The recovery effectiveness for oil spills in ice conditions depends on a complex system and has not been studied in depth, especially not from a system risk control perspective. This paper aims to identify the critical aspects in the oil spill system to enable effective oil spill recovery. First, a method is developed to identify critical elements in a Bayesian Network model, based on an uncertainty-based risk perspective. The method accounts for sensitivity and the strength of evidence, which are assessed for the different Bayesian Network model features. Then, a Bayesian Network model for the mechanical oil spill recovery system is developed for the Finnish oil spill response fleet, contextualized for representative collision accident scenarios. This model combines information about representative sea ice conditions, ship-ship collisions and their associated oil outflow, the oil dispersion and spreading in the ice conditions, and the oil spill response and recovery of the fleet. Finally, the critical factors are identified by applying the proposed method to the developed oil spill response system model. The identified most critical system factors relates collision aspect: Forcing Representative Scenario, Representative Accident Location, Impact Speed, Impact Location, Impact Angle and response aspect: Response Vessel Operability.

在冰况下溢油的恢复效果取决于复杂的系统，尚未进行深入研究，尤其是从系统风险控制的角度来看。本文旨在确定溢油系统中的关键方面，以实现有效的溢油回收。首先，基于基于不确定性的风险观点，开发了一种方法来识别贝叶斯网络模型中的关键要素。该方法考虑了敏感性和证据的强度，并针对不同的贝叶斯网络模型特征进行了评估。然后，为芬兰的溢油应急车队开发了机械溢油回收系统的贝叶斯网络模型，并针对代表性的碰撞事故场景进行了情境化。该模型结合了有关代表性海冰状况的信息，船舶碰撞及其相关的油流出，在冰层条件下的油扩散和扩散，以及溢油响应和船队的恢复。最后，通过将建议的方法应用到已开发的溢油应急系统模型中来确定关键因素。确定的最关键的系统因素与碰撞方面有关：强制代表场景，代表事故地点，撞击速度，撞击位置，撞击角度和响应方面：响应容器可操作性。