Despite the fact that the optimal design for nonlinear model depends on the unknown quantity of parameter to estimate basically, its popularity is growing in bio and engineering statistics area since all those models in the area are virtually nonlinear. In this paper we have dealt with the case when the researcher has multiple objectives in experimentation, decision among the competing models, protection against the departure from the assumed model, and the con icting interests among design criteria. To tackle these issues we attempted several new approaches which are taking advantage of the easiness of constrained optimal design. Several nonlinear models were tested.