A holistic approach to control process safety risks: Possible ways forward

Pursuing process safety in a world of continuously increasing requirements is not a simple matter. Keeping balance between producing quality and volume under budget constraints while maintaining an adequate safety level proves time and time again a difficult task given that evidently major accidents cannot be avoided. Lack of resilience from an organizational point of view to absorb unwanted and unforeseen disturbances has in recent years been put forward as a major cause, while organizational erosive drift is shown to be responsible for complacency and degradation of safety attitude. A systems approach to safety provides a new paradigm with the promise of new comprehensive tools. At the same time, one realizes that risk assessment will fall short of identifying and quantifying all possible scenarios. First, human error is in most assessments not included. It is even argued that determining human failure probability by decomposing it to basic elements of error is not possible. Second, the crux of the systemic approach is that safety is an emergent property, which means the same holds for the technological aspect: risk is not fully predictable from failure of components. By surveying and applying recent literature, besides analysing, this paper proposes a way forward by considering resilience of a socio-technical system both from an organizational and a technical side. The latter will for a large part be determined by the plant design. Sufficient redundancy and reserve shall be kept to preserve sufficient resilience, but the question that rises is how. Available methods are risk assessment and process simulation. It is helpful that the relation between risk and resilience analysis has been recently defined. Also, in a preliminary study the elements of resilience of a process have become listed. In the latter, receiving and interpreting weak signals to boost situational awareness plays an important role. To maintain alertness on the functioning of a safety management system, the process industry is monitoring safety performance indicators. The critical intensity level upon which management must be alarmed is less simple. Risk assessment may be improved, made dynamic, and be a tool of process control by taking account of short-term risk fluctuations based on sensor signals and the influence of human factors with its long-term changes via indicators. Bayesian network can provide the infrastructure. The paper will describe various complexities when applying a holistic control of safety to a process plant in general, and it will more specifically focus on safeguarding measures such as barriers and other controls with some examples.