production Research 50(17), 2012.

21.

Gurcanli

E.G.

and Mungen

, An Occupational Safety Risk Analysis Method at Construction Sites Using Fuzzy Sets, International Journal of Industrial Ergonomics 39(2) (2009), 371–387.

22.

Mehdevari

, Shahriar

and Esfahanipour

, Human Health and Safety Risks Management in Underground Coal Mines Using Fuzzy TOPSIS, Science of The Total Environment 488-489 (2014), 85–99.

23.

Mamdani

E.H.

, Application of fuzzy logic to approximate reasoning using linguistic synthesis, IEEE Transactions on Computers 26(12) (1977), 1182–1191.

24.

Mamdani

E.H.

and Assilian

, An experiment in linguistic synthesis with a fuzzy logic controller, International Journal of Man-machine Studies 7 (1975), 1–13.

25.

Zeng

, An

and Smith

N.J.

, Application of a Fuzzy Basen Decision Making Methodology to Construction Project Risk Assessment, International Journal of Project Management 25 (2007), 589–600.

26.

Ross

T.J.

, Fuzzy Logic Engineering Applications, John Wiley & Sons, Ltd, USA, 2004.

27.

Buckley

J.J.

, Ranking alternatives using fuzzy numbers, Fuzzy Sets Systems 15(1) (1985), 21–31.

28.

Chen

, ve Hwang

, and C. L., Fuzzy Multi Attribute Decision Making: Methods and Applications, Lecture Notes in Economics and Mathematical Systems, Springer-Verlag, New York, 1992.

29.

Hsieh

T.Y.

, Lu

S.T.

and Tzeng

G.T.

, Fuzzy MCDM Approach for Planning and Design Tenders Selection in Public Office Buildings, International Journal of Project Management 22 (2004), 573–584.

30.

Ilbahar

, Karasan

, Cebi

and Kahraman

, A novel approach to risk assessment for occupational health and safety using pythagorean fuzzy AHP & fuzzy inference system, Saf Sci 103 (2018), 124–136.

31.

Karasan

, Ilbahar

, Cebi

and Kahraman

, A new risk assessment approach: Safety and Critical Effect Analysis (SCEA) and its extension with Pythagorean fuzzy sets, Safety Science 108 (2018), 173–187.