Extreme natural events such as floods, storms, mega waves and earthquakes, but also terrorist attacks, have the potential to trigger multiple and simultaneous failures of technological installations (e.g. chemical plants, nuclear facilities, power grids) and consequently becoming a serious hazard for the population and the environment. This complex problem is addressed with an innovative, multi-disciplinary approach with elements from computer science and mathematics.

Even when natural events are implied as causes of loss, such as storms, earthquakes or flooding, human intervention can be found to contribute to the mishap by failing to properly account for their effects. Human error is generically defined as a failure to perform a certain task that leads to an adverse consequence. In order to enhance systems’ safety it is necessary to improve human reliability and this is tightly associated with the development of human error studies and understanding.

In principle, any system with coupled components can be represented as a network. Since the performance and reliability of networks are directly affected by uncertainty, quantitative assessment of uncertainty on systems and networks performance is widely recognized as an important task in practical engineering .The main task of this project is to develop new theory and efficient numerical methods for uncertainty and reliability analysis on large and interconnected systems and networks.

The power grid is one of the largest man-made critical infrastructures and it has been designed to forward electric power from generating units to residential, commercial and industrial end-users. Due to recent trends such as the increasing allocation of uncertain in output renewable generators and environmental changes which are drifting weather scenarios towards extremes, reliability and resilience are becoming major concerns for the future power grid.

Civil engineering structures and engineering systems are subject to degradation by fatigue cracks due to cyclic loading or unloading. In turn when the cracks propagate, the structural system accumulates damage thereby leading to serviceability loss or eventual collapse. These failures can be prevented by appropriate maintenance scheduling and repair despite fluctuations and changes of structural and environmental parameters and conditions.

The safety of nuclear systems and plants relies on the reliability and availability of different safety systems. Under realistic conditions, these systems and components are affected by uncertainties, caused by lack of sufficient knowledge and/or by natural unpredictable external events. The overall aim of the project is to develop a robust methodology for probabilistic safety (risk) assessment. It involves the development of an efficient general purpose computational tool and strategies able to deal with scarce, vague and imprecise information and a methodology that will provide a robust support tool for risk-informed decision making.

Nuclear power plants pose a great risk given the possibility of severe accidents .Given the complexity and high interconnectivity of the systems an initiating event, could cascade to catastrophic consequences. It is, therefore, vital that the vulnerability of the plant to these initiation events and the extent of their consequences be ascertained, to ensure the appropriate mitigating actions are taken. Currently, the tools employed to estimate these crucial quantities are based on legacy techniques like static fault and event tree analyses often associated with unrealistic assumptions that might compromise the accuracy of the results.

Analysing major accidents from a human factor perspective, previous work had demonstrated that design failure is the predominant contributor to human errors in complex technology systems. However, human errors not only apply to decision-making processes of operative labours but also to decision makers at highest hierarchical level. Managers decide on a daily basis which activities should happen, their sequence, and who should perform. For this reason, the influence of managers in safety is also considered in this research, to understand the human errors that lead to unsafe design.

The analysis of complex and realistic systems is in general associated with huge computatiuonal cost. This might make challenging analysing the effect of the uncertainty of the performance of these systems. Hence, meta-modelling tools (e.g. Artificial Neural Network, Gaussian Process Emulator, Kriging models, Response Surface) can be adopted to speed up the required analysis. However, the use of a meta-model can introduce additional uncertainties that needs to be properly accounted for.

Climate change is expected to modify the frequency and intensity of extreme climate events. These new conditions, insert an additional and not negligible element of uncertainty to the vulnerability quantification of technological installations. In order to assess the resilent of critical infrastructures and facilities against natural threats, Enanched Baysian networks are adopted to take into account different natural factors (and the associated uncertainty and imprecision) that could affect facilities safety and performance.

Extreme dynamic events like e.g. earthquakes might produce the failure of structural system. faulte that generate the need for dynamic protective measures of such structures. Hence, it is of paramaunt importance being able to realistically simulate the behaviour the systems and assess the associated risk.
Numerical simulator based on Subset simulation technique is used in order to deal efficiently with a rare event. The proposed procedure has been to adoped to design viscous dampers and to perform the fragility analysis of a nuclear fuel assembly in collaboration with AREVA GmbH.

Central to the engineering and physics design of a new European demonstration fusion power plant (DEMO) is an integrated operating point which respects the limitations placed on performance by all relevant plant systems and their interactions with one another. Such an operating point can be identified and optimised using a systems code. The aim is to develop workflows to assess the uncertainty related to DEMO operating point and use the tools and workflows to develop robust nominal and back-up operating scenarios to increase confidence in the successful creation of such devices.