This is in line with the constructivist basis of the adopted risk perspective of Section

2.3, where the elements in the framework serve to assess the strength of the argumentation. Considering what validity means for the adopted risk perspective, important elements in this context are the completeness of the uncertainty assessment (Aven and Heide, 2009) and the completeness of the bias assessment (Rosqvist and Tuominen, 2004). Thus, the validity framework serves not only to assess how well the model describes the system it intends to describe, but also to systematically reflect on uncertainties and biases underlying its construction. While the various validity concepts can act as sources of model confidence, the extent to which the validity tests fail can be indicated by providing a qualitative uncertainty and bias assessment. This uncertainty and bias assessment RG7204 highlights which parts of the

model would benefit most from a more accurate or comprehensive modeling approach. In the framework application, we focus on face, content and predictive validity. Concurrent validity cannot be established as no other BN models for accidental oil outflow are known to exist. Convergent and discriminant validity require an in-depth comparison of the BN with models of similar, respectively different systems. These are in-principle options but are considered beyond the scope of the current work. In terms of face validity, the presented BN can be considered an appropriate model Y-27632 manufacturer for oil outflow Compound C supplier in ship–ship collision, conditional to impact conditions. This is clear from its construction, which is based on the tank arrangement model by Smailys and Česnauskis (2006) and the collision damage extent model by van de Wiel and van Dorp (2011). The model by Smailys and Česnauskis (2006) has been validated for a number of cases and the analysis in Section 4.2.2 shows that the model leads to a reasonable, conservative estimate of the ship deadweight. The regression model by van de Wiel and van

Dorp (2011) shows a reasonable fit with the cases reported in NRC (2001), see also Section 5.2, and the underlying ship collision mechanics model by Brown and Chen (2002) has been validated for some accident scenarios by Chen (2000). Thus, the BN can be expected to provide reasonable estimates of oil outflows for the intended application in risk assessment for maritime transportation, even when only very limited data about the vessels is available. The oil outflow model includes many, but not all relevant variables for determining the oil outflow. Impact speeds, angle and location and ship masses are important variables in determining the collision damage extent. However, the yaw and sway velocities at the moment of impact also have a certain influence on the collision energy (Ståhlberg, 2010) and damage extent (Wiśniewski and Kołakowski, 2003).