Search

The failure of many System of Systems (SoS) enterprises can be attributed to the inappropriate application of traditional Systems Engineering (SE) processes within the SoS domain, because of the mistaken belief that a SoS can be regarded as a single large, or complex, system. SoS Engineering (SoSE) is a sub-discipline of SE; Risk Management and Modelling and Simulation (M&S) are key areas within SoSE, both of which also lie within the traditional SE domain. Risk Management of SoS requires a different approach to that currently taken for individual systems; if risk is managed for each component system then it cannot be assumed that the aggregated affect will be to mitigate risk at the SoS level. A literature review was undertaken examining three themes: (1) SoS Engineering (SoSE), (2) M&S and (3) Risk. Theme 1 of the literature provided insight into the activities comprising SoSE and its difference from traditional SE with risk management identified as a key activity. The second theme discussed the application of M&S to SoS, providing an output, which supported the identification of appropriate techniques and concluding that, the inherent complexity of a SoS required the use of M&S in order to support SoSE activities. Current risk management approaches were reviewed in theme 3 as well as the management of SoS risk. Although some specific examples of the management of SoS risk were found, no mature, general approach was identified, indicating a gap in current knowledge. However, it was noted most of these examples were underpinned by M&S approaches. It was therefore concluded a general approach SoS risk management utilising M&S methods would be of benefit. In order to fill the gap identified in current knowledge, this research proposed a new model based approach to Risk Management where risk identification was supported by a framework, which combined SoS system of interest dimensions with holistic risk types, where the resulting risks and contributing factors are captured in a causal network. Analysis of the causal network using a model technique selection tool, developed as part of this research, allowed the causal network to be simplified through the replacement of groups of elements within the network by appropriate supporting models. The Bayesian Belief Network (BBN) was identified as a suitable method to represent SoS risk. Supporting models run in Monte Carlo Simulations allowed data to be generated from which the risk BBNs could learn, thereby providing a more quantitative approach to SoS risk management. A method was developed which provided context to the BBN risk output through comparison with worst and best-case risk probabilities. The model based approach to Risk Management was applied to two very different case studies: Close Air Support mission planning and the Wheat Supply Chain, UK National Food Security risks, demonstrating its effectiveness and adaptability. The research established that the SoS SoI is essential for effective SoS risk identification and analysis of risk transfer, effective SoS modelling requires a range of techniques where suitability is determined by the problem context, the responsibility for SoS Risk Management is related to the overall SoS classification and the model based approach to SoS risk management was effective for both application case studies.