Name / Mr. Jukka Lassila / Session / 5
Company / Lappeenranta University of Technology / Block / 2
Address / P.O. Box 20 / Question n° / -
53851 Lappeenranta, Finland / Language used on the floor: English
Phone / +358 5 621 6742 / Accompanying visuals on file ?
e-mail / / LASSILA_FI_author_Alpha5_block2.ppt
DATA ENVELOPMENT ANALYSIS IN THE BENCHMARKING OF ELECTRICITY DISTRIBUTION COMPANIES
1Introduction
This paper focuses on evaluating and developing the DEA-model that is used for the efficiency measurement of distribution companies in Finland. In the present DEA-model there are five factors based on which the DEA-scores are calculated. These factors are operational costs, power quality, distributed energy, the length of the network and the number of customers (see picture below).
Power quality in the DEA-model is measured as the total interruption time experienced by customers. The DEA-model strongly directs the development of distribution systems because the DEA-scores have a direct impact on how much profit companies can make. The allowed return increases if the efficiency score is greater than 0.9 and decreases otherwise. The change in allowed return is calculated as shown in equation (1).
(1)
Basically, there are two ways to improve efficiency score: companies can either decrease operational costs or improve power quality. The latter is done by reliability investments. In addition, these often decrease operational costs. Investments are not taken into account in the DEA-model as a separate factor. Therefore, the model encourages investments but has no means to supervise them. The directing effect focuses on investments that shorten the interruption time because it is the only power quality measure. At the moment there is no incentive to decrease the number of interruptions. This changes if the power quality index is changes so that it takes into account, for example, interruption time, number of interruptions ant the number reclosing operations. Also, the actual harm caused by an interruption should be evaluated. This evaluation can be based on the values of the non-distributed energy (NDE).
2Evaluation of the present DEA-model
Analysis of the DEA-scores and the key figures used in the efficiency benchmarking showed that in general the present DEA-model treated the distribution companies quite equally. That is, the effects of operational costs were predictable and similar for each company. Also, the companies were compared mainly to similar companies, which is one of the basic ideas of the DEA [2]. However, the present DEA-model ignores the fact that companies can significantly affect interruption time by their own actions. This has been commonly taken into account in the network planning and investment calculations. Therefore, the effect of interruption time on the DEA-score has to be predictable and behave steadily from one year to another. In addition, the effect should be similar for each company. These objectives are not met in the present DEA-model.
There are two main problems concerning interruption time in the present DEA-model. Firstly, the directing effects of the model are not equal for all the companies. In the present DEA-model, e.g. the prices of outages differ between 0…500 €/customer,h. Secondly, for many companies the changes in the directing effects vary largely from one year to another. This means that one important economical basis of the network operations planning and investment calculations changes from one year to another. This, in turn, makes the long-term planning more difficult.
In order to solve the problems concerning the interruption time in the DEA-model the use of internal weight restrictions was studied. Restrictions were applied to operational costs and interruption time. The studied restrictions were the relative weight restriction, the direct weight restriction and the virtual product restriction. All these appeared promising at first but in each case further study showed some weaknesses that eventually prevented them from being used in the DEA-model.
3Developments of the present DEA-model
In developing the present DEA-model the goal was to make the interruption time become a meaningful factor in the DEA-model for each company. This was done by studying two methods. In the first method power quality was measured as interruption costs instead of interruption time. The DEA-model itself was not changed. In the second method a principle change in the DEA-model was done: power quality was removed as separate factor from the model. Instead, interruption costs were added to operational costs and the sum of these costs was used as the controlled input of the developed DEA-model.
3.1Interruption costs
Customers of each company were divided into five consumer groups: domestic consumption, agriculture, public consumption, industry and services. Customers belonging to different groups have different values of non-distributed energy.
The number of companies that have insignificant factors in the present DEA-model and the developed DEA-model are shown in the picture. In the present DEA-model power quality in measured as total time of interruption and it is a separate factor in the model. In the developed DEA-model power quality is not a separate factor. It is taken into account in the input of the developed DEA-model by adding it to operational costs.
In the developed DEA-model the significance of all the factors increased when compared to the present DEA-model. Prices of outages obtained when the developed DEA-model was used are shown in next picture.
For most of the companies the range of the prices of outages was quite small, 4…6 €/customer,h. In the present DEA-model this range was 0…500 €/customer,h. In the developed DEA-model the effect of interruption costs remained approximately the same from one year to another for each company.
4Conclusions
The efficiencies of distribution companies in Finland are determined by means of DEA-model. It gives each company an efficiency score, which has a direct impact on how much profit a company can make. Therefore, the efficiency measurement strongly directs the development of distribution networks. The present DEA-model still requires further developments because its directing effects are not equal to all the distribution companies and they lack continuity. This paper introduced a solution to the first problem. The basic idea of the solution was to change the way in which power quality is taken into account in the DEA-model.
In the future, it would be important to take also investments into account in the DEA-model. However, the method by which this could be done was not within the scope of this paper.
LASSILA_FI_Alpha5_block2.doc1/3