The authors specify a network simplex method of the load-flow calculation in a power system. The proposed model is an imaginary and real current one. It is based on a simplifying assumption which allows decoupling of the problem and which is the basis of the algorithm called FLF (flow load-flow algorithm). Power system elements such as lines, transformers, generators and voltage control buses are represented with adapted models. Systematic tests have been performed on some existing power systems, such as the IEEE 24-bus Reliability system. The performance of this method is compared with the results of a decoupled Newton-Raphson load-flow. The accuracy of the FLF algorithm is competitive with the Newton-Raphson method, but its speed still has to be improved