Although many methods are there for solving nonlinear optimization problems, effective constraint handling still remains as a big challenge. Most methods use exterior penalty function like schemes for it, which require incommensurable constraints to be normalized a priori. Although such fixed constraint normalization and subsequent formation of a penalized objective function are theoretically correct, some variable values generated during the optimization process may make a constraint undefined or mislead to an infeasible solution as the optimum. For avoiding such cases, a bounded constraint normalization scheme and a conditional penalized objective function are proposed here, which are to be implemented iteration-wise according to the generated variable values.