Managing the Uncertainty Associated with Hydrogen Gas Hazards and Operability Issues in Nuclear Chemical Plants

The complex and diverse nature of reprocessing and decommissioning operations in existing nuclear chemical plants within the UK results in a variety of challenges. The challenges relate to the quantified risk from hydrogen explosions and how best to manage the associated uncertainties. Several knowledge gaps in terms of the Quantified Risk Assessment (QRA) of hydrogen hazards have been identified in this research work. These include radiolytic hydrogen explosions in sealed process pipes, the failure of ventilation systems used to dilute radiolytic hydrogen in process vessels, the decision uncertainty in installing additional hydrogen purge systems and the uncertainty associated with hold-up of hydrogen in radioactive sludges. The effect of a subsequent sudden release of the heldup hydrogen gas into a vessel ullage space presents a further knowledge gap. Nuclear decommissioning and reprocessing operations also result in operational risk knowledge gaps including the mixing behaviour of radioactive sludges, the performance of robotics for nuclear waste characterisation and control of nuclear fission products associated with solid wastes. Bayesian Belief Networks (BBNs) and Monte Carlo Simulation (MC) techniques have been deployed in this research work to address the identified knowledge gaps. These techniques provide a powerful means of uncertainty analysis of complex systems involving multiple interdependent variables such as those affecting nuclear decommissioning and reprocessing. Through the application of BBN and MC Simulation methodologies to a series of nuclear chemical plant case studies, new knowledge in decommissioning and reprocessing operations has been generated. This new knowledge relates to establishing a realistic quantified risk from hydrogen explosions and nuclear plant operability issues. New knowledge in terms of the key sensitivities affecting the quantified risk of hydrogen explosions and operability in nuclear environments as well as the optimum improvements necessary to mitigate such risks has also been gained.