| Suppose that an experiment must be carried out in a time sequence. The observations may be affected by a certain time trend. A design objective is to choose a run order such that the factor effects are orthogonal to the time trend effects. Another design objective is to choose a run order that minimizes the cost of level changes. In this dissertation, a new construction technique is developed which provides a run order for 2;In practice, level changes of different factors may have different costs. Thus the problem of cost minimization becomes much more difficult. We develop an algorithm to construct minimum cost run orders for fractional factorial design ;Cheng and Steinberg (1991) proved that run orders are trend robust if the factors change level many times for first order autoregressive time trend models. In this dissertation, we obtain a method for constructing run orders with maximum number of level changes in which the main effects are linear trend free for 2... |
