There are many methods for solving an optimal power flow (OPF) problem. Most of them employ one central unit to solve the problem (centralized) while in others, every node/area computes for its coverage and send information to its neighbor(s) every time. The mentioned method is called Distributed OPF. In this thesis, the Distributed OPF aims for a fast and resilient algorithm to solve a DC-OPF problem based on Consensus + Innovation (C+I) method.
The research focuses on developing a faster algorithm in solving an OPF problem for a DC distribution grid. The current C+I method has tuning parameters, all of which are determined by trial and error for every case. They are sensitive parameters that determine the speed of the iteration to converge to the solution. This thesis attempts to form adaptive tuning parameters by understanding their function in every equation for a variable update. By understanding the purpose of each tuning parameter and the physical interpretation, some parameters can be formulated while the other is still determined by estimating the value. The losses and congestion are also taken into account. In the results, by formulating these parameters, they are shown that the iteration number has dropped significantly compared to the previous research.

Regarding the resilience of the algorithm, the distributed approach relies on the information exchange between the nodes and delay on the information exchange will stall the whole iteration process. Therefore, an asynchronous algorithm is implemented to resolve the problem by setting a timeout duration. The timeout duration enables the algorithm to wait for the new information only for the desired duration, and therefore the calculation converges in faster.