Supporting the electricity network operator in the allocation process

In the current design of the electricity sector in the Netherlands, network operators daily receive data from measurement companies about the electricity consumption of their customers for each quarter of an hour. The network operator sorts the data per combination of Program Responsible Party (PRP) and network district region and sends the allocated data to the PRPs. The measurement series sometimes appear to contain errors. These errors need to be corrected in a short time. To avoid forwarding wrong data to the PRPs, the network operator would like to check the data on errors before they are sent. This paper shows that the electricity consumption can easily be predicted because the first-order autocorrelation of the data is very high. A simple prediction method is compared with the use of exponential smoothing models which give only slightly better results at best, but are much more complex to use.