Tool for Rapid Analysis of Monte Carlo Simulations

Abstract

Designing a spacecraft, or any other complex engineering system, requires extensive simulation and analysis work. Oftentimes, the large amounts of simulation data generated are very difficult and time consuming to analyze, with the added risk of overlooking potentially critical problems in the design. The authors have developed a generic data analysis tool that can quickly sort through large data sets and point an analyst to the areas in the data set that cause specific types of failures. The Tool for Rapid Analysis of Monte Carlo simulations (TRAM) has been used in recent design and analysis work for the Orion vehicle, greatly decreasing the time it takes to evaluate performance requirements. A previous version of this tool was developed to automatically identify driving design variables in Monte Carlo data sets. This paper describes a new, parallel version, of TRAM implemented on a graphical processing unit, and presents analysis results for NASA's Orion Monte Carlo data to demonstrate its capabilities.

Aerodynamic parameter calculation and ballistic stability research of special projectile

“…(3) The gun is provided with a directional groove and directional pin, etc., to improve the straightness of the directional groove, the interference fit between the directional groove and the directional pin, and reduce the initial rotational speed when ejecting the muzzle. Considering the coaxiality between the directional groove and the axis of the barrel, for the tailless projectile, it may be possible to reduce the muzzle velocity without slotting [10] .…”

Random Generation of Aerodynamic Derivatives for Monte Carlo Evaluation

“…Because of this, however, it was difficult to find influential uncertain input parameters although MCS outputs are reliable, so methods of detecting influential inputs in MCS have been studied and developed. [6][7][8][9] In Motoda and Miyazawa, 6) "Test Input" vectors were created by randomly selecting the elements of uncertain input vectors that caused MCS "failure" results. MCS was then performed using those Test Input vectors, and a statistical test applied to the results to detect the influential parameters.…”

“…Since practical development requires that the design phase be completed within a given period of time, quick detection methods that skip MCS calculation have been presented. Restrepo and Hurtado 8) applies a pattern recognition algorithm 11) in the early design stages where there tends to be a larger number of failure cases in MCS results. Motoda 9) developed another quick method that is valid for the later design stages where the design has become more refined and there are fewer MCS failures.…”