طرح اولیه سیستم های توزیع برق برای برق رسانی روستایی: یک الگوریتم هیوریستیک برای طراحی شبکه چند سطحی

We describe the first heuristic algorithm that selects the locations and service areas of transformers without requiring candidate solutions and simultaneously builds two-level grid network in a green-field setting. The algorithm we propose minimizes overall cost of infrastructure costs; specifically the combined costs of transformers and the two-tiered network together by solving transformer location problem as well as network design problems in a single optimization framework. In addition, it allows one to specify different costs for the higher throughput lines upstream of the transformer as compared to downstream of the transformer. Simulations are carried out based on real-world spatial distributions of demand points from rural locations in Africa, specifically in places without any pre-existing infrastructure to test the algorithm and generalize the results.

In a technological landscape that is altered by the emergence of off-grid and distributed approaches, there is a need amongst infrastructure planners to evaluate the costs of networked or grid approaches vis a vis off-grid approaches to be able to make rapid assessment of the progress in rural electrification.1 The investment costs of networked approaches are more difficult to estimate than the costs of off-grid approaches because it takes into account both the spatial distribution of demand and the optimal placement of infrastructure to meet that demand. This paper through its algorithm provides a new methodology to estimate the cost of green-field networks rapidly and with high accuracy. The algorithm we present in this paper combines the transformer location problem and the low voltage (LV) and medium voltage (MV) network design problem into a single problem and solves them in a single optimization framework. We propose a heuristic algorithm to design a two-level radial power distribution system. The first level includes the determination of the numbers, locations and capacities of transformers that feed an LV distribution network. The transformers represent load points for an upstream MV network and the MV network is also determined as a part of the first level. The second level includes the determination of the layout of the low voltage network between the transformers and the specified ultimate demand points. Note that the high voltage (HV) network (for that matter source points) further upstream of the MV network are assumed to be known.2 One could have further generalized the problem to include the determination of the HV networks as well, making it a three level problem, but here we consider the HV network as pre-specified for simplicity. The algorithm we propose does not require a set of candidate locations to be considered as transformer locations. The maximum service distance in a low voltage distribution network is also pre-specified and determined from engineering practice. Given these costs, the demand points, the location of the HV network, and the maximum distance of the demand point from the transformer, the algorithm automatically finds the locations and service areas transformers as well as the LV and MV network layout with the goal of minimizing the total costs. Understanding the cost involved with electrification is important in designing a proper smart grid structure. This algorithm can serve as a tool for network engineers and planners to make rapid assessments assisting them with (a) estimates of total cost of distribution, (b) layouts of initial designs and (c) breakdown of total costs into transformer cost and medium and low voltage line costs and giving them a good starting point for more detailed smart grid projects. The methodology we propose ignores transmission losses, load flow considerations and local topography and hence the proposed designs are not meant to replace detailed engineering analyses of grid rollout. Therefore, this tool should be considered as a guide for planning within utilities to be used with large datasets rather than a tool that provides every detail. However, for completeness possible extensions of the algorithm to include more details of power distribution systems are also discussed in Section 5.4. The sections of this paper are outlined as follows: the remainder of this section provides background information including literature review; a more precise statement of the problem is given in Section 2; our approach to the problem is explained in Section 3; algorithm results are provided in Section 4; and a discussion of our algorithm is presented in Section 5.

A new heuristic algorithm for the design of two-level power distribution systems has been introduced. It has been presented that the algorithm finds the number and locations of MV/LV transformers without giving any candidate locations and finds a multi-point low voltage network between demand points and transformers. The algorithm is tested with the real household data digitized from satellite imagery of Sub-Saharan African villages and results are presented as an estimate for investment costs and financial requirements to support electrification problems. The proposed algorithm ignores transmission losses, load flow considerations and local topography. Hence it should be viewed as a quick tool which simplifies a complex problem and provides good starting point for decision makers and practitioners. However; our algorithm is flexible such that it can be simplified to other infrastructure problems (for example; facility location problem) or it can be extended to include more distribution system components such as transformer sizes.