Risk is commonly understood as the product of the probability that a risk will occur and the severity or consequence of that risk being realized. For spacecraft flight dynamics, operators commonly employ a probability of collision (Pc) of 10-4 as a risk mitigation threshold, without considering consequence of a collision. This simplification means that an operator will respond similarly to an event that would produce many long-lived pieces of debris and an environmentally insignificant one. As the number of spacecraft and debris in orbit continues to increase, and the politics of debris creation evolve, flight dynamics teams will face increasing pressure to balance their finite risk mitigation resources in the most efficient and effective manner while accounting for longer term consequences. This paper will assess several potential metrics for collision severity, selected for simplicity and rapid computation, suitable to allow their integration into operational decision cycles. These metrics include Number of Fragments (Nf), Undisposed Mass per Year (UMPY), the Criticality of Space Index (CSI), and several metrics derived from results from Aerospace Corporation’s Debris Assessment and Response Tool (DART).

Beginning with a large database of operational NASA Conjunction Assessment Risk Analysis conjunction data messages, we simulate fragmentation events using the NASA Standard Satellite Breakup Model and compute each potential metric. The MASTER model provides background orbital density information to support calculation of CSI. To efficiently calculate approximate orbital lifetime, we interpolate lifetime from a database of semi-analytic propagations of simulated orbits with different initial orbits, physical properties, and start times relative to a reference solar cycle.

We perform a variety of analyses to assess how the inclusion of a collision consequence metric would influence collision risk reduction decision-making compared to a probability-only approach and to calibrate thresholds for metric-response correlations. Because no “truth” metric exists to quantify severity, we instead perform a crosswalk to explore the extent to which each metric is able to represent severity as expressed in terms of each other metric. We use a genetic algorithm to identify thresholds for each metric to optimize categorizations in a 5x5 NASA Risk Assessment Matrix for either an iso-risk or iso-warning objective. In the iso-risk case, the optimizer minimizes discrepancy in treatment between similar risk events. In the iso-warning case, the optimizer sets thresholds to produce a set of alerts and mitigation actions that is most similar to the operational burden of the current Pc-only based approach. In both cases, we assess potential gains relative to the Pc-only approach. Based on the results of this analysis, we offer recommendations for considering event severity as part of operational collision risk mitigation decision-making.