791. Rational Weight Estimation Based on Statistical Data

Paper

Abstract

Weight engineering includes three phases of weight determination: estimation, calculation, and measurement. To some weight engineers, estimation is the most important and the most challenging of these three. There has been, are, and will be many estimation methods. Many methods are empirical and some are analytical in nature. This paper discusses a method that is in between empiricism and analysis, and considers it superior to both in many practical respects. This method will be referred to as ‘REBOS’, from the title of this paper.
REBOS is based on actual weight data of complete vehicles, standardized functional weight coding, weights sub-divided to the degree that weight parameters will be combined uniquely for each of the sub-divisions, and rationally derived parameter combinations which have a linear relationship with weight. The equations resulting from the weight-parameter linear relationships become usable individually for the weight estimation of sub-divisions, and collectively for the weight estimation of major functions and total vehicle.
A weight assessment is incomplete until vehicle balance and size adequacy are examined. REBOS includes these considerations, plus moments and products of inertia, since few additional parameters are required to include these latter properties. To indicate REBOS accuracy, standard deviations are computed for estimated functional group and vehicle weights, as well as for estimated vehicle balance and inertias, based on the sample data used. Also, estimated size adequacy indications are summarized for the sample data used.
The application of REBOS is practical only when a digital computer is used. Depending on the complexity of the vehicle and the accuracy of the results desired, from 100 to 300 equations must be solved, using a similar quantity of parameters. Average computer time will be from 3 to 15 seconds, depending on the program and the computer used.
Empirical methods are considered unconvincing in appearance and proven to be unreliable for extrapolation. Analytical methods are best used for subdivision weight comparisons, but are too complex for total vehicle estimates. REBOS is based on rationality, and purposely uses only readily available parameters so that accuracy tests can be made relatively easy, using additional actual samples as they become available. REBOS can be used for absolute as well as comparative assessment or estimation. Finally, in order to ensure the user’s constant understanding and control of the program, unnecessary subdivision of weight elements and inclusion of minor parameters in REBOS is to be avoided. A trade-off between program complexity and output accuracy should always be a consideration.