Decision Support System for Mining Machinery Risk Mitigation Driven by Ergonomics and Contextual Theory

Abstract

Despite being very old, the mining industry continues to be one of the major sources of pollution, with more people killed or injured than in all other industries. Prevention of incidents/accidents on machinery in mining pits and the issues of operator safety on mining machinery largely depend on the ergonomic adaptation of the workplace, compliance with safety procedures and policies, and organizational and other influential factors. Evidently, scarce consideration of those factors in the available literature has not given satisfactory results till now. The aim of this paper is to first set up a comprehensive model based on ergonomic factors and contextual theory, which takes into account all the influencing factors on the occurrence of incidents/accidents and represents a complex system of interdependence of influential variables of diverse, mostly stochastic nature, and then design a software solution on the given basis. In this research, based on the extensive data collected, a model was generated using the structural equations modelling methodology, which was then used to design the reasoning logic in the expert system for mitigating the risks of the operation of mining machines. An innovative solution incorporating a mathematical model of the interdependence of influential variables into the stored knowledge base offers a decision support system that provides recommendations for the maintenance of a particular mining machine, depending on the assessment of model factors in a specific decision-making situation at the higher organizational level and ergonomic suitability for the operator at the lower organizational level, and, in that manner, enables the mitigating of risky/unwanted events.

1. Introduction

Mining machines most often work in very complex working conditions, with a dynamic nature and simultaneous involvement of numerous sources of risk, as well as with the logical consequence of high rates of work injuries and fatalities. Accidents resulting from the operation of mining machinery account for 47% of all fatalities in the mining industry [1]. Although there has been a continuous decrease in the number of fatalities in the mining industry, the rate of fatal injuries per 100,000 workers in mining is 16.7, which is more than four times the average rate for all industries of 4.1 [2]. In addition, nowadays, there is an increasing market need for mining machines on the international market with a growth rate of 4% per year [3]. Also, in accordance with the EU strategy until 2030, the performance of industrial safety should be gradually improved by 25% (compared to the base year) in terms of reducing the number of accidents at work, occupational diseases, accidents with environmental consequences, and losses in production, which imposes the necessity of intensive research in the area with the aim of enabling safer operation of the rapidly growing number of mining machines in an ecologically acceptable environment [3,4].

However, accidents and/or incidents in mining pits during the operation of mining machines are still relatively frequent unplanned and unwanted events [1]. Research that considers ergonomic and/or contextual factors in the field of the mining industry is extremely rare (e.g., [5,6]). Hence, this paper outlines possible patterns of failure in the search for solutions to the problems of the mining industry towards context-dependent and ergonomic factors and towards a decision support system based on them.

Therefore, the goal of this work is to set up and check the model of the interdependence of contextual factors (including ergonomic ones), which have an impact on the prevention of accidents and/or incidents and the improvement of the quality of working conditions. The set goal will be achieved in a way that an integrated, analytical–programmatic solution will be offered in the form of a decision support system for mitigating the risks of mining machinery operation.

According to the literature, the most common isolated factors that contribute to the occurrence of accidents and/or incidents during the operation of mining machinery are the following [7,8,9,10,11,12,13]:

• Human factor (human error);
• Environmental factors (influence of natural forces, emergency situations, storms, earthquakes);
• Organizational factors;
• Factors originating from the mining machine, which most often depend on the applied maintenance procedures.

Over 80% of all adverse events involving mining machinery are caused by human error [7]. Namely, the human factor is the key and the most complex cause, considering the various conscious or unconscious behaviours of employees when operating mining machines and the connection between the human and organizational factors [9]. According to rare previous research [8,14,15,16], some of the causes of adverse events caused by the human factor in mining machinery include the physical condition of the driver, overloading the vehicle with cargo, “blind” spots, aggressive driving, non-observance of safety procedures, inadequate distance from other vehicles, fatigue, stress, mental effort due to working in a confined space, underground space, dust, various gases, polluted air, visibility (at daylight, when working on foot, when working in the conditions of reduced visibility). Mosey [17] has stated that with the human element, complex systems are almost an organic process. Also, Mosey [17] has noticed that unlike machines, where outputs are dependent on and proportional to inputs, people can accept, ignore, challenge, or misunderstand the inputs that are fed into the system; inputs and outputs are interdependent, and their exact relationships are unclear. Kahneman’s [18] main idea was to showcase how the brain uses these two systems for thinking and decision-making processes: system 1 operates intuitively and automatically; meanwhile, system 2 uses problem-solving and concentration—we use it to think slowly. This Kahneman research [18] is a valuable contribution to understanding the decision-making process in the operation of mining machines.

Levenson [19,20] asserted that traditional event-based accident and risk models are particularly poor at dealing with human error and decision-making and emphasizes that mental models play a significant role, namely the ability to adapt mental models through experience in interacting with the operating system is what makes the human operator so valuable. Levenson [19,20] pointed out that the sophisticated models of causality based on systems thinking and systems theory present an opportunity to perform more robust accident analysis and, therefore, learn from events. Komljenovic et al. [21] have indicated that organizational factors have not been sufficiently investigated and that it is essential to treat management as a risk control system and incorporate organizational aspects into studies of low-level occurrences. Fa et al. [9] identified the following organizational factors that are also the causes of adverse events when working with mining machinery:

• Resource management;
• Organizational climate;
• Organizational processes, which are further divided into 10 lower-level factors.

Additionally, it is known that modern organization theory requires an analysis of contextual dependence. With the presence of an increasingly strong trend of quantifying the influence of contingent factors and the growing complexity of research with the same goal, empirical verification and theoretical enrichment of the contextual theory have been ongoing for decades [22]. The expectation is that by mastering the effect of contextual factors, the overall effectiveness and efficiency could be increased, which is extremely important for the mining industry.

Brocal et al. [23] have also confirmed that industrial and occupational risks are strongly interlinked, with the human factor being the main link between them. The same authors [24] have suggested a theoretical framework for new and emerging occupational risk modelling. It is especially useful for industrial processes where the risk model is calculated by the formula R = (SR, C, E, CO, L). A risk (R) is a structure consisting of five components: the source of risk (SR), causes (C), events (E), consequences (CO), and the likelihood (L). From this point, the authors have developed the technique to identify and characterize the new and emerging risks through five conditions (C1–C5). The problem of determining uncertainties is also one part of the complexity of the risk calculation process. Some researchers suggest complex models for determining uncertainties, which can take the form of risk formulas, while others, for risk calculation, take the likelihood to be the same as for past events [25]. However, Dekker et al. [26] noticed that accidents should be seen as complex phenomena, and investigations about them should focus on gathering multiple narratives from different perspectives instead of looking for causes of accidents.

One of the most agreed-upon concepts nowadays is the Industry 4.0 concept, which opens numerous possibilities and advantages. The concept of I4.0 is a certain way to collect and explore data using tools such as machine learning algorithms, artificial intelligence, and neural networks. For complex systems, such as the open pit machinery operator, the concept I4.0 is a further solution to be explored. Open pit mining machinery can be tested by the multi-sensor measuring environment, unlike the underground mining machinery, due to lack of internet connection in all latter cases.

Some studies that focused on the analysis of the ergonomic adjustment of the workplace of the mining machines operator point to the actuality of the research [22,27,28] but also to insufficiently researched factors of ergonomic adaptation. Horbery et al. [28] have indicated how the human factor and ergonomics contribute to risk reduction in mining practice. Dempsey et al. [29] argued that the advantage of the application of smart technologies in cooperation with the application of ergonomic solutions enables significant progress in the near future in the field of risk reduction in mining. Also, the studies of Haas et al. [30] and Spasojević-Brkić and others [6] refer to the recommendations for measurements and research of safety climate and ergonomic adjustment in the mining sector in order to increase safety and health at work while indicating the need for a simultaneous increase in efficiency and productivity. Evidently, previous research had most often covered an individual or possibly a narrow range of contextual and ergonomic factors.

Industrial systems are complex socio-technical systems, so it is important to look into the ways in which the operator–machine interaction affects industrial safety as a category of contextual factors. However, the literature that is currently available in academic publications shows that there has been little research conducted on the subject and that the fact has had very little impact on technical standards and norms [22]. However, despite the challenges and complexity, it is acknowledged that the regulation’s criteria will only provide satisfactory outcomes if the needs of users of industrial machinery and equipment are taken into consideration. Therefore, the identification and setting of a methodology for quantifying the impact of the significant operator–machine interaction factors on industrial safety with a focus on mining machinery is of primary importance for this industrial sector [31].

Similar to the need to investigate the influence of a group of significant contextual factors of operator–machine interaction, i.e., the human factor on industrial safety, the limited literature in the field [32] also indicates the need to research the impact of contextual factors of an organizational nature on industrial safety. The importance of effective assessment of the impact of the management system was recognized for the first time in 1990 in the document API’s Recommended Practice 750—Management of Process Hazards (API’s Recommended Practice 750, Management of Process Hazards), related to the maintenance of process equipment and assessment of its integrity, with the indication that focusing only on technical factors is not enough for an adequate risk assessment. Two years later, in 1992, OSHA published the document “Process Safety Management of Highly Hazardous Chemicals”, which also recognized the importance of the influence of the management factor but without proposing a methodological framework. Also, in API 581 [33], intended for the petrochemical industry, it was indicated that the generic plant failure frequencies should be corrected for differences in the process safety management system. However, despite the importance of safety in the context of mining machines, the literature that considers the influence of organizational factors has, to date, been limited. The maintenance model settings in other, more or less complex technical systems do exist, but it should be emphasized that they also do not take into account the importance of organizational factors. A kind of review in the field is presented [34,35] provide an overview of maintenance models and failure analysis methodologies for a bucket-wheel excavator. Evidently, the connection between human and organizational factors remains insufficiently researched. Accordingly, there is an evident need to develop a methodology for quantifying the impact of significant organizational and human factors on the industrial safety of mining machinery.

According to the facts given above, there is an evident need to establish an innovative solution to the problem of mitigating the risk of the operation of mining machines in a way that includes as many interdependent contextual factors as possible. In this sense, the goal of this research paper is to create a model of interdependence of the contextual factors affecting the occurrence of human error, which originate from the ergonomic maladjustment of the workplace to the needs of mining machinery operators and an inadequate organizational/safety climate in the company. The model is created in order to prevent accidents and/or incidents and improve quality working conditions, and, on its basis, an integrated analytical tool and a software solution will be developed that makes up a system for mitigating the risks of mining machinery operations.

The designed integrated analytical tool, as a decision support system aimed at mitigating the risks of the operation of mining machinery, is directed at efficiently implementing the improvements in terms of the following:

• Raising the level of safety and health at work;
• Raising the level of productivity;
• Reductions in maintenance costs.

After following the recommendations of the system, an innovative integrated analytical tool for a complete description and analysis of the problem of risk management of mining machines and the selection of an adequate maintenance strategy can have a great impact on the future development of all industrial branches in which mining machines work. Additionally, the software solution has the goal of creating the application of the designed tool that is understandable, applicable, practical to use, and web-based in order to enable better overall performance of mining machines and enable greater satisfaction of all stakeholders in the mining industry.

2. Applied Methods

2.1. Research Framework Setting

The setting of the theoretical model of the innovative integrated analytical tool for the complete description and analysis of the problem of risk management of mining machines is based on the analysis of ergonomic and contextual factors, including the defined human, technical, and organizational factors, significant for the assessment of workplace’s risks. In the development of the theoretical model of an ergonomic and contextual adaptive system for mitigating the risks of the operation of mining machines, the following was carried out:

• Identification, evaluation, and ranking of all impact factors specific to different conditions of the context from the point of view of the risk of interruption to the work process, safety and health of working employees, and impact on the near and far environment of the technical system;
• Determining the measures that should be taken to reduce significant risks, that is, to reduce the probability and/or consequences of those events, with acceptable costs.

For each element of the man–machine system (in the specific case of the operator–manager–mining machine), the impact, i.e., the importance in terms of performing the task of the system, was assessed, i.e., fulfilling its functional requirements. In particular, critical elements or parts of machines whose failure leads to an unacceptable level of consequences for safety, health, and the environment and a significant level of economic consequences, especially in terms of large financial and other company losses due to direct and indirect costs arising as a result of incidents during work operations, have been identified. Also, these failures could lead to damage to materials, as well as other equipment, entailing unplanned costs of frequent equipment repairs and consequent reductions in productivity, disruption of the material transport plan, increased workload of other equipment, increased downtime, increased frequency of failure, as well as maintenance costs with rarely available spare parts.

The first phase of the research was the data collection process, which was carried out in the field. Data were collected from the production and technical documentation of the company. In addition, a survey was conducted using a questionnaire and, if necessary, employees were additionally interviewed on a sufficiently large sample of mining machines.

The prepared data will be processed by the application of the structural equation modelling technique as a new paradigm of multivariate analysis of a confirmatory character. Through systems of equations, it successfully describes one-way or two-way influences of manifest and latent variables on each other.

Interpretation of statistically valid results of the new theoretical framework for a contextual adaptive system for risk mitigation of mining machines in industrial enterprises led to making conclusions and setting up a new theoretical framework oriented towards practical use. The backbone of the applicability of the theoretical framework oriented towards the practical application of the solution refers to the storage of knowledge in the decision support system in order to preserve the conclusions of the analysis, shortening the analysis time in order to mitigate the risk, which prevents an unwanted event and raises the quality of the work process within the framework of improving safety and health at work.

2.2. Methodology

The design methodology of the decision support system for mining machinery risk mitigation is driven by ergonomics and contextual theory. In accordance with the previously presented methodological steps, it implies the following steps:

• Generation of structured questionnaires;
• Data collection, which also includes the use of software to support survey processes, data storage, and processing based on questionnaires;
• Modelling with structural equations;
• Development of a contextual adaptive decision support system for mitigating the risk of mining machine operation;
• Testing and implementation of the decision support system.

2.3. Instruments

The first step towards developing a decision support system for mitigating the risk of the operation of mining machines was establishing a theoretical and methodological framework of ergonomic and contextual factors important for risk assessment of the operation and maintenance of mining machines. This primarily implied the development of a measuring instrument for the purposes of data collection.

In order to develop a measuring instrument (Measuring instrument; Table S1, Supplementary Materials), we performed a literature review, focusing on a number of terms and concepts related to occupational safety and health (OHS). In particular, the literature on specific factors such as technical, human, organizational, and sustainability factors was used only to identify the factors to be measured for the assessment of workplace safety climate in mining companies. To this end, prominent theoretical models of workplace safety were consulted to identify relevant factors [1,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66].

The developed measurement instrument enables the identification and quantification of all technical, human, and organizational factors and factors related to sustainability as predictors. As such, it represents an innovative base for setting up a decision support system for mining machinery risk mitigation driven by contextual theory. All data collected by means of the measuring instrument are saved in a Microsoft Excel table for further analyses and preparation for use in the software for structural equation modelling.

2.4. Sampling and Data Collecting

On the basis of the set measuring instrument (Measuring instrument; Table S1, Supplementary Materials), we developed a survey questionnaire aimed at ergonomics, i.e., the state of working conditions, and a survey questionnaire for organizational factors of importance for safety and health at work, with a five-level verbal–numerical scale (where grade 1 indicates a negative statement regarding the question asked and grade 5 is the maximum favourable grade).

A total of 149 respondents filled out the questionnaire for occupational safety and health at work. All 149 respondents are employees in several open pit mines in Serbia and Montenegro. Some of the respondents were operators of mining machines, some were auxiliary workers, and some of them were also managers. In the sample of 149 respondents, there were 65 operators on mining machines who could answer the questions about the ergonomic adjustment of the mining machine cabin to their needs. Therefore, we asked these 65 operators to fill out the questionnaire on the state of working conditions. Only for the operators on mining machinery, age, height, weight, and work experience were collected.

In the Supplementary Materials, there are both survey questionnaires (Survey S1: Survey for operators on minining machinery; Survey S2: Survey on the state of safety and health at work) which respondents filled out in paper form. In these survey questionnaires, there were no empty fields because the respondents were told that answers must be provided to all questions.

The sampling procedure for collecting the mining machine downtime data was conducted at the same open pit mines where operators (the excavator operator, damper operator, etc.) were employed. During the research period, the auxiliary worker at the open pit mine manually wrote down on paper all records about time in downtime, the cause of downtime, and type of downtime, as well as the type of mining machines. This was the only way to connect the operator assessment of working conditions and the causes and types of downtimes. By setting up the databases, we linked the database on the operator workplace and machine downtimes that occurred in that particular operator workplace. Some of the recorded mining machine downtimes had zero duration in downtime but were included in further analyses. Some of the recorded downtimes were caused by factors that were not related to our investigations and were, hence, excluded from the sample.

3. Results

Table 1. Descriptive statistics of the research sample on working conditions.

	Operators
	Age (Years)	Height (cm)	Weight (kg)	Work Experience (Years)
Sample size	65	65	65	65
Minimum	19.000	166.000	60.000	1.000
Maximum	54.000	190.000	150.000	38.000
Range	35.000	24.000	90.000	37.000
Median	35.000	180.000	90.000	9.000
Mean value	34.846	179.415	91.092	10.631
Sample Variance	74.776	31.843	277.161	93.864
Standard Error	8.647	5.643	16.648	9.688

shows descriptive statistics for operators on mining machines.

The key results of the survey questionnaire on the working conditions are summarized in the following way.

When it comes to the questions about the possibility of adjusting the seat, the operators responded as follows:

• A total of 15.38% of mining machine operators answered that the seat was not height-adjustable;
• A total of 10.77% of mining machine operators answered that the seat cannot be adjusted horizontally;
• A total of 9.23% of mining machine operators answered that the seat was not set at the proper height.

When it comes to back support, the operators responded as follows:

• A total of 50.77% of mining machine operators answered that the seat did not have lumbar support, while 15.38% answered that there was no back support at all.

When it comes to arm support, the operators responded as follows:

• A total of 58.46% of mining machine operators stated that there were no armrests;
• A total of 60.00% of mining machine operators expressed dissatisfaction with the adjustability of armrests;
• A total of 56.92% of mining machine operators expressed dissatisfaction with the proper height of armrests.

Additionally, mining machine operators did not unfavourably rate the following:

• Vibrations through the seat, floor, or controls;
• The easily accessible pedal;
• The size of the cabin;
• Noise in the cabin;
• Thresholds and handholds;
• Easy entry and exit from the cabin;
• Visibility in the work area;
• Window reflection in the cabin.

The factor that was rated unfavourably, aside from those listed above, is the following:

• Mining machine operators stated that their view was obstructed by obstacles (70.00% of pit mining machine operators).

Open pit mining machine operators did not have a favourable opinion regarding their motivation by management (38.46% believe that management did not motivate or reward them), while 52.61% of pit mining machine operators believe that failures were caused by factors other than human or organizational factors.

Finally, the key factors that affect the reduction of quality of work conditions for the pit mining machine operators were identified:

• The seat cannot swivel—87.69% of respondents expressed dissatisfaction with this issue;
• Absence from work due to poor working conditions (sick leave)—83.08% of respondents expressed dissatisfaction with this issue;
• Armrests are not adjustable—60.00% of respondents expressed dissatisfaction with this issue;
• There are no armrests—58.46% of respondents expressed dissatisfaction with this issue;
• Armrests are not at the proper height—56.92% of respondents expressed dissatisfaction with this issue.

Table 2. Descriptive statistics of the research sample on occupational safety and health.

	Mean	Standard Error	Median	Standard Deviation	Sample Variance	Range
SC1-1	4.256757	0.071185	4	0.865999	0.749954	4
SC1-2	4.337838	0.066118	5	0.804359	0.646994	3
SC1-3	4.094595	0.068026	4	0.827568	0.684869	3
SC1-4	4.195946	0.074056	4	0.900935	0.811684	4
SC1-5	4.398649	0.071583	5	0.870842	0.758366	4
SC2-1	3.925676	0.068533	4	0.833738	0.695119	3
SC2-2	3.851351	0.074493	4	0.906251	0.821291	4
SC2-3	3.824324	0.080056	4	0.97392	0.94852	4
SC2-4	4.060811	0.077435	4	0.942037	0.887433	4
SC3-1	2.844595	0.088247	3	1.073571	1.152556	4
SC3-2	2.898649	0.093324	3	1.135332	1.288978	4
SC3-3	2.945946	0.083467	3	1.015417	1.031072	4
SC4-1	3.506757	0.088654	4	1.078526	1.163219	4
SC4-2	2.608108	0.108744	2	1.322928	1.750138	4
SC4-3	3.304054	0.089255	3	1.085832	1.179031	4
SC4-4	2.567568	0.103781	3	1.262554	1.594043	4
SC4-5	3.391892	0.086097	3	1.047414	1.097077	4
SC4-6	4	0.080789	4	0.982846	0.965986	4
SC4-7	3.297297	0.088635	3	1.078292	1.162714	4
SC5-1	3.121622	0.094384	3	1.148234	1.318441	4
SC5-2	3.060811	0.089547	3	1.089382	1.186753	4
SC6-1	2.621622	0.091667	2.5	1.115173	1.243611	4
SC6-2	3.689189	0.081836	4	0.995578	0.991175	4
SC6-3	4.141892	0.0614	4	0.746968	0.557961	3
SC6-4	3.831081	0.077531	4	0.943207	0.88964	4
SC7-1	4.256757	0.086356	5	1.050569	1.103696	4
SC7-2	4.47973	0.07205	5	0.876524	0.768294	4
SC7-3	3.878378	0.078425	4	0.954085	0.910278	4
SC8-1	3.722973	0.080116	4	0.974651	0.949945	4
SC8-2	3.783784	0.091232	4	1.109885	1.231844	4
SC8-3	3.344595	0.091319	3	1.11094	1.234188	4
SC9-1	3.837838	0.071141	4	0.865468	0.749035	4
SC9-2	3.925676	0.071165	4	0.86576	0.74954	4
SC9-3	3.331081	0.087028	3	1.058741	1.120932	4
SC9-4	3.256757	0.091021	3	1.107314	1.226145	4
SC10-1	4.141892	0.069808	4	0.84925	0.721226	3
SC10-2	2.952703	0.090114	3	1.09628	1.201829	4
SC10-3	4.060811	0.071894	4	0.874634	0.764984	4
SC10-4	3.945946	0.072883	4	0.886664	0.786174	4
SC11-1	3.5	0.092214	4	1.12183	1.258503	4
SC11-2	3.648649	0.080713	4	0.98191	0.964148	4
SC11-3	3.513514	0.080497	3	0.979285	0.959	4
SC11-4	3.47973	0.096581	4	1.174963	1.380539	4
SC11-5	3.587838	0.08464	4	1.029689	1.060259	4
SC11-6	3.540541	0.09115	4	1.10889	1.229638	4

presents the descriptive statistics for the collected data on occupational safety and health.

According to correlation matrices (Supplementary Materials—Correlation matrices S1: Correlation matrices Ergonomic factors; Correlation matrices S2: Correlation matrices Safety and health at work), the factors which are not correlated are eliminated in future analysis and renamed as Q10–Q21 in Figure 1.

In addition to survey questionnaires, during a period of six months, data on the operation and stoppages of mining machines that were located at different locations—surface mines in Serbia and Montenegro—were recorded. The operation of 14 types of mining machines (excavators, bulldozers, drills, dumpers, combine loaders, loaders, bucket-wheel excavators, etc.) was monitored. For each stoppage/failure, the cause of the stoppage/failure and the time spent in the stoppage/failure were recorded. A total of 9363 downtimes were recorded manually on paper during the investigation period. The recorded data consist of the type of downtime, time in downtime, and cause of downtime. All these data were used for creating the database of machine downtimes.

Table 3. Descriptive statistics for mining machine downtimes.

	Technical Downtimes	Electrical Downtime	Inadequate Management (Misuse)	Organizational Downtime	Level of Danger
Number of downtimes (some of the downtimes had one or more causes at the same time)	8185	706	259	772	9366
Sum (minutes)	259,250	39,510	44,225	57,380	57,647
Average	31.67379	55.96317	170.7529	74.32642	6.154922
Median	5	30	60	30	6
Maximum	1200	480	600	7000	10
Minimum	0	5	5	5	1
Standard error	90.5483	48.16778	218.7768	338.8762	1.580656
Variance	8198.994	2320.135	47,863.29	114,837.1	2.498474

shows descriptive statistics for the collected mining machines’ downtimes. Operator cabins in all machines were similar, but there were some differences regarding the age of the machine.

3.1. Structural Equation Model

The grouped data in the form of three databases (the database resulting from the survey on working conditions, the database resulting from the collection of data on safety and health factors at work, and the database on downtimes/failures of machines), which were described in the previous chapter, the structural equation modelling method was carried out to establish the cause-and-effect relationships between the observed variables (the SPSS AMOS ver 21 software package was used).

Firstly, a correlation matrix was performed for each data group; these matrices are provided in Supplementary Materials. For further analyses, only the variables with significant correlation coefficients were used. Secondly, three individual models were developed in structural equations: one for working conditions (verified in [67]), one for safety and health at work, and one for the machine downtime causes. Finally, all three models were joined into one model.

The construct EO (ergonomic characteristics of the system, operator–workplace) represents the group of constructs that are related to the ergonomic adjustment of operator workplace adjustments. The construct SHW (safety and health at work) represents the group of constructs that are related to the contextual factors originating from the safety climate in the organization. Constructs RM1 and RM2 are related to the risk of possible mining machine failure. The cause-and-effect relation between these four constructs was the subject of plenty of testing and searching for a structure that finally provided satisfactory results (parameters χ², df, GFI, AGFI, NFI, CFI, TLI, RMSEA).

These results are given in Supplementary Materials (Supplementary Materials—Figure S1: Analysis summary, esimates and model fit), where, for the final model, there was significance (p-value) between constructs.

The final structural equation model was obtained by SEM analysis, as shown in Figure 1, where the labels have the following meanings from constructs and indicators.

Constructs and their indicators in the model are as follows:

• ES—Ergonomic adjustments of the seat of the mining machine:
  • Q1 Rating of the height adjustability of the mining machine operator’s seat;
  • Q2 Rating of the horizontal adjustability of the mining machine operator’s seat;
  • Q3 Assessment of the ergonomic suitability of the seat according to the height of the mining machine.
• EC—Ergonomic adaptations of the control organs of the mining machine:
  • Q4 Assessment of the adjustability of the location of the control organs of the mining machine;
  • Q5 Evaluation of the controllability of the mining machine’s control organs;
  • Q6 Evaluation of the possibility of easily reaching the controls or levers;
• EE—Ergonomic evaluation of the working conditions of the mining machine.
• AC—Anthropometric characteristics of a mining machine operator:
  • HO—Height of the operator in cm;
  • AR—Age of the respondent in years;
  • WE—Working experience of the respondent in years.
• EO—Ergonomic features of the operator system and workplace:
  • Q7 Assessment of absenteeism due to poor working conditions;
  • Q8 Evaluation of the negative impact of exhaust gases on operators.
• RM1—Mining machine risk of the first group of causes of downtime (risk of abuse and electric and technical nature):
  • IF—The impact factor of downtimes originating from inadequate management (misuse) of the mining machine;
  • EF—The influence factor of electrical downtime on the mining machine.
• RM2—Risk of the mining machine of the second group of causes of downtime (risk of an organizational and technical nature):
  • TF—The impact factor of technical downtime on the mining machine;
  • OF—Impact factor of organizational downtime on the mining machine.
• SAC—Safety Awareness and Competencies:
  • Q10 Assessment of awareness of safety and health responsibilities in the workplace;
  • Q11 Evaluation of the comprehensibility of rules on safety and health at the workplace;
  • Q12 Evaluation of the effectiveness of solving health and safety problems in the workplace.
• SC—Safety Communication:
  • Q13 Assessment of the supervisor’s awareness of safe work practices;
  • Q14 Assessment of communication with the supervisor about safety rules;
  • Q15 Evaluation of the respondent’s awareness of work safety in the company.
• RP—Rules and Procedures:
  • Q16 Evaluation of the respondent’s awareness of the safety and health policy at work in the company;
  • Q17 Evaluation of the respondent’s understanding of the occupational health and safety policy in the company;
  • Q18 Assessment of the respondent’s involvement in improving the occupational health and safety policy in the company.
• SP—Safety Policy:
  • Q19 Respondents’ assessment of the representation of the safest ways of working in occupational safety and health rules and procedures;
  • Q20 Evaluation of respondents in the implementation of safety improvement in the shortest possible time;
  • Q21 Respondent’s assessment of the compliance of practices with occupational health and safety rules and procedures.
• SHW—Safety and health at work:
• EO—Ergonomic characteristics of the system, operator–workplace:

The model adequacy indices are χ² = 909.57, df = 313, GFI = 9.898, AGFI = 0.826, NFI = 0.906, CFI = 0.908, TLI = 0.905, RMSEA = 0.066.

The structural equation modelling method obtained the mathematical dependencies among the observed variables. Based on these, a contextual adaptive decision support system was designed, which carries out expertise and provides recommendations on risk mitigation measures.

Reflective models can be estimated through Equation (1) [68], where the latent variable is assumed to underlie or cause its related causal indicators. Simple linear regression Formula (1) provides $x_i$—causal indicator, ${\pi }_{i0}$—constant intercept, ${\pi }_i—\mathrm{item} \mathrm{loading}, \xi —\mathrm{latent} \mathrm{variable}, {\epsilon }_1—\mathrm{measurement} \mathrm{error}, i=\mathrm{1,2},\dots n.$

$$\begin{gathered} x_1={\pi }_{10}+{\pi }_1\xi +{\epsilon }_1 \\ {x}_2={\pi }_{20}+{\pi }_2\xi +{\epsilon }_2 \\ {x}_n={\pi }_{n0}+{\pi }_n\xi +{\epsilon }_n \end{gathered}$$

(1)

$$\left[\begin{array}{c}HO\\ AR\\ WE\\ Q3\\ Q2\\ Q1\\ Q4\\ Q6\\ Q5\\ Q7\\ Q8\end{array}\right]=\left[\begin{array}{cccc}-0.65& 0& 0& 0\\ 0.96& 0& 0& 0\\ 0.94& 0& 0& 0\\ 0& 0.72& 0& 0\\ 0& 0.91& 0& 0\\ 0& 0.79& 0& 0\\ 0& 0& 0.50& 0\\ 0& 0& 0.79& 0\\ 0& 0& 0.88& 0\\ 0& 0& 0& 0.63\\ 0& 0& 0& 0.94\end{array}\right]·\left[\begin{array}{c}AC\\ ES\\ EC\\ EE\end{array}\right]+\left[\begin{array}{c}22.366\\ 7.595\\ 10.374\\ 1.048\\ 0.504\\ 1.056\\ 1.792\\ 0.350\\ 0.239\\ 1.636\\ 0.326\end{array}\right]$$

(2)

$$\left[\begin{array}{c}Q10\\ Q11\\ Q12\\ Q13\\ Q14\\ Q15\\ Q19\\ Q20\\ Q21\\ Q16\\ Q17\\ Q18\end{array}\right]=\left[\begin{array}{cccc}0.89& 0& 0& 0\\ 0.74& 0& 0& 0\\ 0.69& 0& 0& 0\\ 0& 0.56& 0& 0\\ 0& 0.77& 0& 0\\ 0& 0.68& 0& 0\\ 0& 0& 0.53& 0\\ 0& 0& 0.80& 0\\ 0& 0& 0.77& 0\\ 0& 0& 0& 0.74\\ 0& 0& 0& 0.73\\ 0& 0& 0& 0.74\end{array}\right]·\left[\begin{array}{c}SAC\\ SC\\ RP\\ SP\end{array}\right]+\left[\begin{array}{c}0.191\\ 0.329\\ 0.393\\ 0.647\\ 0.426\\ 0.561\\ 0.516\\ 0.423\\ 0.533\\ 0.498\\ 0.584\\ 0.586\end{array}\right]$$

(3)

$$\left[\begin{array}{c}SAC\\ SC\\ RP\\ SP\end{array}\right]=\left[\begin{array}{c}0.52\\ 0.76\\ 0.89\\ 0.93\end{array}\right]·\left[SHW\right]+\left[\begin{array}{c}0.541\\ 0.127\\ 0.043\\ 0.084\end{array}\right]$$

(4)

$$\left[\begin{array}{c}AC\\ ES\\ EC\\ EE\end{array}\right]=\left[\begin{array}{c}0.55\\ -0.89\\ -0.91\\ 0.67\end{array}\right]·\left[EO\right]+\left[\begin{array}{c}67.103\\ 0.348\\ 0.134\\ 1.320\end{array}\right]$$

(5)

$$\left[\begin{array}{c}TF\\ OF\\ IF\\ EF\end{array}\right]=\left[\begin{array}{cc}0.55& 0\\ 0.49& 0\\ 0& 0.72\\ 0& 0.71\end{array}\right]·\left[\begin{array}{c}RM2\\ RM1\end{array}\right]+\left[\begin{array}{c}\mathrm{16,170,832.782}\\ \mathrm{263,486.489}\\ \mathrm{119,620.860}\\ \mathrm{85,556.736}\end{array}\right]$$

(6)

$$\left[RM1\right]=-0.25·\left[SHW\right]+\left[\mathrm{4,850,076.534}\right]$$

(7)

$$\left[\begin{array}{c}RM2\\ RM1\end{array}\right]=\left[\begin{array}{c}0.49\\ 0.21\end{array}\right]·\left[EO\right]+\left[\begin{array}{c}\mathrm{4,850,076.534}\\ \mathrm{124,523.940}\end{array}\right]$$

(8)

3.2. Decision Support System for Mining Machinery Risk Mitigation

On the basis of the structural equation model that describes the contextual dependence of the risk of the operation of mining machines, the designed decision support system (DSS) is programmed in the C# language with a dual function.

The first module of the designed DSS refers to the rule base mechanism that observes each factor individually in the operator–mining machine system and gives a recommendation for risk mitigation in the event that one of the observed values exceeds the permitted threshold level (Supplementary Materials—Knowledge Base S1: Knowledge base).

The second module of the designed DSS refers to the implementation of mathematical Equations (1)–(8) in the inference mechanism, which, based on the numerical values of the entered observed variables, obtains the values of the constructs whose value is assessed from the aspect of risk for the occurrence of unwanted events in the human–mining machine system. Formula (1) describes the general form of mathematical structural equations, while Formula (2) describes the relations of ergonomic features of the operator system and workplace (EO) with four constructs (AC, ES, EC, EE). These four constructs represent the influence of the anthropometric characteristics of the operator, ergonomic adjustments of seats, control organs, and working conditions. Formula (5) gives the mathematical relation between EO and AC, ES, EC, and EE constructs for the final extraction of the EO segment.

Formula (2) represents safety and health at work, also using four constructs (SAC, SC, RP, SP). These four constructs take into account the mining operators’ attitudes in light of safety awareness, competencies, safety communications, and compliance with safety rules, procedures, and policies. Formula (4) describes the calculation of SHW—safety and health at work—on the basis of four complementary constructs (SAC, SC, RP, SP).

Formula (6) represents the mining machine causes of downtime: the impact factor of downtimes originating from inadequate management (misuse) of the mining machine (IF), the influence factor of electrical downtime on the mining machine (EF), the impact factor of the technical downtime on the mining machine (TF), and the impact factor of organizational downtime on the mining machine (OF). These four constructs use the collected data for each workplace and its impact is calculated as the risk of mining machine downtimes. The risk of occurrence of the mining machine-specific downtime is calculated as the product of the duration of the observed downtime, its level of danger, and the frequency of occurrence.

Formula (7) describes the relation between the first group of the mining machine causes of downtime and safety and health at work. We have found that only the first group of causes of downtime was interrelated with the SHW construct. That is the reason why we divided the causes of the mining machine downtime break in two groups.

Formula (8) describes the relation between the mining machine causes of downtime (divided into two groups—RM1 and RM2) and ergonomic characteristics of the operator system and workplace (EO). This equation provides a view on the indirect link between the anthropometric characteristics of the operator, ergonomic adjustments of seats, control organs, and working conditions and mining machine causes of downtime, where, in some of the causes, the human factor is dominant (abuse of machine, organizational factor, inadequate maintenance, etc.)

The contextual, ergonomic, adaptive decision support system is designed as a web-based tool, which the user accesses online. Then, the user is interactively surveyed on the questions related to ergonomic working conditions in the cabin, age, years of work experience, height, weight, compliance with safety rules and procedures and health at work, the type of mining machinery on which he/she works, the age of the observed mining machine, the frequency and type of stoppages that occur on the observed mining machine along with the degree of danger of the observed stoppages by type. The entire system of responses received from an individual user in the designed DSS is managed as a single case (case study), for which a model is formed based on the mathematical formulas described in (1). Then, it is stored in the database.

In general, the concept of the designed DSS has been conceived in such a way that, based on the extensive research sample described in Section 4, the data for a new case study were compared to establish the degree of risk. The risk is determined on the basis of the cause-and-effect covariance relationships and equations given in Formulas (2)–(8).

Figure 2 shows the segments of the user interface of the DSS for mitigating the risks of the operation of mining machines.

4. Discussion

In the discussion of research results, we need to refer to the fact that, in the existing sources, there is no similar contextual model that simultaneously combines the factors of ergonomic suitability of the cabin for the operator of the mining machine, working conditions of the operator, compliance with safety procedures and rules and the occurrence of various types of downtime in mining machinery (although it is known that the human factor plays an important role in downtimes).

Due to the fact that the area of operator–machine interoperability has been unexplored, the highest percentage of injuries and adverse events in mining machinery is caused by the human factor, and hence, the collected databases give us the possibility to explore the cause-and-effect relationships in an operator–machine system. In the analysis of the collected data on the ergonomic adjustment of the cabin to the operator, the majority of respondents (87.69%) expressed dissatisfaction with the seat (swivelling capability and armrests not adjustable or non-existent). Additionally, a significant proportion of the surveyed operators pointed to frequent absence from work due to poor working conditions (83.98%) (questions Q1–Q3, construct ES, questions Q7–Q8, construct EE in the SEM model).

Through the analysis of the collected data on safety and health operators’ attitudes, we have discovered that only four groups of questions had significant influence according to SEM modelling. These groups include safety awareness and competencies, safety communication, rules and procedures, safety policy, and cover range of questions Q10–Q21.

Cause-and-effect relationships were found between the constructs: EO and ES (−0.89), EO and EC (−0.91), EO and EE (0.67), EO and RM2 (0.49), SHW and SAC (0.52), SHW and SC (0.76), SHW and RP (0.89), SHW and SP (0.93). Based on these cause-and-effect relationships between constructs, we can notice that the relation between EO and the risk of machine downtimes is more significant than the relation between SHW and the risk of machine downtimes. This conclusion highlights ergonomic adjustments to operators’ cabins as an important factor in the prevention of unwanted mining machine downtimes. For easier use of the obtained mathematical model (Equations (1)–(8)), a DSS tool was designed with the inference logic that indicates risk prevention measures. The tool interface is user-friendly, and a list of preventive measures is compiled in a simple and comprehensible way in order to mitigate work risks. On the other hand, the value of the risk related to individual constructs in the multivariable system that follows from the structural equation model is important for the employees in the mining industry, risk assessors, local residents near mining companies, and beyond.

5. Conclusions

This survey has served to investigate the influential factors that lead to the occurrence of unwanted events during the operation of mining machinery, originating from the ergonomic adjustment of the cabin to the operator of the mining machinery, compliance with safety procedures and rules, organizational climate, the appearance of downtime on the mining machinery due to the human factor and/or organizational factors, etc. Collecting data to investigate the interdependence of contextual factors in the operator–manager–mining machine system was a unique challenge solved by participative ergonomics, given that it is not easy to collect data and identify samples of various types of downtime in mining pits due to the pace of work. Therefore, a valuable database containing data on stoppages on various mining machines (excavators, loaders, bucket-wheel excavators, dumpers, and other machines) and a database on operators’ attitudes towards the ergonomic suitability of the workplace, working conditions, organizational climate, and the like, made it possible to prove the hypotheses about the interdependencies between the contextual factors and cause-and-effect relationships from the impact on the occurrence of risks in the operation of mining machines. The structural equation model that describes the contextual dependence of the risk of the operation of mining machines shows that there are significant interdependencies, which, in the designed DSS, allows us to predict the occurrence of unwanted events for the newly observed operator–manager–mining machine system. The DSS online allows the user to suggest preventive and/or mitigation measures for the occurrence of incidents/accidents in the operator–manager–mining machine system. The DSS has been tested on open pit mining sites in Serbia, and it may be argued that it provides expertise. Still, the question remains whether the human factor, as a key element in the system, will implement the recommended preventive or mitigation measures. Thus, in the further development of the contextual adaptive DSS for mitigating the risk of working with mining machinery, it is necessary to develop an alarm system to prevent the operator from further work in the event that preventive and/or mitigation measures have not been implemented. Significant work also lies ahead in order to integrate useful explainability into this DSS, as well as to allow users to understand why a system has produced a given output, which will extend its usage, in our opinion.