The appropriate management of water resources in the world is a policy that can be used in the line of optimum water consumption. One of these management instructions is optimum designing of canal sections for transferring water in irrigation networks. Since water losses in an irrigation network include seepage and evaporation losses, and regarding that the seepage losses depends on the geometry of canal and evaporation losses depends on the free water level, the designing of canal section will be done with the minimum water loss which include minimizing the seepage and evaporation from the canal under uniform flow. Until now, the optimum designing has been done based on relations some of which have high errors in exact estimation of seepage value, so it is necessary to apply a more exact method in optimization process for estimating the seepage value in canal. In this study, the canal seepage value for different states have been modeled by using SEEP/W software package, then the most appropriate soft model have been extracted between the input data (the geometry of canal and the technical characteristics of the bed soil) and output data (seepage). The extracted soft model had been compared to other experimental relations of seepage estimation, and beside confirming its high exactness in estimating seepage compared to other methods, it also have been used as a more efficient method for estimating seepage in optimization process with the aim of minimizing water losses in canals. In the proposed structure, Manning equation has been introduced as the major constraint, allowed minimum velocity, and Froude number as the minor constraint of the problem. The results of this study have been presented in form of dimensionless diagrams that made it possible to design the canal’s dimensions simply. Investigating the results of present study and previously conduced researches indicate the significant changes of canal’s optimum dimension in the present study compared to other re
