694. Mass Properties Uncertainty Analysis of Aerospace Vehicle Hardware

Paper

Abstract

Weight engineering has progressed as an analytical displine to now answer not only the question of how much but also the question of within what tolerance. The purpose of this paper is to dissect the latter question and arrive at a rational approach to the Uncertainty Analyses of Aerospace Verhical Hardware. The basis for the discussion must be built around a consistent set of definitions which are as follows:
(a) Nominal Weight – The most probably weight based upon the specific requirements and criteria. It includes contingency for in-scope design development changes. Specific allowances for ‘growth’ are included to the extent specified inthe requirements and criteria.
(b) Contingency – That weight which must be added to an estimate to account for lack of detail in the system and/or the type of method used to make the estimate.
(c) Growth – Changes in weight due to revisions in requirements and criteria, or concept, and/or solution of major unanticipated development problems.
(d) Uncertainty – The variation of weight (or other mass property) about the nominal due to routine but random nature of design development problems, manufacturing tolerances, and natual variations of possible design approaches.
Each of the above definitions sets apart a separate subtechnology, however, these subtechnologies cannot be worked as completely separate areas of interest for each has a definite interface with the other. Industry-wide adoption of and adherence to these definitions would allow a base from which future work in the areas of Uncertainty, Growth, Contingency and Weight prediction could build.
This paper presents a statistical analysis of all mass properties as related to the definition of Uncertainty above, while being aware of the interrelationships of Nominal Weight, Contingency and Growth. Element uncertainties are established by one of two methods: 1) design analyzed components are evaluated by the historical pattern of components of similar design development, 2) semi-analytical estimates as developed in the weight prediction work are evaluated by a standard statistical method, the standard deviation assuming a normal distribution. (Note how this ties to the job of estimating nominal weights.) Sufficient data has been gathered to use the method on two McDonnell Douglas Corp. programs with satisfactory success. The major problem areas are: a) exact definitions on element size to meet the criterion of independent errors, b) obtaining more historical data to insure consideraiton of all error sources, and c) the presentation of the rather complex technical data in a clear and concise manner to obtain management acceptance.