Resource Management for Maintenance Work in Electricity Networks

Stefan Ravizza, Bachelor Thesis, Summer 2007

Supervisors: Christian Balderer, Michael Guarisco

In industrialized countries, a reliable supply of electrical energy is taken for granted. Grid operators are responsible for an adequate quality of supply but also expected to minimize investment and operating costs to allow for low network access fees. Resource management is therefore a major challenge, in particular to find a configuration of resources which not only guarantees the required quality of supply but also minimizes cost.

Operation of the grids and the equipment includes maintenance and restoration of supply after failures. Whereas restoration work, which is driven by stochastic failures, only amounts for about 5% of the yearly volume, the remaining 95% are due to maintenance work whose tasks are deterministic and can be planned in advance.

One aspect of the quality of supply concerns the availability of electricity to customers, which is usually measured in terms of the non-availability of supply. In the medium- and low-voltage power grid, most failures infer an interruption of supply. Hence it is the restoration work which mainly influences theachieved quality of supply. However, maintenance work also affects the quality of supply as it allows for a reliable operation and prevents further failures.

In this project we consider the problem of finding a configuration of resources which covers the given maintenance work at minimum cost. This covering problem can be formulated as a mixed integer program (MIP). Each type of resources has a limited working capacity and its characteristic efficiency for each type of work. The possibility of outsourcing certain tasks can be incorporated by additional types of resources. However, the outsourcing of tasks often infers additional work (e.g. control, quality assurance) for another type of resources. A feasible solution to the MIP fulfills these constraints and assures that all maintenance tasks are accomplished. The goal of this bachelor project was to implement, test and analyze the described MIP.

In the first part of the project, a program for solving the described problem was implemented in C++. To allow for a convenient pre- and post-processing of input parameters and results, an adequate set of text files was defined and integrated. The actual MIP was solved by the commercial solver MOSEK.

In the next part, the given MIP was investigated analytically. It was shown that the linear relaxation of a slightly modified problem can be solved analytically. The optimal solution can be obtained by a simple sorting algorithm.

Finally, the model was tested in a small case study with real-world data of RWE Rhein-Ruhr Netzservice GmbH. The solutions of the MIP and its linear relaxation were compared and discussed. A detailed sensitivity analysis based on the costs of outsourcing a certain task was performed. The results of this analysis provide guidelines on adequate levels and costs of outsourcing.