Abstract
Design and analysis of non-orthogonal sequential experiments with ordinal factors by ANOVAmodel are proposed.At the 2nd series,design-points would be shifted in the direction of the better lattice point solution (BLPS)or preferable conditions,under considering the results of the 1st series. And, it’s reasonable that experiments are replicated at the near area of BLPS or the
important design-points. This is extremely effective to reduce the size of experiments, under enough level of degrees of freedom for error.If sizes of factors’effects were twice as much as standard deviation, it was confirmed that this method was practically effective by numerical simulation results about the power test.Merits of this work are as follows.1)Some ingenuities according to both of the experimenters’aim and thoughts shall be included. 2) The size of experiments would be reduced,and variances for both error and between-series are estimated.3) Time,cost and labor will be saved,and time-for-delivery will be also reduced.As variance of between-series is certainly removed from error term, treatment-mean and error variance are unbiasedly estimated.It is different from Response surface method,Direct sum method or Evolutionary operation.
