(4) Risk Estimating Relationships

Based on experience with spacecraft cost models typically used in industry, it was determined that the Risk Estimating Relationships (RERs) should be developed using regression techniques similar to those used in the cost estimating relationship models of packages such as the Aerospace Corporation’s Small Satellite Cost Model (SSCM). Basing the RER development on techniques already existing in the community provides a solid starting point of development.

This research focuses on seven mission risks as identified in my AIAA and JoSS paper, each with between four and seven root causes for a total of 34 root causes. Twelve function forms were studied for each root cause. The Sum of Standard Deviations (SSD) value was calculated for each function following the General Error Regression (GER) methodology and an Excel Macro implemented the Excel Solver function for each of the function forms of the given root cause. This solver routine finds the coefficients of the function which minimize the SSD value while keeping the bias calculation near zero. Figure 1 shows the Excel spreadsheet for one of the root causes of one of the mission risks. The spreadsheet ranks the SSD values from lowest to highest, and the generalized R^2 value from highest to lowest, since it is desired to have a low SSD and a high R^2. To eliminate function forms with really low SSD values, but also really low R^2 values, a combined rank score was established. It is this combined rank score which determines the function form that is most representative of the data, and will thus be used as the risk estimating relationship for the given root cause. In the case that the R^2 value is incredibly low, it is the author’s discretion to select the next best function, based on combined rank. In the case that there is a tie between functions, the function with the lowest SSD value is selected.