The article presents an approach to increasing of the energy efficiency of the parallel technological systems based on optimal coordinated processes control and the principle of minimization of the energy consumption based on the optimal distribution of the capacities of the parallel processes with account of the local disturbances. The principle is based on the assumption that in order to reduce the energy consumption of a parallel system it is necessary to decrease its processes' capacities with a high level of local disturbances, preserving the required total system productivity by increasing the capacities of the processes with the low levels of the local disturbances. The assumption is based on the fact that, as a rule, the processes with a high level of the local disturbances have higher local energy consumption than the processes with a low level of disturbances. It is shown that the problem of the optimal capacities distribution is a task of a multiparameter optimization with constraints. The paper discusses implementation of this principle on the basis of a two-level control system with a coordinator. The coordinator's task is to compute the values of the optimal local capacities in real time based on information concerning the current values of the local disturbances. The values of the optimal local capacities are fed into the local automatic control systems, which implement the appropriate control actions concerning the local processes. The analytical and iterative algorithms for the optimal coordination of the local capacities values is considered. The analytical algorithm is based on Lagrange's method, while the iterative algorithm is based on the search for an active gradient solution. The paper considers application of the proposed approach for saving the energy consumption of an oil-production wells group. Here, the parallel technological system includes the oil-production wells as the local processes. The results of the numerical experiments show feasibility of saving energy by more than 5 % in comparison with a technological system without the optimal distribution of the productive capacities. Thus, the proposed approach can reduce significantly the energy consumption of a parallel technological system without an expensive modernization of its technology.