Response surface methods for force balance calibration modeling

A "modern design of experiments" (MDOE) approach to balance calibration at NASA Langley Research Center focuses on the application of formal experimental design techniques to address weaknesses in the current calibration methodologies. The Single-Vector Balance Calibration System (SVS) has been developed as an innovative mechanical load application system specifically designed for the efficient and accurate execution of a formal experimental design. This paper emphasizes practical applications of response surface methodology with the analyses of experimental data. Calibration experimental design concepts including the estimation of the required data volume and an evaluation of the model prediction capability are presented. Randomization, replication, and blocking are proposed as means of tactical defense against systematic errors present in all calibration systems. Response surface methods are implemented in obtaining an adequate model with the minimum number of terms and partitioning of the unexplained variance. A systematic approach to augmenting a second order model with higher order terms is discussed. Applying formal experimental design techniques to force balance calibration provides a suite of sophisticated and elegant tools that advance balance calibration technology.