Approaches to modeling of accidents in socio-technical complex systems

Introduction
The Increasing complexity of technology in processes and activities related to nuclear power plants, chemical and petroleum, mining and oil extraction, etc. are creating new security problems connected to errors and failures that trigger the disasters or major accidents. In complex systems, the command and control systems are semantically complex usually require a large amount of time to obtain relevant knowledge if we add that the operations are carried out under the pressure of production goals or limitations resources, the process are very dangerous and demand especial attention. 

The research was focused on documenting new theoretical approaches of socio-systems and techniques for modeling and analysis of accidents in safety-critical systems. 

Methods
The study included analysis of 12 cases of major accidents with the meaningful information that was possible to compile related to these accidents. The information was verified and contrasted by means of detailed checklists, including socio-technical and human factors, with the decision that triggered each event. Inside The accident analysis methods was used the tree of causes and a situation analysis that focused on the Organization

Results
Regardless of the limitations and biases of the feedback that differs from the experimental, in 9 of the 12 accidents studied show multiple causes and conditions, as well as interdependence between technology and organizational systems. In 8 cases the causes identified can be assessed as determinants of accidents.

Discussion
Generally, accident analysis models used to predict accidents during the development of safety critical systems, is based on sequential models. By contrast with systems models can analyze accidents caused by emergent phenomena that arise due to the complex nonlinear interactions between system components. 
On the other hand, traditional security techniques for analyzing risks, such as, fault tree analysis and probabilistic, are not sufficient to explain the complexity of modern sociotechnical systems, including self-understanding of accident causation.