An intelligent algorithm for performance evaluation of job stress and HSE factors in petrochemical plants with noise and uncertainty

https://doi.org/10.1016/j.jlp.2012.10.004Get rights and content

Abstract

This study presents an intelligent algorithm based on Adaptive Neuro-Fuzzy Inference System (ANFIS) and statistical methods for measuring job stress in noisy and complex petrochemical plants. Job stress is evaluated against health, safety, environment and ergonomics (HSEE) program in the integrated algorithm. The algorithm is composed of seventeen distinct steps. To achieve the objectives of this study, standard questionnaires with respect to HSEE are completed by operators. The average results for each category of HSEE are used as inputs and job stress is used as output for the algorithm. Moreover, operators' stress level with respect to HSEE is evaluated by the algorithm. Finally, operators with weak stress level are identified. The advantage and superiority of the intelligent algorithm are shown by error analysis in contrast with conventional regression approaches. This is the first study that introduces an integrated intelligent algorithm for assessment and improvement of job stress and HSEE in noisy, complex and uncertain environment.

Highlights

► An intelligent algorithm is proposed to rank job stress with respect to HSEE. ► The proposed algorithm is ideal for complex and noisy petrochemical plants. ► HSEE and stress factors are considered as input and output variable, respectively. ► The algorithm is applied to control room operators. ► The algorithm is capable of handling complexity and uncertainty.

Introduction

HSE programs attempt to decrease workplace injuries, health issues and severe effects to environment. Also, effective application of ergonomics factors in workplace could create a balance between human operators and job design. This in turn could increase human productivity, safe workplace and job satisfaction. The principal of HSE is now well recognized in most petrochemical plants. Several industries use the term HSE to describe health, safety and environment as one entity (Deng, 1999). Various studies have shown positive influences of applying ergonomics rules to the workplace including machine, job and environmental design (Abou-Ali & Khamis, 2003; Ayoub, 1990a, 1990b; Azadeh, Mohammad Fam, Khoshnoud, & Nikafrouz, 2008; Blanchard & Fabrychy, 1998; Shikdar & Das, 2003; Shikdar & Sawaqed, 2004). Studies in ergonomics have created data and instructions for industrial applications (Blanning, 1984; Bryden & Hudson, 2005; Burri & Helander, 1991). However, there is still a low level of acceptance and few applications in industry. Lack of utilization of the ergonomics rules could bring inefficiency to the workplace. Moreover, ergonomically deficient workplaces could cause physical and emotional stress, low productivity and poor quality of working conditions (Azadeh, Mohammad Fam, Khoshnoud, et al., 2008; Burri & Helander, 1991; Cabrero-Canosa et al., 2003; Caldwell, Breton, & Holburn, 1998). The ergonomics deficiencies are main cause of health hazards in workplaces and decreased workers' productivity (Champoux & Brun, 2003). However, ergonomics applications have not achieved significant momentum in developing countries (Azadeh, Mohammad Fam, Khoshnoud, et al., 2008).

By considering health, safety, environment and ergonomics (HSEE) encourages organizations to adopt a healthy and safe life-style. It insists on ecological efficiency by continuously improving energy consumption and decreasing waste. It optimizes the relation between human operators, machines and work environment (Azadeh, Mohammad Fam, Khoshnoud, et al., 2008; Changchit & Holsapple, 2001; Chen & Yang, 2004). It also provides considerable benefits by decreasing the costs associated with workplace accidents and injuries.

There have been several studies on the impact of HSE and Ergonomics issues in manufacturing systems. Saksvik and Nytr (1996) presented an implementation of internal control (IC) of health, environment and safety (HES) in Norwegian enterprises. IC involves systematic actions that reduce stress and occupational ill-health which will, in turn, prevent injuries and workplace absenteeism (Saksvik & Nytr, 1996). Eklund (1997) presented the relationships between ergonomics and several factors such as work conditions, product design, ISO 9000, continuous improvements and TQM (Eklund, 1997). Azadeh, Nouri, and Mohammad Fam (2005) evaluated the impact of total system design factors (TSD) on human performance in a power plant (Azadeh et al., 2005). In addition Azadeh, Keramati, Mohammad Fam, and Jamshidnejad (2006) described an integrated macroergonomics model for operation and maintenance of power plant (Azadeh et al., 2006). Torp and Moen (2006) presented the effects of implementing or improving occupational health and safety management system. They considered management on the work environment, occupational health and safety behavior and musculoskeletal health of workers in small and medium-sized companies (Torp & Moen, 2006). Mohammad Fam, Azadeh, and Azam Azadeh (2007) used non-parametric statistical analysis to investigate the impacts of total ergonomics factors on local factors (Mohammad Fam et al., 2007). Azadeh, Mohammad Fam, Sadjadi, and Hamidi (2008) presented an integrated framework for designing and development of the integrated health, safety and environment (HSE) model in a gas refinery in Iran. It was shown that total ergonomics model is superior to the conventional ergonomics approach (Azadeh, Mohammad Fam, Sadjadi, et al., 2008). Moreover, Azadeh, Mohammad Fam, and Nouri (2008) present a framework for development of integrated intelligent human engineering environment in complex critical systems. By integration of conventional Health, safety and environment (HSE) with job systems, health, safety, environment and ergonomics (HSEE) is developed. In fact, re-engineering organizational structures and teamwork through electronic data interchange are considered in their study (Azadeh, Mohammad Fam, & Nouri, 2008). Duijm, Fiévez, écileGerbec, Hauptmanns, and Konstandinidou (2008) showed that HSE management would benefit greatly from guidance on how to use existing management systems efficiently and also from the further development of meaningful safety performance indicators that identify the conditions prior to accidents and incidents (Duijm et al., 2008). Mohammad Fam, Azadeh, Faridan, and Mahjoub (2008) used behavior sampling (SBS) technique to evaluate the workers safety behavior in an Iranian gas treatment company (Mohammad Fam et al., 2008). Again, Azadeh, Mohammad Fam, and Azadeh (2009) implemented a study on a gas treatment company to show the need for and superiority of HSEE over conventional HSE and by responding to a questionnaire ergonomics in addition to HSE are evaluated in this refinery. HSEE integrated the structure of human and organizational systems with a conventional HSE system and it is caused enhancing reliability, availability, maintainability and safety through the proposed integrated framework (Azadeh et al., 2009). Hivik, Moen, Mearns, and Haukelid (2009) reported a qualitative interview study of 31 employees, with and without leadership responsibility, employed in a Norwegian petroleum company to gain insight into how the workers conceptualized the HSE concept, different aspects of HSE culture and differences between the informants (Hivik et al., 2009). Hassim and Hurme (2010) presented an Inherent Occupational Health Index has been developed for assessing the health risks of process routes during process research and development stage. The method takes into account both the hazard from the chemicals present and the potential for the exposure of workers to the chemicals (Hassim & Hurme, 2010). Besides, the certification and implementation of occupational health and safety management systems have become a priority for many organizations. To investigate the status of implementing occupational health and safety management systems (OHSMSs), and to explore important performance indicators for OHSMSs' performance appraisal in the printed circuit board (PCB) industry in Taiwan, Chen, Wu, Chuang, and Mac (2009) administered a survey to eleven PCB manufacturers, all of which have been certified as compliant to the guidelines on OHSMS of the Occupational Health and Safety Assessment Series (OHSAS) 18001, and twenty six OHSAS specialists from the academe. Chang and Liang (2009) developed a model to evaluate the performance of process safety management systems of paint manufacturing facilities. The model was constructed based on a three level multi-attribute value model (MAVT) approach. Einarsson and Brynjarsson (2008) suggested an approach for a human factor program as a learning experience through case studies from incidents and accidents in Iceland and Netherlands. From these observations, a more holistic system view is proposed involving authorities and contractors. Azadeh et al. (2011) presented an adaptive neural network algorithm for assessment and improvement of job satisfaction with respect to HSE and ergonomics program in a gas refinery. However, the stated study is not capable of handling noise and uncertainty associated with large and complex systems such as petrochemical plants. The present study however presents a robust algorithm to cover noise and uncertainty associated with job stress and HSE factors in petrochemical plants.

Workplace stress is the harmful physical and emotional response that occurs when there is a poor match between job demands and the capabilities, resources, or needs of the workers. Job stress results from the interaction of the worker and the conditions of work. “Views differ on the importance of worker characteristics versus working conditions as the primary cause of job stress (Niosh, 1999)”. The differing viewpoints suggest different ways to prevent stress at work. According to one school of thought, differences in individual characteristics such as personality and coping skills are very important in predicting whether certain job conditions will result in stress. In other words, what is stressful for one person may not be a problem for someone else. This viewpoint underlies prevention strategies that focus on workers and ways to help them cope with demanding job conditions. Sauter, Murphy, and Hurrell (1990) introduced the means to change the organization to prevent job stress through workload balance, job design, work definition, participative management, work schedule, etc.

There are several studies about stress in literature. There are many behavioral disturbances in modern man as well as psychosomatic disturbances such as stress. In a highly competitive job market, where the means of work and information are rapidly transformed, one of the stress factors is the professional insecurity. There are also other factors linked to the organization of modern production such as: the space factor and the psycho-social and inter-subjective relations of the working environment. The modern organization of work produces stressing conditions which may develop psychosomatic disturbances in the productive agents. Although the professionals feel capable of acting in some places of work, the fast transformations in modern world create some insecurity which provokes anxiety leading to stress. Due to this process, the cognitive ergonomics is formally integrated to the productive system; integrating to the business charts and assuming a technical feature to the extent that it has helped in the transformation of knowledge into production force, so as to measure the work for the man (Moiseichyk, De Campos, & Bento, 2000).

The nature and feasibility of preventive measures taken to reduce work stress in the construction industry are discussed by Van der Molen and Hoonakker (2000). Employers and employees in 18 construction companies were surveyed about feasible measures to reduce work stress. From the viewpoint of work stress, working with teams is the solution as it is an innovative measure. The measures include the use of older employees to act as coaches for younger personnel, and counseling and brief therapies.

In an organizational context, participatory ergonomics (PE) programs set out to involve employees in ergonomic design and analysis efforts in order to promote a safer, more user-friendly, and productive workplace. While the purpose of a PE program is to involve employees, currently there is no quantitative method to evaluate the effectiveness of PE programs from the employee's perspective. “Based on a review of the recent published literature, five key constructs were identified as critical evaluative components: employee knowledge base, employee involvement, employee support, perceptions of managerial support, and employee stress due to ergonomic changes (Matthews & Gallus, 2005)”. With the help of an organizational management–labor committee, a content validity approach was used to generate items. The proposed five-dimension measure was administered to 63 participants working in large manufacturing plant with an established PE program. Initial empirical support for the proposed dimensions was found. Potential uses of the developed measure in a larger context are discussed (Matthews & Gallus, 2005).

A mental stress assessment tool box was developed by Hackl-Graber, Haiden, and Wittmann (2000) in order to reduce mental stress and assist in improving an employee's work environment, particularly the organizational structure. The box consists of tools which support occupational health and safety experts with analyzing mental stress loads in organizations and implementing changes for reducing mental stress. The assessment consists of three steps: description of organization, rough analysis, and detailed analysis. The detailed analysis is in the form of a workshop. This approach enables the employees to learn about their stress and allow them to participate in the change process.

Although several studies have investigated stress or HSEE issues, however, there is shortage of research works that consider stress with respect to HSEE. Most of the previous studies used simple approaches to investigate stress (Glazer, Stetz, & Izso, 2004; Holman et al., 2006; Hsu et al., 2007; Landa, López-Zafra, Martos, & Aguilar-Luzón, 2008; Lapane & Hughes, 2007). Other studies used causal models and simple surveys and statistical methods for assessment of stress and there is a lack of using more complex tools in this issue to deal with noise, uncertainty and complexity associated with job stress and HSEE (Lee & Shin, 2010). This is particularly important when collected data may be corrupted, nonlinear and of uncertain nature. This study presents an effective algorithm to determine stress level among operators with respect to HSEE factors in complex and uncertain environments. This is the first study that considers HSEE and stress factors in an integrated manner in noisy and complex environments. Moreover, this study proposes an adaptive intelligent algorithm for measuring and improving stress among operators with respect to HSEE in a large complex petrochemical plant. In fact, job stress with respect to HSEE factors has been reviewed in the present study and its efficiency among operators has been ranked. Standard questionnaires have been distributed to operators (3 shifts) in which variety of questions related to HSEE and stress were presented to operators. Adaptive Neuro-Fuzzy Inference System (ANFIS) has been used to predict the efficiency of operators by comparing ANFIS outputs with actual outputs. Consequently operators are ranked and their efficiencies are identified with respect to HSEE factors as inputs and stress as output. In other words, 4 groups of inputs (health, safety, environment and ergonomics) and 1 main output (stress) are defined by the questionnaires. The main objective of the proposed algorithm is to contribute to the use of ANFIS, statistical and mathematical methods in the efficiency measurement of stress and HSEE factors in a petrochemical unit with 115 operators. The efficiencies are obtained by taking the ratio between the observed and predicted values for the output(s) of each decision (Athanassopoulos & Curram, 1996).

Fuzzy logic has been introduced by L. Zadeh in 1965 to deal with possibility theory (Zadeh, 1965). Concept of partial truth and some types of the uncertainty in the real are the point of L. Zadeh departure in order to introduce fuzzy logic. Fuzzy logic belongs to the uncertainty theory actually. Also artificial neural networks are introduced based on properties of biological neurons in order to efficiently use in control, estimation, classification, clustering, classify and other area of artificial intelligence. In the field of artificial intelligence, neuro-fuzzy refers to combinations of artificial neural networks and fuzzy logic. Neuro-fuzzy was proposed by Jang et al. (1997). Neuro-fuzzy hybridization results in a hybrid intelligent system that synergizes these two techniques by combining the human-like reasoning style of fuzzy systems with the learning and connectionist structure of neural networks. Neuro-fuzzy hybridization is widely termed as fuzzy neural network (FNN) or neuro-fuzzy system (NFS) in the literature.

Fuzzy systems and neural networks are both very popular techniques that have seen increasing interest in recent decades. At a first glance, they seem to be totally different areas with merely marginal connections. However, both methodologies belong to the soft computing area. Soft computing includes approaches to human reasoning and learning that try to make use of the human tolerance for incompleteness, uncertainty, imprecision and fuzziness in decision-making processes. Many different structures for fuzzy neural networks (FNNs) have been proposed (Zhou, Li, & Jin, 2002). Among them ANFIS is a neural-network based on fuzzy approach, in which the learning procedures are performed by interleaving the optimization of the antecedent and conclusion parts parameters (Aliyari Shoorehdeli, Teshnehlab, Khaki Sedigh, & Ahmadieh Khanesar, 2009). ANFIS uses a feed forward network to search for fuzzy decision rules that perform well on a given task. Using a given input–output dataset, ANFIS creates a FIS whose membership function parameters are adjusted using a back-propagation algorithm alone or a combination of a back-propagation algorithm with a least squares method. This allows the fuzzy systems to learn from the data being modeled. Consider a first order Takagi–Sugeno fuzzy model with a two input, one output system having two membership functions for each input. Then, the functioning of ANFIS is a five-layered feed-forward neural structure, and the functionality of the nodes in these layers can be summarized as:o1,i=μAi(x)o1,i=μBi2(y)where x or y is the input to the node, Ai or Bi2 is a fuzzy set associated with this node. At the first layer, for each input, the membership grades in the corresponding fuzzy sets are estimated. At the second layer, all potential rules between the inputs are formulated by applying fuzzy intersection (AND). The product operation is used to estimate the firing strength of each rule.o2,i=wi=μAi(x)×μBi(y),i=1,2

The third layer is used for estimation of the ratio of the ith rule's firing strength to the sum of all rule's firing strengths.o4,i=w¯ifi=w¯i(pix+qiy+ri)where w¯i is the output of layer 3 and {pi, qi, ri} is the parameter set. Parameters in this layer will be referred to as consequent parameters. The final layer computes the overall output as the summation of all incoming signals from layer 4.Overalloutput=o5,i=iw¯ifi=iwifiiwi

Optimizing the values of the adaptive parameters is of vital importance for the performance of the adaptive system. Jang et al. (1997) developed a hybrid learning algorithm for ANFIS which is faster than the classical back-propagation method to approximate the precise value of the model parameters. The hybrid learning algorithm of ANFIS consists of two alternating phases: (1) gradient descend which computes error signals recursively from the output layer backward to the input nodes, and (2) least squares method, which finds a feasible set of consequent parameters. We observe that, given fixed values of elements of premise parameters, the overall output can be expressed as a linear combination of the consequent parameters.

There are different advanced fuzzy inference techniques. In this study, we have used one kind of them that generates a Sugeno-type FIS structure using subtractive clustering and requires separate sets of input and output data as input arguments. When there is only one output, it may be used to generate an initial FIS for ANFIS training. It accomplishes this by extracting a set of rules that model the data behavior. The rule extraction method first uses the subclust function to determine the number of rules and antecedent membership functions and then uses linear least squares estimation to determine each rule's consequent equations. This function returns an FIS structure that contains a set of fuzzy rules to cover the feature space. The arguments for the advanced fuzzy inference technique are as follows:

  • Xin is a matrix in which each row contains the input values of a data point.

  • Xout is a matrix in which each row contains the output values of a data point.

  • radii is a vector that specifies a cluster center's range of influence in each of the data dimensions, assuming the data falls within a unit hyper box.

Membership function (MF) is a curve that defines how each point in the input space is mapped to a membership value (or degree of membership) between 0 and 1. The input space is sometimes referred to as the universe of discourse, a fancy name for a simple concept. There are different kinds of membership functions as follows:

  • Triangular membership function

  • Trapezoid membership function

  • Gaussian membership function

  • Bell membership function

  • Linear membership function

The default input membership function type in this study is Gaussian, and the default output membership function type is linear. Table 1 summarizes the inference methods and their types.

Section snippets

Methodology

An integrated ANFIS algorithm is proposed to measure stress with respect to HSEE. It should be stated that ANFIS is capable of concurrently handling noise and complexity. Noise arises from input data due to subjectivity of data. Also, complexity arises from the fact that comprehensive database does not exist for such studies. Furthermore, in developing countries existence of uncertainty and complexity are inevitable due to social, economical and political issues. Locating the sources and types

Experiment: the intelligent algorithm

The applicability of the proposed adaptive intelligent algorithm is experimented in an actual petrochemical unit. It is shown how each step is applied to assess the relationship between job stress and HSEE program in the control rooms of this unit. Moreover, the efficiency of the proposed integrated ANFIS algorithm is shown by mean absolute percentage error (MAPE). Moreover, it is compared with conventional regression approaches to show its superiorities. Also, the qualitative features of the

Computational results and analysis

  • Steps 1 and 2: To prove the reliability and validity of the questionnaire, Cronbach's Alpha and factor analysis have been used. The reliability analysis of a questionnaire determines its ability to yield the same results on different occasions and validity refers to the measurement of what the questionnaire is supposed to measure (Cooper & Schindler, 2003). In order to assess the reliability of instrument, Cronbach's alpha for all criteria of research variables [HSEE] has been calculated.

Conclusion

A highly unique and intelligent ANFIS algorithm was proposed to measure and rank job stress with respect to HSEE. The proposed algorithm is ideal for complex and noisy petrochemical plants because of nonlinearity of ANFIS in addition to its universal approximations of functions and its derivates, which makes them highly flexible. Moreover, HSEE factors were considered as input variables and stress was considered as output variable. The proposed algorithm is composed of 17 distinct steps. To

Acknowledgments

The authors are grateful for the valuable comments and suggestion from the respected reviewers. Their valuable comments and suggestions have enhanced the strength and significance of our paper. This study was supported by a grant from University of Tehran (Grant No. 8106013/1/09). The authors are grateful for the support provided by the College of Engineering, University of Tehran, Iran.

References (61)

  • C.Y. Chen et al.

    A comparative analysis of the factors affecting the implementation of occupational health and safety management systems in the printed circuit board industry in Taiwan

    Journal of Loss Prevention in the Process Industries

    (2009)
  • J.R. Chen et al.

    A predictive risk index for safety performance in process industries

    Journal of Loss Prevention in the Process Industries

    (2004)
  • A. Costa et al.

    Evaluating public transport efficiency with neural network models

    Transportation Research

    (1997)
  • H. Deng

    Multicriteria analysis with fuzzy pairwise comparison

    International Journal of Approximate Reasoning

    (1999)
  • N.J. Duijm et al.

    Management of health, safety and environment in process industry

    Safety Science

    (2008)
  • S. Einarsson et al.

    Improving human factors, incident and accident reporting and safety management systems in the Seveso industry

    Journal of Loss Prevention in the Process Industries

    (2008)
  • B.B. Flynn et al.

    A framework for quality management research and an associated measurement instrument

    Journal of Operations Management

    (1994)
  • S. Glazer et al.

    Effects of personality on subjective job stress: a cultural analysis

    Personality and Individual Differences

    (2004)
  • M.H. Hassim et al.

    Inherent occupational health assessment during process research and development stage

    Journal of Loss Prevention in the Process Industries

    (2010)
  • G.T. Holman et al.

    Using surveys to identify stressors in generalized jobs: a direct clustering method

    International Journal of Industrial Ergonomics

    (2006)
  • H.C. Hsu et al.

    Work stress among nursing home care attendants in Taiwan: a questionnaire survey

    International Journal of Nursing Studies

    (2007)
  • K.L. Lapane et al.

    Considering the employee point of view: perceptions of job satisfaction and stress among nursing staff in nursing homes

    Journal of the American Medical Directors Association

    (2007)
  • Y. Lee et al.

    Job stress evaluation using response surface data mining

    International Journal of Industrial Ergonomics

    (2010)
  • Z.H. Qaisar et al.

    A safe regression testing approach for safety critical systems

    Advances in Engineering Software

    (August 2011)
  • A.A. Ramli et al.

    Possibilistic regression analysis of influential factors for occupational health and safety management systems

    Safety Science

    (2011)
  • P. Saksvik et al.

    Implementation of internal control (IC) of health, environment and safety (HES) in Norwegian enterprises

    Safety Science

    (1996)
  • A.A. Shikdar et al.

    Occupational health and safety in the oil industry: a managers' response

    Computers and Industrial Engineering, Ergonomics

    (2004)
  • S. Torp et al.

    The effects of occupational health and safety management on work environment and health: a prospective study

    Applied Ergonomics

    (2006)
  • L.A. Zadeh

    Fuzzy sets

    Information Control

    (1965)
  • Y. Zhou et al.

    A new fuzzy neural network with fast learning algorithm and guaranteed stability for manufacturing process control

    Fuzzy Sets and Systems

    (2002)
  • Cited by (0)

    View full text