This study proposes an improved methodology for solving the semi-controlled ventilation network problem. After analysis and consideration of the objective and legal, technical and operation convenience factors, the semi-controlled network problem is formulated through a nonlinear nonconvex programming model. Using the special ordered sets variables, the nonconvex problem is linearized and then optimized by the modified branch and bound procedure where the automatic interpolation technique is used to improve the accuracy to which the each nonlinear function is approximated. The global optimality of the methodology is on the computational the~ry basis. And the applicability of the methodology to the generalized ventilation problem is investigated, and it is demonstrated with a number of examples. This study also compares the several methods which have been used so far to optimize the underground ventilation networks. / Ph. D.